Vol.62 KHTN-F

SCIENTIFIC JOURNAL OF HANOI METROPOLITAN UNIVERSITY − VOL.
62/2022 1
2 HANOI METROPOLITAN UNIVERSITY
SCIENTIFIC JOURNAL OF HANOI METROPOLITAN UNIVERSITY − VOL.62/2022 3
1. NOVEL INCREMENTAL ALGORITHMS FOR ATTRIBUTE REDUCTION FROM

DYNAMIC DECISION TABLES WHEN ADDING OBJECT SET ................................... 5
Về thuật toán gia tăng rút gọn thuộc tính khi bổ sung tập đối tượng trong bảng quyết định
thay đổi
Ho Thi Phuong, Truong Duc Phuong
2. DESIGN SOME GRADE 6 MATH TESTS TO ASSESS THE DEVELOPMENT OF
MATHEMATICAL COMPETENCIES .............................................................................. 13
Thiết kế một số đề kiểm tra đánh giá môn Toán lớp 6 theo hướng phát triển phẩm chất, năng lực
Nguyen Van Anh, Le Thi Van Dung, Tran Anh Phuong, Nguyen Thi Hong
3. SOME REAL PROBLEMS AND APPLICATION OF THE GREATEST COMMON DIVISOR,
THE LEAST COMMON MULTIPLE .................................................................................... 24
Một số bài toán thực tế và áp dụng của ước chung lớn nhất, bội chung nhỏ nhất
Hoang Minh Giang, Nguyen Thi Thu Hoa, Nguyen Thi Hong
4. APPLICATION OF LAGRANGE MULTIPLIER METHOD IN THE PROBLEM OF
FINDING THE MINIMUM AND MAXIMUM VALUES ................................................ 34
Ứng dụng phương pháp nhân tử Lagrange trong bài toán tìm giá trị lớn nhất và giá trị nhỏ nhất
Nguyen Duc Nguyen, Nguyễn Thi Hong
5. SOME APPLICATION MODELS OF FIRST-ORDER DIFFERENTIAL EQUATIONS IN
FACT ................................................................................................................................... 44
Một số mô hình ứng dụng của phương trình vi phân cấp một trong thực tế
Nguyen Thi Thu Huyen
6. A DEEP LEARNING MODEL THAT DETECTS THE DOMAIN GENERATED BY THE
ALGORITHM IN THE BOTNET ....................................................................................... 52
Một mô hình học sâu phát hiện tên miền được tạo bởi thuật toán trong mạng Botnet
Nguyen Trung Hieu, Cao Chinh Nghia
7. RUIN PROBABILITIES FOR A GENERAL INSURANCE RISK MODEL ................................ 64
Xác suất thiệt hại đối với mô hình rủi ro bảo hiểm tổng quát
Nguyen Thi Thuy Hong
8. THE DEPOSITION OF Zn-DOPED SnO2 THIN FILMS BY A COMPRESSION
SPRAYER FOR ALCOHOL VAPOR SENSOR ................................................................ 71
Chế tạo màng mỏng SnO2 pha tạp Zn bằng phương pháp phun áp suất ứng dụng cho cảm
biến nhạy hơi cồn
Pham Van Vinh, Dang Tran Chien, Pham Thi Lan Huong
9. FABRICATION AND CHARACTERISTICS OF NANOCOMPOSITE TiO2/GQDs THIN

FILMS .................................................................................................................................. 79
Chế tạo và khảo sát các tính chất đặc trưng của màng mỏng nanocompozit TiO2/GQDs
Dang Tran Chien, Phan Xuan Thien
10. ASSESS THE PERCEPTION OF CLIMATE CHANGE STUDENTS AT HANOI
METROPOLITAN UNIVERSITY ..................................................................................... 88
Khảo sát, đánh giá nhận thức về biến đổi khí hậu của sinh viên trường Đại học Thủ đô Hà Nội
Nguyen Xuan Trinh
11. STUDY ON NATURAL FLOOR CLEANER FROM SOME HERBALS ............................. 95
Nghiên cứu tổng hợp nước lau sàn hữu cơ từ một số loại thảo dược thiên nhiên
Ngo Thanh Son, Bui Van Nam
12. RESEARCH ABILITY TO CREATE CALLUS AND REGENERATION PANAX
BIPINNATIFIDUS (PANAX BIPINNATIFIDUS) IN VITRO CULTURE .................... 105
Nghiên cứu khả năng tạo mô sẹo và tái sinh cây sâm vũ điệp (panax bipinnatifidus) trong
môi trường nuôi cấy invitro
Nguyen Nhu Toan, Luu Ngoc Sinh, Nguyen Thi Binh, Tran Đang Khanh
13. USINGSOME BIOINFORMATIC TOOLS TO MINING GENES CODING
CELLOBIOHYDROLASE FROM METAGENOME DATA OF THE BACTERIA
SURROUNDING WHITE-ROT FUNGI (Trametes versicolor) IN CUC PHUONG
NATIONAL PARK ........................................................................................................... 119
Sử dụng một số công cụ tin sinh để khai thác gen mã hóa enzyme cellobiohydrolase từ dữ
liệu metagenome của khu hệ vi khuẩn quanh nấm mục trắng (trametes versicolor) ở vườn
quốc gia Cúc Phương
Nguyen Thi Binh, Le Thi Thu Hong, Truong Nam Hai
NOVEL INCREMENTAL ALGORITHMS FOR

ATTRIBUTE REDUCTION FROM DYNAMIC DECISION
TABLES WHEN ADDING OBJECT SET
Ho Thi Phuong1*, Truong Duc Phuong2

1
Tay Nguyen University, 2Hanoi Metropolitan University
Abstract: In recent years, attribute reduction methods following fuzzy rough set approach
have attracted the attention of researchers because they improve the accuracy of the
classification model. However, most of the proposed methods are performed on the unchanged
decision table. In this paper, we build an incremental algorithm to find the approximate reduct
according to the combined filter-wrapper approach. Experimental results on a number of
sample datasets show that the proposed incremental algorithm is more efficient than some
other incremental algorithms following the filter approach in terms of the number of reductive
set attributes and classification accuracy.
Keywords: Fuzzy rough sets, fuzzy distances, incremental algorithms, decision tables,
attribute reduction.
Received 15 June 2022

Revised and accepted for publication 23 August 2022
(*) Email: htphuong@ttn.edu.vn
1. INTRODUCTION
In fact, decision tables are often large in size and are constantly changing and updating [1,
2]. The application of reductive set finding algorithms according to the traditional rough set
approach and extended rough set models faces many challenges [3]. In case the decision tables
are changed, these algorithms had to recompute the reduct on the entire decision table after the
change, so the cost of computation time increases significantly. In case the decision table is
large, the implementation of the algorithm on the entire decision table will be difficult in terms
of execution time [4]. Therefore, splitting the decision table to find the reduct on each part is
the proposed solution. However, calculating the reduct based on the reducts of each part is a
problem to be solved. Therefore, the researchers proposed an incremental computational
approach to find the reduct [5, 6, 7]. In case the decision table is changed, the incremental
algorithm does not recompute the reduct on the entire decision table, but only updates the
existing reduct based on the changed data composition. In the case of a large decision table, the
incremental algorithm finds the reduct on a fragmented component, then updates the reduct
when adding the remaining components. In theory, the incremental algorithm is capable of
minimizing the execution time and is capable of performing on large decision tables.
The main objective of the paper is to reduce the number of reductive set attributes and
improve the classification accuracy compared to the published incremental algorithms. In this
paper, Incremental Filter-Wrapper Algorithm for Fuzzy Partition Distance based Attribute
Reduction When Add Objects, called IFW_FDAR_AdObj algorithm, is propose to find the
approximate reduct of the decision table using fuzzy distance measure in the case of addition
of the feature set. The proposed algorithm based on the combined filter-wrapper approach
consists of two stages: the filter stage finds the candidates for the reduct each time the attribute
with the greatest importance is added, called the approximate reduct, with the stopping
condition that the fuzzy distance measure is preserved; Wapper stage finds the reduct with the
highest classification accuracy. Experimental results on sample data sets show that the
suggested incremental algorithm is more efficient than the non-incremental filter-wrapper
algorithm in terms of execution time. Moreover, the proposed algorithm is more efficient than
the published filter incremental algorithms in terms of the number of attribute sets and
classification accuracy by selecting the candidate with the best classification accuracy in the
wapper stage.
The rest of the paper is organized as follows. Section II develops IFW_FDAR_AdObj
algorithm to find the approximate reduct of the decision table using fuzzy distance measure in
the case of addition of the feature set. Section III presents result. Conclusions are drawn in
section IV.
2. IFW_FDAR_ADOBJ ALGORITHM (INCREMENTAL FILTER-WRAPPER
ALGORITHM FOR FUZZY PARTITION DISTANCE BASED ATTRIBUTE
REDUCTION WHEN ADD OBJECTS)
In this section, we propose a filter-wrapper incremental algorithm by using FPD when
adding object set into the decision table.
Algorithm IFW_FDAR_AdObj
Input:
1. A decision table 𝐷𝑆 = (𝑈, 𝐶 ∪ 𝐷) with 𝑈 = {𝑥1 , 𝑥2 , . . . , 𝑥𝑛 }, a FER 𝑅̃ , the reduct 𝐵 ⊆ 𝐶.
2. Fuzzy equivalent matrices
𝑀𝑈 (𝑅̃𝐵 ) = [𝑏𝑖𝑗 ]𝑛×𝑛 , 𝑀𝑈 (𝑅̃𝐶 ) = [𝑐𝑖𝑗 ]𝑛×𝑛 , 𝑀𝑈 (𝑅̃𝐷 ) = [𝑑𝑖𝑗 ]𝑛×𝑛
3. Added set of objects 𝛥𝑈 = {𝑥𝑛+1 , 𝑥𝑛+2 , . . . , 𝑥𝑛+𝑠 }
Output: The approximation reduct 𝐵𝑏𝑒𝑠𝑡 of 𝐷𝑆 ′ = (𝑈 ∪ 𝛥𝑈, 𝐶 ∪ 𝐷) with highest
classification accuracy.
Sep 1: Initialization
1. 𝑇: = ∅; // T contains the candidates for best reduct

2. Compute fuzzy equivalent matrices on 𝑈 ∪ 𝛥𝑈
𝑀𝑈∪𝛥𝑈 (𝑅̃𝐵 ) = [𝑏𝑖𝑗 ](𝑛+𝑠)×(𝑛+𝑠) , 𝑀𝑈∪𝛥𝑈 (𝑅̃𝐷 ) = [𝑑𝑖𝑗 ](𝑛+𝑠)×(𝑛+𝑠) ;
Step 2: Check the added set of objects

3. Set 𝑋: = 𝛥𝑈;
4. For 𝑖 = 1 to 𝑠 do
5. If [𝑥𝑛+𝑖 ]𝐵̃ ⊆ [𝑥𝑛+𝑖 ]𝐷̃ then 𝑋: = 𝑋 − {𝑥𝑛+𝑖 } ;
6. If 𝑋 = ∅ then Return 𝐵0; //Approximation reduct does not change
7. Set 𝛥𝑈: = 𝑋; 𝑠: = |𝛥𝑈|; //Reset the object set
Step 3: Finding the best reduct
8. Compute original FPDs
𝐹𝑃𝐷𝑈 (𝛷(𝑅̃𝐵 ), 𝛷(𝑅̃𝐵∪𝐷 )) ; 𝐹𝑃𝐷𝑈 (𝛷(𝑅̃𝐶 ), 𝛷(𝑅̃𝐶∪𝐷 ));
9. Compute FPDs using incremental formulas:
𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐵 ), 𝛷(𝑅̃𝐵∪𝐷 )) ; 𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐶 ), 𝛷(𝑅̃𝐶∪𝐷 ))
//Filter stage: finding candidates for reduct
10. While 𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐵 ), 𝛷(𝑅̃𝐵∪𝐷 )) ≠ 𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐶 ), 𝛷(𝑅̃𝐶∪𝐷 )) do
11. Begin
12. For each 𝑎 ∈ 𝐶 − 𝐵 do
13. Begin
14. Compute 𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐵∪{𝑎} ), 𝛷(𝑅̃𝐵∪{𝑎}∪𝐷 )) by using cremental formulas;
15. Compute 𝑆𝐼𝐺𝐵 (𝑎) = 𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐵 ), 𝛷(𝑅̃𝐵∪𝐷 )) −

𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐵∪{𝑎} ), 𝛷(𝑅̃𝐵∪{𝑎}∪𝐷 )) ;
16. End;
17. Select 𝑎 ∈ 𝐶 − 𝐵 satisfying 𝑆𝐼𝐺𝐵 (𝑎𝑚 ) = 𝑀𝑎𝑥 {𝑆𝐼𝐺𝐵 (𝑎)};
𝑎∈𝐶−𝐵
18. 𝐵: = 𝐵 ∪ {𝑎𝑚 };
19. 𝐵0 : = 𝐵0 ∪ {𝑎𝑚 };
20. 𝑇: = 𝑇 ∪ 𝐵0;
21. End;
//Wrapper stage: Finding the reduct with the highest classification accuracy
22. Set 𝑡: = |𝑇| //t is the number of T, 𝑇 = {𝐵0 ∪ {𝑎1 }, 𝐵0 ∪ {𝑎1 , 𝑎2 }, . . . , 𝐵0 ∪

{𝑎1 , 𝑎2 , . . . , 𝑎𝑡 }};
23. Set 𝑇1 : = 𝐵0 ∪ {𝑎1 }; 𝑇2 : = 𝐵0 ∪ {𝑎1 , 𝑎2 }; . . . ; 𝑇𝑡 : = 𝐵0 ∪ {𝑎1 , 𝑎2 , . . . , 𝑎𝑡 };
24. For j:= 1 to t do
25. Calculate the classification accuracy on 𝑇𝑗 by using 10-fold classifier;
26. 𝐵𝑏𝑒𝑠𝑡 : = 𝑇𝑗𝑜 in which 𝑇𝑗𝑜 has the highest classification accuracy;
Return 𝐵𝑏𝑒𝑠𝑡 ;
3. EXPERIMENTAL RESULTS
Compared with two fuzzy rough set based incremental algorithm ( IV-FS-FRS-2 , IARM
) and two rough set based incremental algorithm (ASS-IAR , IFSA ). Specifically, IV-FS-
FRS-2 is a filter algorithm based on fuzzy discernibility matrix, while IARM is a filter
algorithm based on relative discernibility relation. ASS-IAR is a filter algorithm based on active
sample selection, while IFSA is a filter algorithm based on dependency function.
3.1 Data set
This subsection introduces experiments for evaluating the classification accuracy of
IFW_FDAR_AdObj algorithm compared with two fuzzy rough set based incremental algorithm
( IV-FS-FRS-2 [5,8], IARM [6,9]) and two rough set based incremental algorithm (ASS-IAR
[7,10], IFSA [8,11]). Specifically, IV-FS-FRS-2 is a filter algorithm based on fuzzy
discernibility matrix, while IARM is a filter algorithm based on relative discernibility relation.
ASS-IAR is a filter algorithm based on active sample selection, while IFSA is a filter algorithm
based on dependency function. Experiments are deployed on some benchmark datasets from
UCI [9,12] as in Table 1.
For the algorithms IV-FS-FRS-2 and IARM by fuzzy rough set approach, all real value
attributes are normalized into the values in interval [0, 1] on each dataset [8]:
𝑎(𝑥 )−𝑚𝑖𝑛(𝑎)
𝑎′ (𝑥𝑖 ) = 𝑚𝑎𝑥(𝑎)−𝑚𝑖𝑛(𝑎)
𝑖
(1)
in which, max(a) and min(a) are the maximum and minimum of a. FER 𝑅̃𝑎 [8] on a is defined
as:
𝑅̃𝑎 (𝑥𝑖 , 𝑥𝑗 ) = 1 − |𝑎(𝑥𝑖 ) − 𝑎(𝑥𝑗 )| where 𝑥𝑖 , 𝑥𝑗 ∈ 𝑈 (2)
For each attribute 𝑎 ∈ 𝐶 with nominal or binary value, the FER 𝑅𝑎 is defined where
𝑥𝑖 , 𝑥𝑗 ∈ 𝑈:
1, 𝑎(𝑥𝑖 ) = 𝑎(𝑥𝑗 )
𝑅𝑎 = { (3)
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
On decision attribute {𝑑}, we use FER 𝑅{𝑑} . For 𝑥𝑖 , 𝑥𝑗 ∈ 𝑈
1, 𝑑(𝑥𝑖 ) = 𝑑(𝑥𝑗 )
𝑅{𝑑} = { (4)
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
The partition 𝑈/𝑅{𝑑} = {[𝑥𝑖 ]{𝑑} }, where 𝑥𝑖 ∈ 𝑈 and [𝑥𝑖 ]{𝑑} = {𝑥𝑗 ∈ 𝑈|𝑅{𝑑} (𝑥𝑖 , 𝑥𝑗 ) = 1} is
an equivalent class. Then, equivalent class [𝑥𝑖 ]𝑑 is considered as a fuzzy equivalent class,
denoted by [𝑥𝑖 ]𝑑̃ . The membership function is defined as 𝜇[𝑥𝑖 ]𝑑̃ (𝑥𝑗 ) = 1 if 𝑥𝑗 ∈ [𝑥𝑖 ]𝑑 and
𝜇[𝑥𝑖 ]𝑑̃ (𝑥𝑗 ) = 0 if 𝑥𝑗 ∉ [𝑥𝑖 ]𝑑 .
For the algorithms ASS-IAR and IFSA by traditional rough set approach, we use a fuzzy
C-mean clustering (FCM) to discretize real-valued data before attribute reduction.
Each data set is divided into two approximately equal parts: original data set (Column 5 in
TABLE 3.1) and incremental data set (Column 6 in TABLE 3.1). Original data set is denoted
as U0. Incremental data set is randomly separated into 5 equal parts, each part is denoted by U1,
U2, U3, U4, U5 respectively. To applying incremental algorithm IFW_FDAR_AdObj, IV-FS-
FRS-2, IARM, ASS-IAR and IFSA, at first we perform this algorithm on original data set. Next,
this algorithm is executed when sequentially adding from the first part to the fifth part of
incremental dataset.
Table 3.1. Description of data sets when adding object set
Number of condition Number
Original Increment attributes
Number of
ID Data Description number al number
of objects Nominal Real- decision
of objects of objects All
value valued class
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Libras
1 Libra 360 180 180 90 0 90 15
movement
Wisconsin
WDB
2 diagnostic 569 284 285 30 0 30 2
C
breast cancer
3 Horse Horse colic 368 183 185 22 15 7 2
4 Heart Statlog (heart) 270 135 135 13 7 6 2
5 Credit Credit approval 690 345 345 15 9 6 2
Germa German credit
6 1000 500 500 20 13 7 2
n data
Contraceptive
7 Cmc 1473 733 740 9 7 2 3
Method Choice
8 Wave Waveform 5000 2500 2500 21 0 21 3
3.2. Computation time of IFW_FDAR_Adobj, IV-FS-FRS-2 IARM, ASS-IAR AND IFSA

Figure 3.1 shows that the execution time of IFW_FDAR_AdObj is higher than the
execution time of IV-FS-FRS-2 and IARM on all datasets. Although the calculation of fuzzy
distances in IFW_FDAR_AdObj is simpler than others measures in IV-FS-FRS-2, IARM,
ASS-IAR and IFSA, the algorithm IFW_FDAR_AdObj needs more time to run the classifier.
The execution time of ASS-IAR is smallest because of the elimination of useless incomming
samples in incremental computation.
Figure 3.1. Time of IFW_FDAR_AdObj, IV-FS-FRS-2 IARM, ASS-IAR và IFSA on Libra

(second)
3.3 Classification accuracy and reduct cardinality of IFW_FDAR_Adobj, IV-FS-FRS-2,
IARM, ASS-IAR AND IFSA
We use CART classifier (CART – Classification And Regression Tree) to compute the
classification accuracy in the wrapper stage of IFW_FDAR_AdObj. We also use CART
classifier to compute the classification accuracy for IFW_FDAR_AdObj, IV-FS-FRS-2, IARM,
ASS-IAR after attribute reduction. The 10-fold cross-validation technique is also used. We
divide the dataset into 10 approximately equal parts. One part is selected randomly for testing,
the others are used for training. This progress is repeated 10 times. We denote the accuracy of
classification as 𝑣 ± 𝜎 where 𝑣 is the mean of 10 runs and 𝜎 is standard error. All experiments
are installed on PC Core(TM) Intel (R) i7-3770CPU, 3.40 GHz, Windows 7 using Matlab.
0.65
Độ chính xác phân lớp
0.6
IFW-FDAR-AdObj
0.55 IV-FS-FRS-2
IARM
0.5
ASS-IAR
0.45 IFSA
U0 U1 U2 U3 U4 U5
Các tập đối tượng của dữ liệu Libra
Figure 3.2. Classification accuracy of IFW_FDAR_AdObj, IV-FS-FRS-2, IARM, ASS-IAR và IFSA

The result of classification accuracy obtained by our algorithm are presented in Figure 3.2.
As in this figure, for the cardinality of reduct at each incremental step, the proposed filter-
wrapper algorithm IFW_FDAR_AdObj is much smaller than IV-FS-FRS-2, IARM, ASS-IAR
and IFSA. As a result of this paper, the accuracy and the generality of classification rule set on
reduct of IFW_FDAR_AdObj are better than those of IV-FS-FRS-2, IARM, ASS-IAR and
IFSA. Moreover, because of the selection of the reduct with highest classification accuracy in
wrapper stage, the classification accuracy of IFW_FDAR_AdObj is higher than IV-FS-FRS-2,
IARM, ASS-IAR and IFSA on all data sets. The classification accuracy of IV-FS-FRS-2,
IARM by fuzzy rough set approach is higher than that of ASS-IAR, IFSA by traditional rough
set approach. For each data set, we can see that the classification accuracy does not increase
when adding incremental data set. This is because there are some noise objects in incremental
data sets that decrease the classification accuracy of learning algorithms.
4. CONCLUSION
The paper proposed a solution to find the reduct of the decision table according to the
combined filter-wrapper approach in the case of adding object sets to minimize the number of
reductive set attributes and improve improve the accuracy of the classification model. In this
paper, IFW_FDAR_AdObj algorithm was introduced to solve the problem. The experimental
results are compared with the other algorithms, that has shown that IFW_FDAR_AdObj
algorithm is efficient. We will continue to study the incremental algorithms that find the reduct
of the decision table in the case of adding and removing the attribute set in the future.
REFERENCES
1. Demetrovics, J., Thi, V.D., & Giang, N.L. (2014), Metric Based Attribute Reduction in Dynamic
Decision systems, Annales Univ. Sci. Budapest., Sect. Comp, Vol. 42, 157-172.
2. Huong, N. T. L., &Giang, N. L. (2016), “Incremental algorithms based on metric for finding reduct
in dynamic decision systems”, Journal on Research and Development on Information &
Communications Technology, Vol.E-3, No.9, 26-39.
3. Z. Pawlak (1991), Rough sets: Theoretical Aspects of Reasoning about Data, Kluwer Academic
Publisher, London.
4. Q.H. Hu, D.R. Yu, Z.X. Xie (2016), Information-preserving hybrid data reduction based on fuzzy-
rough techniques, Pattern Recognit. Lett. 27(5), pp. 414-423, 2016.
5. J.H. Dai, H. Hu, W.Z. Wu,Y.H. Qian, D.B. Huang (2018), “Maximal Discernibility Pairs Based
Approach to Attribute Reduction in Fuzzy Rough Sets”, IEEE Transactions on Fuzzy Systems, Vol.
26, Issue 4, pp. 2174-2187.
6. J.H. Dai, Q.H. Hu, H. Hu, D.B.Huang (2017), “Neighbor inconsistent pair selection for attribute
reduction by rough set approach”, IEEE Transactions on Fuzzy Systems, Vol. 26, Issue 2, pp. 937-950.
7. L.J.Ping, Z. W. Xia, T.Z. Hui, X.Y. Fang, M. T. Yu, Z.J. Jing, Z. G. Yong, J. P. Niyoyita (2020),
“Learning with fuzzy rough set-based attribute selection”, Expert Systems with Applications, Vol.
139, pp. 1- 17.
8. Y.Y. Yang, D.G. Chen, H. Wang, X.H. Wang (2017), “Incremental perspective for feature selection
based on fuzzy rough sets”, IEEE Transactions on Fuzzy Systems, Vol. 26, Issue 3, pp. 1257-1273,
9. Y.Y. Yang, D.G. Chen, H. Wang, Eric C.C.Tsang, D.L. Zhang, “Fuzzy rough set based incremental
attribute reduction from dynamic data with sample arriving”, Fuzzy Sets and Systems, Volume
312, pp. 66-86, 2017.
10. Y.Y. Yang, D.G. Chen, H. Wang (2017), “Active Sample Selection Based Incremental Algorithm
for Attribute Reduction With Rough Sets”, IEEE Transactions on Fuzzy Systems, Vol. 25, Issue 4,
pp. 825-838.
11. W.H. Shua, W.B. Qian, Y.H. Xie (2019), “Incremental approaches for feature selection from
dynamic data with the variation of multiple objects”, Knowledge-Based Systems, Vol. Vol. 163, pp.
320-331.
VỀ THUẬT TOÁN GIA TĂNG RÚT GỌN THUỘC TÍNH KHI BỔ
SUNG TẬP ĐỐI TƯỢNG TRONG BẢNG QUYẾT ĐỊNH THAY ĐỔI
Tóm tắt: Trong mấy năm gần đây, các phương pháp rút gọn thuộc tính theo tiếp cận tập thô
mờ đã thu hút sự quan tâm của các nhà nghiên cứu vì chúng nâng cao độ chính xác của mô
hình phân lớp. Tuy nhiên, phần lớn các phương pháp đề xuất đều thực hiện trên bảng quyết
định không thay đổi. Trong bài báo này, chúng tôi xây dựng thuật toán gia tăng tìm tập rút
gọn xấp xỉ theo hướng tiếp cận kết hợp filter-wrapper. Kết quả thử nghiệm trên một số bộ số
liệu mẫu cho thấy, thuật toán gia tăng đề xuất hiệu quả hơn một số thuật toán gia tăng khác
theo tiếp cận filter về số lượng thuộc tính tập rút gọn và độ chính xác phân lớp.
Từ khoá: Tiếp cận tập thô mờ thuật toán gia tăng đề xuất, bảng quyết định, rút gọn thuộc tính.
DESIGN SOME GRADE 6 MATH TESTS TO ASSESS THE

DEVELOPMENT OF MATHEMATICAL COMPETENCIES
Nguyen Van Anh, Le Thi Van Dung, Tran Anh Phuong, Nguyen Thi Hong*
Hanoi Metropolitan University
Abstract: In this article, we study the innovations in testing and assessing learning and
educational outcomes in the direction of developing students' qualities and competencies
according to the 2018 general education program. From that, we design the regular and
periodical test in the direction of developing qualities and competencies for 6th grade students
according to the new general education program. In each test, we have developed test
specifications, matrix of questions, and grading instructions.
Keywords: Competencies, periodical test, regular test, developing qualities and competencies,
6th grade students, assessing.
Received 20 May 2022

Revised and accepted for publication 26 July 2022
(*) Email: nthong@daihocthudo.edu.vn
1. INTRODUCTION
1.1. Modern tendencies in testing and assessment of learning and education result to
develop student’s quality and competence
1.1.1. Purpose of the assessment
To provide accurate and timely information, to determine the actual situation, learning
achievements, and training according to the level of meeting the required requirements of the
general education program and the progress of students, to adjust teaching and learning
activities to improve the quality of education.
1.1.2. The conception of assessment in the direction of developing student’s quality and
competence
Evaluating learning outcomes in the direction of developing quality and competence should
focus on outputs that are the qualities and competencies of learners, represented in the level and
competency to creatively apply knowledge in solving tasks and requirements of diverse and
vivid reality. This method attaches importance to assessing the learning process of students,
regularly and continuously throughout the teaching process. Competency assessment focuses
on the goal of assessing the progress of learners compared to themselves rather than others.
In essence, there is no contradiction between assessment of competence and assessment of

knowledge and skills. Competence assessment is a higher development step, not entirely based
on the educational program of each subject, where teachers create opportunities for students to
solve problems in real-life situations. To do this, students must apply both knowledge and skills
learned and self-study, in school and in society, not only in terms of knowledge but also
emotionally and ethically.
1.1.3. Principles of testing and assessment in the direction of developing student’s quality
and competence
Current inspection and evaluation need to follow the following principles: ensuring
objectivity, comprehensiveness, systematicity, authenticity, practicality, and efficiency. In
particular, ensuring authenticity, authenticity, and efficiency are the most important. Only in
this way will it be possible to reflect whether the learner's competency to progress, qualities,
and competencies are suitable to the human conditions and needs of the society.
1.1.4. The tendency of completing the assessment
For the test and assessment of students to achieve good results, it is necessary to pay
attention to a few points as follows:
- Guide students to develop self-examination and self-assessment skills.
This comes from the trend of student-centered teaching. Train students in learning methods
to prepare for lifelong and continuous self-study.
- The examination and assessment of students must following requirements:
• Reproduction of knowledge.
• Practice skills and techniques.
• Developing cognitive competence, especially creative thinking competency.
• Create a real change in students’ attitudes and behaviors.
• Train them to detect and solve problems arising in reality.
1.1.5. Competencies that need to be developed
- Aids and tools competency (CC) – being able to make use of and relate to the aids and tools
of mathematics
- Communicating competency (GT) – being able to communicate in, with, and about mathematics
- Problem tackling competency (GQVĐ) – formulating and solving mathematical problems
- Modelling competency (MHH) – being able to analyze and build mathematical models
concerning other areas
- Mathematical thinking competency (TD) – mastering mathematical modes of thought
2. CONTENT
According to the 2018 high school program, the assessment method has also been changed
from a score-based assessment to an assessment based on students’ competencies. After that
change, teachers, as well as educators, will wonder if the test of students’ competency
development? How to design such a test? How to build a test that is closest to the assessment
requirements? Understanding that, we have researched this issue and given 2 test questions: a
regular test and a periodical test for Math 6. We design these 2 tests with the hope to help
teachers more easily build a math test by the requirements of the Ministry of Education and
Training and also serve to develop the competence of students.
The following two tests (including 1 regular test and 1 periodical test) were designed in the
direction of assessing the quality and competency of 6th-grade students. The test samples are
complete: test questions, grading instructions, matrix of questions, exam specifications to
orient, illustrate in reality for teachers the standards of student assessment in the direction of
quality and competence. The scope of knowledge is carried out according to the new 6th-grade
textbook program "Cánh diều" to ensure it is consistent with the current curriculum of the
Ministry of Education and Training.
2.1. Regular test
Time: 45 minutes
Lesson: Powers with natural exponents
Objective: This test is to provide the competency to apply the properties and laws of
exponents and exponents with natural exponents in solving problems; assessment of learning
results, competency to memorize knowledge about exponentiation with natural exponents of
students.
A. Matrix of exam questions
Application
Recognization Understanding Total
Low High
MC Essay MC Essay MC Essay MC Essay

1. Some - Identify some
special special powers.
powers.
Number of 2 2
questions
Time 1 1
(minutes)
Points
1 1
Percentage
10% 10%
Competency TD
element
- Apply properties - Apply properties

2. Properties of powers and of exponents
of exponents. special powers. - Flexible
calculation
transformation
Number of 4 1 1 1 7
questions
Time 5 5 2 7 19
(minutes)
Points
2 2 0,5 3 7,5
Percentage
20% 20% 5% 30% 75%
Competency GQVĐ GQVĐ GQVĐ GQVĐ
element TD
3. Laws for -Find out the rules

large powers. for great powers of
power
-Change regular
expressions
Number of 1 1 2
questions
Time 5 15 20
(minutes)
0,5 1 1,5
Points
5% 10% 15%
Percentage
Competency GQVĐ GQVĐ
element TD TD
Total number 2 4 1 1 1 1 1 11
of questions
Total time
Total score 1 5 5 2 7 5 15 40
Percentage 1 2 2 0,5 3 0,5 1 10
10% 20% 20% 5% 30% 5% 10% 100%
B. Test
I. Multiple choice (4 points)
Question 1 : (Recognization) 𝑎0 (𝑎 ≠ 0) is equal to:
1
A.1 B. a C.0 D.𝑎
Question 2 : (Recognization) Find the natural number n > 0, knowing that: 𝑛2017 = 𝑛
A.2017 B.2016 C.1 D.2
310 .11+310 .5
Question 3: (Low application) Calculate the value of the expression: A= 39 .24
A.768 B.3 C.728 D.36

Question 4: (Understanding) The result of the calculation 105 . 102 is:
A. 103 B. 1010 C. 108 D.107
Question 5: (Understanding) (23 )6 is equal to:
A. 29 B. 218 C. 212 D.23
Question 6: (Recognization) 𝑎𝑚 . 𝑏 𝑚 is equal to:
A. (𝑎𝑏)𝑚 B. (𝑎𝑏)𝑚𝑚 C. (𝑎𝑏)𝑚+𝑚 D.(𝑎 + 𝑏)𝑚
Question 7: (Understanding) Write the following results as a power: 324 : 86
A. 22 B. 4 C. 23 D. 8
Question 8: (High application) Find the last digit of 22013
A. 2 B. 4 C. 6 D. 8
II. Essay (6 points )
Question 1: (Understanding) (2 points ) Find 𝑥, know :
a) (2𝑥 + 1)3 = 125;
b) 3𝑥 + 25 = 26.22 + 2.30 .
Question 2: (Low application) (3 points) Do the following calculations in the most logical way:
a) (217 + 172 )(915 − 315 )(24 − 42 );
b) (82017 − 82015 ): (82104 . 8);
c) 81. (27 + 915 ): (35 + 332 ).
Question 3: (High application) (1 point )
𝐴 = 1 + 3 + 32 + ⋯ + 32006
a) Find the last digit of A,
b) Prove that A = (32007 − 1): 2.
C. Scoring Guide
Multiple choice (4 points)
1 2 3 4 5 6 7 8
A C C D B A A A
Question Answer The point

1 a) 2 1 point
(2 points) b) 4 1 point
a) 0 1 point
2
b) 63 1 point
(3 points )
c) 9 1 point
Essay a) Find the last digit of A:
(0.5 points )
(6 points) 𝐴 = 1 + 3 + 32 + ⋯ + 32006
⇒ The last digit of A is 3.
3
(1 point) b) Prove :𝐴 = (32007 − 1): 2 (0.5 points )
⇒ 3𝐴 − 𝐴 = (3 + 32 + 33 + ⋯ + 32007 ) − (1 +
3 + 32 + ⋯ + 32006 )
⇒ 𝐴 = (32007 − 1): 2
2.2. Periodic test
Time: 90 minutes
Aims: The test aims to provide competence and progress of students after a semester; The
lesson is divided into 30% multiple-choice and 70% essay, the essay section accounts for the
majority of student’s thinking and presentation, recognizing the competency to apply math
knowledge in combination with other subjects and use mathematical tools to solve problems as
well as real-world problems; diagnose strengths and weaknesses of learners to promptly adjust
teaching objectives and teaching methods accordingly.
The periodic test includes knowledge of 3 chapters: chapter 1: natural numbers, chapter 2:
integers, and chapter 3: intuitive geometry. In these 3 chapters, we focus the most on chapter 2.
Because chapter 1 is the part of knowledge that students have learned, familiarized, and applied
a lot in previous classes, hence in the test they put the questions in recognization, low
application form to help students to remind knowledge and a high application question which
improves students’ thinking ability. In terms of chapter 2, this is a new piece of knowledge
compared to students, thus we put questions from understanding to low application to help
students become more fluent when working with this knowledge. Especially when calculating
with negative integers, students still make many mistakes. Therefore, adding more questions to
this knowledge section also helps teachers to realize outstanding problems and creates
opportunities for students to practice with this knowledge. As for chapter 3, this chapter mainly
helps students to review their knowledge of the formula for calculating the perimeter and area
of plane geometry and provides more properties of those shapes. Hence in general, the children
are familiar with this knowledge, therefore we only put 2 questions in this section, including 1
understanding question: help review the properties of specific shapes and 1 question about the
real life problem. The practical application helps them develop problem solving capacity and
apply the knowledge they have learned to specific situations in life.
A. Matrix of test questions

Level
Applications
Recognization Understanding Total
Topic
(Content, Low High
chapter)
MC Essay MC Essay MC Essay MC Essay
-Write a set - Apply divisibility - Apply divisible Apply
- Identify prime properties and sign exponentiatio
numbers definition of prime - Find GCD, n to natural
Chapter 1: Natural
numbers to analyze analyze natural exponents
numbers
a natural number numbers
- Apply divisible
sign
2 1 1 4
Number of questions
(Q1,2) (Q11) (Q13)
Time (minutes)
Score
1 1.5 0.5 3
Ratio %
10% 15% 5% 30%
GQV
GQV
Competency Element TD GQVĐ GQVĐ GQVĐ Đ
Đ
TD
Apply the rule of
brackets, the
Use negative theory of negative
Chapter 2: Integers integers to integers,
represent operations with
integers to
calculate
Number of questions 1 3 1 1 1 7
Time (minutes) (Q4) (Q5,6,7) (Q9) (Q8) (Q10) 5.5
Score 0.5 1.5 1 0.5 2
Ratio % 5% 15% 10% 5% 20% 55%
MHH GQV
Competency Element
TD Đ
Determine the
Calculate area, the
Chapter 3: Intuitive symmetry axes of
perimeter of some
geometry some plane
plane geometry
geometry
Number of questions 1 1 2
Time (minutes) (Q3) (Q12 1.5
Score 0.5 1 15%
Ratio % 5% 10%
GT TD
Competency Element CC GQV
Đ
Total number of questions 3 4 1 1 3 1 13
Total time 1.5 2 1 0.5 4.5 0.5 10
Total score
Ratio % 15% 20% 10% 5% 45% 5% 100%
B. Test
I. Multiple choice: (4 points )
Question 1: (Recognization) A is a set that concludes the natural numbers not equal to 0
and less than 8 can be written:
A. A ={𝑥| 𝑥 ∈ ℕ∗ | 𝑥 < 8}
B. A ={𝑥| 𝑥 ∈ ℕ | 𝑥 < 8}
C. A ={𝑥| 𝑥 ∈ ℕ | 𝑥 ≤ 8}
D. A ={𝑥| 𝑥 ∈ ℕ∗ | 𝑥 ≤ 8}
Question 2: (Recognization) How many prime numbers are less than 18?
A. 8 B. 7 C. 6 D. 9
Question 3: (Understanding) Which of the following clauses is wrong?
A. The rectangle with different side lengths has 2 symmetry axes.
B. The hexagon with equal sides has 3 symmetry axes.
C. The rhombus an with angle not equal to 90° has 2 symmetry axes.
D. The isosceles triangle has 1 symmetry axis.
Question 4: (Recognization) Which of the following is the integer indicating the year of
the event "The first Olympics took place in 776 BC"?
A. – 1776 B. 776 C. -776 D. 1776
Question 5: (Understanding) If 𝑥 − 𝑐 = 𝑎 − (𝑎 + 𝑐 + 𝑏), then 𝑥 is equal to:
A. 𝑥 = −𝑏 B. 𝑥 = 𝑎 − 𝑏 + 𝑐
C. 𝑥 = 𝑎 + 𝑏 − 𝑐 D. 𝑥 = −𝑎
Question 6: (Understanding) Choose the correct equality:
A. −24 + 4 = 21 B. 54 + (−6) = 48
C. 4.3 + (−5) = 6 D. −9 + 32 = 16
Question 7: (Understanding) In the discussion, An, Binh, and Chi made the following statements:
An: “The sum of two positive integers is always a positive integer”
Binh: “The sum of two negative integers is always a negative integer”
Chi: “The sum of two integers with same sign always has the same sign as those two integers”
Which statement is correct, which one is incorrect?
A. An, Binh are correct, Chi is wrong. B. An is correct, Binh and Chi are wrong.
C. All of them are correct. D. All of them are wrong.
Question 8: (Low application) Let 𝑥 ∈ ℤ and (−154 + 𝑥) ⋮ 3 then:
A. 𝑥 divided by 3, remainder 1
B. 𝑥 is divisible by 3
C. 𝑥 divided by 3, remainder 2
D. inconclusive is divisible by 3 of 𝑥
II. Essay: (6 points)
Question 9: (Understanding) (1 point) Find 𝑥, know:
a. −17 − (−2 + 𝑥) = 3
b. (−2𝑥 + 2). (−4𝑥 − 23 ) = 0
c. −7. (5 − 𝑥) − 2. (𝑥 − 10) = 15
d. 𝑥 − 15 = −20 − 4𝑥
Question 10: (Low application) (2 points) Quick calculate (if can):
a) 24.82 + 24. 18 – 100 b) [131 − (13 − 4)2 ] − 24 . 5
c) 12 + 3. [39 – (5 – 2)2 ] d) 20180 − {152 : [175 + (23 . 52 − 6.25)]}
Question 11: (Low application) (1.5 points) A military unit of 576 men and 360 women.
Come to the Central region to help people overcome the consequences of floods. They intend
to divide into many groups to serve many communes, in which the distribution of men and
women equally among the groups. How many groups can be divided at most? At that time, each
group has how many men and how many women.
Question 12: (Low application) (1 point) The floor of a factory is rectangular. They want
to cover the floor of that factory with bricks. Assume a brick has a length and width of 9m and
5m. The price of each brick is 35,000 VND. Know the width of the factory is 75m. The width
is half of its length.
a) Calculate the floor area of the factory.
b) How much money does it take to cover the floor of the factory?
Question 13: (High application) (0.5 points)
Let A = 20213 and B= 2020.2021.2022
Do not specifically calculate the value of A and B. Compare A and B.
C. Scoring guide
Multiple choice (4 points)
1 2 3 4 5 6 7 8
A B B C A B C C
Sentence Answer Point

a) 𝑥 ∈ {−18} 0.25 points
Question 9 b) 𝑥 ∈ {−2; 1} 0.25 points
(1 point) c) 𝑥 ∈ {6} 0.25 points
d) 𝑥 ∈ {−1} 0.25 points
a) 2300 0.5 points
Question 10 b) -30 0.5 points

(2 points) c) 102 0.5 points
d) 0 0.5 points
Find GCD(576, 360) is equal to 72. 0.5 points
Find the number of men in each group 0.5 points
Question 11
is 8.
(1.5 points)
Essay Find the number of women in each 0.5 points
(7 points) group is 5.
2. The length of the factory floor is 0.25 points
150 m.
The floor area of the factory is 11250 0.5 points
𝑚2 .
Question 12
3. The area of one brick is 45 𝑚2 . 0.25 points
(2 points)
The number of bricks needed to cover
the floor is 250 tablets. 0.5 points
The amount to spend to cover the
floor is 8750000 dong. 0.5 points
Find:
Question 13 A =20213
0.5 points
(0.5 points) B =20213 − 2021
⇒A > B
4. CONCLUSION
In this article, we have built two test questions to assess the student’s ability in the direction
of developing the quality of learners’ competence for grade 6 students according to the 2018
general education program, of which one is: one is the regular test (45 minutes) and the other
is the periodic test (90 minutes).
REFERENCES
1. B. V. Tuyên, Bài tập nâng cao và một số chuyên đề Toán 6, Hà Nội: Nhà xuất bản giáo dục Việt Nam.
2. Bộ GD&ĐT, “Chương trình phổ thông tổng thể 2018”.
3. Deborah V. Mink, Earlene J. (FRW) Hall (2009), “Problem – Solving Overview,” in Strategies for
Teaching Mathematics, Shell Education, pp. 187 – 237.
4. J. P. Minda (2015), The psychology of thinking reasoning, decision – making & problem – solving,
SAGE.
5. National Council for Teachers of Mathematics (2000), Principles and Standards for School
Mathematics.
6. S. Frei(2008), Teaching Mathematics today, Shell Education.
7. TS Nguyễn Sơn Hà,TS Phạm Xuân Chung, Th.S Nguyễn Thu Hằng, PGS. TS Nguyễn Thị Thanh
Hồng, PGS. TS Nguyễn Thị Thanh Huyền, Th.S Đỗ Quang Long, Th.S Đào Minh Thư, TS Nguyễn
Ngọc Tú, Th.S Vũ Thị Phương Thảo (2020), “Tài liệu hướng dẫn bồi dưỡng giáo viên phổ thông cốt
cán Mô đun 3: Kiểm tra đánh giá học sinh trung học cơ sở theo hướng phát triển phẩm chất, năng
lực môn Toán,” Hà Nội.
THIẾT KẾ MỘT SỐ ĐỀ KIỂM TRA ĐÁNH GIÁ MÔN TOÁN LỚP 6

THEO HƯỚNG PHÁT TRIỂN PHẨM CHẤT, NĂNG LỰC
Tóm tắt: Trong bài báo này chúng tôi nghiên cứu những đổi mới trong kiểm tra đánh giá kết
quả học tập, giáo dục theo hướng phát triển phẩm chất, năng lực của HS theo chương trình
giáo dục phổ thông 2018. Từ đó chúng tôi thiết kế các đề kiểm tra thường xuyên và định kỳ
theo hướng phát triển phẩm chất, năng lực cho học sinh lớp 6 theo chương trình giáo dục phổ
thông mới. Trong mỗi đề kiểm tra chúng tôi có xây dựng đặc tả đề kiểm tra, ma trận đề, và
hướng dẫn chấm cho từng đề.
Từ khoá: Năng lực; bài kiểm tra định kỳ; bài kiểm tra thường xuyên; phát triển năng lực;
phát triển phẩm chất, học sinh lớp 6; đánh giá.
SOME REAL PROBLEMS AND APPLICATION OF THE GREATEST

COMMON DIVISOR, THE LEAST COMMON MULTIPLE
Hoang Minh Giang, Nguyen Thi Thu Hoa, Nguyen Thi Hong*
Abstract: One of the contemporary learning and teaching orientations in the general
education curriculum in 2018 is to connect math to everyday lives and focus on applying
mathematics concepts to solve real-world issues. As a result, we studied a variety of practical
problems involving the greatest common divisor and least common multiple. Furthermore, we
analyzed typical application problems and the RSA algorithm in cryptography. In particular,
we came up with real-life problems involving the greatest common divisor and least common
multiple and implemented the RSA algorithm in Maple.
Keywords: cryptosystem, RSA algorithm, key generation, key distribution, encryption,
decryption, greatest common divisor, least common multiple, common divisor, common
multiple, common factor, prime factor.
Received 8 May 2022

1. INTRODUCTION
The greatest common divisor and the least common multiple have many applications in
practice as well as in mathematics. For example in finding the common denominator of two or
more fractions; finding a way to evenly distribute the number of gifts to the students; queuing
in sports,… More recently, its applications have also been found in cryptography, and computer
science (see [2;3;5] and the reference therein). Here are some definitions and how to find the
greatest common divisor and least multiple.
1.1. Divisor and multiple
Definition 1.1 ([2],p.1)
If natural number 𝑎 is divisible by natural number 𝑏, we say 𝑎 is a multiple of 𝑏 and we
also call that 𝑏 is a divisor of 𝑎.
Notation: 𝐷(𝑎) is the set of divisors of 𝑎.
𝑀(𝑎) is the set of multiples of 𝑎.
1.2. Common divisor and greatest common divisor

Definition 1.2 ([6],p.51 &p.54)
A common divisor of two or more numbers is the divisor of all the numbers.
The greatest common divisor of two or more numbers is the greatest number in the set of
their common divisors.
Notation: 𝐶𝐷(𝑎; 𝑏) is the set of common divisors of 𝑎 and 𝑏.
𝐺𝐶𝐷(𝑎; 𝑏) is the greatest common divisor of 𝑎 and 𝑏.
How to find the greatest common divisor ([6],p.55)
We find the greatest common divisor by factoring numbers into prime factors. So finding
𝐺𝐶𝐷 of two or more numbers greater than 1, we do the three steps as follows:
Step 1: Factor each number into prime factors
Step 2: Choose common factors
Step 3: Make a product of chosen factors, with the smallest exponent for each factor.
That product is 𝐺𝐶𝐷 which we have to find
1.3. Common multiple and least common multiple
Definition 1.3 ([6], p.52 &p.57)
A common multiple of two or more numbers is the multiple of all the numbers.
The least common multiple of two or more numbers is the leasr non-zero number in the set
of their common multiples.
Notation: 𝐶𝑀(𝑎; 𝑏) is the set of common multiples of 𝑎 and 𝑏.
𝐿𝐶𝑀(𝑎; 𝑏) is the least common multiple of 𝑎 and 𝑏.
How to find the least common multiple ([6], p.58)
We find the least common multiple by factoring numbers into prime factors. So finding
𝐿𝐶𝑀 of two or more numbers greater than 1, we do the three steps as follows:
Step 1: Factor each number into prime factors
Step 2: Choose common and individual factors
Step 3: Make a product of chosen factors, take for each factor with its highest exponent.
That product is 𝐿𝐶𝑀 which we have to find.
2. SOME REAL PROBLEMS ABOUT THE GREATEST COMMON DIVISOR
AND THE LEAST COMMON MULTIPLE
2.1. Problem about “the area”
Example 1: Uncle Tan has a rectangular field with 70𝑚 long and 42𝑚 wide. He wants to
divide this field into the equal squares with the units of meters in order to grow flowers and
fruits. So with this division, what is the maximum length of this square side?
Analysis: Uncle Tan wants to divide the rectangular field into the equal squares so the
length and the width of the rectangle field is divisible by the length of the square side. Therefore,
the length of the square side is the common divisor of 70 and 42.
Solution:
Let 𝑎 be the length of the square side (𝑚𝑒𝑡𝑒𝑟; 0 < 𝑎 < 42).
According to the exercise 1, uncle Tan divides the rectangular field into the equal squares
in order to grow flowers and fruits so 𝑎 is the common divisor of 70 and 42.
If the length of the square side is the maximum, then 𝑎 must be the greatest common
divisor of 70 and 42. 𝑆𝑜 𝑎 = 𝐺𝐶𝐷(70; 42) .
Now, we find the greatest common divisor of 70 and 42. We have: 70 = 2.5.7; 42 = 2.3.7.
Hence 𝐺𝐶𝐷(70; 42) = 2.7 = 14 ⇒ 𝑎 = 14 (satisfied).So the length of the largest square
side is 14 meters.
2.2. Problem about dividing the number of people or things
Example 2: The homeroom teacher at class 5A of Cu Khoi primary school wants to divide
36 notebooks, 18 boxes of crayons and 60 ballpoint pens into some equal rewards to give to
students on the year-end celebration. What is the greatest number of rewards that can be divided
into? How many notebooks, boxes of crayons and ballpoint pens are there for each reward?
Analysis: Because the homeroom teacher at class 5A of Cu Khoi primary school wants to
divide 36 notebooks, 18 boxes of crayons and 60 ballpoint pens into some equal rewards so the
number of notebooks, boxes of crayons and ballpoint pens in each prize is the common divisor
of 36; 18; 60.
Solution:
Let 𝑎 be the number of rewards which the homeroom teacher at class 5A of Cu Khoi
primary school gives to students on the year-end celebration (𝑎 ∈ ℕ∗ ; 𝑎 < 18).
Because the homeroom teacher at class 5A of Cu Khoi primary school wants to divide 36
notebooks, 18 boxes of crayons and 60 ballpoint pens into some equal rewards so 𝑎 is the
common divisor of 36; 18 and 60.
If the reward is the maximum, then 𝑎 must be the largest number such that 36 ⋮ 𝑎; 18 ⋮
𝑎; 60 ⋮ 𝑎. Therefore: 𝑎 = 𝐺𝐶𝐷(36; 18; 60).
Now, we find the greatest common divisor of 36; 18 and 60. We have:
36 = 22 . 32 ; 18 = 2.32 ; 60 = 22 . 3.5
Thus 𝐺𝐶𝐷(36; 18; 60) = 2.3 = 6 ⇒ 𝑎 = 6 (satisfied). Therefore, it is possible to divide at
most into 6 rewards are 6 notebooks, 3 boxes of crayons and 10 ballpoint pens.
2.3. The problem about queuing

Example 3: The number of students in grade 6 of Cu Khoi middle school is divided into
three classes: 6A, 6B and 6C. The class 6A has 45 students, the class 6B has 50 students and
the class 6C has 55 students. During the flag-raising ceremony on the first of the week, both of
them line up in some equal vertical rows to salute the flag without any odd people in each class.
Find the maximum number of vertical rows that can be lined.
Analysis: Because the number of students in grade 6 of Cu Khoi middle school line up in
some equal vertical rows to salute the flag without any odd people in each class so the number
of vertical rows is the common divisor of 54; 42 and 48.
Solution:
Let 𝑎 be the number of vertical rows that can be lined (𝑎 ∈ ℕ∗ ; 𝑎 < 45).
According to the exercise 3, the number of students in grade 6 of Cu Khoi middle school
is divided into three classes: 6A, 6B, 6C and during the flag-raising ceremony on the first of the
week, both of them line up in some equal vertical rows to salute the flag without any odd people
in each class so 𝑎 is the common divisor of 54; 42 amd 48.
If the vertical row is the maximum, 𝑎 must be the largest number such as 54 ⋮ 𝑎; 42 ⋮
𝑎; 48 ⋮ 𝑎. Therefore: 𝑎 = 𝐺𝐶𝐷(54; 42; 48).
Now, we find the greatest common divisor of 54; 42 and 48. We have:
54 = 2.33 ; 42 = 2.3.7; 48 = 24 . 3
Hence 𝐺𝐶𝐷(54; 42; 48) = 2.3 = 6 ⇒ 𝑎 = 6 (satisfied). So the maximum number of
vertical rows is 6.
2.4. Mathematics problems about calculating the number of days or hours for some
objects to work together.
Example 4: Mai and Lan are students of Bat Trang middle school but in two different
classes. Mai scans the classroom every 15 days and Lan scans the classroom every 9 days. At
the first time, both of them scan the classroom on the same day. At least how many days do
they scan the classroom on the same day again?
Analysis: Because Mai and Lan are students of Bat Trang middle school but in two different
classes, Mai scans the classroom every 15 days and Lan scans the classroom every 9 days so
the number of days which both of them scan the classroom on the same day is the common
multiple of 15 and 9.
Solution:
Let 𝑎 be the number of days which both of them scan the classroom on the same day
(𝑎 ∈ ℕ∗ ; 𝑎 > 15).
Because Mai and Lan are students of Bat Trang middle school but in two different classes,
Mai scans the classroom every 15 days and Lan scans the classroom every 9 days so 𝑎 is the
common multiple of 15 and 9.
If the number of days is the maximum, 𝑎 must be the least common multiple of 15 and 9.
Therefore: 𝑎 = 𝐿𝐶𝑀(15; 9).
Now, we find the least common multiple of 15 and 9. We have:15 = 1.3.5; 9 = 1.32 .
Therefore 𝐿𝐶𝑀(15; 9) = 1.32 . 5 = 9.5 = 45. Hence 𝑎 = 45 (satisfied).
So at least 45 days they scan the classroom on the same day again.
3. APPLICATION OF THE GREATEST COMMON DIVISOR AND THE
LEAST COMMON MULTIPLE
RSA algorithm
GCD can be used for several applications in modular arithmetic. One of the applications is
RSA algorithm. Derived in 1977 by Ron Rivest, Adi Shamir and Leonard Adleman, RSA has
become a commonly used public-key cryptosystem.
There are four steps in implementing RSA algorithm: key generation, key distribution,
encryption and decryption.
Key generation.
Being an asymmetric cryptosystem, RSA involves a public key and a private key.
The key generation is based on idea of creating a one-way function with intention that the
ciphertext would only be decrypted in a 28hần2828ntial amount of time using the private key
(see Fig.1).
Fig 1. One way functions are easy to compute but it is extremely hard to compute their
inverse functions. (© Bill MacKenty)
The keys are generated as follow:
1. Choose distinct large prime numbers 𝑝 and 𝑞. These numbers are kept secret.
2. Compute 𝑛 = 𝑝𝑞. This is the modulus for public key and private key.
3. Calculate 𝜙(𝑛) and 𝜙(𝑛) is kept secret.
4. Choose 𝑒 such that 1 < 𝑒 < 𝜙(𝑛) and gcd(𝑒, 𝜙(𝑛)) = 1.
𝑒 is the public key exponent.
5. Determine the private key exponent 𝑑 = 𝑒 −1 (𝑚𝑜𝑑 𝜙(𝑛)) using the extended
Euclidean algorithm. According to Bézout’s identity, there exist 𝑠, 𝑡 such that
𝑠 ⋅ 𝑒 + 𝑡 ⋅ 𝜙(𝑛) = gcd(𝑒, 𝜙(𝑛)) = 1.
We can set 𝑑 = 𝑠 if 𝑠 > 0 and 𝑑 = 𝜙(𝑛) − 𝑠 if 𝑠 < 0.
The public key is (𝑛, 𝑒) and the private key is (𝑝, 𝑞, 𝑑).
Key distribution
We suppose that Bob wants to send a message to Alice and among them, there are
intermediaries such as wifi points, internet service providers and their email servers and
eavesdroppers. To use RSA, Alice transmits her public key (𝑛, 𝑒) to Bob and keeps her private
key 𝑑. Bob then uses Alice’s public key to encrypt the message. After receiving the ciphertext,
Alice can use her private key to decrypt it.
Encryption
The message is encoded to an integer 𝑚 such that 0 ≤ 𝑚 < 𝑛.
Using Alice’s public exponent key 𝑒, the ciphertext 𝑐 = 𝑚𝑒 (𝑚𝑜𝑑 𝑛).
Bob then transmits the ciphertext to Alice.
Decryption
Alice can decrypt the ciphertext 𝑐 by using her private key exponent 𝑑 by computing
𝑐 𝑑 = (𝑚𝑒 )𝑑 = 𝑚 (𝑚𝑜𝑑 𝑛).
Then, Alice can decode 𝑚 to get the original message. [7][9]
Fig 2. Encrypting and decrypting data using RSA

We implemented RSA algorithm in Maple 2020 as follow:
# RSA(Rivest-Shamir-Adleman) algorithm
restart:
with(StringTools):
list_to_str := proc(list_of_int)
local list_of_str:
list_of_str := map(x -> convert(x, string), list_of_int):
return Join(list_of_str, “”)
end proc:
list_hex_format:= proc(dec_list)
local list_of_hex:
list_of_hex := map(x -> convert(x, ‘hex’), dec_list):
return Join(list_of_hex, “”)

end proc:
min_public_exp := proc(totient_n)
local min_exponent_e:
for min_exponent_e from 2 to totient_n do
if gcd(min_exponent_e, totient_n) = 1 then
break
end if:
end do:
return min_exponent_e
end proc:
private_exponent := proc(public_exponent, totient)
local s, t, private_exp_key:
description “compute private key exponent d:” \
“d * e = 1 (mod phi_n) using extended Euclidean algorithm” \
“Bezout’s identity gcd(e, phi_n) = 1 = d * e + t * phi_n”:
igcdex(public_exponent, totient, ‘s’, ‘t’):
if s > 0 then
private_exp_key := s:
else
private_exp_key := totient + s:
end if:
return private_exp_key
end proc:
encrypting_str := proc(msg, pub_key)
description “description function f(message, public_key) “:
return msg ^ pub_key[2] mod modulus_n:
end proc:
decrypting_str := proc(cipher, private_exp)
description “decrypting function g(c, k’) = c^d mod n = original_msg” \
“private key k’ = (p, q, d) or (n, d) -> decrypt” :
return cipher^private_exp mod modulus_n
end proc:
decrypted_txt := proc(decrypted_dec)
description “convert list of decimal equivalents back to text”:
return Join(map(x -> Char(x), decrypted_dec), “”);
end proc:
# RSA algorithm
# secret message m
# msg_str := “The quick brown fox jumps over the lazy dog”;
# convert the message to decimal equivalents
msg_decimal := map(Ord, Explode(msg_str));
msg_decimal_str := list_to_str(msg_decimal);
“Message in hex format” = list_hex_format(msg_decimal);
# 1. Choose different large random prime numbers p & q
private_key_p := 257;
private_key_q := 263;
# 2. Calculate n = p * q = modulus for public key & private key
modulus_n := private_key_p * private_key_q;
# 3. Euler’s totient function: (private) phi(n) = (p – 1) * (q – 1)
phi_n := (private_key_p – 1) * (private_key_q – 1);
# phi_n := NumberTheory:-Totient(n);
# 4. Choose public key exponent 1 < e < phi(n), gcd(e, phi(n)) = 1
# short bit-length, small Hamming weight -> more efficient encryption
# (smallest/fastest/ least secure) public key exponent
public_exponent_key := min_public_exp(phi_n);
# 5. Compute private key exponent d: d * e = 1 (mod phi_n)
private_exponent_key := private_exponent(public_exponent_key, phi_n);
# 6. Public_key k = (modulus_n, public_exponent_key) -> encrypt
public_key := [modulus_n, public_exponent_key]:
public_key_hex := map(x -> convert(x, ‘hex’), public_key);
# 7. Apply the encrypting function on (message, public_key)
encrypted_decimals := map(x -> encrypting_str(x, public_key),
msg_decimal);
encrypted_decimal_str := list_to_str(encrypted_decimals);
“Cipher text” = list_hex_format(encrypted_decimals);
# 8. Apply the decrypting function on (c, d)

# the same as msg_decimal
decrypted_decimals := map(x -> decrypting_str(x, private_exponent_key),
encrypted_decimals);
“Decrypted message” = list_hex_format(decrypted_decimals);
decrypted_str = decrypted_txt(decrypted_decimals);
The important results obtained are in Fig 4 and Fig 5. Here, the message to be sent is “The
quick brown fox jumps over the lazy dog”. It is a popular English pangram that contains all
English alphabets. We then convert the alphabets to its ASCII decimal equivalents and display
the message in hex format. The public key is (𝑛, 𝑒) = (67591, 3) is used to encrypt the
message, then the cipher text is displayed in hex format. After using the private exponent key
is 𝑑 = 44715, we obtain the original message.
Fig 4. Encryption process
Fig 5. Decryption process

4. CONCLUSION
The article presents some real problems about the greatest common divisor and the least
common multiple in practice. Moreover we analyzed typical application problems and the RSA
algorithm in cryptography.
REFERENCE
1. R.L. Rivest, A. Shamir, and L.M. Adleman (1978), “A Method for Obtaining Digital Signature and
Public-key Cryptosystems”, Communications of the ACM, Vol. 21, No. 2, pp. 120-126.
2. Hengki Tamando Sihotang et al (2020), “Design and Implementation of Rivest Shamir Adleman’s
(RSA) Cryptography Algorithm in Text File Data Security”, Journal of Physics: Conference Series.
3. George E. Andrews (1971), Number theory, Saunders company.
4. Hà Huy Khoái (tổng chủ biên), sách giáo khoa toán 6 tập 1 bộ kết nối tri thức với cuộc sống, Nhà
xuất bản giáo dục Việt Nam.
5. Lê Đức Thuận – Tạ Ngọc Trí (đồng chủ biên), Sách củng cố và ôn luyện Toán 6 tập 1, Nhà xuất bản
đại học quốc gia Hà Nội.
6. Bộ Giáo Dục và Đào Tạo, sách song ngữ việt – anh toán 6 tập 1, Nhà xuất bản giáo dục Việt Nam.
MỘT SỐ BÀI TOÁN THỰC TẾ VÀ ÁP DỤNG CỦA ƯỚC CHUNG

LỚN NHẤT, BỘI CHUNG NHỎ NHẤT
Tóm tắt: Một trong những định hướng của chương trình giáo dục phổ thông môn toán năm
2018 đó chính là kết nối toán học với thực tiễn, vận dụng toán học vào thực tiễn. Nhằm góp
phần phát triển mục tiêu của chương trình giáo dục phổ thông, chúng tôi nghiên cứu một số
bài toán thực tiễn và áp dụng của ước chung lớn nhất và bội chung nhỏ nhất. Hơn nữa, chúng
tôi trình bày năm dạng bài toán thực tế liên quan đến ước chung lớn nhất, bội chung nhỏ nhất
và một áp dụng của ước chung lớn nhất trong thuật toán RSA trong ngành mật mã học. Cụ
thể chúng tôi xậy dựng được các ví dụ về các bài toán thực tế và triển khai được thuật toán
RSA trong 33hần33.
Từ khoá: hệ thống mật mã, thuật toán RSA, tạo khóa, 33hần phối khóa, mã hóa, giải mã, ước
chung lớn nhất, bội chung nhỏ nhất, ước chung, bội chung, thừa số chung, thừa số nguyên tố.
APPLICATION OF LAGRANGE MULTIPLIER METHOD

IN THE PROBLEM OF FINDING THE MINIMUM
AND MAXIMUM VALUES
Nguyen Duc Nguyen, Nguyễn Thi Hong*

Abstract: In this paper, we study the Lagrange multiplier method of the extrema of a
multivariable function, from which we apply to solve problems of finding the maximum and
minimum values of general math. Moreover, from the maximum value, the smallest value
found, we can find a primary solution to the problem. In addition, we also create new math
problems for secondary and high school students.
Keywords: Maximum problem, minimum problem; Lagrange multiplier method.

1. INTRODUCTION
In mathematics, the Lagrange multiplier method (named after the mathematician Joseph
Louis Lagrange) is a method for finding the local minimum or maximum of a function of many
variables satisfying certain given conditions. The problems of finding the maximum and the
minimum value at the high school level also often have many variables, and often the variables
satisfy some given condition. We find that the problem of finding the maximum value, the
minimum value with many variables is like the problem of finding the conditional extrema of
a function of many variables. Therefore, a question arises about whether it is possible to use the
Lagrange multiplier method to solve problems about the maximum and minimum values?
On the other hand, extreme problems at the high school level are one of the most difficult
types of math. It often appears in test questions for good students, entrance exams for grade 10,
etc. Usually, we will find a solution. Of the problem if the extrema are known and the equal
sign occurs. Therefore, we study and use the Lagrange multiplication method to find the
extreme values, then we rely on the found extremes to find elementary solutions for these
problems. To have a closer look at the Lagrange multiplication method, we refer readers to
works [1, 2, 3] and the references therein.
2. CONTENT
2.1. Directional derivatives
Definition 2.1
The directional derivative of 𝑓 at (𝑥0 , 𝑦0 ) in the direction of a unit vector 𝑢 = 〈𝑎, 𝑏〉 is
f ( x0 + ha , y0 + hb ) − f ( x0 , y0 )
Du f ( x0 , y0 ) = lim ,
h→0 h
if this limit exists.
Theorem 2.2.[2,p. 912]
If 𝑓 is a differentiable function of 𝑥 and 𝑦 , then 𝑓 has a directional derivative in the
direction of any unit vector 𝑢 = 〈𝑎, 𝑏〉 and
Du f (x , y) = fx (x , y)a + fy (x , y)b .
Definition 2.3.[2,p.913]
If 𝑓 is a function of two variables 𝑥 and 𝑦, then the gradient of 𝑓 is the vector function ∇𝑓
defined by
f f
f (x , y) = fx (x , y), fy (x , y) = i+ j .
x y
Definition 2.4.[2,p. 914]
The directional derivative of 𝑓 at (𝑥0 ; 𝑦0 ; 𝑧0 ) in the direction of a unit vector
𝑢 = 〈𝑎, 𝑏, 𝑐〉 is
f ( x0 + ha , y0 + hb, z0 + hc ) − f ( x0 , y0 , z0 )
Du f ( x0 , y0 , z0 ) = lim ,
h→0 h
if this limit exists.
If we use vector notation, then we can write both definitions (2.1 and 2.4) of the directional
derivative in the compact form
f ( x 0 + hu ) − f ( x 0 )
Du f ( x 0 ) = lim ,
h→0 h
where 𝑥0 = 〈𝑥0 , 𝑦0 〉 if n = 2 and 𝑥0 = 〈𝑥0 , 𝑦0 , 𝑧0 〉 if n = 3 . This is reasonable because
the vector equation of the line through 𝑥0 in the direction of the vector 𝒖 is given by 𝑥 = 𝑥0 +
𝑡𝒖 and so 𝑓(𝑥0 + ℎ𝒖) represents the value of 𝑓 at a point on this line.
If 𝑓(𝑥, 𝑦, 𝑧) is differentiable and 𝒖 = 〈𝑎, 𝑏, 𝑐〉, then the same method that was used to prove
Theorem 2.2 can be used to show that
Du f (x , y , z) = fx (x , y , z)a + fy (x , y , z)b + fz (x , y , z)c . (2.6)
For a function 𝑓 of three variables, the gradient vector, denoted by ∇𝑓 or grad 𝑓, is
f (x , y , z) = fx (x , y , z), fy (x , y , z), fz (x , y , z)
or, for short,

f f f
f = fx , fy , fz = i+ j+ k .
x y z
Then, just as with functions of two variables, Formula 2.6 for the directional derivative can
be rewritten as
Du f (x , y , z) = f (x , y , z)  u . (2.7)
2.2. Maximum and minimum values
One of the main uses of ordinary derivatives is in finding maximum and minimum values.
In this section we see how to use partial derivatives to locate maxima and minima of functions
of two variables.
Look at the hills and valleys in the graph of 𝑓 shown in Figure 1. There are two points
(𝑎, 𝑏) where 𝑓 has a local maximum, that is, where 𝑓(𝑎, 𝑏) is larger than nearby values of
𝑓(𝑎, 𝑏). The larger of these two values is the absolute maximum. Likewise, 𝑓 has two local
minima, where 𝑓(𝑎, 𝑏) is smaller than nearby values. The smaller of these two values is the
absolute minimum.
Definition 2.5.[2,p. 923]

A function of two variables has a local maximum at (𝑎, 𝑏) if 𝑓(𝑥, 𝑦) ≤ 𝑓(𝑎, 𝑏) when is
near (𝑎, 𝑏). [This means that 𝑓(𝑥, 𝑦) ≤ 𝑓(𝑎, 𝑏) for all points (𝑥, 𝑦) in some disk with center
(𝑎, 𝑏). The number 𝑓(𝑎, 𝑏) is called a local maximum value. If 𝑓(𝑥, 𝑦) ≥ 𝑓(𝑎, 𝑏) when (𝑥, 𝑦)
is near (𝑎, 𝑏) , then 𝑓 has a local minimum at (𝑎, 𝑏) and 𝑓(𝑎, 𝑏) is a local minimum value.
If the inequalities in Definition 2.5 hold for all points (𝑥, 𝑦) in the domain of 𝑓, then 𝑓
has an absolute maximum (or absolute minimum) at (𝑎, 𝑏) .
Theorem 2.6.[2,p.923]
If 𝑓 has a local maximum or minimum at (𝑎, 𝑏) and the first-order partial derivatives of 𝑓
exist there, then 𝑓𝑥 (𝑎, 𝑏) = 0 and 𝑓𝑦 (𝑎, 𝑏) = 0.
2.2. Lagrange multipliers
It’s easier to explain the geometric basis of Lagrange’s method for functions of two
variables. So we start by trying to find the extreme values of 𝑓(𝑥, 𝑦) subject to a constraint of
the form 𝑔(𝑥, 𝑦) = 𝑘. In other words, we seek the extreme values of 𝑓(𝑥, 𝑦) when the point
(𝑥, 𝑦) is restricted to lie on the level curve 𝑔(𝑥, 𝑦) = 𝑘. Figure 1 shows this curve together
with several level curves of 𝑓. These have the equations 𝑓(𝑥, 𝑦) = 𝑐, where 𝑐 = 7, 8,9,10,11.
To maximize 𝑓(𝑥, 𝑦) subject to 𝑔(𝑥, 𝑦) = 𝑘 is to find the largest value of 𝑐 such that the level
curve 𝑓(𝑥, 𝑦) intersects 𝑔(𝑥, 𝑦) = 𝑘. It appears from Figure 1 that this happens when these
curves just touch each other, that is, when they have a common tangent line. (Otherwise, the
value of 𝑐 could be increased further.) This means that the normal lines at the point (𝑥0 , 𝑦0 )
where they touch are identical. So the gradient vectors are parallel; that is,
f ( x0 , y0 ) = g ( x0 , y0 ) for some scalar  .
This kind of argument also applies to the problem of finding the extreme values of
𝑓(𝑥, 𝑦, 𝑧) subject to the constraint 𝑔(𝑥, 𝑦, 𝑧) = 𝑘. Thus the point (𝑥, 𝑦, 𝑧) is restricted to lie on
the level surface S with equation 𝑔(𝑥, 𝑦, 𝑧) = 𝑘. Instead of the level curves in Figure 1, we
consider the level surfaces 𝑓(𝑥, 𝑦, 𝑧) = 𝑐 and argue that if the maximum value of 𝑓 is
𝑓(𝑥0 , 𝑦0 , 𝑧0 ) = 𝑐 , then the level surface 𝑓(𝑥, 𝑦, 𝑧) = 𝑐 is tangent to the level surface
𝑔(𝑥, 𝑦, 𝑧) = 𝑘 and so the corresponding gradient vectors are parallel.
This intuitive argument can be made precise as follows. Suppose that a function f has an
extreme value at a point 𝑃(𝑥0 , 𝑦0 , 𝑧0 ) on the surface S and let C be a curve with vector equation
𝑟(𝑡) = 〈𝑥(𝑡), 𝑦(𝑡), 𝑧(𝑡)〉 that lies on S and passes through P. If 𝑡0 is the parameter value
corresponding to the point P, then 𝑟(𝑡0 ) = 〈𝑥0 , 𝑦0 , 𝑧0 〉 . The composite function ℎ(𝑡) =
𝑓(𝑥(𝑡), 𝑦(𝑡), 𝑧(𝑡)) represents the values that 𝑓 takes on the curve C. Since 𝑓 has an extreme
value at (𝑥0 , 𝑦0 , 𝑧0 ), it follows that ℎ has an extreme value at 𝑡0 , so ℎ′ (𝑡0 ) = 0. But if 𝑓 is
differentiable, we can use the Chain Rule to write
0 = h (t0 ) = fx ( x0 , y0 , z0 ) x  (t0 ) + fy ( x0 , y0 , z0 ) y  (t0 ) + f2 ( x0 , y0 , z0 ) z  (t0 )
= f ( x0 , y0 , z0 )  r ( t0 ) .
This shows that the gradient vector ∇𝑓(𝑥0 , 𝑦0 , 𝑧0 ) is orthogonal to the tangent vector
′
𝑟 (𝑡0 ) to every such curve C. But we already know that the gradient vector of 𝑔, ∇𝑓(𝑥0 , 𝑦0 , 𝑧0 ),
is also orthogonal to 𝑟 ′ (𝑡0 ) for every such curve. This means that the gradient vectors
∇𝑓(𝑥0 , 𝑦0 , 𝑧0 ) and ∇𝑔(𝑥0 , 𝑦0 , 𝑧0 ) must be parallel. Therefore, if∇𝑔(𝑥0 , 𝑦0 , 𝑧0 ) ≠ 0, there is a
number  such that
f ( x0 , y0 , z0 ) = g ( x0 , y0 , z0 ) .
The number  in Equation 1 is called a Lagrange multiplier. The procedure based on

Equation 1 is as follows.
Method of Lagrange multiplers
To find the maximum and minimum values of 𝑓(𝑥, 𝑦, 𝑧) subject to the constraint
𝑔(𝑥, 𝑦, 𝑧) = 𝑘 (assuming that these extreme values exist and ∇𝑔 ≠ 0 on the surface
𝑔(𝑥, 𝑦, 𝑧) = 𝑘):
4. Find all values of 𝑥, 𝑦, 𝑧, and  such that
f (x , y , z) = g(x , y , z) .
And
g(x , y , z) = k .
(b) Evaluate 𝑓at all the points (𝑥, 𝑦, 𝑧) that result from step (a). The largest of these
values is the maximum value of 𝑓; the smallest is the minimum value of 𝑓.
Example 2.5. [1,p. 149] Use Lagrange multipliers to find the maximum and minimum
values of the function subject to the given constraint 𝑥 2 + 𝑦 2 = 10.
For this problem, 𝑓(𝑥, 𝑦) = 3𝑥 + 𝑦 and 𝑔(𝑥, 𝑦) = 𝑥 2 + 𝑦 2 − 10.
Let’s go through the steps:
• f = 3,1 .
• g = 2x ,2y .
This gives us the following equation:
3,1 =  2x ,2y .
We break up the above equation and consider the following system of 3 equations with 3
unknowns (x , y , ) .
Now, we plug  back into our original equations and get 𝑥 = ∓3 and 𝑦 = ∓1. We get the
following extreme points (3,1); (−3, −1).
We can classify them by simply finding their values when plugging into 𝑓(𝑥, 𝑦).
• f (3,1) = 9 + 1 = 10 .
• f (−3, −1) = −9 − 1 = −10 .
So the maximum happens at (3,1) and the minimum happens at (−3, −1).
2.3. Application of Lagrange Multipliers method in the problem of finding the Minimum
and Maximum value
Problem 3.1 . Given positive real numbers 𝑥, 𝑦 such that 𝑥 + 𝑦 = 3. Find the minimum
value of the expression.
25 1
T = x 2 + 2y + − .
4y 2
Solution:
Application of Lagrange Multipliers:
25 1
We have:𝑓(𝑥, 𝑦) = 𝑥 2 + 2𝑦 + 4𝑦 − 2 and 𝑔(𝑥, 𝑦) = 𝑥 + 𝑦 − 3.
25
• f = 2 x ,2 − .
4y2
• g = 1,1 .
25
 2 x ,2 − =  1,1 .
4y2
1 5 1 5 29
So, we have the critical point (2 , 2). Hence: 𝑓 (2 , 2) = .
4
Solving by the elementary method:
We have:
 1 25 3
T =  x2 − x +  + y + + (x + y) − .
 4 4y 4
2
 1 25 3
T =  x −  + y + + (x + y) − .
 2 4y 4
2
 1
Moreover:  x −   0 for all 𝑥, so:
 2
25 3
T y+ + (x + y) − .
4y 4
25
Application Cauchy’s inequality to two positive numbers 𝑦 and :
4𝑦
25 3 29
T  2 y. +3− = .
4y 4 4
29 1 5
Hence the minimum value of the expression is when 𝑥 = 2 , 𝑦 = 2 .
4
Problem 3.2. [TS 10 Vĩnh Long 2019-2020]
Given positive real numbers x, y such that 𝑥 + 𝑦 = 1. Find the minimum value of the
expression.
1
A = 2x 2 − y 2 + x + + 1 .
x
Solution:
Application of Lagrange Multipliers
1
We have: 𝑓(𝑥, 𝑦) = 2𝑥 2 − 𝑦 2 + 𝑥 + 𝑥 + 1 and 𝑔(𝑥, 𝑦) = 𝑥 + 𝑦 − 1.
1
• f = 4 x + 1 − , −2y .
x2
• g = 1,1 .
1
 4x + 1 − , −2y =  1,1 .
x2
 1
4 x + 1 − x 2 =   1
 x =
 −2y =   2.
x + y = 1 y = 1


1
So, we have the critical point 𝑀( , 1) of A.
2
15 1
Hence the minimum value of the expression is reach at 𝑥 = 𝑦 = 2 .
4
Since 𝑥 + 𝑦 = 1 implies 𝑦 = 1 − 𝑥, plugging into A, we have:
1 1
A = 2x 2 − (1 − x)2 + x + + 1 = x 2 + 3x + .
x x
 1  1 1
A =  x2 − x +  +  4 x +  − .
 2  x 4
2
 1  1 1
A =  x −  +  4x +  − .
 2  x 4
1 1 15
A  0 + 2 4x  − = .
x 4 4
15 1
Hence the minimum value of the expression is reach at 𝑥 = 𝑦 = 2 .
4
7
Problem 3.3. Let 𝑎, 𝑏 > 0 such that 𝑎 + 𝑏 ≤ 2 𝑎𝑏 . Find the minimum value of the
expression:
1 1
Q = 9a + 16b + + .
a b
Solution:
7 1 1 7
We have: 𝑎 + 𝑏 ≤ 2 𝑎𝑏 ⇒ 𝑎 + 𝑏 ≤ 2.
1 1 9 16
We set: 𝑥 = 𝑎 , 𝑦 = 𝑏 and 𝑄 = 𝑥 + 𝑦 + 𝑥 + .
𝑦
9 16 7
We have: 𝑓(𝑥, 𝑦) = 𝑥 + 𝑦 + 𝑥 + and 𝑔(𝑥, 𝑦) = 𝑥 + 𝑦 − 2. .
𝑦
Let’s go through the steps :
9 16
• f = 1 − 2
,1 − 2 .
x y
• g = 1,1 .
9 16
1− 2
,1 − 2 =  1,1 .
x y
We consider the following system of 3 equations with 3 unknowns 𝑥, 𝑦and .
 9
1 − x 2 =   3 4

 16 x = y


x =
3
1 − 2 =     2.
 y x + y = 7 y = 2
 7 
 2
x + y =
 2
3 3 35
So: (2 , 2) is the critical point and 𝑓 (2 , 2) = .
2
35 2 1
Hence the minimum value of the expression is reached at 𝑎 = 3 , 𝑏 = 2.
2
9 16
We have: 𝑄 = (4𝑥 + 𝑥) + (4𝑦 + ) − 3(𝑥 + 𝑦)
𝑦
9 16
Apply Cauchy’s inequality to positive numbers (4𝑥, 𝑥) and (4𝑦, 𝑦 ):
9 16 7 35
Q  2 4 x. + 2 4 y. − 3. = .
x y 2 2
35 2 1
Hence the minimum value of the expression is reached at 𝑎 = ,𝑏 = .
2 3 2
7 28 108
Problem 3.4 Let 𝑎, 𝑏, 𝑐 > 0 such that + + ≤ 75. Find the minimum value of the
𝑎 𝑏 𝑐
expression:
a b 7c
T= + + .
2a + 1 b + 1 2c + 3
Solution:
1 1 1
We set: 𝑥 = 𝑎 , 𝑦 = 𝑏 , 𝑧 = 𝑐 with 𝑥, 𝑦, 𝑧 > 0.
1 1 7
⇒ 7𝑥 + 28𝑦 + 108𝑧 ≤ 75 and 𝑇 = 2+𝑥 + 1+𝑦 + 2+3𝑧.
1 1 7
We get: 𝑓(𝑥, 𝑦, 𝑧) = 2+𝑥 + 1+𝑦 + 2+3𝑧 and 𝑔(𝑥, 𝑦, 𝑧) = 7𝑥 + 28𝑦 + 108𝑧 − 75.
−1 −1 −21
• f = , , .
(2 + x) (1 + y) (2 + 3z)2
2 2
• g = 7,28,108 .
−1 −1 −21
, , =  7,28,108 .
(2 + x) (1 + y) (2 + 3z)2
2 2
We consider the following system of 4 equations with 4 unknowns (x , y , z , ) :

 −1
 (2 + x)2 = 7 
 x = 1
 −1 = 28  x + 2 = 2 + 2y 
   1
 (1 + y)  7 + 7y = 6 + 9 z  y = .
2
 −21 7 x + 28y + 108 z = 75  2

 = 108   1
 (2 + 3z)  z = 2
2
7 x + 28y + 108 z = 75

1 1 1 1
So we get (𝑥, 𝑦, 𝑧) = (1, 2 , 2) is the critical point and 𝑓 (1, 2 , 2) = 3
Hence the minimum value of the expression is 3 reached at 𝑎 = 1, 𝑏 = 𝑐 = 2.
Solution 2:
We have:
1 1 4 1 4 7 1 38
T = (2 + x) + + (1 + y) + + (2 + 3z) + − (7 x + 28y + 108 z) −
9 2+ x 9 1+ y 7 2 + 3z 63 21
Apply Cauchy’s inequality, we get:
2+ x 1 4(1 + y) 1 4(2 + 3z) 7 1 38
T 2 . +2 . +2 . − .75 − .
9 2+ x 9 1+ y 7 2 + 3z 63 21
2 4 75 38
T  + +4− − =3.
3 3 63 21
Hence the minimum value of the expression is 3 reached at 𝑎 = 1, 𝑏 = 𝑐 = 2.
3. CONCLUSION
The article mentions the application of The Lagrange multiplier method in solving
conditional extrema problems. I give some problems close to high school students and
innovative solutions. Moreover, it helps students practice thinking and creativity effectively.
Besides, through the Lagrange method, we can find the elementary solution.
REFERENCE
1. James Stewart (2008), Calculus Early Transcendentals, 6e, McMaster University, 912-914, 923.
2. Vũ Tuấn (2015), Giáo trình Giải tích Toán học tập Hai, Nhà xuất bản Giáo Dục Việt Nam, 160.
3. S. Jamshidi (2013), ”Multivariate Calculus; Fall 2013-Lagrange Multipliers ”, 148-149.
4. Đề thi tuyển sinh vào 10 Vĩnh Long năm 2019-2020.
ỨNG DỤNG PHƯƠNG PHÁP NHÂN TỬ LAGRANGE TRONG BÀI

TOÁN TÌM GIÁ TRỊ LỚN NHẤT VÀ GIÁ TRỊ NHỎ NHẤT
Tóm tắt: Trong bài báo này, chúng tôi nghiên cứu phương pháp nhân tử Lagrange
của cực trị hàm nhiều biến, từ đó chúng tôi áp dụng giải các bài toán tìm giá trị lớn
nhất, giá trị nhỏ nhất của toán phổ thông. Hơn nữa, từ giá trị lớn nhất, giá trị nhỏ
nhất tìm được chúng tôi đua ra lời giải sơ cấp cho bài toán. Ngoài ra chúng tôi cũng
sáng tạo thêm các bài toán mới cho học sinh phổ thông.
Từ khoá: Bài toán GTLN, GTNN, phương pháp nhân tử Lagrange.
SOME APPLICATION MODELS OF FIRST-ORDER

DIFFERENTIAL EQUATIONS IN FACT
Nguyen Thi Thu Huyen

Abstract: Differential equations is a field of Mathematics that has many practical

applications. This paper presents the application of first-order differential equations with five
models: Model describing radioactive atoms, model for Spread of Disease, model of
population growth, Model of bank interest rate, Criminal investigation with specific examples
for readers to easily access.
Keywords: Differential equations, decomposition rate, disease spread, populationc growth.

Email: ntthuyen2@daihocthudo.edu.vn
1. INTRODUCTION
Differential equations is a field of mathematics that has always attracted the strong interest
of mathematicians and applied scientists. The theory of differential equations becomes an
effective tool, especially in describing and analyzing practical problems not only in science and
technology but also in many different fields such as medicine, biology, economics,
environment... The article presents examples to clarify some models of real-world problems for
differential equations, helps readers better understand the role and application of differential
equations in life.
2. CONTENT
2.1. Some basic concepts [1] [2]
2.1.1. Definitions
A first - order differential equation is an equation of the general form:
F ( x, y , y ' ) = 0 (1)
where the function F is defined in the domain D  R3
If in domain D, from equation (1) we can solve y ' :
y ' = f ( x, y )
then we get the first - order differential equation solved to the derivative.
2.1.2. Solution of the differential equation

The function y =  ( x ) is determined and differentiable on the interval I = ( a; b ) is called
a solution of equation (1) if the following two conditions are satisfied:
a) ( x,  ( x ) ,  ' ( x ) )  D  x  I
b) F ( x,  ( x ) ,  ' ( x ) )  0 in the domain I
2.1.3. The Existence and Uniqueness Theorem

Cauchy Problem: The solution of a first order differential equation is infinite. The set of
solutions of a first order differential equation depends on an arbitrary constant c. In practice,
people are often interested in solutions of first - order differential equations that satisfy certain
conditions.
We consider the following problem for equation (2), which is called the Cauchy problem (or
initial value problem):
 y ' = f ( x, y )
Find y ( x ) satisfied:  (2) where ( x0 , y0 )  D is called the initial condition.
 y ( x0 ) = y0
D  R2 . Function f satisfies
Lipschitz's condition: Let the function f ( x, y ) is defined on a set
a Lipschitz condition in the variable y on a set D if a constant L  0 exists with:
f ( x, y1 ) − f ( x, y2 )  L y1 − y2
whenever ( x, y1 ) , ( x, y2 )  D . The constant L is called a Lipschitz constant for f .

The Existence and Uniqueness Theorem: Suppose the function f ( x, y ) in (2) is continuous
and satisfies the Lipschitz condition for the variable y on the rectangle:
D = ( x, y )  R 2 / x − x0  a, y − y0  b
Then the solution of Cauchy's problem exists and is unique in the interval I =  x0 − h, x0 + h 
with h = min  a,  and M = max f ( x, y ) .

b
 M ( x , y )D
2.1.4. Separable differential equations

Separable differential equations is a differential equation that has one of the following forms:
dy
= f ( x) g ( y)
dy f ( x) dy g ( y )
= , = or
dx g ( y ) dx f ( x) dx
where f ( x), g ( y) are continuous functions on ( a, b )  R .
Solution: We put the differential equation in the form: f ( x ) dx + g ( y ) dy = 0
Integrate both sides, then the general solution of the equation is:
 f ( x ) dx +  g ( y ) dy = C
2.2. Application models

2.2.1. Model describing radioactive atoms
Let R0 be the mass of radium initially at t = 0 and R(t) be its mass at any time t.
dR
Then the decay rate is . This speed is a negative quantity because R decreases over time
dt
According to the problem condition, we have:
dR
= −kR, ( R  0 ) where k is the scaling factor ( k  0 )
dt
Integrating the above equation, we have:
ln R = −kt + ln C1 hay R ( t ) = C.e− kt ( C = C1 )
I see that R = R0 when the t =0

Substituting these values into the expression of the general solution we find C = R0
Therefore, R ( t ) = R0 e− kt
Example 1: It is known that the decay rate of radium is directly proportional to its quantity at
each given instant. Find the decay law of radium, given its initial mass and the time T needed
to decay half of the original radium mass. The half – life of radium is 1600 years. What
percentage of a given amount of the radium will remain after 100 years?
R0
According to the problem condition R = when t = T . We have 𝑘:
2
1
ln
ln 2
k= 2=
−T T
ln 2
− t
So, the decay law of radium is expressed by the formula R = R0 e T
Therefore: R(100) = R0e−0,043t
R(100) = e−0,043 = 0.958 = 95.8%

So after 100 years, 4.2 percent of radium will be remain.
2.2.2. Model for Spread of Disease
The rate of spread is proportional to the number of people infected and the number of
people who are not infected.
Let N ( t ) be the number of infected people at time t, P be the total number of people
(constant); k is the growth constant. Because the rate of spread is proportional to the number of
infected and uninfected people, we have the equation
dN (t )
= kPN (t ) − kN 2 (t )
dt
Solving the differential equation, we have

dN (t ) dN (t )
= kPN (t ) − kN 2 (t )  = kdt
dt N (t ).( P − N (t ))
Integrating the above equation, we have:
1  1 1  1 N (t )
  +  dN (t ) =  kdt  ln
P  N (t ) P − N (t )  P P − N (t )
= kT + C
N (t ) N (t )
 ln = Pkt + C  = e Pkt .C
P − N (t ) P − N (t )
P
So N (t ) =
1 + C.e− Pkt
Example 2: There is a model of the spread of the Covid - 19, in which the rate of spread
is proportional to the number of infected and uninfected people. In a remote town of 5000
residents, the number of people with covid at the beginning of the week was 160 and this
number had increased to 1200 by the end of the week. How long does it take for 80% of the
town's residents to be infected with Covid -19?.
With P = 5000 and N (0) = 160 , we have 160 = 5000 0  C = 121

1 + C.e 4
Because after 7 days the infected population increases to 1200, we have:
5000 1  363 
1200 = − k .5000.7
k= ln  
1 + C.e 35000  38 
Because 80% of the inhabitants of the town are infected, the equation is as follows
5000
80%.5000 =  t = 14.875
1 + C.e− k .5000.t
Thus, after about 15 days, 80% of the town's residents were infected with Covid -19.
2.2.3. Model of population growth [3]
The logistic equation was introduced (around 1840) by the Belgian mathematician and
demographer P.F. Verhulst as a possible model for human population growth.
Where P(t) is the number of people at time t , K is the capacity of the medium, r is a
constant.
(Note: K is called the carrying capacity if it is the maximum population that the
environment can sustain over a long period of time)
Using the description by the logistic function, we have a mathematical model that predicts
population growth as follows:
dP  P (t )   P(t ) 
= rP (t ) 1 −  or P '(t ) = rP (t ) 1 − 
dt  K   K 
Solving this differential equation by the method of dissociation of variables with 𝑃0 being
the number of people at time t = 0 (or the number of people at the beginning of time t), we get:
dP  P (t )  dP K dP
K  
=  dt  
r P(t ) ( K − P(t ) ) 
= rP(t ) 1 −  = dt
dt   P(t ) 
rP (t ) 1 − 
 K 
P(t ) K
 ln = rt + C or P(t ) =
P(t ) − K 1 + C.e− rt
Example 3: In 1990, the world population was about 5.28 billion people and in 2000, the
world population reached 6.07 billion people. Assume that the relative growth rate decreases
as the population increases and starts to become negative when the population size exceeds its
capacity to accommodate K, which is the maximum population the environment can tolerate in
the long run (also called the logistic model). Assuming the capacity is 100 billion people, use a
logistic model to predict the world population in 2025?
Because the capacity is 100 billion people, K = 100 billion
In 1990, the world population was 5.28 billion, so for t = 0 then P0 = 5, 28 tỉ
K
Applying the results of the general problem P(t ) = with t = 0, P(t ) = P0 = 5, 28
1 + C.e− rt
100
billion and K = 100 billion, we have 5, 28 =  C  18
1 + C.e0
In 2000, the world population was 6,07 billion, so for t = 10 then P1 = 6, 27 tỉ
With t = 10, P(t ) = P1 = 6, 27 billion, C = 18 and K = 100 billion, we have:
K 100
P(t ) = − rt
 6,07 =  r  0,015
1 + C.e 1 + 18.e−10 r
In the year 2025, we have t = 35, , C = 18 and K = 100 billion, we have:
K 100
P(t ) = − rt
 P ( 35) =  8,551 billion people
1 + C.e 1 + 18.e−0,015.35
So, in 2025, the world population is estimated at 8.551 billion people.
2.2.4. Model of bank interest rate
Problem 1: Suppose you initially have an amount S 0 of millions. You make a plan to
deposit that money in the bank at a r% interest rate for one year. What is the amount received
after t years?
The symbol S(t) is the amount of money you have after a period of t years. We have:
S ( t + t ) = S ( t ) + r %tS ( t ) where r tS ( t ) is the amount of interest generated after a period
of t , k t is the amount you pay in addition.
S ( t + t ) − S ( t )
Then = r %.S ( t )
t
Let t → 0 we get S ' ( t ) = r %.S ( t )
dS ( t ) dS ( t )
 = r %.S ( t )  = r %dt
dt S (t )
Integrating both sides we get ln S ( t ) = r %t + C  S ( t ) = er %t +C = C.er %t
Solving the differential equation with the initial condition S ( 0 ) = S0 , we have
S ( t ) = S0 e r %t
Example 4.1: Nam deposits an amount of VND 8 million in a bank with an interest rate of
0.9% for 1 month. After exactly 5 years, how much is the amount in the book knowing that
during that time Nam did not withdraw a single penny.
With an interest rate of 0.9% for 1 month, we will have 10.8% for one year.
According to the formulation of problem 1, we have
S ( 5 ) = 8.106.e10,8%.5 = 13728054,9 million
Problem 2: Suppose you plan to save money for the future.
Every year you deposit k million and you deposit it in the bank with compound interest
of r % for one year (unchanged). We will construct the initial value versus balance over time
problem as follows:
The symbol S(t) is the amount of money you have after a period of t years. Then
S ( t + t ) = S ( t ) + r %tS ( t ) + k t where r %tS ( t ) is the amount of interest generated after
the time t , k t is the amount you pay in addition.
S ( t + t ) − S ( t )
Then = r %S ( t ) + k
t
Let t → 0 we get S ' ( t ) = r %.S ( t ) + k
dS ( t ) dS ( t )
 = r %.S ( t ) + k  = dt
dt r %.S ( t ) + k
Integrating both sides, we get:
1
r%
ln r %S ( t ) + k = t + C  r %S ( t ) + k = e r %t +C  S ( t ) =
1
r%
( Ce r %t − k )
Solving the differential equation with the initial condition S ( 0 ) = 0 , we have

k r %t k
S (t ) = e −
r% r%
Example 4.2: Suppose at the age of 30, you plan to save money for your children, every
year you deposit k million and you deposit it in a bank with an interest rate of 6% for one year
(unchanged), you retire at the age of 65. For you to have at least a billion in savings for your
children when you retire, how much is k equal to?
According to the formula in problem 2, we have:
k 6%.35 k
S ( t )  109  e −  109  k  8372673.371
6% 6%
So, every year, you need to deposit at least 8372673.371
2.2.5. Criminal investigation [3]
Newton's law of cooling: The rate of heat loss of a body is directly proportional to the
difference in the temperatures between the body and its environment.
It is assumed that the body temperature T ( t ) obeys the law of cooling with the expression
dT ( t )
= k (T ( t ) − M ) where T is the body temperature at time t (hours), M is the constant
dt
temperature of the environment, and k is a constant that depends on the material properties of
the object.
Solving the differential equation, we have:
dT ( t )
= kdt  ln T ( t ) − M = kt + C  T ( t ) = Ce kt + M
T (t ) − M
Scope of application of the law: This law has many applications ranging from determining
the time it takes for a cup of coffee to cool to a drinkable temperature, to its use in forensic
science, to determine how long a corpse has been dead.
Example 5: At one house, a victim was killed. Police were called at 10:56 am. The police
0
took the victim's body temperature at that time was 31 . An hour later, the police took the
0
victim's temperature again and the victim's body temperature at that time was 30 . The
0
temperature in the house was 20 . What time was the victim killed?
We have:
T ( 0 ) = 31 31 = C + 22 C = 9
    .
T (1) = 30 30 = C.e + 22 k = −0,11778
k
So T (t ) = 9.e kt + 22
0
When the victim is killed, the victim's body temperature is 37 , We have the time the
ln15 − ln 9
victim is killed by the time the police arrive: t = = −4,337 ( hours) = 4 hours 20
k
minutes. So the victim was killed at 6:36
3. CONCLUSION
Thus, in this article, I have presented a number of applications of differential equations in
the fields of economics, demography, medicine, etc. From there, it helps readers see the
relationship of mathematics in general, differential equations in particular to other sciences. In
each model, the article has shown the relationship between the factors in the problem to lead to
differential equations, solving differential equations along with specific problems so that
readers have a better approach.
REFERENCES
1. Pham The Hoan, Pham Phu (2014), The basis of differential equations and stability theory,
Education Publishing.
2. Jeffrey R.Chasnov (2019), Differential Equations for Engineer, pp.23-24, University of Miami.
3. Mersha Amdie Endale (2015), Some Applications of first order differential equations to real world
system, pp.13-32, MSc Graduate Seminar - Haramaya University.
4. Verhulst, Piere-Francois (1838), Correspondance mathématique et Physique, pp.113-120, A.
Quetelet.
5. James Stewart (2002), Calculus concepts and Contexts, edition, pp.539, Brooks-Cole.
MỘT SỐ MÔ HÌNH ỨNG DỤNG CỦA PHƯƠNG TRÌNH VI PHÂN

CẤP MỘT TRONG THỰC TẾ
Tóm tắt: Phương trình vi phân là một lĩnh vực Toán học có rất nhiều ứng dụng trong thực tế.
Trong bài báo này tôi trình bày ứng dụng của phương trình vi phân cấp một với năm mô hình:
Mô hình mô tả lượng nguyên tử phóng xạ, mô hình lây lan dịch bệnh, mô hình tốc độ tăng
trưởng dân số, mô hình lãi suất ngân hàng, mô hình sử dụng trong điều tra tội phạm với các
ví dụ cụ thể để người đọc có thể tiếp cận một cách dễ dàng.
Từ khoá: Tốc độ phân rã, lây lan dịch bệnh, tăng trưởng dân số, lãi suất, điều tra.
A DEEP LEARNING MODEL THAT DETECTS THE DOMAIN

GENERATED BY THE ALGORITHM IN THE BOTNET
Nguyen Trung Hieu(*), Cao Chinh Nghia

Faculty of Mathematics - Informatics and application of science and technology in crime
prevention, The People's Police Academy
Abstract: Domain Generation Algorithm (DGA) is the group of algorithms that generate
domain names for attack activities in botnets. In this paper, we present a Bi-LSTM deep
learning model based on Attention mechanism to detect DGA-generated domains. Through
the experimental process, our model has given good results in detecting DGA-generated
domains belong to the Post and Monerodownloader family. In general, the F1 measure of the
model in the multi-class classification problem reaches 90%. The micro average (macro avg)
efficiency is 86% and the average (weighted avg) efficiency is 91%.
Keywords: Bi-LSTM deep learning network; deep learning; malicious URL detection;
Attention mechanism in deep learning.

(*) Email: hieunt.dcn@gmail.com
1. INTRODUCE
Botnet Attacks
The development of Internet has brought many benefits to users, but it is an environment
for cybercriminals to operate also.
Botnet Attack is one of the common attacks. Each member of the botnet is called a bot. A
bot is a malicious software created by attackers that control infected computers remotely
through a command and control server (C&C server). The bot has a high degree of autonomy
and is equipped with the ability to use communication channels to receive commands and
update malicious code from the control system. Botnets are commonly used to transmit
malware, send spam, steal sensitive information, phishing, or create large-scale cyberattacks
such as distributed denial of service (DdoS) attacks [1].
The distribution’s widespread of bots and the connection between bots and control servers
often requires the Internet. The bots need to know the IP address of the control server to access
and receive commands. In order to avoid detection, command and control servers do not register
static domain names, instead of continuously change addresses and different domains at
different intervals. Attackers use Domain Generation Algorithm (DGA) to generate different
domain names for attacks [2] aimed at masking these control and control servers.
Identifying the attack of a malicious domain can effectively determine the purpose of the
attack, the tools and malware used, and take preventive measures to greatly reduce the damage
caused by the attack induced attack.
Domain Generation Algorithm
The Domain Generation Algorithm (DGA) can use operators in combination with ever-
changing variables to generate random domain names. The variables can be day, month, year
values, hours, minutes, seconds or other keywords. These pseudo-random strings are
concatenated with the Top-level domain (.com, .vn, .net...) to generate the domain names. The
algorithm of the Chinad malware written in Python [3] shows the input seed includes letters
from a-z and numbers from 0-9 and combines the values of days, months, five. The results are
combined with the TLDs ('.com', '.org', '.net', '.biz', '.info', '.ru', '.cn') to form the complete
domain name.
Table 1. Some DGA samples
Conflicker gfedo.info Bigvikto support.showremote-conclusion.fans

ydqtkptuwsa.org r turntruebreakfast.futbol
bnnkqwzmy.biz speakoriginalworld.one
Cryptolock nvjwoofansjbh.ru Bamital cd8f66549913a78c5a8004c82bcf6b01.i

er qgrkvevybtvckik.org nfo
eqmbcmgemghxbcj.co. a024603b0defd57ebfef34befde16370.o
uk rg
5e6efdd674c134ddb2a7a2e3c603cc14.
org
Chinad qowhi81jvoid4j0m.biz
29cqdf6obnq462yv.co
m
5qip6brukxyf9lhk.ru
A DGA can generate a large number of domains in a short time, and bots can select a small
portion of them to connect to the C&C server. Table 1 shows some examples of domain names
initialized with DGA [4]. The
Chinad malware can generate 1000
domain names per day with the
letters a-z and numbers 0-9.
Bigviktor combines 3 to 4 different
words from 4 predefined lists
(dictionaries) that can generate 1000
domains per month.
Figure 1 depicts the connection
process between the C&C server and
the DGA domains [5]. The attacker
uses the same DGA and initial
Figure 1. DGA-based botnet communication mechanism kernels for the C&C server and the
bot to generate the same domain
dataset. The attacker needs to select a domain name only from the generated list and register it
for the C&C server 1 hour before performing the attack. The bots on the victim's machine will
in turn send the domain name resolution requests in the generated list of domains to the Domain
Name System (DNS). The DNS system will return the IP address of the corresponding C&C
server, then the bots begin to communicate with the server to receive the command. If the C&C
server is not found in the previous domain, the bots will query the next set of domains generated
by the DGA until an active domain name is found [6].
2. MAIN CONTRIBUTION OF THE ARTICLE
The main of contributions of the paper include:
1 - Introduce a deep learning approach using Bidirectional-Long Short Term Memory (Bi-
LSTM) model based on Attention mechanism in detecting domains created by DGA. Our model
has worked well in the problem of detecting malicious URLs [7].
2 - Presenting experimental results shows a significant improvement compared to previous
techniques with the use of open data sets.
The remainder of paper is organized as follows: Section 2 presents related studies. Our
deep learning network architecture and solution is presented in Section 3. Section 4 presents
our experimental process, including the steps to select the database and the results obtained.
Finally, Section 5 is the conclusion, comments on the results achieved as well as the future
direction of the paper.
2.1. Related studies

In recent years, much research on Botnet detection has been published. Nguyen Van Can
and colleagues [8] proposed a model to classify benign domains and DGA domains based on
Neutrosophic Sets. Testing on 3 data sets of Alexa, Bambenek Consulting [9] and 360lab [4]
shows that the model has an accuracy result of 81.09%.
R. Vinayakumar et al [10]. have proposed a DGA detection method based on analyzing the
statistical features of DNS queries. Feature vectors are extracted from domain names by text
representation method, optimal features are calculated from numerical vectors using deep
learning architecture in Table 2. The results show that the model has high accuracy with 97.8%.
Table 2. DBD deep architecture [10]
Layers Output Shape
Embedding (None, 91, 128)
Conv1D (None, 8, 764)
MaxPooling1 (None, 2, 164)
LSTM (None, 70)
Dense (None, 1)
Activation (None, 1)
Yanchen Qiao et al [2]. have proposed a method to detect DGA domain names based on
LSTM using Attention mechanism. Their model is executed on the data set from Bambenek
Consulting [9], with an accuracy of 95.14%, overall precision of 95.05%, recall of 95.14% and
F1 score of 95.48%.
Duc Tran [11] built an LSTM.MI model that combines binary classifier and multiclass
classifier with an unbalanced dataset. In which, the original LSTM model is applied a cost-
sensitive adaptation mechanism. Cost items are included in the back-to-back learning process
to account for the importance of delineation between classes. They demonstrate that LSTM.MI
provides at least 7% improvement in accuracy and macro mean recall over the original LSTM
and other modern cost-sensitive methods. It can also maintain high accuracy on non-DGA
generated labels (0.9849 F1 points).
2.2. Proposed model
Our proposed model includes: input layer, embedded layer, two Bi-LSTM layers, one
attention layer and output layer. The architecture of the model is shown in Figure 2 [7].
The detection module will take as input a data set of

𝑇 domain addresses with a structure of the form
{(𝑢1 , 𝑦1 ), … . , (𝑢𝑇 , 𝑦𝑇 )} . Where, xt is a pair
(𝑢𝑡 , 𝑦𝑡 )where u t (with t = 1, …, 𝑇) is a domain in the
training list and 𝑦𝑡 an associated label.
Each domain, in its raw form, before being trained,
is processed in two steps to form the input vector:
- Step 1: Cut off the TLD part of the domain name
then tokenize the raw data – convert the string of
characters in the rest to encrypted data in the form of an
integer using Keras's Tokenizer library;
- Step 2: Normalize the size of the encrypted data in
step 1 to the same length. This way we can convert the
original domain string into the input vector V = {v1 , v2 ,
Figure 2. Bi-LSTM network architecture. v3 , …vT } . Each vector has a fixed length. Any missing
vector, add the value 0 to give the length enough.
Next, we use a bidirectional LSTM network (Bi-LSTM) to model URL sequences based
on a word vector representation. In Bi-LSTM architecture, there are two layers of nodes hidden
from two separate LSTMs, two LSTMs capturing distant dependencies in two different
directions. Since the output vector of the embedded layer is V = {v1, v2, v3, …vT }, the forward
LSTM will read the input from v1 to vT and the backward LSTM will read the input from v T
⃗⃗⃗𝑖 𝑣à ℎ
to v 1 . Meanwhile a pair of hidden states ℎ ⃖⃗⃗⃗𝑖 is initialized. We can get the output of the Bi-
LSTM layer by combining the two hidden states according to the formula:
𝑻
⃗⃗⃗𝒊 , 𝒉
𝒉𝒊 = [𝒉 ⃖⃗⃗⃗𝒊 ] (1)
It uses two layers of Bi-LSTM and the experimental data set is quite large. Therefore, Batch
Normalization layer will be used to normalize the data in batch layers to a normal distribution
to stabilize the learning process and greatly reduce the number of epochs needed to train the
network, thereby increasing the speed of training. training.
As described in this paper, the hidden states at all locations are considered with different
Attention weights. We apply Attention mechanism to capture the relationship between ⃗⃗⃗ℎ𝑖 𝑣à ⃖⃗⃗⃗
ℎ𝑖 .
This information is aggregated with respect to the feature from the output of the second Bi-
LSTM network. This helps the model to focus only on the important features instead of the
confounding or less valuable information.
Initially, the weights 𝑢𝑡 are calculated based on the correlation between the input and
output according to the following formula:
𝒖𝒕 = 𝒕𝒂𝒏𝒉(𝑾𝒉𝒕 + 𝒃) (2)
These weights will be renormalized to the Attention weight vector 𝛼𝑡 using the softmax function:
𝐞𝐱𝐩 (𝒖𝑻𝒕 𝒖)
𝜶𝒕 = (3)
∑𝒕 𝐞𝐱𝐩 (𝒖𝑻𝒕 𝒖)
Then the vector 𝑐𝑡 is calculated based on the Attention weight vector and the hidden states
ℎ1 … ℎ𝑇 as follows:
𝒄𝒕 = ∑ 𝜶𝒕 𝒉𝒕 (4)
𝒕
value 𝑐𝑡 , the more important the feature 𝑥𝑡 plays in detecting the DGA domain .
Finally, to predict a domain, the calculation results are passed through a Dense layer with
1 hidden neuron using the activation function sigmoid to receive a return value between 0 and
1. The resulting y will be helps determine if a domain is benign or DGA. Thus, the input domain
name will be normalized into a vector form, this vector will be passed through the Embedding,
Bi-LSTM, Batch Normalization, Bi-LSTM, Attention layers before giving the output result. In
addition, the model uses adam optimization algorithm with default parameters in keras. And to
prevent the model from falling into overfitting state (overfitting) compared with the real model
of the data, we use more Dropout technique for Bi-LSTM layers. The mechanism of Dropout
is that in the process of training the model, with each time we update the weights, we randomly
remove the number of nodes in the layer so that the model cannot depend on any node of the
previous layer, but instead which tends to spread evenly.
2.3. Experiment
In this paper, we conduct 2 experiments:
1- Experimentally check the accuracy of the model in 2-class classification: Domains
generated by DGA algorithm and normal domains
2- Experiment to check the accuracy of the model in multi-class classification: Detect
different DGA algorithms in a given data set.
DGA Regular domain name
Count 30000 30000
Mean 14.245103 9.623797
Std 4.337851 3.300294
Min 6 6
25% 12 7
50% 13 9
75% 16 11
Max 25 25
2.4. Evaluation Dataset

In this paper, we use a dataset consisting of DGA domains collected from Bambenek
Consulting [9] and normal domains obtained from Alexa. With two different tests, we use two
different data sets.
Table 2. Summary of the collect dataset
Domain Type Sample Domain Type Sample Domain Type Sample
tinba 64313 Chinad 1484 unknownjs 172
ramnit 62227 P2P 985 beautiful baby 161
necurs 30789 Volatile 966 pandabanker 94
murofet 24562 proslikefan 750 cryptowall 91
Post 21881 Sphinx 733 an 67
qakbot 19258 Pitou 749 Unknowndroppr 59
shiotob 14451 Dircrypt 699 sisron 59
monerodownloader 14422 Fobber 572 kingminer 29
ranbyus 13417 padcrypt 551 gozi 24
kraken 5529 Zloader 555 dromedan 2
Cryptolocker 5780 Geodo 557 madmax 2
locky 3869 MyDoom 333 g01 1
vawtrak 3022 Beebone 291 mirai 1
qadars 2309 tempedreve 242
ramdo 1932 Vidro 188
Dataset for test 1: Consists of 30000 DGA domains with label 1 and 30000 normal domains
with label 0. This dataset is randomly shuffled, then divided into two small sets as training
dataset and test dataset. In which there are 46 different types of DGA domain names with the
number given in Table 2. The distribution parameters of character length of each type of domain
name given in Table 3. In which, the sample with the smallest length is 6, the maximum length
is 25 and the average length of the DGA domain name is 14.2, the normal domain name is 9.6.
Table 4. Label assessment to types
Domain type Labels Domain type Labels

Post 0 Qakbot 6
Kraken 1 Ramnit 7
Legit 2 Ranbyus 8
Monerodownloader 3 Shiotob/urlzone/bebloh 9
Murofet 4 Tinba 10
Necurs 5
Dataset for test 2: With the goal of testing the multi-class classification, the types of DGA
domains used include the families Post, Kraken, Monerodownloader, Murofet, Necurs,
Shiotob/urlzone/bebloh, Qakbot, Ramnit, Ranbyus, Tinba are labeled according to Table 4. The
number of domain names for test 2 includes 25,000 normal domain names and 25,000 domain
names belonging to DGA families.
3. PERFORMANCE METRIC
The performance of the algorithms is evaluated using the confusion matrix. In there:
• True negatives (TN) – are benign sites that are predicted to be benign.
• True Positives (TPs) – are malicious sites that are expected to be malicious.
• False negatives (FN) – are malicious sites that are expected to be benign.
• False positives (FPs) – are benign but expected to be malicious sites.
From there we have the measures:
Accuracy:
(𝑻𝑷+𝑻𝑵)
ACC =(𝑻𝑷+𝑻𝑵+𝑭𝑷+𝑭𝑵) (5)
The article also uses the measures of F-measure, precision, and recall, which are shown in
the following formulas:
𝑻𝑷
𝑷𝒓𝒆𝒄𝒊𝒔𝒊𝒐𝒏 = (6)
𝑻𝑷 + 𝑭𝑷
𝑻𝑷
𝑹𝒆𝒄𝒂𝒍𝒍 = (7)
𝑻𝑷 + 𝑭𝑵
𝟐 ∗ 𝑹𝒆𝒄𝒂𝒍𝒍 ∗ 𝑷𝒓𝒆𝒄𝒊𝒔𝒊𝒐𝒏
𝑭𝟏 = (8)
𝑹𝒆𝒄𝒂𝒍𝒍 + 𝑷𝒓𝒆𝒄𝒊𝒔𝒊𝒐𝒏
A high Precision value means that the accuracy of the points found is high. A high recall
means a high TP rate, which means that the rate of missing really positive points is low. The
higher the F1, the better the classifier. In addition, Table 5. Parameter of model
In experience no.1
we also use the loss function binary cross entropy
Layers Output Shape
(BCE) to calculate the difference between two embedding (None, 38, 128)
quantities: 𝑦̂- the label of the predicted URL and bidirectional (None, 38, 128)
y - the correct label of each URL. Loss function is batch_normalization (None, 38, 128)
like a way to force the model to pay a penalty for bidirectional_1 (None, 38, 128)
attention_with_context (None, 38, 128)
each wrong prediction, and the number of
addition (None, 128)
penalties is proportional to the severity of the dense (None, 1)
error. The smaller the loss value, the better the
model shows that the prediction results are good, on the contrary, if the prediction results differ
too much from reality, the larger the loss value.
̂) + (𝟏 − 𝒚) 𝒍𝒐𝒈(𝟏 − 𝒚
𝑩𝑪𝑬 = −(𝒚𝒍𝒐𝒈(𝒚 ̂)) (9)
3.1. Experimental results
Table 6. Experimental results no. 1 The model is built on the basic configuration of
Loss 3.2705e-04 the Kaggle platform with Keras kernel, Tensorflow
ACC 0.9999 backend. Which uses ModelCheckPoint to save the
Precision 0.9999
training process and EarlyStopping to immediately
Recall 0.9998
F1 0.9999 stop the training process when the best value is found.
3.1.1. Experiment number 1
The parameters of the model in the first experiment are showed in Table 5.
Table 7. Parameter of the model in experience no. 2
Layers Output Shape
embedding (None, 33, 128)
bidirectional (None, 33, 128)
batch_normalization (None, 33, 128)
bidirectional_1 (None, 33, 128)
attention_with_context (None, 33, 128)
addition (None, 128)
dense (None, 11)
With the binary classification problem between the DGA domain and the normal domain,
the model gives the results in Table 6 with an accuracy of up to 99%. With this result, we
assume that there is a difference coming from the distribution of the domain's length. We will
run other tests to further test the stability of the model.
3.2. Experiment number 2
Table 8. Results of experiment 2
Class Precision Recall F1 Support
0 1.00 0.99 1.00 3986
1 0.78 0.76 0.77 1114
2 0.98 1.00 0.99 50086
3 1.00 1.00 1.00 2665
4 0.85 0.59 0.69 4383
5 0.81 0.85 0.83 5746
6 0.52 0.82 0.64 3572
7 0.76 0.95 0.84 11535
4 0.85 0.59 0.69 4383
5 0.81 0.85 0.83 5746

6 0.52 0.82 0.64 3572
7 0.76 0.95 0.84 11535
8 0.89 0.88 0.89 2432
9 0.97 0.9 0.94 2602
4 0.85 0.59 0.69 4383
ten 0.91 0.59 0.71 11879
accuracy 0.90 100000
macro avg 0.86 0.85 0.85 100000
weighted avg 0.91 0.90 0.90 100000
In this experiment, we test the multi-class detection ability of the model with three
measures of precision, recall and f1. The parameters used in the model are presented in Table
7. Due to multi-class classification, as an output, we use a hidden layer of size 11 corresponding
to 11 labels to be classified. The experimental results are presented in Table 8. With the normal
domain (labeled as 2), the Precision is 98%, F1 is 99%. Our model gives the best results when
classifying labels for DGA domains belonging to the Post family (label number 0) and
Monerodownloader (label number 3). In contrast, the model gave the worst results on the
Qakbot family (label 6) when the rate of classifying a benign site as a malicious site with a
Precision measure of 52%. For the Murofet family (label number 4) and Tinba (label number
10), the model gives false classification results when evaluating the DGA domain name into a
benign domain with a recall measure of 59%. In general, the F1 measure of the model in the
multi-class classification problem reaches 90%. The micro average (macro avg) efficiency is
86% and the average (weighted avg) efficiency is 91%.
4. COMPARISON WITH OTHER DGAS DETECTION METHODS
The evaluation was performed on a dataset from the same source [9] as the studies being
compared. The results compared with the study of Chanwoong Hwang and colleagues are
shown in Table 9 showing that our model has a higher detection capacity.
Table 10 compares the ability to detect DGA domains

Table 9. Comparison
Chanwoong Proposed labeled 4,5,6,7,8,9,10 in the study of Yanchen Qiao and
Hwang model
Duc Tran. Yanchen Qiao [2] using LSTM with Attention
Accuracy 88.77% 90%
mechanism. Duc Tran's model [11] is a cost-sensitive
Precision 89.01% 91%
original LSTM . Cost items are class dependent, taking into
Recall 88.77% 90%
account the importance of classification between classes.
F1-score 88.695% 90%
Our model exhibits good detectability across four DGA
families: Necurs, Qakbot, Ramnit, Ranbyus and lesser on the Shiotob family, tinba.
Table 10. Results comparring with Yanchen Qiao and Duc Tran
Yanchen Qiao Duc Tran Our Model
La Fa
TABLE III. F TABLE VI. F TABLE IX. F
bel mil TABLE I. P TABLE II. R TABLE IV. P TABLE V. R TABLE VII. P
TABLE VIII. R
1- 1- 1-
s y recision ecall recision ecall recision ecall
score Score Score
Mu 0.7641 0.7207 0.7418 0.5330 0.7423 0.6205 0.85 0.59 0.69
4 rofe
t
Nec 0.6651 0.1722 0.2735 0.5248 0.1104 0.1824 0.81 0.85 0.83
5
urs
Qak 0.7862 0.5013 0.6122 0.7716 0.4350 0.5564 0.52 0.82 0.64
6
bot
Ra 0.4688 0.7525 0.5777 0.6068 0.8062 0.6925 0.76 0.95 0.84
7 mni
t
Ran 0.4672 0.8455 0.6018 0.3617 0.7073 0.4787 0.89 0.88 0.89
8 byu
s
Shi 0.9751 0.9251 0.9494 0.9741 0.9004 0.9358 0.97 0.90 0.94
9 oto
b
Tin 0.9259 0.9920 0.9578 0.8951 0.9961 0.9429 0.91 0.59 0.71
10
ba
5. CONCLUSION
In this paper, we have presented an approach using Bi-LSTM deep learning network based
on Attention mechanism. [7] to solve the problem of detecting domains generated by algorithms
in Botnet, too. The model further shows a strong ability to detect DGA domains. The model
with 2 layers of Bi-LSTM combined with Attention gives results when detecting DGA domains
with 90 % accuracy. In the future, we will continue to improve the model, and at the same time
evaluate the model on larger, more complex datasets to verify the accuracy of the proposed
model. The research results in this direction can be integrated into the DNS domain name
filtering systems to automatically discover the domains of the Botnet network.
REFERENCES
1. Soleymani and F. Arabgol (2021), "A Novel Approach for Detecting DGA-Based Botnets in DNS
Queries Using Machine Learning Techniques," Journal of Computer Networks and Communications.
2. Y. Qiao, B. Zhang, W. Zhang, A. K. Sangaiah and H. Wu (2019), "DGA Domain Name Classification
Method Based on Long Short-Term Memory with Attention Mechanism," Applied Sciences, vol. 9,
no. 20.
A. Qi, J. Jiang, Z. Shi, R. Mao and Q. Wang (2018), "BotCensor: Detecting DGA-Based Botnet Using
Two-Stage Anomaly Detection.," in 2018 17th IEEE International Conference On Trust, Security
And Privacy In Computing And Communications/ 12th IEEE International Conference On Big Data
Science And Engineering (TrustCom/BigDataSE).
A. K. Sood and S. Zeadally (2016), "A Taxonomy of Domain-Generation Algorithms," IEEE Security
& Privacy, vol. 14, no. 4, pp. 46-53, 05 August 2016.
3. N. T. Hiếu and T. N. Ngọc (2020), "Phát hiện URL độc hại sử dụng mạng học sâu Bi-LSTM dựa
trên cơ chế Attention," in Hội thảo quốc gia lần thứ XXIII: Một số vấn đề chọn lọc của Công nghệ
thông tin và truyền thông, Quảng Ninh.
B. N. T. T. A. T. H. V. L. L. H. S. N. T. K. S. Nguyen Van Can (2020), "A new method to classify
malicious domain name using Neutrosophic sets in DGA botnet detection," Journal of Intelligent &
Fuzzy Systems, vol. 38, p. 4223–4236.
4. S. K. P. P. A. M. J. A. Vinayakumar R (2019)., "DBD: Deep Learning DGA-Based Botnet
Detection," in Advanced Sciences and Technologies for Security Applications Deep Learning
Applications for Cyber Security, T. M. Alazab M., Ed., Switzerland, Springer, Cham, pp. 127-149.
5. H. M. V. T. H. A. T. L. G. N. Duc Chan (2018), "A LSTM based framework for handling multiclass
imbalance in DGA botnet detection," in Neurocomputing, 2018, p. 2401–2413.
MỘT MÔ HÌNH HỌC SÂU PHÁT HIỆN TÊN MIỀN ĐƯỢC TẠO
BỞI THUẬT TOÁN TRONG MẠNG BOTNET
Tóm tắt: Thuật toán khởi tạo tên miền (DGA) là một nhóm các thuật toán tạo ra các tên miền
phục vụ cho các hoạt động tấn công trong mạng botnet. Trong bài báo này, chúng tôi trình
bày một mô hình học sâu bi-lstm dựa trên cơ chế attention để phát hiện các tên miền dga. Qua
quá trình thực nghiệm, thuật toán cho kết quả tốt trong việc phát hiện các tên miền dga thuộc
họ post và monerodownloader. Về tổng thể, độ đo f1 của mô hình trong bài toán phân loại đa
lớp đạt 90%. Hiệu suất trung bình vi mô (macro avg) đạt 86% và hiệu suất trung bình
(weighted avg) đạt 91%.
Từ khóa: Mạng học sâu Bi-LSTM, miên DGA, phát hiện URL độc hại, cơ chế Attention trong
học sâu.
RUIN PROBABILITIES FOR A GENERAL INSURANCE RISK MODEL
Nguyen Thi Thuy Hong

Abstract: In this paper, we proved the formula calculating the ruin probability for a general
risk model. We generalized the Picard-Lefvre formula (see [6]) for ruin probability of general
risk model as well as the results of Stephane Claude Lefvre and Loisel (see [1]). In their
studies, the authors gave the formula of ruin probability for classical risk model only while in
our study, we figured out the formula for a general risk model.
Keywords: Ruin probability, premium, claims, general risk models.

Email: ntthong05@gmail.com
1. INTRODUCTION
In recent years, the finance services in general and insurance industry in particular have
played a crucial role in economy helping to adjust and promote all other economic activities.
This area has also become a platform for launching various ideas and intellectual applications.
Over the past few decades, questions in insurance and finance have been among the focus of
many mathematicians in the area of probability theory and statistics. Financial investment is an
integral part of any business sector that wants to make a profit. However, investments always
carry risk elements in itself, which might even result in loss or bankruptcy. While the nature of
insurance is to share risks, the underlying part of an insurance company is also an investment,
and therefore carries risk itself. Assessment of risk level and the likelihood of occurrence is
apparently a much needed research to mitigate risk and minimize possible loss recently, the risk
theory has been studied widely, especially in insurance and financial risks. One of the key issues
that the theory deals with is the Ruin Probability in the risk models.
This paper aims to provide a formula for the probability of ruin for a general risk model.
Picard and Lefvre [6] recently derived an explicit formula (hereinafter referred to P-L formula)
for the finite non-ruin probability in a compound Poisson model where the claim amounts are
integer-valued. Such a case is important because discretization of claim amounts is often
required for numerical calculations in practice. Many (see De Vylder [2],[3] and Ignatov [4],[5])
have pointed out the importance of the P-L formula as well as its wide applications. Claude
Lefvre and Stephane Loisel (see [1]) have recently extended P-L formula for the compound
binomial and compound Poisson risk models. Moreover, the ruin probability formula is also
given explicitly (which we can calculate exactly given that assumptions can be satisfied in
practice). The key idea in the two author’s proof is to make use of ballot theorem.
This paper also extends the ruin probability to the general model. Algorithms are also
established for numerical calculations of the ruin probability formula.
2. CONTENT
2.1 The exact formula for the probability of risk
Let us consider an insurance company that evaluates its revenues, expenses and costs,
losses, profits at fixed periodic times (for example month, quarter or year).
Assuming that the insurance company’s initial capital is u > 0. At the end of each period t,
t = 1, 2, 3, ..., we denote:
X t is the claim size in the tth period
Yt is the premium in the tth period.
The reserves U t of the company at time t are given by
t t
U t  u +  Yi −  X i
i =1 i =1
At the end of each period t, t = 1, 2, 3, ..., the ruin would occur if U t < 0.
The probability of ruin within finite time Ψ(u, t), defined by:
Ψ(u, t) = P( U t < 0, for some t = 1, 2, 3, ..., T; 1 ≤ T < +∞).
Tu stands for the first time time of ruin.
We have
Ψ(u, t)=1 − P( Tu ≥ t + 1),
where P( Tu ≥ t + 1), denotes the probability of non-ruin within finite time. We give the
formula for calculating the probability of non-ruin within finite time P( Tu ≥ t + 1), as follow.
Theorem 2.1. Assuming that the insurance company’s initial capital is u. The insurer’s
capital at the end of period t, t = 1, 2, 3, ..., is a random variable U t :
t t
U t  u +  Yi −  X i (2.1)
i =1 i =1
Where X i ; Yi are the claim size and premium of the ith period, respectively.
Without loss of generality we can assume that
Let ( X i )i 1 and (Yi )i 1 be two sequences of independent and identically distributed (i.i.d.)
non-negative integer-valued random variables. Furthermore, ( X i )i 1 and are assumed to be
independent (Yi )i 1
There exists a positive integer M < ∞ such that P(Y1  M ) = 1 and P( X 1 M) 1

(because the total premium income and claim size are only finite).
M
Denote pk = P( X 1 = k ) ; qk = P(Y1 = k ), k  M ,  qk = 1.
k =0
Then we have the exact formula for the probability of non-ruin within finite time:
𝛲(𝛵𝑢 ≥ 𝑡 + 1) = ∑ 𝑞𝑘1 𝑞𝑘2 −𝑘1 … 𝑞𝑘𝑡−𝑘𝑡−1 ( ∑ 𝑝𝑖1 𝑝𝑖2 … 𝑝𝑖𝑡 ) (2.2)

0<𝑘𝑖 −𝑘𝑖−1 ≤𝑀 0≤𝑖1 <𝑘𝑡 +𝑢
1≤𝑖≤𝑡 0≤𝑖1 +𝑖2 ≤𝑘2 +𝑢
[ 𝑘0 =0 0≤𝑖1 +...+𝑖𝑡 <𝑘𝑡 +𝑢 ]
Proof. For convenience, we denote the formula (2.1) in the following form:
Ut u Vt St
𝑡
Where 𝑉𝑡 = ∑ 𝑌𝑖 are the total premium (at the end of the ith period).
𝑖=1
𝑡
Where 𝑆𝑡 = ∑ 𝑋𝑖 are the total claim size (at the end of the ith period).
𝑖=1
Tu stands for the first time of ruin.

Our goal is to calculate the non ruin probability of P(Tu t 1) : That is, until at the end
of period t, the company is not at risk.
Obviously, the first we have
Tu  t + 1 = U i  0, i = 1,2,..., t (2.3)
We have
(Tu  t + 1) = (U i  0,1  i  t ) = (Si  Vi + u,1  i  t )
t iM
=  (Si  k + u )(Vi = k ) (2.4)
i =1 k =1
since
P(0 Yi M) 1 P(0 Vi Y1 Y2 ... Yi iM ) 1, i :1 i t
from (2.4) we have the following:

P(Tu t 1) P [( S1 u )(V1 0) ( S1 1 u )(V1 1) ... ( S1 M u )(V1 M )]
( S2 u )(V2 0) ( S2 1 u )(V2 1) ... ( S 2 2M u )(V2 2M ) ...
( St u )(Vt 0) ( St 1 u )(Vt 1) ... ( St tM u )(Vt tM )
= P[( S1  u )  ( S2  u )  ...  ( St  u )(V1 = 0)(V2 = 0)(Vt = 0)]  ...  ...
P{[( S1 k1 u ) (S2 k2 u ) ... ( St kt u )](V1 k1 )(V2 k 2 )(Vt kt )} (2.5)

0 ki ki 1 M
1 i t
k0 0
We have (2.5) due to the following properties of Vi , notice that Yi are nonnegative integers.
If i j and ki k j then
P[(Vi = ki )(V j = k j )] = P[(Y1 + Y2 + ... + Yi = ki )(Y1 + Y2 + ... + Yi + ... + Y j = k j )]
P[(Vi ki )(Yi 1 ... Y j kj ki )] 0
From (2.5), we rewrite
𝛲(𝛵𝑢 ≥ 𝑡 + 1) = ∑ 𝑃[(𝑆1 < 𝑘1 + 𝑢)(𝑆2 < 𝑘2 + 𝑢). . . ( 𝑆𝑡 < 𝑘𝑡 + 𝑢)]𝑃[(𝑌1 = 𝑘1 )

0<𝑘𝑖 −𝑘𝑖−1 ≤𝑀
1≤𝑖≤𝑡
[ 𝑘0 =0 ]
(Y2 = k2 − k1 )...(Yt = kt − kt −1 )]
= ∑ 𝑞𝑘 𝑞𝑘 … 𝑞𝑘𝑡−𝑘 𝑃[(𝑆1 < 𝑘1 + 𝑢)(𝑆2 < 𝑘2 + 𝑢). . . ( 𝑆𝑡 < 𝑘𝑡 + 𝑢)] (2.6)

1 2 −𝑘1 𝑡−1
0<𝑘𝑖 −𝑘𝑖−1 ≤𝑀
1≤𝑖≤𝑡
[ 𝑘0 =0 ]
Now we calculate the probability P[( S1  k1 + u )( S2  k2 + u )...( St  kt + u )] ,

note that
( S1  k1 + u ) = ( X 1 = i1 ) (a .s)
0i1  k1 + u
( S 2  k2 + u ) = ( X 1 + X 2  k2 + u ) = ( X 1 = i1 , X 2 = i2 ) (a.s )
i1 + i2  k2 + u
0i1 ,i2
….
( St  kt + u ) = ( X 1 + X 2 + ... + X t  kt + u ) = ( X 1 = i1 , X 2 = i2 ,..., X t = it ) (a.s )
i1 + i2 +...+ it  kt + u
0i1 ,i2 ,...,it
So we have
P[( S1 k1 u )( S 2 k2 u )...( St kt u )]
=𝑃 ⋃ (𝑋1 = 𝑖1 , 𝑋2 = 𝑖2 , … , 𝑋𝑡 = 𝑖𝑡 )
0≤𝑖1 <𝑘1 +𝑢
0≤𝑖1 +𝑖2 <𝑘2 +𝑢
[0≤𝑖1 +𝑖2 +...+𝑖𝑡 <𝑘𝑡 +𝑢 ]
= 
0i1  k1 + u
P ( X 1 = i1 , X 2 = i2 ,..., X t = it ) (2.7)
0i1 + i2  k2 + u
................
0i1 + i2 +...+ it  kt + u
Note that ( X i )i 1 is are sequence of independent and identically distributed variables.

Combining the results (2.6), (2.7), we have formula (2.2) in theorem 2.1.
This ends proof of the Theorem 2.1.
2.2 A numerical example
Here, we will illustrate the results of the exact formula for the probability of P(Tu  t + 1)
in the formula (2.2) in the case of:
M = 10
Sequences: q : 0.2, 0.1, 0.05, 0.1, 0.05, 0, 0, 0.2, 0.2, 0.1
p : 0.3, 0, 0.1, 0.1, 0.05, 0.05, 0, 0.2, 0.2, 0
t u=0 u=5 u = 10 u =15
2 P(Tu  t + 1) 0.3942 0.7329 0.9489 0.9936
3 P(Tu  t + 1) 0.3445 0.6691 0.9007 0.9766
4 P(Tu  t + 1) 0.3135 0.6238 0.8603 0.9563
5 P(Tu  t + 1) 0.2920 0.5898 0.8270 0.9360
Calculation schedule
Data:
- Arrays p, q: probability series P, Q.
- Variables t, u, M.
- Temporary variable Sp: Sum of multiplications p[i] for each time caculate multiply of
factors q[k].
- Variable Prob: Final result.
- Arrays k, i: Indexs k[n], i[m] với n = 1..t, m = 1..t
Main Procedure:
Begin
Input data for t, u, M, p, q;
Initialize for array k (k[n] = 0, n = 1..t);
Call MultiplyQ(1);
End Main Procedure;
Procedure MultiplyQ(n): //This procedure is used to generate data for indexs k for each
time calculate multiplication of factors q
Begin
If n = 1 {Loop k[1] := 1; k[1] <= M; k[1] := k[1] + 1 {Call MultiplyQ(2);}} Else {Loop
k[n] := k[n-1] + 1; (k[n] - k[n - 1] <= M) and k[n] <= n * M; k[n] := k[n] + 1; {If n < t {Call
MultiplyQ(n+1);} Else {Set Tq := 1; Loop l := 1; l <= t; l := l + 1; {Tq := Tq * q[k[l] - k[l -
1]];}
Sp := 0;
Initialize for array i (i[m] = 0, m = 1..t);
Call MultiplyP(1);
Prob = Prob + Tq * Sp;}}}
End Procedure;
Procedure MultiplyP(m): //This procedure is used to generate data for indexs i for each
time calculate multiplication of factors q
Begin
If m = 1 {Loop i[1] := 1; (i[1] <= k[1] + u) và i[1] <= M; i[1] = i[1] + 1 {Call
MultiplyP(2);}} Else {Set x = 0; Loop l := 1; l <= m - 1; l := l + 1; {x := x + i[l];}
Loop i[m] := 1; (i[m] < k[m] + u - x) and i[m] <= M; i[m] := i[m] + 1; {If m < t {Call
MultiplyP(m+1);} Else {Set Tp := 1; Loop l := 1; l <= t; l := l + 1; {Tp := Tp * p[i[l]];}
Sp := Sp + Tp;}}}
End Procedure;
3. CONCLUSION
In this paper, we proved the formula calculating the ruin probability for a general risk
model. Algorithms are also established for numerical calculations of the ruin probability
formula. From the calculated results in the above table (see numerical example ) we see that:
When the initial capital increases, the probability P(Tu  t + 1) also increases. With the same
initial capital, as t increases, P(Tu  t + 1) decreases. This is consistent with the results of
Theorem 2.1.
REFERENCES
1. Claude Lefvre and Stephane Loisel (2018), On finite-time Ruin probabilities for classical risk
models, Scandinavian Actuarial Journal, 1 (2018), 41-60.
2. F. E. De Vylder (1997), La formule de Picard et Lefvre pour la probabilit de ruine en temps fini,
Bulletin Francais d’Actuariat, 1(1997), 30-41.
3. F.E. De Vylder (1999), Numerical finite-time ruin probabilities by the Picard-Lefvre formula,
Scandinavian Actuarial Journal,2(1999), 375-386.
4. Z. G. Ignatov, V. K. Kaishev and R. S. Krachunov (2001), An improved finite-time ruin probability
formula and its Mathematica implementation, Insurance: Mathematics and Economics, 29, 375-386.
5. Z. G. Ignatov and V. K. Kaishev (2014), A finite-time ruin probability formula for continuous claim
severities, Journal of Applied Probability, 41(2014), 570-578.
6. Ph. Picard and Cl. Lefvre (1997), The probability of ruin in finite time with discrete claim size
distribition, Scandinavian Actuarial Journal, (1997) 58-69.
XÁC SUẤT THIỆT HẠI ĐỐI VỚI MÔ HÌNH RỦI RO

BẢO HIỂM TỔNG QUÁT
Tóm tắt: Nội dung chính của bài báo là tìm ra công thức tính xác suất thiệt hại (xác suất rủi
ro) cho mô hỉnh rủi ro tổng quát. Chúng tôi đã mở rộng công thức Picard-Lefèvre (xem [6]
)cho xác suất rủi ro (không rủi ro) với mô hình rủi ro tổng quát. Kết quả này là mở rộng đáng
kể kết quả trước đó của Claude Lefèvre và Stephane Loisel (xem [1] ). Trong các nghiên cứu
của các tác giả trên chỉ xét mô hình rủi ro cổ điển, trong khi đó chúng tôi nghiên cứu mô hình
rủi ro tổng quát chung và tìm ra công thức tính chính xác xác suất thiệt hại cho mô hình này.
Từ khóa: Xác suất rủi ro, phí bảo hiểm, yêu cầu bồi thường, mô hình rủi ro tổng quát.
THE DEPOSITION OF Zn-DOPED SnO2 THIN FILMS BY A

COMPRESSION SPRAYER FOR ALCOHOL VAPOR SENSOR
Pham Van Vinh1*, Dang Tran Chien2, Pham Thi Lan Huong3
1
VNU University of Engineering and Technology
2
Hanoi University of Natural Resources & Environment
3
Vietnam Military Medical University
Abstract: The thin films of SnO2 and Zn-doped SnO2 were deposited on glass wafer substrate
by a compression sprayer system using SnCl2.2H2O and Zn(CH3COO)2.2H2O as precursors.
The influence of the deposition temperature and Zn-doped contents on SnO2 and Zn-doped
SnO2 crystal phase were investigated by XRD, respectively. The results showed that the films
of SnO2 start to crystallize in the form of tetragonal structural at temperature of 370oC. SnO2
films were crystalized well at temperature above 400oC. SnO2 still retained the crystal
structure when the Zn impurity content was up to 2%. Further increasing in Zn content, the
films had more crystalline phases of zinc oxide. The surface morphology of the films observed
by SEM showed that the shape of the SnO2 crystals with size less than 100nm was fairly
uniformly. The crystal size decreased slightly with increasing impurity content. The impurities
influenced greatly on the alcohol vapor sensitivity of the films. The sensitivity of pure films
increased with an increase in operation temperature even in high temperature. The sensitivity
of the Zn-doped films increases with the increase in operation temperature up to 350 oC.
Further increase in temperature decreases the sensitivity. At the optimum operation
temperature, the films with 2 % Zn showed the best sensitivity.
Keywords: SnO2, Zn-doped SnO2, alcohol sensitivity, compression sprayer.
Received 8 May 2022

(*) Email: vinhpv@vnu.edu.vn
1. INTRODUCTION
Organic solvents including ethanol, methanol, acetone… are potential hazards to human
health. They have the ability to stimulate nerves, cause dizziness, headaches and even lose the
ability to control behavior [1]. Unfortunately, these chemicals are commonly used in industrial
processes where a large number of workers are working. Besides, drinking alcoholic beverages
can impact the driving ability of driver, causing an accident or traffic insecurity. It is necessary
to have highly sensitive sensors to detect the solvent vapors in the air and measure the
concentration of alcohol in the breath. Gas sensors based on semiconductor metal oxide (SMO)
thin films such as SnO2, ZnO, TiO2, Fe2O3 and WO3… are well suited for this purpose due to
their ability to change resistance upon interaction with toxic gases [2, 3, 4, 5]. Among the
SMOs, SnO2 that has good chemical stability and stably in high temperature conditions has
been widely used for gas sensor. Because of oxygen vacancies, SnO2 is an n-type semiconductor
and it donates electrons to other element in most reaction process [6, 7]. In air, oxygen will take
electrons from SnO2 to become adsorbed oxygen on the surface, resulting in an increase in the
resistance of the film. Organic vapors are usually reducing agents so that they can react with
the adsorbed oxygen and return the electrons to SnO2, causing the decrease of its electrical
resistance. Thanks to this property, SnO2 thin films can be used as organic vapor sensitive
sensors [8]. However, pure SnO2 usually gives low selectivity and high operating temperature
[9, 10].This has been a major hindrance in practical applications. Recently, there have been
many attempts to improve its gas sensitivity and selectivity. For this purpose, the studies have
been carried out mainly following three directions: i) adding doped elements; ii) combining
with other SMOs; iii) creating novel structure. Among these research directions, doping can
modify band structure to change optical and electrical properties of the material [11] e.g. Zn
dopant can create the defect or traps in the band gap of SnO2 to improve its optical properties
[12]. Zn2+ has a radius (0.074 nm) close to that of Sn4+ (0.070nm) which is a favorable condition
for Zn2+ to substitute for Sn2+ in the crystal lattice. The substitution elements break the bonds
to create an acceptor energy levels or states near the valence band resulting in the increase in
hole concentration by receiving electrons from valence band. This process is very useful to gas
sensing applications [13]. Zn-doped SnO2 has been indeed studied for humidity sensor [14],
sulfur hexafluoride vapor sensor [15], formic acid gas sensor [16] and ammonia sensor [17]. In
this study, pure SnO2 and Zn-doped SnO2 thin films are deposited on glass substrate by
compressive sprayer deposition apparatus for alcohol vapor sensor.
2. CONTENT
2.1. Experimental
Deposition method: The thin films of SnO2 and Zn-doped SnO2 were deposited by a
compress sprayer under the control of computer. The schematic diagram of the experimental
setup published elsewhere [18] as showed in fig.1. The spray solution for pure SnO2 was
prepared by dissolving SnCl2.2H2O into C2H5OH solvent. After 30 mins stirring, an
appropriated amount of HCl was dropped slowly into the solution. The dropping process had
finished when the pH of solution was appropriate 5 and solution became transparent.
Figure 1. Schematic diagram of experimental apparatus [18]

(1)-Compressor, (2)-Spray nozzle, (3)-Solution tank, (4)-Heater
A predetermined amount of Zn(CH3COO)2.2H2O was added in to the solution to deposit

the doped samples. After that, the solutions were sprayed on hot glass substrate with different
temperature. Thanks to the chemical reactions, the films of pure SnO2 and Zn-doped SnO2 were
formed on the substrate. Deposition temperature was carefully studied to find the optimum
temperature to deposit the SnO2 films. This temperature was then used to deposit Zn-doped
SnO2 thin films. Characterization methods: The crystal structures was studied by X-ray
diffractometer (D8 ADVANCE BRUCKER) with Cu Kα radiation (λ = 0.154056nm). Surface
morphology was observed by SEM (HitachiS-4800). The alcohol sensitivity was investigated
by static method using a homemade system in which the resistance was measured by a Keithley
2000 multi-meter as showed in the figure 2.
Figure 2. Schematic diagram of gas sensor measurement system

2.2. Result and discusión
The XRD results in fig.3 showed that no XRD peaks were found for the films deposited
below 250oC. This indicated that the films were amorphous. Three XRD peaks corresponding
to the tetragonal structural of SnO2 (refer to JCPDS No. 77-0447) appeared for the films
deposited at 370oC and the number of XRD peaks increased to eight with further increase in
temperature. This indicated that the crystallization of the films was improved as the increase of
deposition temperature. The films were crystalized well at temperature above 400oC. Therefore,
the temperature of 400oC had been used for further experiments.
250
SnO2(110)
200
150
Intensity (a.u)
SnO2(200) SnO2(211)
SnO2(101) SnO2(220)
100
SnO2(310)
SnO2(301)
50
d
c
0 b
a
20 30 40 50 60 70
o
Angle 2 ( )
Figure 3. XRD pattern of SnO2 deposited on substrates at temperatures: a) 250°C;

b) 370°C; c) 400°C d) 460°C.
250
SnO2(110)
200
SnO2(200)
SnO2(211)
SnO2(101)
150
SnO2(220)
Intensity (a.u)
SnO(110)
SnO(112)
ZnO(102)
100
50 (a)
(b)
(c)
0 (d)
20 30 40 50 60

Angle 2 ( )
Figure 4. XRD pattern of Zn-doped SnO2: a) 0%Zn; b) 2%Zn; c) 7% Zn; d) 10% Zn

Figure 4 is Zn-doped SnO2 films that deposited at 400oC with different Zn contents. The
results showed that SnO2 still retained the crystal structure when the Zn impurity content was
up to 2%. Further increasing in Zn content, the films had more crystalline phases of zinc oxide.
The diffraction peaks were shifted to the small angle when increasing the Zn content. The peak
shift was believed to the substitution of Zn2+ for Sn4+ in the SnO2 crystal lattice. Indeed, the
ionic radius of Zn2+ (0.074 nm) is larger than that of Sn4+ (0.070 nm). The substitution should
increase the distance between the lattice planes, resulting in shifting XRD peaks toward the
small angle.
Figure 5. SEM images of Zn-doped SnO2: A) 0% Zn; B) 2% Zn; C) 7%Zn; D) 10%Zn.

SEM images of Zn-doped SnO2 with different Zn content were showed on Fig. 5. The crystal
size was less than 100 nm and decreased slightly as the increase of Zn doping content. The
crystals were tetragonal crystalline shape that reconfirmed the crystal structure of SnO2.
10
Sensor response Rair/Rg

6
4 0% Zn
2% Zn
7% Zn
2 10% Zn
0
200 250 300 350 400 450
o
Temperature ( C)
Figure 6. Influence of temperature on sensor response of pure SnO2 and Zn-Doped SnO2
The influence of temperature on the alcohol gas sensivity of pure SnO2 and Zn-doped SnO2
films was investigated with the alcohol concentration of 0.4 mg/L (fig.6). For the pure films,
their sensor responses have increased with temperature in the temperature range up to 425°C.
The sensitivity of the film has not been studied at higher temperatures because it is no longer
relevant for practical applications. For Zn-doped SnO2 films, the sensor responses have
increased with temperature up to 350oC. At higher temperature, the sensor responses decreased.
There are two physico-chemical processes affecting sensitivity that occur simultaneously. The
temperature promotes chemical reactions between the alcohol and the adsorbed oxygen that
increase the sensitivity. On the other hand, temperature also promotes desorption of oxygen on
the surface of the films that decrease the sensitivity. At high temperature, the desorption process
dominates, so the sensitivity tends to decrease. This is the reason why the optimum operation
temperature of the films for detecting alcohol is 350oC.
Zn impurity in SnO2 crystals influence strongly on the sensitivity of doped films as
described in fig.6. The substitution of Zn2+ for Sn4+ provides more oxygen vacancies for SnO2,
resulting in more oxygen species adsorbed on the surface of SnO2 and more surface oxygen
vacancies in the Zn-doped SnO2. Therefore, the Zn-doped SnO2 sensor shows a higher response
[19]. XRD pattern in fig.4 showed that the Zn impurity content of 2% exhibited the optimum
substitution of Zn2+ for Sn4+. This is the reason why the films doped with 2% Zn showed the
best sensor response.
8.0
7.5
7.0
6.5
6.0
5.5
5.0
-1 0 1 2 3 4 5 6 7 8 9 10 11
Zn content (%)
Figure 6. Influence of Zn impurity content on the sensor response of the films at 350oC
The influence of alcohol vapor on the sensivity of 2% Zn- doped SnO2 was studied at
temperature of 350oC. The sensor response increases with the increase in alcohol vapor
concentration up to 1.2 mg/L. Further increase in the concentration, the sensor response tends
to approach to saturation value. The reaction between alcohol vapor and adsorbed oxygen
causes the change of the sensor response. At low vapor concentrations, the chemical reactions
are weak, so the sensor response is low. Increasing the vapor concentration should promote
chemical reactions, causing an increase in sensor response. However, if the vapor concentration
is increased to a value at which all of adsorbed oxygen reacts completely with the alcohol vapor,
the response will be saturated.
12
10
2
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Concentration(mg/L)
Figure 7. Influence of alcohol vapor on the sensivity of 2% Zn- doped SnO2
3. CONCLUSION
The Zn-doped SnO2 crystals were successfully deposited on hot substrate by compressed spray
method. The films were crystalized well at 400oC with body centered tetragonal structure. The
crystals have tetragonal shape with the size was less than 100 nm and decreased slightly as the
increase of Zn doping content. Zn-doped with 2% exhibited an optimum content due to the fine
structure and the best sensitivity. The operation temperature and the saturation alcohol vapor
concentration that gave the best film sensitivity is 350oC and 1.2 mg/L, respectively.
Acknowledgment
This paper is funded by VNU University of Engineering and Technology under grant
number of CN22.02.
REFERENCES
1. Pridgway,T.Enixon, J.PLeach, (2003), “Occupational exposure to organic solvents and long-term
nervous system damage detectable by brain imaging, neurophysiology or histopathology”, Food and
Chemical Toxicology, vol. 41: pp. 153-187.
2. Xiaoying Kang, Nanping Deng, Zirui Yan, Yingwen Pan, Wei Sun, Yaofang Zhang, (2022),
“Resistive-type VOCs and pollution gases sensor based on SnO2: A review”, Materials Science in
Semiconductor Processing, vol. 138: p. 106246.
3. Mariane A.Franco, Patrick P.Conti, Rafaela S.Andre, Daniel S.Correa, (2022) “A review on
chemiresistive ZnO gas sensors”, Sensors and Actuators Reports, Vol. 4: p.100100.
4. Xu Tian, Xiuxiu Cui, Tingrun Lai, Jie Ren, Zhichao Yang, Mingjing Xiao, Bingsen Wang, Xuechun
Xiao, Yude Wang, (2021), “Gas sensors based on TiO2 nanostructured materials for the detection of
hazardous gases: A review”, Nano Materials Science, vol. 3: pp. 90-403.
5. Nasir Rahman, Juan Yang, Zulfiqar, Mohammad Sohail, Rajwali Khan, Anwar Iqbal, Chanez
Maouche, Abid Ali Khan, Mudasser Husain, Shaukat Ali Khattak, Saima Naz Khan, Aurangzeb
Khan, (2021), ”Insight into metallic oxide semiconductor (SnO2, ZnO, CuO, α-Fe2O3, WO3)-carbon
nitride (g-C3N4) heterojunction for gas sensing application”, Sensors and Actuators A: Physical, vol.
332: p.113128.
6. P. Chetri, B. Saikia, A. Choudhury, (2013), “Structural and optical roperties of Cu doped
SnO2 nanoparticles: An experimental and density functional study” J. Appl. Phys. vol. 113:
p.233514.
7. A. Azam, A.S. Ahmed, M. Chaman, A.H. Naqvi, (2010), “Investigation of electrical properties of Mn
doped tin oxide nanoparticles using impedance spectroscopy”, J. Appl. Phys. vol. 108: p. 094329.
8. Gurpreet Singh, Nipin Kohli, Ravi Chand Singh, (2017), “Sensitive and selective ethanol sensor based
on Zn-doped SnO2 nanostructures”, J Mater Sci: Mater Electron vol: 28, pp. 3013–13023.
9. Chi W. S., Lee C. S., Long H., Oh M. H., Zettl A., Carraro C., Kim J. H., Maboudian R., (2017),
“Direct Organization of Morphology-Controllable Mesoporous SnO2 Using Amphiphilic Graft
Copolymer for Gas-Sensing Applications”, ACS Appl. Mater.Interfaces,
vol. 9: pp. 37246 –37253.
10. Lee S.-H., Galstyan V., Ponzoni A., Gonzalo-Juan I., Riedel R., Dourges M.-A., Nicolas Y.,
Toupance T. Finely (2018), “Tuned SnO2 Nanoparticles for Efficient Detection of Reducing and
Oxidizing Gases: The Influence of Alkali Metal Cation on Gas-Sensing Properties”, ACS Appl.
Mater. Interfaces, vol. 10: pp. 10173–10184.
11. M. Kumar, B. Singh, P. Yadav, V. Bhatt, M. Kumar, K. Singh, A.C. Abhyankar, A. Kumar, J.-H.
Yun, (2017), “ Effect of structural defects, surface roughness on sensing properties of Al doped ZnO
thin films deposited by chemical spray pyrolysis technique” Ceram. Int., vol.43:p. 3562.
12. P. Baraneedharan, S.I. Hussain, V.P. Dinesh, C. Siva, P. Biji, M. Sivakumar (2015), “Lattice doped
Zn–SnO2 nanospheres: A systematic exploration of dopant ion effects on structural, optical, and
enhanced gas sensing properties”, Appl. Surf. Sci. vol. 357: p.1511
13. J.S. Bhat, K.I. Maddani, A.M. Karguppikar, (2006), “Influence of Zn doping on electrical and optical
properties of multilayered tin oxide thin films”, Bull. Mater. Sci. vol. 29: p. 331.
14. N.D. Md Sin , Noorlaily Samsudin, S. Ahmad, M.H. Mamat, M. Rusop, (2013), “Zn-doped SnO2
with 3D cubic structure for humidity sensor”, Procedia Engineering, vol.56: pp. 801-806.
15. FangmingDeng, YigangHe, GuolongShi, BingLi, XiangWu, (2016), “Low-temperature
cataluminescence sensor for sulfur hexafluoride utilizing coral like Zn-doped SnO2 composite”,
Sensors and Actuators B: Chemic, vol. 237: pp. 120-126.
16. Nantikan Tammanoon, Anurat Wisitsoraat, Adisorn Tuantranont, Chaikarn Liewhiran, (2021),
“Flame-made Zn-substituted SnO2 nanoparticulate compound for ultra-sensitive formic acid gas
sensing”, Journal of Alloys and Compounds, vol. 871: p. 159547.
17. M.Boomashri, P.Perumal, AslamKhan, Ahmed M.El-Toni, Anees AAnsari, Ravindra K.Gupta,
Prashantha Murahari, K. Deva Arun Kumar, (2021), “Zinc influence on nanostructured tin oxide
(SnO2) films as ammonia sensor at room temperature”, Surfaces and Interfaces, vol. 25: p. 101195.
18. Pham Van Vinh and Vu Kim Thai, (2021), “Deposition and study of alcohol vapor sensitivity of
SnO2/ZnSnO3 thin films”, Scientific Journal of Hanoi Metropolitan University, Vol. 46: p. 75-84.
19. Qinqin Zhao, Dianxing Ju, Xiaolong Deng, Jinzhao Huang, Bingqiang Cao & Xijin Xu (2015),
“Morphology-modulation of SnO2 Hierarchical Architectures by Zn Doping for Glycol Gas Sensing
and Photocatalytic Applications”. Scientific Reports Vol.5, P. 7874.
CHẾ TẠO MÀNG MỎNG SnO2 PHA TẠP Zn BẰNG PHƯƠNG PHÁP
PHUN ÁP SUẤT ỨNG DỤNG CHO CẢM BIẾN NHẠY HƠI CỒN
Tóm tắt: Màng mỏng SnO2 và SnO2 pha tạp Zn được phủ trên đế thủy tinh bằng hệ phun áp suất
sử dụng muối SnCl2.2H2O và Zn (CH3COO) 2.2H2O làm tiền chất. Ảnh hưởng của nhiệt độ chế
tạo và nồng độ tạp chất Zn lên cấu trúc tinh thể của màng SnO2 và SnO2 pha tạp Zn đã được
khảo sát bằng XRD. Kết quả cho thấy màng SnO2 với cấu trúc tứ giác đã bắt đầu kết tinh ở nhiệt
độ 370oC và kết tinh tốt ở nhiệt độ trên 400oC. SnO2 vẫn giữ nguyên cấu trúc tinh thể khi nồng
độ tạp chất Zn lên đến 2%. Nếu tiếp tục tăng nồng độ Zn tăng hơn nữa thì sẽ có thêm nhiều pha
tinh thể của oxit kẽm. Hình thái bề mặt của các màng quan sát bằng ảnh SEM cho thấy các tinh
thể SnO2 phân bố khá đồng đều với kích thước nhỏ hơn 100nm. Kích thước tinh thể giảm nhẹ
khi nồng độ tạp chất tăng. Tạp chất ảnh hưởng rất nhiều đến độ nhạy hơi cồn của màng. Độ
nhạy của màng tinh khiết tăng lên khi nhiệt độ làm việc của nó tăng lên. Tính chất này vẫn còn
duy trì ngay khi ở vùng nhiệt độ khá cao. Độ nhạy của màng pha tạp Zn tăng lên khi nhiệt độ
làm việc của nó tăng lên đến 350oC. Nhiệt độ tăng hơn nữa sẽ làm giảm độ nhạy. Ở nhiệt độ
làm việc tối ưu, các màng pha tạp với nồng độ 2% Zn thể hiện độ nhạy tốt nhất.
Từ khóa: SnO2, SnO2 pha tạp Zn, độ nhạy hơi cồn, hệ phun áp suất.
FABRICATION AND CHARACTERISTICS OF

NANOCOMPOSITE TiO2/GQDs THIN FILMS
Dang Tran Chien1*, Phan Xuan Thien2

1
Hanoi University of Natural Resources and Environment,
2
Institute of Material Sciences, Vietnam Academy of Science & Techonology
Abstract: This work carries out the study and fabrication of TiO2/Graphene Quantum
Dots (GQDs) nanocomposite thin films to be applicated as an electronic collection
material layer in nanostructured solar cells. The different concentrations of added GQDs
by the spreading method were loaded on TiO2/GQDs nanocomposite films. The
fabricated films were investigated for their structural and morphological characteristics
by X-ray diffraction spectroscopy; Ramam spectra and Scanning Electron Microscopy
(SEM) imaging. The optical properties were investigated through absorption and
fluorescence spectroscopy. The photoelectrical properties of the film were investigated
on measuring I-V characteristics in the dark and when illuminated by the solar simulator
AM 1.5G. The obtained results show that adding GDQs has significantly improved the
photoelectronic properties of the materials.
Keywords: TiO2, GQDs, nanostructure solar cells.
Received 3 May 2022

(*) Email: dtchien@hunre.edu.vn
1. INTRODUCTION
TiO2 is one of the promising materials for applications ranging from water spliting, solar
cells, energy storage devices, optoelectronic devices, photocatalysis. Nanostructured TiO2 films
are used as the electron-collecting material layer and plays an important role to the parameters
of the devices. However, TiO2 is a wide band gap semiconductor material so it absorbs light in
the ultraviolet region, in addition, the conductivity of TiO2 is also very low, which limits a lot
to improving the performance of solar cells as well as being less efficient when used as a
photocatalyst under sunlight. On the other hand, the high recombination rate of the carrier is
also a reason for the efficiency of solar cell devices as well as the low efficiency of
photocatalysis. To overcome those disadvantages, the adding of semiconductor quantum dots
(QDs), such as CdS, CdSe, PbS, especially GQDs onto the TiO2 material films has been
proposed to expanding the working spectrum to the visible light region, increasing the ability
to separate charge carrier pairs as well as increasing the ability to conduct electrons to the
electrodes [1, 2, 3]. Among the quantum dots that are attracting researchers' attention, GQDs is
a new type of material that has been researched and fabricated recently and it has many special
physical and chemical properties such as high electrical conductivity, strong luminescence,
luminescence wavelength changes with the excitation light wavelength [4, 5], capable of
forming bonds with TiO2 crystals [1, 6]. In addition, GQDs have more advantages than
conventional semiconductor QDs that are easy to fabricate and very environmentally friendly.
Therefore, the combination of GQDs with TiO2 nanocrystals is considered to be able to give
the material system with better optical absorption, reducing the recombination process due to
the charge transfer between the nanocrystals TiO2 and GQDs, increasing the conductive
property of the material system. This promises to provide a material system with suitable
properties for making 3rd generation solar cell devices as well as increasing their photocatalytic
ability. In this work, we fabricated TiO2/GQDs nanocomposite thin film used as an electronic
collection layer in nanostructured solar cell devices. The different concentrations of added
GQDs by the spreading method were loaded on TiO2/GQDs nanocomposite films. The
fabricated films were investigated for their structural and morphological characteristics by X-
ray diffraction spectroscopy; Ramam spectra and SEM imaging. The optical properties were
investigated through absorption and fluorescence spectroscopy. The photoelectric properties of
the film were investigated on measuring I-V characteristics in the dark and when illuminated
by the solar simulator AM 1.5G. The obtained results show that adding GDQs has significantly
improved the photoelectronic properties of the material that can be applied to improve the
performance of nanostructured solar cell devices.
2. CONTENT
2.1. Experimental setup
The chemicals used in this work are included: ethanol C2H5OH (99.7%,), aceton CH3CHO
(99.5%), zinc powder Zn (90%), hydrochloric acid HCl (1M), decon 90 (Fisher), deionized
water, titanium dioxide (TiO2-P25), graphene quantum dots (GQDs). GQDs were chemically
prepared by initial carbon precursors trinitropyrene, sodium hydroxide with an average size of
about 2-5 nm, have absorption spectra in the range of 300 -500 nm and luminescence with
emission wavelengths from ultraviolet to 600 nm. The detail of the procedure was published in
[7, 8]. Titanium dioxide TiO2 P25 were added to ethanol, distilled water and GQDs, stirring for
15 minutes, then the mixture was ultrasonically vibrated for 30 minutes, as shown in Figure 1.
Titanium-dioxide (TiO2) nanoparticles suspension concentration in ethanol by volume was
5% then the GQDs solution were added to TiO2 solution with volume concentration at 0%, 5%,
10%, 20%, and 30%.
Fig.1. The procedure of fabrication of TiO2/GQDs nanocomposite films

The samples fabricated for the study include: Sample 0%: TiO2 powder / ethanol and
samples of nanocomposite TiO2, GQDs/2ml ethanol in the ratios: 0.05%, 0.1%, 0.2%, 0.3% of
GQDs in terms of mass. The solution mixtures were stirred for 15 minutes and then
ultrasonically vibrated for 30 minutes so that TiO2 was uniformly distributed with GQDs in
water- ethanol medium. To investigate the photocurrent, the FTO/TiO2/GQDs samples were
fabricated as follows: Fluorine-doped Tin Oxide (FTO)/glass substrates were prepared using
HCl solutions 1M and zinc powder. Firstly, cover the FTO substrates to be retained with
vacuum tape. Secondly, a thin layer of zinc powder is evenly coated on the surface to be etched,
after 15 minutes, the FTO layer completely corroded. The FTO substrates were cut into small
pieces of the dimensions 4.5 cm x 1.5 cm and then etched as shown in Figure 2 then use
vacuum tape to cover the parts without TiO2/GQDs loaded on. The coated film is 1.5 cm x 0.5
cm in size. Drop 20 µl of TiO2/GQDs solution on the prepared substrate, then it was placed on
a ultrasonic vibrator to spread the membrane evenly. The sample was annealed at temperature
of 80°C to evaporate the solvent. Finally, the sample was annealed at 450°C for 30 minutes.
Fig. 2. Schematics of TiO2/GQDs composite films synthesis by spreading method

and the photogragh of TiO2/GQDs composite films with volume concentration of GQDs
at 0%, 5%, 10%, 20%, and 30%.
2.2. Results and discussion

Figure 3 shows the SEM image of TiO2/GQDs samples spreaded on silicon substrate with
different concentrations of GQDs, annealed at 450oC for 30 minutes. As shown in Figure 3, the
fabricated TiO2/GQDs films are quite porous with 18-25 nm TiO2 particles. The images of the
GQDs can not be observed because they are so small that the SEM method can not
recognize them.
Fig.3. SEM images of TiO2/GQDs with GQDs concentrations a) 0%, b) 5%, c) 10%, d) 20%, e) 30%
Figure 4 shows the XRD spectrum of the TiO2/GQDs films after annealing. As can be seen
on the Figure, the peaks on the XRD spectrum corresponded with anatase phase TiO2 and no
strange peaks appeared. This proves that after being annealed at 450oC for 30 minutes, TiO2
films are completely become anatase phase. When adding GQDs to TiO2 material, the
diffraction peaks of TiO2 remain unchanged and strange peaks did not appear, indicating that
the addition of GQDs does not change the crystal structure of TiO2.
Fig.4. XRD diffraction patterns of TiO2/GQDs with GQDs concentrations a) 0%, b) 5%, c)
10%, d) 20%, e) 30%
In addition, it can be seen that the thin film included only pure TiO2, the TiO2 particles are
lacking in high porosity. After being mixed with GQDs (Figure 3 b, c, d, e), it seems that the
presence of GQDs enhances the binding between TiO2 and TiO2 particles, so the film has an
attachment cohesive and less porous.
TiO2 - GQDs 5%
TiO2 - GQDs 10%
TiO2 - GQDs 20%
104 TiO2 - GQDs 30%
TiO2
Intensity (counts)
103
102
101
200 400 600 800 1000 1200 1400 1600 1800 2000
Raman shift (cm-¹)
Fig. 5. Raman spectra of TiO2/GQDs with different concentrations of GQDs
Fig. 6. Absorption spectra of TiO2/GQDs with different concentrations of GQDs

Due to the reason that the existence of GQDs in the TiO2/GQDs composite films were not
observed by SEM images and X-ray diffraction patterns, Raman spectroscopy was conducted
on the fabricated samples. The results are as shown in Figure 5.
From Figure 5, one can see that the TiO2/GQDs composite films with different
concentrations of GQDs all appear strong peaks corresponding to the number of waves at about
395 cm-1, 521 cm-1, 638 cm-1 which characterized the properties of anatase TiO2 material. In
addition, there are some peaks with much smaller intensity at the wave number position of about
1376 cm-1 and 1675 cm-1, corresponding to the vibrational mode that responsible for the D and
G bands of the quantum dot graphene material. With those results, the presence of GQDs in
the TiO2 films can be confirmed so TiO2/GQDs nanocomposite films have been successfully
fabricated.
The TiO2/GQDs films on the glass substrate were heat treated for UV-VIS absorption
spectroscopy. The results are shown in Figure 6. It can be seen that the pure TiO2 film absorbs
in the ultraviolet region, but the absorption tail extends to the region above 500 nm. This is
because the TiO2 material has a high porosity, which causes light to be scattered strongly, so
the transmittance of the film is reduced. With adding GQDs, the absorption intensity of
TiO2/GQD films at wavelength below 500 nm was significantly enhanced, especially in the
ultraviolet range from 350 to 400 nm. TiO2/GDQs samples with 5% and 10% GQDs
concentration showed the strongest absorption. This is consistent with the fact that both TiO2
and GQDs materials strongly absorb in the UV region. In addition, GQDs also partially absorb
light in the 500 nm region. Above 500 nm of wavelength, the films containing GQDs have
reduced absorption intensity compared to the TiO2 sample. This is because the overlap of GQDs
between TiO2 nanoparticles increases the bonding ability between the materials. Therefore, the
film is not too porous like pure TiO2 film, thus reducing the scattering phenomenon in the film.
This is consistent with the results of SEM image studied as presented in the previous section.
The optical properties of TiO2/GQDs nanocomposites are also studied through fluorescence
spectrometry. Figures 7a and 7b are the fluorescence spectra of TiO2/GQDs film samples with
different concentrations excited by lasers with wavelengths of 325 nm and 420 nm,
respectively. In the case of excitation at 325 nm (Figure 7a), the emission peak is observed at
about 400 nm with weak intensity in all samples. Those peaks are related to a direct transition
(band - band). A strong spectral band at longer wavelengths from about 450 nm to 750 nm
occurs both in pure TiO2 and GQDs-doped films. The strong luminescence of TiO2 films in the
visible light is thought to be emission processes from surface states due to oxygen vacancies
and defects on TiO2. This result coincides with the results of Yu Chen et al. [9] and many other
studies on TiO2 nanomaterials. When adding GQDs to the TiO2 film, the luminescence intensity
of the samples corresponding to the band-band emissions and the emission due to the surface
state increased significantly compared to the pure TiO2 film and the spectral form was
completely unchanged.
Fig.7. Photoluminenscence spectra of TiO2/GQDs with different concentrations of GQDs: a)

325 nm and b) 420 nm excited wavelengths
This phenomenon can be explained as follows: The electrons of GQDs, after being excited
to the conduction band, have shifted to the conduction band of TiO2 due to the difference in
energy levels of the two materials. The electron then recombines with the hole in the valence
band of TiO2 leading to emitting luminescence. From Figure 7, it can be seen that when the
GQDs concentration changes, the photoluminescence (PL) spectrum also changes but not much
and the sample with 20% GDQs gives the highest intensity of PL spectral peaks.
With excitation wave at 420 nm which has a lower energy than the band gap of TiO2 so
there is a indirect transition band to band in the material system. Luminescence emitting due to
surface states at wavelengths in the range 470 nm to 750 nm. When the concentration of GQDs
increased, the intensity of the PL spectrum increased and reached the maximum in the sample
with 10% GQDs, then began to decrease gradually. These results indicate that the
nanopcomposite TiO2/GQDs films have a charge transfer between GQD and TiO2 and the
transfer depends not too much on the content of GQDs particles.
Fig. 8. I-V characteristic curve of TiO2/GQDs in dark and under ullimination

To investigate the photoelectrical properties of the TiO2/GQDs films, I-V measurement
under illumination and in dark was conducted. The measured potential ranges linearly from -
1V to 1V. The results are as shown in Figure 7 and Table 1. It can be seen that when illuminated,
the photocurrent intensity of all TiO2 films with GQDs increases and has a larger value than
that of TiO2 films without GQDs. This proves that the nanocomposite films exhibits better
photoelectric properties than pure TiO2 films. The highest current is 10% TiO2/GQDs sample.
This is completely consistent with the results of fluorescence spectroscopy studies of the
materials. 10% samples with 10 % GQDs show better charge transfer and has maximum PL
spectral intensity.
Table 1. Curent and resistance of the TiO2/GQDs films at a potential of 1V
The resistance of the films without illumination is smaller than that of the TiO2 films,
indicating that GQDs reduce the resistance of the TiO2 films and increase the photoelectric
properties of the films. These results are shown in Table 1 in which one can see the 10% sample
gives the largest change rate and reaches a value of 3.29 times. Moreover, the resistance of
TiO2/GQDs films is also significantly smaller than that of pure TiO2 film. This result shows
that TiO2/GQDs can improve the performance of solar cells.
3. CONCLUSION
TiO2/GQDs nanocomposite films have been successfully fabricated by spreading method
with different GQDs concentrations from 0 to 30% by volume. TiO2/GQDs films increase the
light absorption capacity of materials in the ultraviolet region. The good charge transfer process
between GQDs and TiO2 contributes to increasing charge separation of the material system.
The sample with 10% GQDs reaches a value of 3.29 times, the resistance of TiO2/GQDs films
is also significantly smaller than that of pure TiO2 film. With properties such as strong light
absorption (especially in the ultraviolet) and high electrical conductivity, GQDs can overcome
the disadvantages of TiO2 material. Nanocomposites TiO2/GQDs layers can be used as electron
collectors for nanostructured solar cells with high efficiency.
Acknowledgement
This work has been done at the Lab of Material & Energy Devices, Institute of Material
Sciences, Vietnam Academy of Science & Techonology, 18 Hoang Quoc Viet Str., Cau Giay
Dist., Hanoi under the supervision of Associate Professor Pham Duy Long.
REFERENCES
1. P. Sudhagar , I. H. Cardona , H. Park , T. Song , S. H. Noh , S. Gimenez, I. M. Sero , F. F. Santiago, J.
Bisquert, C. Terashimae, U. Paik, Y. S. Kang, A.Fujishimae, T.H. Han (2016), “Exploring Graphene
Quantum Dots/TiO2 interface in photoelectrochemical reactions: Solar to fuel conversion”,
Electrochimica Acta, vol. 187: pp. 249-255.
2. G. Rajende, J. Kumar, P.K. Giri (2018), “Interfacial charge transfer in oxygen deficient TiO2-graphene
quantum dot hybrid and its influence on the enhanced visible light photocatalysis”, Applied Catalysis
B: Environmental, vol. 224: pp. 960-972.
3. H. Wang, P. Sun, S. Cong, J. Wu, L. Gao, Y. Wang, X. Dai, Q.Yi, G. Zou (2016), “ Nitrogen-Doped
Carbon Dots for “ green ” Quantum Dot Solar Cells”, Nanoscale Research Letters, vol 11: pp.27.
4. S.Zhu, Y. Song, X. Zhao, J. Shao, J. Zhang, B. Yang (2015), “The photoluminescence mechanism
in carbon dots (graphene quantum dots, carbon nanodots, and polymer dots): Current state and future
perspective”, Nano Research, vol. 8(2): pp. 355–381.
5. B. Zheng, Y. Chen, P. Li, Z. Wang, B. Cao, F. Qi, J. Liu, Z. Qiu, W. Zhang (2017), “Ultrafast
ammonia-driven, microwave-assisted synthesis of nitrogen-doped graphene quantum dots and their
optical properties”, Nanophotonics, vol. 6(1): pp. 259–267.
6. Y. Zhang, N. Zhang, Z. R. Tang, Y.J. Xu (2012), “Improving the photocatalytic performance of
graphene–TiO2 nanocomposites via a combined strategy of decreasing defects of graphene and
increasing interfacial contact”, Physical Chemistry Chemical Physics, vol. 14: pp.9167-9175.
7. P.N. Thang, L.X. Hung, V. D. Chinh, P. N. Hong, D. N. Tu, P. T. Nga (2018) “Plasmon effect of
graphene quantum dots on corrugated silver film”, Electric Electron Tech Open Acc. J , vol. 2(6):
pp.367‒370.
8. S. Yu, Y. Zhong, B. Yu, S. Cai, L. Wu, Y. Zhou (2016) “Graphene quantum dots to enhance the
photocatalytic hydrogen evolution efficiency of anatase TiO2 with exposed {001} facet”, Physical
Chemistry Chemical Physics, vol. 18: pp. 20338-20344.
9. Xuqiang Hao, Zhiliang Jin, JingXua Shixiong Min, Gongxuan Lu (2016), “Functionalization of
TiO2 with graphene quantum dots for efficient photocatalytic hydrogen evolution”, Superlattices and
Microstructures, vol.94: pp.237-244.
CHẾ TẠO VÀ KHẢO SÁT CÁC TÍNH CHẤT ĐẶC TRƯNG

CỦA MÀNG MỎNG NANOCOMPOZIT TiO2/GQDs
Tóm tắt: Trong bài báo này chúng tôi nghiên cứu và chế tạo màng mỏng na nô compozit
TiO2/GQDs (Chấm lượng tử Graphene) ứng dụng làm lớp vật liệu thu điện tử trong pin mặt
trời cấu trúc nano. Nồng độ khác nhau của GQDs được phủ bằng phương pháp trải trên màng
mỏng na nô compozit TiO2/GQDs. Đặc điểm cấu trúc và hình thái của các màng mỏng na nô
compozit TiO2/GQDs được khảo sát bằng phương pháp nhiễu xạ tia X, phổ Ramam và chụp ảnh
SEM. Tính chất quang học được khảo sát bằng phương pháp đo phổ hấp thụ và phổ huỳnh
quang. Tính chất quang điện của màng mỏng được nghiên cứu thông qua đo đặc trưng I-V trong
bóng tối và khi được chiếu sáng bởi đèn mô phỏng ánh sáng Mặt trời AM 1.5G. Kết quả thu
được cho thấy việc bổ sung GDQs đã cải thiện đáng kể các đặc tính quang điện tử của vật liệu.
Từ khóa: Ôxít Titan (TiO2), Chấm lượng tử Graphene (GQDs); Pin mặt trời cấu trúc nano.
ASSESS THE PERCEPTION OF CLIMATE CHANGE

STUDENTS AT HANOI METROPOLITAN UNIVERSITY
Nguyen Xuan Trinh

Ha Noi Metropolitan University
Abstract: The main cause of climate change is believed to be an increase in activities that
generate greenhouse gas emissions, over-exploitation of greenhouse gas sinks such as
biomass, forests, ecosystems, etc. other marine, coastal and inland ecosystems (mainly CO2
and Methane CH4). Greenhouse gas emissions cover the earth and will trap the sun's heat.
This leads to global warming and climate change. The world is warming faster than at any
other time in recorded history.
Keywords: Climate change, greenhouse gas (greenhouse gas: CO2; CH4; CFC; ..) ecological
system fossil fuel.

Email: nxtrinh@hnmu.edu.vn
1. INTRODUCTION
Step Into the XXI century, As the technology grows like stormily, the industry is
promoted, bring the nations a huge profit. Therefore, the nations have developed the industry
to achieve national interests, which have forgotten the interests of all mankind. By the way,
mankind faces a series of survival challenges, one that challenges the environment and climate
change. Earth's climate change is the change of the climate system including the armament,
hydrosphere, biosphere, lithosphere, and cryosphere at present and in the future by natural and
man-made nuclei in a certain period. Measured in records or millions of years. A sea changes
can be a change in military weather or a change in the classification of events around a mean.
Climate change may be limited to a certain region or may occur globally. In recent years,
especially in the policy environment, climate change often refers to the current climate change,
collectively referred to as global warming. The main causes of Earth's climate change are
increased activities that generate greenhouse gases, over-exploitation of sinks and reservoirs of
greenhouse gases such as biomass, forests, marine ecosystems, coastal areas and adjacent areas.
The change in the representative species, distribution and coverage of the vegetation can
happen due to climate change, this is very noticeable. In any event, a slight climate change
could lead to increased precipitation or snowfall and increased levels of warmth, leading to
improved plant growth and subsequent uptake of more CO2 from the air. Climate change is
causing us to suffer a variety of consequences such as poverty, death, disease, education-related
problems, flooding in plains, coastal landslides,...
Vietnam is a country located in the tropical monsoon region of Southeast Asia, with a long
coastline and low riverine delta areas, the economy depends mainly on natural resources,
Vietnam is therefore one of the countries seriously affected by climate change. To deal with
this problem, many measures to improve the capacity to prevent natural disasters and adapt to
climate change have been oriented quite early, especially for students and students (Ministry of
Education and Training, 2012). This is an important group of subjects trained under the national
education program with many subjects showing the natural-socio-economic-environment
interaction in real life. However, in order to properly assess this group's understanding of the
impacts of climate change, the need to choose an effective approach in the field of education is
extremely urgent. Surveying, assessing students' awareness and learning programs about
climate change phenomenon, the results obtained after the assessment survey help the author to
have a specific orientation in the process of teaching students with specialisation its industry.
Hanoi Metropolitan University is a multidisciplinary university. Therefore, scientific research
projects and access to climate change of students in the school are also very interested.
2. CONTENT
2.1. Research subjects
Research on awareness and attitudes of students at Hanoi Capital University on climate
change today. Freshman and senior students aged 19 -22 years old.
2.2. Research scope
Space scope: Hanoi Metropolitan University
Research is limited to students of Education, Preschool, Economics, Tourism Culture,
Foreign Languages, and Natural Sciences at Hanoi Metropolitan University.
Time range: from March 2019 - October 2019
Content limitations: Focusing on students' perceptions of climate change.
2.3. Research content
Assess students' general knowledge about climate change.
Assess students' perceptions and attitudes about issues related to climate change: The level
of interest and anxiety about the impact of climate change
+ Students' awareness about the responsibilities of sectors / fields for climate change
+ Students' expectations in climate change mitigation
+ The level of student participation in low-carbon activities
Proposing solutions to raise students' awareness about climate change.

2.3.1. Research purposes
Assess students' perceptions, learning programs, and attitudes toward climate change.
Proposing some solutions to raise awareness about climate change for students.
2.3.2. Request
Honest and objective data to properly assess the reality of students' perceptions. The
proposed solutions must be suitable to the conditions of the school as well as the locality, and
be highly practical.
2.4. Overview
2.4.1 Scientific basis
Climate change (defined by the Climate Convention) is directly or indirectly attributed to
human activities changes the composition of the global atmosphere and contributes to natural
climate variability in comparable time periods. Climate change is "a change of climate that is
either directly or indirectly human activities that change the composition of the global and
external atmospheres which are automatically variable of the climate. closely over a long period
of time. Factors that can cause climate change to occur are changes in atmospheric radiation,
including processes such as changes in solar radiation, deviations of Earth's orbit, mountain
tectonics, tectonics, etc. continental drift and changes in greenhouse gas concentrations.
Various environmental responses to climate change can enhance or reduce initial changes.
Some components of the climate system, such as the oceans and ice caps, respond slowly to
changes in solar radiation because of their high mass. As a result, the climate system may take
centuries or longer to fully respond to external changes.
Global sea level change over the past several centuries has been estimated using tide
gauges, measurements collated over long periods of time to give a long-term average water
level. More recently, the altimeter - combined with the precise positioning of satellite orbits -
has provided an improved method of measuring global sea level change. Scientists have
observed, measured and found that the ice in the ice island of Greenland has lost a large amount,
directly affecting the island nations or coastal countries. It is estimated that if the ice continues
to melt, the sea level will rise by at least 6m by 2100. At this rate, most of Indonesia's islands,
and many other coastal cities will completely disappear.
2.4.2. Reason
Group of objective causes (due to the change of nature) include: changes of earth orbit
parameters, changes in the continental - sea distribution of the earth's surface, variation in the
sun's emission and the absorption of earth's radiation, volcanic activity. Group of subjective
causes (due to human influence) stemming from the change of land use purpose and water
source, the increase of greenhouse gas emissions.scientific review of the Intergovernmental
Panel on Climate Change (IPCC) shows that the consumption of energy from burning fossil
fuels in the energy production industry, industry, transportation, construction,… contribute
about half (46%) to global warming; tropical deforestation contributes about 18%; agricultural
production is about 9%; chemical industries (CFC, HCFC) about 24%; The remaining (3%) is
from other activities. In the context of climate change, man-made factors also affect the climate.
The widely agreed scientific view of climate change is that "the climate is changing, and these
changes are in large part due to human impact”. The element management system in which
there is a change in the main content of nature including the change in the activity of the sun,
the orbit of the earth, the transition of the continents... also has a significant impact.
2.4.3. Expression
Climate change manifestations include (IPCC, 2007):
Global average temperature increases due to global warming.
Sea level rise due to thermal expansion and melting.
With the use of mechanical measuring tools, scientists have determined sea level elevation
through traces on coral reefs, coastal sediments, on the sea floor, and grains in rocks. Lime and
archaeological remains near the coast. The most advantageous dating methods are the uranium-
thorium and radiocarbon methods, while the cosmic nuclear dating method is sometimes used
to determine the age of surfaces that have undergone a reduction in ink levels. Sea level has
been rising at an average rate of 1.8 mm/year over the past century, and more recently, in the
era of using altimeter satellites to determine sea level.
In the context of climate change, man-made factors also affect the climate. The widely
agreed scientific view of climate change is that "the climate is changing, and these changes are
in large part due to human impact." Changes in composition and atmospheric quality. Changes
in the activity intensity of the atmospheric circulation, the water cycle in nature and other
biochemical cycles.
[1] Earth's current temperature is causing species to disappear or be in danger of extinction.
About 50% of plant and animal species will face extinction by 2050 if the earth's temperature
increases by another 1.1 to 6.4 degrees Celsius. This loss is due to habitat loss to fallow land,
deforestation and warming seas. Biologists have noticed that some animals have migrated to
the poles to find habitats with suitable temperatures. For example, the red fox, which used to
live in North America, has now moved to the Arctic.
Humans are no exception. Wasteland and rising sea levels also threaten our habitat. And
when plants and animals are lost, our food, fuel and income sources are also lost. However, the
increase in global average temperature and sea level rise are often considered as two main
manifestations of climate change.
2.4.4. Consequence
According to SOS Environment (2014), there are 10 terrible consequences of climate
change including:
Climate change and increasing carbon dioxide levels are "challenging" our ecological
system. It is estimated that about 50% of plant species face absolute risk by 2050 if the earth's
temperature increases by another 1.1 to 6.4 degrees Celsius. Habitat loss is caused by fallow
land, deforestation and warming seas.
The ecosystem is destroyed, loss of biodiversity, war and conflict, economic impacts,
disease, drought, storm and flood, the heat wave intense, glaciers and glaciers are shrinking,
sea levels are rising.
A recent forecast by Climate Central (USA) shows that many coastal cities around the
world, including Ho Chi Minh City, are at risk of being submerged by sea water earlier than
previously forecast due to climate change. Ice at the poles melts with speed alarm.
Increasing temperature on earth causes sea level to gradually rise. Increasing the
temperature causes the glaciers, sea ice or continental ice on the earth to melt and increase the
amount of water that flows into the seas and oceans.
Each radiation pollution, erratic weather changes, heavy rain and wind, prolonged hot sun,
smoke and dust signal ... have a great impact on human health, life expectancy is reduced, skin-
related diseases, respiratory tract Cardiovascular disease is increasing.
2.4.5. The reality of education on climate change in Vietnam
[2] Recognizing the enormous and serious impacts caused by climate change, the Prime
Minister of the Socialist Republic of Vietnam has approved the National Target Program to
respond to climate change (Decision No. 1670 / QD -TTg issued on October 31, 2017).
To implement the national target program to cope with climate change, the Minister of
Education and Training approved the Action Plan to cope with climate change of the Education
sector for the period 2016 – 2020 and decision 329 / QD- MOIT issued on 25/01/2014 on
approving the project “Information, propaganda on response to climate change and disaster
prevention in schools in the period of 2013-2020”.
University of Natural Resources and Environment is the only unit in Vietnam that has a
bachelor’s degree program on climate change and sustainable development.
Ministry of Natural Resources and Environment, Central of Vietnam Association of Clean
Water and Environment, Journal of Environment and Life organize contest "Climate change
with life".
2.5. Research Methods
2.5.1. Data collection methods
Information about the situation and changes of global climate change as well as in Vietnam
through articles, the internet, ... especially the information that is regularly and continuously
updated, the latest information.
2.5.2. Sampling method
[3] Students' knowledge and perceptions about climate change may depend on their age,
gender, family characteristics, and specialized subjects of each faculty, so the selected sample
should be representative. The number of classes selected for the survey is as follows:
Department Course Number of classes

Preschool First year 1
Last year 1
Students of Education First year 1
Last year 1
Economics First year 1
Last year 1
Tourism Culture First year 1
Last year 1
Foreign Languages First year 1
Last year 1
Natural Sciences First year 1
Last year 1
Total 12
Interviewed students and the results of the survey showed that: Students' awareness of the
consequences of environmental pollution leading to the consequences of climate change is
limited and unclear. From there, it is necessary to have a specific action plan to bring the
message to the students of Hanoi Capital University in particular and students in general who
need more attention and specificity. [5]
2.5.3. Interview method
Directly interview students using structured questionnaires. The questionnaire is built so
that students can easily understand and fill out the information. The content of the Questionnaire
covers the entire content of the study:
Questionnaire content Target Kind of question
General information about Evaluate the factors The multiple choice
the student affecting knowledge and question
student awareness
General knowledge of Assess students' Comments / knowledge on
students about climate knowledge climate change are made and
change the question is "true" or
"wrong".
Awareness of students on Assess students' Comments given and
issues related to climate perceptions on issues related to questions on a likert scale (5
change climate change point scale)
2.6. Methods of data analysis
Using SPSS software
- Calculate% of students who answer questions
- Calculate the average score of students by department, by course
- Correlation analysis and assessment of factors affecting students' knowledge and
perceptions about climate change
[4] Data processing and analysis: Data were cleaned and entered using Epidata software.
SPSS software was used to analyze the obtained data. Descriptive statistical analysis using
frequency and proportion distributions for categorical variables. Classification of students'
perception: Correct perception: Correctly answered ≥75% of the questions. Incorrect
perception: Correct answer
3. CONCLUSION
Research topic on awareness and attitude of students at Hanoi Capital University about
climate change in the current period. The study shows that the percentage of students with
correct awareness about climate change is high, but awareness about the impact of climate
change on health is still limited. In particular, the correct awareness of the groups of diseases
caused by the impact of climate change such as mental health-related diseases, malnutrition,
diseases that are available in the country and groups of people vulnerable to climate change
such as people with a low standard of living. low socio-economic status, ethnic minorities, local
people, etc. account for a low percentage. Measures to raise awareness about impacts of climate
change on health for students should be implemented in the coming time, focusing on
improving knowledge about health issues caused by climate change impacts and groups of
people. [3]
REFERENCES
1. Ministry of Natural Resources and Environment (2014), Law on Environmental Protection.
2. Ministry of Natural Resources and Environment (2016), Climate change scenarios and sea level rise
for Vietnam.
3. Decision No. 1670 / QD-TTg dated October 31, 2017 approving the target program to cope with
climate change and green growth in the period 2016-2020.
4. Ministry of Education and Training (2012), Decision No. 1612/QD-BGDDT on approving a set of
reference documents on climate change response and disaster risk reduction.
5. Ministry of Education and Training (2015), A guide to teaching and learning about climate change
response. The project "Adaptation to climate change with children in the center".
KHẢO SÁT, ĐÁNH GIÁ NHẬN THỨC VỀ BIẾN ĐỔI KHÍ HẬU
CỦA SINH VIÊN TRƯỜNG ĐẠI HỌC THỦ ĐÔ HÀ NỘI
Tóm tắt: Biến đổi khí hậu phản ánh sự thay đổi dài hạn về nhiệt độ và các hiện tượng thời
tiết. Những thay đổi này là tự nhiên, nhưng kể từ những năm 1800, tác động của con người là
nguyên nhân chính gây ra biến đổi khí hậu, chủ yếu do đốt nhiên liệu hóa thạch (như lượng
khí thải gây hiệu ứng nhà kính bao trùm trái đất và sẽ giữ lại nhiệt của mặt trời. Điều này dẫn
đến hiện tượng nóng lên toàn cầu và biến đổi khí hậu. Thế giới đang nóng lên nhanh hơn bất
kỳ thời điểm nào khác trong lịch sử được ghi lại.han, dầu và khí đốt) tạo ra khí giữ nhiệt.
Từ khóa: Biến đổi khí hậu; khí gây hiệu ứng nhà kính (khí nhà kính: CO2;CH4;CFC;..) các
hệ sinh thái.nhiên liệu hóa thạch
STUDY ON NATURAL FLOOR CLEANER FROM SOME HERBALS
Ngo Thanh Son*, Bui Van Nam

Abstract: The study presents synthesizing natural floor cleaner from soapberry, essential oil
of cinnamon, coriander, and lemongrass. Different composite ratios are given and surveyed
through user opinions to determine the optimal ratio. The natural floor cleaner product
achieves the criteria of being clean, antiseptic, non-sticky, and slight scent at the optimal rate
of 1L of floor cleaning solution, including 300ml of soapberry: 300ml of cinnamon: 380ml of
distilled water: 10ml glycerol: 5ml coriander essential oil: 5ml lemongrass essential oil. The
optimal floor cleaning sample was analyzed according to TCVN 12589:2018- Floor cleaner
– technical requirements and test methods.
Keywords: floor cleaner, soapberry, cinnamon, coriander, lemongrass.

(*) Email: ntson@daihocthudo.edu.vn
1. INTRODUCTION
Today, chemical floor cleaners are popular, with ingredients from chemical compounds,
with many advantages, such as the ability to clean and disinfect quickly and cleanly [1-4]. In
addition, it has a long-lasting scent and good preservation ability. However, according to many
studies, using chemical floor cleaners for a long time can affect humans, for example, causing
allergies, skin irritation (especially for children), breathing problems, some liver and lung
diseases, and potential cancer risk. Moreover, the wastewater after using these floor cleaners
can affect the water environment and the wastewater system. Many herbs have cleansing,
antiseptic and scenting properties [5], but the following have outstanding properties such as:
Soapberry: Soapberry contains Sapindus saponin (C41H61O13) which has strong cleansing
power. So, soapberry is often used to make some detergents such as washing clothes, washing
dishes, cleaning floors, washing hair...
Cinnamon: The most crucial ingredient in cinnamon is essential oil. The cinnamon
essential oil contains no less than 85% cinnamic aldehyde and several other compounds, such
as diterpenoids, phenyl glycosides, flavonoids, tannins, and coumarins. The cinnamon essential
oil has antiseptic effects, stimulates the nervous system and digestion, enhances blood
circulation, and promotes intestinal peristalsis.
Coriander: The composition of coriander essential oil includes compounds such as
borneol, cineole, cymene, dipentene, linalool, phellandrene, pinene, terpineol, and terpinolene,
which have the effect of reducing stress, relaxing, reducing pain, eliminating unpleasant odors.
Lemongrass: Lemongrass essential oil has a fresh lemon scent mixed with lemongrass.
Lemongrass essential oil contains 65-85% citral content, vitamins and minerals. It should have
antibacterial and reduce headaches.
The research aims to create a user-friendly and environmentally friendly floor cleaning
product; it has investigated the properties of some natural herbs such as soapberry, cinnamon,
coriander, and lemongrass.
2. MATERIALS AND METHODS
2.1. Methods
2.2.1. Literature review
Overview of primary and secondary documents related to:
- Chemical floor cleaner, organic/natural floor cleaner.
- Natural herbs with cleansing, antiseptic, and fragrance properties
- Some processes for synthesizing organic floor cleaners
- Characteristics and properties of soapnut extract, cinnamon essential oil, lemongrass
essential oil, coriander essential oil
2.2.2. Experiments: pre-treatment material
• Soapberry:
- Soapberry has a high Saponin content, about 52-60% [6], completely soluble in water and
without impurities, was imported for research use.
- Soapberry will then experiment to determine the dilution ratio with distilled water with
the following criteria: the foaming level after shaking is moderate; the stone and wood floor
must be clean, no residue, no stickiness. All criteria will be determined by sense.
- Soapberry dilutes with the following ratio of water/soapnut extract: 1:100; 1:200; 1:300;
1:400; 1:500.
Based on the use and visual evaluation results after testing (mentioned in section 3), it is
determined the optimal high-dilution ratio of soapnut and water.
• Cinnamon essential oil:
- Cinnamon bark is washed and then boiled in water at a temperature of 100oC, high
pressure for 5 hours. Then cool and filter, and a cinnamon essential oil solution is obtained
[6,7].
- Cinnamon essential oil does not need pretreatment but is used in experiments.
• Lemongrass and coriander essential oil
- Preparation of essential oils by direct distillation method.
2.2.3. Synthetic natural floor cleaning solution
Two main components determine natural floor cleaner: The solution has cleaning and
disinfecting properties: diluted soapberry solution and cinnamon essential oil.
Fragrance part: Includes lemongrass and coriander essential oil
For solutions with cleaning and disinfecting properties: two solutions from soapberry and
cinnamon essential oil in the ratio of volume: 1:1; 1:2; 2:1; and 1:3. Based on sensory evaluation
and opinion of the research team, the optimal ratio will be proposed. After having the optimal
ratio of soapberry and cinnamon essential oil solution, the research team further diluted it with
distilled water at the ratios 1:0.5; 1:1; 1:1,3; 1:1.5, and 1:2. When finding the optimal ratio will
add lemongrass essential oil and coriander to create aroma. The organic floor cleaner was tested
at the ratios of soap : cinnamon : distilled water : lemongrass essential oil : coriander essential
oil as follows 30:30:38:1:0; 30:30:38:0:1; 30:30:38:1:1; 30:30:38:1:2; 30:30:38:2:1 and
30:30:38:2:2. On that basis, the organic cleaning solution has been determined the optimal ratio
through laboratory experiments. However, because many senses express the results, the authors
made a questionnaire to survey users.
2.2.4. Questionnaire
• Number of survey questionnaire: 15
• Floor cleaner at the ratio 30:30:38:1:0; 30:30:38:0:1; 30:30:38:1:1; 30:30:38:1:2;
30:30:38:2:1 and 30:30:38:2:2 are denoted as A, B, C, D, E and F, respectively.
• The survey form has the following contents:
I. Sensory evaluation of cleaning effect, repelling insects and scent
• Foam level when mixing 100ml of solution with 5L of water? (Less – Moderate – Many
– Other)
• Is the surface of the material clean after wiping? (Yes- Not - Other)
• Is there any stickiness or grit after cleaning? Use hand to feel (Sticky - Nonsticky - Other)
• Is the scent pleasant? is it hot? (Comfortable 1- Comfortable 2 - Comfortable 3-
Uncomfortable - Other, where 1 to 3 represents an increasing degree of comfortable)
• Does the scent last long? (15 minutes - 30 minutes - 1 hour - 2 hours - 3 hours)
II. Changes in use over time
• Under good storage conditions, does the scent decrease? (Reduce a lot - Reduce a little
- No reduction - Other)
• Does the bottom of the bottle see sediment? (Many – Less – None – Other)
III. Impact on user's health
• Does it cause skin irritation? (Yes – No – Other)
• Children have irritation, rash? (Yes – No – Other)
• Have breathing problems (within 1 week of use) (Yes – No - Other)

2.2.5. Sampling analysis
The floor cleaning sample is shaken well, bottled 1L, sent for analysis at the Technical
Center for Standards and Quality 1 (Quatest1), according to TCVN 12589:2018
3. RESULTS AND DISCUSSION
3.1. Composition and properties of natural floor cleaner
3.1.1. Dilution ratio of soapberry and cinnamon essential oil
Soapberry was diluted with distilled water in different proportions and the results shown
in Table 1.
Table 1. Dilution ratio
Ratio 1:100 1: 200 1:300 1:400 1:500
Foam level Many Many Moderate Less Less
Stone and Clean, Clean, a little Clean and Clean and dry Clean and
wood stickiness stickiness dry dry
surface after and sticky
wiping marks
Colour Dark brown Dark brown Pale brown Pale brown Pale brown
In Table 1, it can be seen that when increasing the dilution between soapberry and distilled
water, the foaming level decreases, specifically at 1:100 and 1:200 samples give much foam
level, while at 1:400 and 1:500, there is little foam. Therefore, with the criterion of moderate
foam, the 1:300 sample rate is the most moderate. In addition, the 1:300 sample gives better
cleanliness, dryness, and color than the rest. Therefore, a dilution ratio of 1:300 was chosen for
the subsequent experiments.
3.1.2. Natural floor cleaner ingredients
Cleaning and disinfecting ingredients
From the following table, it shows that the cleaning effect at the ratio 1:1 and 2:1 gives
good effect, however, the foaming ratio at the ratio 2:1 is high, so the 1:1 ratio was chosen for
the next experiments.
Table 2. Effective ratio of soapberry and cinnamon solution
Ratio Colour Cleaning Foam level Smell
(soapberry/cinnamon) ability
1:1 Sepia Good Moderate Light
1:2 Red brown Weak Less Strong smell
of cinnamon
1:3 Red Weak Less Strong smell
of cinnamon
2:1 Brown Good Many Not clear
The cleaning efficiency of soapberry and cinnamon diluent is shown in the following table:
Table 3. Effective ratio of soapberry, cinnamon and water solution
Ratio Cleaning Foam
Colour Smell
(soapberry/cinnamon/water) ability level
1:0,5 Sepia Good Many Light
1:1 Sepia Good Moderate Light
1:1,3 Sepia Good Moderate Light
1:1,5 Sepia Moderate Less Light
1:2 Sepia Moderate Less Less
After being diluted with distilled water at different rates, soapberry and cinnamon solutions
retain their reddish-brown color. The cleaning efficiency is inversely proportional to the amount
of distilled water, the more diluted, the lower the efficiency. The excellent effect is shown in
the ratios 1:0.5, 1:1, and 1:1,3. The degree of foaming is also inversely proportional to the
degree of dilution. At the ratio of 1:1 and 1:1.3 for medium and moderate foaming. In addition,
at these two ratios, there is a slight odor. For economic optimization, the research team chose a
1:1.3 ratio of soap, cinnamon, and distilled water solutions for flavoring experiments in the
following section.
Ingredients for cleaning, disinfecting and flavoring
The following table shows the effectiveness of organic floor cleaner at different ratios of
lemongrass essential oil: coriander essential oil. It can be seen that only the smell of organic
floor cleaners changes when adding flavoring ingredients. Only the aroma ratio sample is
30:30:38:1:1, the aroma is pleasant while the other samples show the strong smell of lemongrass
or smell. With the given results, the research team found that the ratio of 30:30:38:1:1 achieved
optimal efficiency in terms of cleaning, disinfecting and aromatherapy effects.
Table 4. Effective natural organic floor cleaner
Ratio Colour Cleaning ability Foam level Smell
30:30:38:1:0 Sepia Good Moderate Strong smell of
lemongrass
coriander
30:30:38:1:1 Sepia Good Moderate Pleasant
coriander
lemongrass
lemongrass
3.2. User survey results

30 survey questionnaires were distributed and 15 questionnaires were collected. In which,
12 votes were for wooden floor (accounting for 40%) and 18 votes were for stone floor
(accounting for 60%).
Sensory evaluation of cleaning effect, repelling insects and scent
The foaming when mixed with 100ml of solution with 5L of water of 6 test samples is
shown in the following figure.
100
90
80
Foaming level (%)
70
60
50 Less
40 Moderate
30
Many
20
10
0
A B C D E F
Sample
Figure 1. Foaming level of 6 samples

All six samples were assessed as having moderate foaming and accounted for more than
80% of the respondents, of which samples C and E had more than 90%.
Figure 2 shows the cleanliness and stickiness of the 15 rating sheets. 100% of the survey
samples were rated as clean efficiency. 5/6 samples did not have a sticky or gritty feeling after
wiping, while sample F had 1 vote for having a sticky feeling.
E
Sample
D Clean
C Not clean
B Sticky
Non-Sticky
A
0 50 100 150 200 250

Clean and sticky level (%)
Figure 2. Degree of cleanliness and stickiness after wiping

The degree of the pleasantness of the scent is shown in Figure 3. No questionnaire was
filled out as unpleasant. Samples A, B, C, and E are rated as having a pleasant scent at the
highest level of 3, in sample C is found to be the most pleasant by 80% of users. This assessment
is vital because the samples differ in the percentage of aromas, so this is an essential factor in
evaluating the odor effectiveness of organic floor cleaners.
90
80
70
Confortable (%)
60
50 Comfortable-1
40 Comfortable-2
30 Comfortable-3
20 Uncomfortable
10
0
A B C D E F
Sample
Figure 3. Comfort level when using 6 samples

Figure 4 shows the level of incense retention in 6 samples. 100% of samples do not show
fragrance retention from 1 to 3 hours. The scent lasts up to about 30 minutes. This time can be
quite long because the floor area is often large, so it is easy to smell. The scent may last longer
in a narrow space environment such as the floor of a toilet or bathroom. Sample B is more than
90% rated as the longest-lasting scent, possibly due to the nature of the scent essence.
100
90
80
70
A
Insence (%)
60
B
50
C
40
D
30
E
20
F
10
0
15 mins 30 mins 1 hours 2 hours 3 hours
Time
Figure 4. The level of scent retention of 6 samples

Changes in use over time

According to Figure 5, none of the samples was evaluated as having a significant reduction
in scent. Over 80% of samples scents did not decrease under good storage conditions. Samples
C and E nearly 100% agree that the scent is not reduced.
120
Reduce incense retention (%)
100
80
60 Reduce a lot
Reduce a little
40
No reduction
20
0
A B C D E F
Sample
Figure 5. Degree of scent change

100% of the survey questionnaires found little sediment in 6 cleaning samples when left
from the first to the last day of the survey. The residues are minimal in size and do not cause
discomfort to the user.
Impact on user's health
The survey questionnaires did not find any health effects after using 6 samples of water for
a long time, namely, no skin irritation; Children do not have irritation, rash, no respiratory
problems (within one week of use).
In addition to the above survey ideas, people feel that combining the scent of samples with
lemongrass and coriander also makes the mind more comfortable and the body more
comfortable when using chemical products. Especially when wiping in an air-conditioned
room, the scent feels longer and more pleasant.
Thus, after surveying 30 people using 6 product samples, the research team found that
sample C (ratio of soapberry: cinnamon : distilled water: lemongrass essential oil: coriander
essential oil is 30:30:38: 1:1) is a highly rated product. Based on the survey results, the research
team made the following changes to the product:
- Conduct 2-layer filtration with the solution to reduce the rate of sediment
- Add Glycerol to increase preservation and reduce the evaporation rate
Then determine the sample phase ratio as follows:
1L floor cleaning solution includes: 300ml soapberry: 300ml essential oil: 380ml distilled
water: 10ml Glycerol: 5ml coriander essential oil: 5ml lemongrass essential oil
3.3. Test results according to TCVN

The test results for the floor cleaning sample are shown in the following table. The results
in the table show that the organic floor cleaner sample did not exceed the allowable standards,
TCVN 12589:2018, Floor cleaner – technical requirements and test methods
Table 5. Test results
No. Target name Unit Test method Results
1 Status Liquid, homogeneous,
without layering or
precipitation at temperatures
below 20oC
2 Color, smell Brown, clear
Characteristic scent
3 Asen Conc. (As) mg/kg ACM THA 05+ No detect (LOD = 0,1)
AOAC 2015.01
4 Metal Conc. mg/kg TCVN 6971:2001 <2,0
TCVN 12589:2018
5 Residue content, % g/ml ASTM D 1290 0,035
mass
6 Content of non- g/ml ASTM D 2834
volatile matter (total
solids), % by mass
7 pH pH meter 6,5
8 Stability at 50oC for No precipitation
8h
4. CONCLUSION
Organic floor cleaner is synthesized from soapberry, cinnamon, lemongrass essential oil,
and coriander essential oil, which are ingredients of natural origin, containing active ingredients
capable of cleaning and scent. Fragrant, suitable for human health, such as saponins, cinnamic
aldehydes, citral, and geraniol. The optimal ratio for floor cleaning water is: 1L of floor cleaning
solution including 300ml soapberry: 300ml essential oil: 380ml distilled water: 10ml Glycerol:
5ml coriander essential oil: 5ml lemongrass essential oil.
Natural floor cleaner with ingredients of natural origin does not contain harmful chemicals,
does not cause skin irritation or some respiratory diseases, and is safe for users, especially
children. The scent of organic floor cleaner comes from natural ingredients and aromatic
essential oils; although not durable over time, it is entirely non-toxic and suitable for human
health. In addition, compounds in organic floor cleaning water, when discharged into the
environment, are easy to decompose, causing no pollution to water sources and the
environment.
REFERENCES
1. Do Tat Loi (2004), Vietnamese medicinal plants and herbs, Times Publishing House.
2. Thai Doan Tinh, Vu Quoc Trung (2005), Experimental organic chemistry, Vietnam Education
Publishing House.
3. Le Thi Anh Dao, Dang Van Lieu (2010), Practice organic chemistry, Pedagogical University
Publishing House, 2010
4. Duong Tan Nhut (2011), Plant biotechnology: Basic and applied research, Agriculture Publishing House.
5. Nguyen Minh Thao (2014), Organic synthesis, Hanoi National University Publishing House.
6. Nguyen Van Han (2015), Medicinal Extraction Technique, Medicine Publishing House.
7. A.T Soldatenkov, N.M Kolyadina, Le Tuan Anh (2017), Applied Organic Chemistry, Knowledge
Publishing House.
8. TCVN 12589:2018, Floor cleaner – technical requirements and test methods
9. TCVN 6971:2001, Synthetic washing water for the kitchen.
NGHIÊN CỨU TỔNG HỢP NƯỚC LAU SÀN HỮU CƠ

TỪ MỘT SỐ LOẠI THẢO DƯỢC THIÊN NHIÊN
Tóm tắt: Nghiên cứu trình bày quy trình tổng hợp nước lau sàn hữu cơ từ cao bồ hòn, tinh dầu
quế, tinh dầu mùi già và tinh dầu sả chanh. Các tỷ lệ tổng hợp khác nhau được đưa ra và khảo
sát qua ý kiến người sử dụng để đưa ra tỷ lệ tối ưu. Sản phẩm nước lau sàn hữu cơ đạt được tiêu
chí sạch, sát khuẩn, không bám dính và mùi hương nhẹ lưu lau ở tỷ lệ 1L dung dịch lau sàn bao
gồm: 300ml dd cao bồ hòn: 300ml dd vỏ quế: 380ml nước cất: 10ml Glycerol: 5ml tinh dầu Mùi
già: 5ml tinh dầu sả chanh. Mẫu nước lau sàn tối ưu đã được phân tích theo Tiêu Chuẩn Việt
Nam (TCVN) 12589:2018, Nước lau sàn – yêu cầu kỹ thuật và phương pháp thử
Từ khóa: Nước lau sàn, cao bồ hòn, tinh dầu quế, tinh dầu mùi già và tinh dầu sả chanh
RESEARCH ABILITY TO CREATE CALLUS AND

REGENERATION PANAX BIPINNATIFIDUS (PANAX
BIPINNATIFIDUS) IN VITRO CULTURE
Nguyen Nhu Toan1*. , Luu Ngoc Sinh1, Nguyen Th Binh1, Tran Đang Khanh2
1
2
Agricultural Genetics Institute
Abstract: Vu Diep ginseng is known to people as Tam That Wild, Tam That Leaf Split, Hoang
Lien That, Tam That Lobe split bird feathers twice, Vu Diep Tam That, Ginseng Twice Split,
Bamboo Blood Ginseng but no ginseng. Many international scientists note research on it.
Studies show that Vu Diep ginseng contains a number of medicinal substances that are
beneficial to health such as: saponin triterpen, Saponin A, B, C, D, reducing sugar, oleanolic
acid and 16 amino acids such as lysine, cysteine, histidine, valine, phenylalanine, leucin,
isoleucin, proline and inorganic substances such as Fe, Ca. In which, experts said that Vu
Diep ginseng contains many compounds similar to ginseng. In particular, the leaves and roots,
and flowers of Vu Diep ginseng contain saponoside compounds of the dammaran group.
Vietnam is researching as well as producing, trying to awaken the medical and economic
value of Ginseng Vu Diep. Our studies have initially determined the environment, influencing
factors and the ability to create callus as well as the regeneration process of Invitro-
environmental plants.
Keywords: Panax, Invitro, Callus, Embryo, Invitro, MS.

(*) Email: nntoan@daihocthudo.edu.vn
1. INTRODUCTION
1.1. Materials and Methods
Ginseng Vu Diep is known to people as Tam That Wilderness… but not many international
scientists pay attention to research on it. Studies show that Vu Diep ginseng contains a number
of medicinal substances that are beneficial to health such as: saponin triterpen, Saponin A, B,
C, D, reducing sugar, oleanolic acid and 16 amino acids such as lysine, cysteine, histidine,
valine, phenylalanine, leucin, isoleucin, proline and inorganic substances such as Fe, Ca. In
which, experts said that Vu Diep ginseng contains many compounds similar to ginseng. In
particular, the leaves and roots, and flowers of Vu Diep ginseng contain saponoside compounds
of the dammaran group. Vietnam is researching as well as producing, trying to awaken the
medical and economic value of Ginseng Vu Diep. In the research directions, the direction of
tissue culture has really brought agriculture to an advanced stage, so if you want to research
and develop Vu Diep ginseng in a modern direction, bringing high economic efficiency, you
cannot ignore it. through this technique. In fact, Vu Diep ginseng has been successfully
propagated from seeds and tubers... But going one step further to produce ginsenoside Vu Diep
ginseng by tissue culture, almost no research works have been published. . With the desire to
learn about this plant of economic value along with the cell culture technologies that have been
and are being implemented for the purpose of propagation and production of compounds of
economic value, we develop Research on the topic: "Study on the ability to create callus and
regenerate seedlings of Vu Diep ginseng (Panax bipinnatifidus) in invitro culture medium".
The required purpose of the topic: Determining the ability to create callus, regenerate
shoots and root in the process of creating seedlings of Panax ginseng (Panax bipinnatifidus) by
tissue culture method; Creating quality seedlings, serving the needs of mass production of
medicinal ginseng in a number of mountainous districts (Ba Vi, Soc Son) of Hanoi city and
northern mountainous provinces
1.2. Material
Parts of Vu Diep Ginseng.
- Biological characteristics of Vu Diep ginseng plant
+ Scientific name: Panax bipinnatifidus
+ Family: Araliaceae family
+ Other names: Tam That leaves sawed, Hoang Lien ventricular, Tam That lobe split bird
feathers twice, Vu Diep Tam That, Ginseng twice split, Bamboo details ginseng
It is a perennial herbaceous plant with a height of 10-20 cm, sometimes growing to a height
of 50 cm. Compound leaves with stalks 6-8 cm long, hairless. Flowers grow in clusters at
axillary stalks, white. The berries are a type of berry that usually grows in clusters and has a
spherical shape. Inside the fruit contains 1-2 seeds and when ripe is red. The tubers are long,
the inner intestine is yellow, white or purple. Wild tamarind is usually found in moist forests
with altitudes from 1900 to 2400 m. The tree is commonly distributed in North Vietnam (many
in Lao Cai) and Southern China. Parts Used: Root tubers. Harvesting and processing: The roots
of perennial plants after being harvested will be washed and then dried or dried. Wild sage
contains many saponins. In addition to these components, the plant also contains many
medicinal substances similar to those in Ngoc Linh ginseng.
Environment:
Using MS background environment. Also added: 1.0 mg/l 2,4-D and 0.2 mg/l TDZ.
The culture medium was adjusted to pH = 5.8. The medium was sterilized by autoclave at
1210C, 1atm pressure.
Equipment and tools:
Room for preparation of medium, sterilization of culture medium, preservation of mother

solution. Includes environmental autoclave, refrigerator, electric stove, analytical balance,
measuring tube, pipette, pH meter.
- Aseptic inoculation room includes plant cabinets, UV lamps, autoclaved instruments.
- Cold room for culture includes iron shelves, lights, thermometers, air conditioners.
- Tools include alcohol lamp, plate, sample cutter, scissors, cotton ball, 250 ml, 500 ml
sterile test tube bottles, sterile paper, elastic band.
Chemistry:
Alcohol 960, 700; Sterile distilled water; Javel solution; Dilute soap solution.
- Subsequent studies use the results of previous studies such as callus, shoots, etc.
Figure 2.1. Sam Vu Diep callus Figure 2.2. Vu Diep Ginseng Buds
Figure 2.3. Vu Diep Ginseng Root Figure 2.4. Sam Vu Diep clonal embryo
2. RESEARCH METHODS
- The experiments are deployed and conducted according to the general procedure
including:
+ Scar tissue culture
+ Regeneration of shoots from callus
+ Root culture
+ Cultivation of biomass
- Plant regeneration through somatic embryogenesis
Statistical analysis
+ Figures are calculated using Excel software.

+ Applying SAS software (2008) to analyze and compare experimental results.
+ The means were separated on the basis of the least significant differences (LSD) at the
0.05 probability level.
3.1. Investigate the conditions affecting the culture of Vu Diep ginseng
3.1.1. Effects of some disinfectants on the culture:
For Sam Vu Diep, the experimental part is the head (germ, stem, and root).
Select straight or lateral shoots as culture material with a length of 2-3 cm, remove all
leaves, treat with 70% alcohol for 1 min in a sterile incubator, then rinse 3 times with distilled
water. then treat the sample with calcium hypochlorite or HgCl2 solution and continue rinsing
with sterile distilled water several times to remove all disinfectant. After sterilization, cut the
green head, length 0.4 cm, and place it in the medium.
Table 3.1. Effect of different types of sterilization concentration
Sample
Sample HgCl2
number
10 0.2% 0.3% 0.4%
Sam Vu Diep 5 10 15 20 5 10 15 20 5
(Sprout, stem, Time
root) (minute) 100% 100% 100% 20% 40% 36% Chết Chết Chết
Sample
Sample Ca(OCl)2
number
10 5% 10% 15%
Sam Vu Diep 10 15 20 5 10 15 5 10
(Sprout, stem, Time
root) (minute) 100% 100% 100% 42% Chết Chết Chết Chết
Vu Diep ginseng treated with HgCl2 at a concentration of 0.2% for 20 minutes, the
infection rate was low (20%) and the infection rate was not high (80%). As for the treatment
with Ca(OCl)2 at 10% concentration, for 5 minutes, the infection rate was low (42%) and the
no infection rate was high (58%). (Table 3.1).
3.1.2. The influence of hormone combinations on morphogenesis

Using samples of sprouts, stems and roots of Vu Diep ginseng. Co-culture on different
combinations of media to evaluate morphogenesis.
Table 3.2. Effects of hormone combinations on morphogenesis of Ginseng Vu Diep
Subject Implant parts Hoccmon The result of

morphogenesis
IBA IAA NAA K 2ip G
0.2 0.2 Died
0.5 0.5 Died
0.5 1 Callus
Ginseng Sprouts, Stems,
0.2 0.5 Callus
Vu Diep Roots
0.5 0.25 Roots
0.5 1 1 Somatic embryo
0.5 3 0.5 Bud embryo
6 0.4 Roots
1 3 Buds
The data in Table 3.2 shows that:

- Combinations of IBA 0.2 + 0.2 K and IBA 0.5 + 0.5 K cause death, combination IBA 0.5
+1.0K and combination IAA 0.2 + 0.5K create Callus. The remaining combinations such as
IBA 0.5 + 2ip 1 hormone combination, then Sam Vu Diep generates somatic embryos; The
combination of hormones NAA 1 + G 3 produces shoot embryos and the combination of
hormones NAA 6 + 2ip 0.4 produces roots.
3.1.3. Effect of lighting conditions on the ability to create callus from leaves and petioles
The best medium for initial callus formation from leaf and petiole samples was used to
investigate lighting conditions. Samples were placed in two conditions of complete darkness
and light (16 h/day).
Depending on the type of explant, light may or may not be needed during callus formation.
For leaf samples, in most cases, callus formation in the dark was usually better than in the light.
However, in some cases, the explants produced better callus in bright conditions.
The results in Table 3.3 show that the rate of callus formation on leaf and petiole samples
is almost equivalent between the two light and dark conditions, but the amount of callus in the
dark condition is less and the callus quality is also poor. due to vitreous phenomenon, especially
the medium with 3.0 mg/l 2,4-D.
Table 3.3. Effect of lighting conditions on the ability to create callus from leaves and petioles
Percentage of callus formation (%)
2,4-D (mg/l) Part
Lighting (16 hours/day) Totally dark
0.5 20 30
1.0 Leaves 90 80
2.0 90 90
3.0 80 80
0.5 100 100
1.0 Petiole 100 100
2.0 100 100
3.0 100 100
3.1.4. Effect of initial explant size on callus proliferation
Callus after proliferation was used for shoot regeneration and adventitious roots.
Callus was cut in three different sizes, respectively: KT1, KT2, KT3. Callus samples with
defined size were inoculated into rapid multiplication medium.
Table 3.4. Effect of initial explant size on callus proliferation
KT1 KT2 KT3
Observation criteria
(0.5*0.5) (0.8*0.8) (1.0*1.0)
Original fresh weight (mg) 139 ± 8 268 ± 12 488 ± 19
Dimensions (cm) 1.1*0.9 1.2*1.0 1.4*1.2

Biomass after
4 weeks of Fresh weight (mg) 626 ± 38 812 ± 32 1516 ± 62
culture
Dry weight (mg) 51.9 ± 31 55.8 ± 2.3 112.6 ± 4.7
Dry matter percentage (%) 8.18 7.08 6.59
Dry biomass growth rate 5.44 3.28 2.64
The explant size is an important factor in in vitro propagation. When investigating the effect
of initial explant size on callus proliferation, we found that the smallest size (KT1) gave the best
effect in terms of both biomass growth and dry weight, while not there was a big difference in
proliferation ability between KT2 and KT3 (Table 3.4). This correlation can be derived from the
correlation between the explant size - the ability to obtain nutrients from the medium and due to the
influence of endogenous waste products of the callus during the culture process.
3.1.5. Effect of auxin on the ability to initiate callus from leaves and petioles
Studies available on subjects of the genus Panax have shown that the callus initiation phase
often involves a combination of cytokinin and auxin.
After sterilization, leaf and petiole samples were inoculated into MS medium supplemented
with 0.2 mg/l TDZ and auxins 2,4-D, IBA, NAA, with concentrations varying from 0.5; 1.0;
2.0 and 3.0 mg/l. The leaf specimen was placed face up on the medium and the petiole was also
placed face up (Cut facing up). The results obtained after 8 weeks of culture are recorded in
Table 3.5. Of the three types of auxins added to the medium, only 2,4-D was able to stimulate
leaves and petioles to create callus. On medium supplemented with 1.0 mg/l 2,4-D, the explants
had the highest rate of callus formation (reaching 90% for leaves and 100% for petioles), with
the highest amount of scar formation. , firm structure and bright yellow color. At the
concentration of 3.0 mg/l 2,4-D, the scar tissue started to show vitreous phenomenon.
Therefore, at a concentration of 2,4-D of 3.0 mg/l or more, it is not suitable for callus generation.
Table 3.5. Effect of auxin on the ability to initiate callus from leaves and petioles
Scar tissue formation rate (%)
Auxin Auxin concentration (mg/l)
Petiole Leaves
0.5 100 20
2,4-D 1.0 100 90

2.0 100 90
3.0 100 80
0.5 0 0
1.0 0 0
IBA
2.0 0 0
3.0 0 0
0.5 0 0
1.0 0 0
NAA
2.0 0 0
3.0 0 0
3.1.6. Effect of auxin on the ability to proliferate the callus of Sam Vu Diep
Callus samples generated from the initiation stage were inoculated into MS medium
supplemented with 0.2 mg/l TDZ and auxins 2,4-D, IBA and NAA with concentrations varying
from 0.5; 1.0; 2.0; 3.0 and 5.0 mg/l under irradiation conditions 16 h/day. The results obtained
in Table 3.6 showed that: After the proliferation process, callus cultured on medium with 0.5
mg/l IBA had the highest dry matter ratio of 9.62% but the highest dry mass growth rate was
4.56 times. was obtained in callus on medium with 2,4-D at a concentration of 1.0 mg/l. It is
possible that the combination of auxin and cytokinin increased the ability of the callus to obtain
sugars and other nutrients from the callus environment and led to the proliferation of the callus,
especially the dry matter ratio. IBA may be an auxin that stimulates nutrient uptake from the
environment better when combined with TDZ than NAA and 2,4-D. As a result, the dry matter
ratio of callus cultured on medium containing IBA was the highest among the three auxins used.
Although IBA gave the highest percentage of callus with the highest percentage of dry matter,
2,4-D had the highest dry matter growth rate (4.56 times) and a relatively high dry matter rate
(8.18%). On the other hand, callus on 2,4-D medium has the best morphology, which is a form
of callus with high regenerative capacity.
Table 3.6. Effect of auxin on the ability to proliferate the callus of Sam Vu Diep
Biomass after 4 weeks of culture
Original Dry
Concentration
fresh biomass
Auxin (mg/l) Fresh Dry Dry matter
weight growth
weight weight percentage
(mg) rate
(mg) (mg/l) (%)
43.3 ±
0.5 203 ± 16 584 ± 34 7.42 3.18
2.5
66.2 ±
1.0 212 ± 14 809± 37 8.18 4.56
2,4-D 3.0
2.0 204 ± 17 711 ± 32 52.4 ± 2.4 7.37 3.73
3.0 205 ± 9 508 ± 24 36.6 ± 2.2 7.21 2.65
5.0 201 ± 13 493 ± 38 34.6 ± 1.7 7.01 2.50
0.5 197 ± 18 474 ± 23 45.6 ± 2.2 9.62 3.45
1.0 203 ± 19 532 ± 29 48.6 ± 2.7 9.14 3.56
IBA 2.0 207 ± 13 631 ± 32 49.5 ± 2.5 7.84 3.63
3.0 203 ± 15 552± 26 41.1 ± 1.9 7.45 3.10
5.0 209 ± 12 531 ± 23 35.3 ± 1.5 6.66 2.53
0.5 218 ± 8 485 ± 13 41.2 ± 1.1 8.49 2.81
1.0 212 ± 14 548 ± 21 45.0 ± 1.8 8.22 3.33
NAA 2.0 206 ± 15 588 ± 18 46.6 ± 1.4 7.92 3.37
3.0 199 ± 7 602 ± 32 45.7 ± 2.4 7.60 3.38
5.0 205 ± 14 720 ± 48 51.6 ± 3.4 7.20 3.77
3.2. Some factors affect the ability to regenerate shoots from callus
3.2.1. Effect of BA and NAA on shoot regeneration from callus
Callus obtained in the callus rapid multiplication experiment were separated and
transferred into ½ MS medium supplemented with BA and NAA with concentrations in Table
3.6. The ratio between auxin and cytokinin is essential for shoot regeneration, cytokinin usually
promotes shoot formation and this process is often stimulated by the addition of auxin at low
concentrations. In the trial, when using BA in combination with NAA, the results showed that
different combinations of NAA and BA, the combination of 1.0 mg/l BA and 1.0 mg/l NAA
gave the highest number of shoots at 6.3 buds/ sample and the mean weight was 0.185 g (Table
3.7).
Table 3.7. The ability to regenerate shoots from callus on MS . medium with additional BA
and NAA
Number of
BA (mg/l) NAA (mg/l) Bud weight (g)
shoots/sample
0.5 5.0 0.106
1.0 6.1 0.141
0.5 1.5 4.6 0.193
2.0 3.3 0.197
2.5 3.0 0.094
0.5 5.5 0.163
1.0 6.3 0.185
1.0 1.5 5.9 0.158
2.0 3.9 0.148
2.5 3.7 0.157
0.5 4.2 0.152
1.0 5.5 0.141
2.0 1.5 2.9 0.144
2.0 2.8 0.112
2.5 2.7 0.108
0.5 3.3 0.154
1.0 3.0 0.122
4.0 1.5 2.6 0.122
2.0 0.8 0.108
2.5 0 0
3.2.2. Effect of BA on shoot growth of Vu Diep ginseng invitro
The best shoots after collection were separated and transferred to ½ MS medium
supplemented with 1.0 g/l activated carbon, 30 g/l sucrose, 0.5 mg/l NAA and BA (0.5; 1.0;
2.0; 4.0 mg/l).
Table 3.8. Effect of BA on growth Vu Diep ginseng buds invitro
BA (mg/l) Trọng lượng tươi (g) Chiều cao chồi (cm) Số lượng lá/ chồi
0.5 0.61 5.66 3.0

1.0 0.87 6.16 3.3
2.0 0.72 4.11 4.0
4.0 0.71 4.33 3.9
Of the BA concentrations used, a concentration of 1.0 mg/l BA combined with 0.5 mg/l
NAA resulted in the best shoot growth with shoot fresh weight of 0.87 g and shoot height of
6.16 cm (Table). 3.8). Therefore, the culture medium supplemented with 1.0 mg/l BA and 0.5
mg/l is best for shoot growth.
3.2.3. Effect of sugar concentration on shoot growth
The best shoots in the experiment were separated and transferred to ½ MS medium
supplemented with 0.5 mg/l NAA, 1.0 mg/l BA, pH = 5.7 and sugar with sugar concentrations
of 10; 20; 30; 40; 50; 60 g/l.
Table 3.9. Effect of sugar concentration on shoot growth
Number of
Sucrose (g/l) Bud weight (g) Bud height (cm)
leaves/buds
10 0.49 4.4 2.2

20 0.55 5.4 2.5
30 0.68 5.7 2.6
40 1.06 5.8 3.2
50 1.46 6.1 3.5
60 1.28 6.1 3.2
The test results show that sucrose is the predominant soluble carbohydrate and the
commonly used concentration is in the range of 30 - 120 g/l sucrose. Studying the effect of
sucrose on shoot growth of Vu Diep ginseng after 90 days of culture showed that the addition
of sucrose to the culture medium had a positive effect on shoot growth. The increase in sucrose
concentration in the medium not only stimulates the growth of Vu Diep ginseng shoots but also
has a strong effect on their weight change. A concentration of 50 g/l sucrose gave the best
results in terms of weight, height and number of leaves (Table 3.9).
3.2.4. Effect of activated carbon on shoot growth in vitro
The best shoots in the experiment were separated and transferred to ½ MS medium
supplemented with 0.5 mg/l NAA, 1.0 mg/l BA, pH = 5.7 with activated carbon concentrations
of 1.0, respectively; 2.0; 3.0 and 4.0 g/l.
Activated carbon is not a plant growth regulator, but it has the ability to change the
composition of the medium. Activated charcoal regulates the pH of the environment, absorbing
substances that interfere with tissue growth. The obtained results showed that when the
concentration of activated carbon increased, there was a clear change in the weight as well as
the height of the shoots, but the number of leaves did not change significantly.
The highest shoot weight on the medium containing 2.0 g/l activated carbon was about 1.01
g/bud, an increase of 1.9 times compared to the control (Table 3.10). So the concentration of
2.0 g/l activated carbon is the most suitable for the proliferation of Vu Diep Ginseng buds.
Table 3.10. Effect of activated carbon on shoot growth in vitro

Activated carbon (g/l) Bud weight (g) Bud height (cm) Number of leaves/buds
0 0.53 3.3 3.6
1.0 0.61 4.6 3.7
2.0 1.01 5.3 3.3
3.0 0.97 6.8 2.7
4.0 0.94 8.5 3.1
3.3. Factors affecting the possibility of uncertain rooting from callus
3.3.1. Effect of IAA, IBA, NAA on the ability of uncertain rooting from callus
Table 3.11. Effect of IAA, IBA, NAA on ability indeterminate rooting from callus
Maximum
Concentration Rooting rate Amount Root
Auxin length of
(mg/l) (%) roots / samples mass/sample (%)
roots (mm)
1.0 30 3.0 ± 0.3 5.98 18
3.0 100 8.7 ± 0.1 21.88 13
NAA
5.0 70 2.6 ± 0.1 6.23 9
7.0 50 2.1 ± 0.1 12.21 8
1.0 0
3.0 0
IAA
5.0 10 0.2 ± 0.2
7.0 0
1.0 70 1.6 ± 0.1 7.83 16
3.0 80 4.0 ± 0.3 5.21 21
IBA
5.0 100 4.8 ± 0.3 15.81 18
7.0 60 3.5 ± 0.1 8.06 1.7
Callus was inoculated into rooting medium containing auxins (NAA, IBA, IAA) at
concentrations of 1.0, respectively; 3.0; 5.0; 7.0 mg/l. During the investigation of the effects of
the above three types of auxins, we found that IAA is not suitable for the rooting of Vu Diep
ginseng from callus, because this auxin hardly stimulates the callus to take root indeterminately.
NAA and IBA are the opposite. Concentration of 3.0 mg/l NAA gave the best results with the
rooting rate up to 100%, the largest number of roots/sample (8.7 roots/sample), the largest root-
mass/sample ratio (21.88 %). The maximum length of the roots is 13 mm (Table 3.11). IBA
concentration at 5.0 mg/l gave a 100% rooting rate, an average number of roots/sample of 4.8
samples, a mass ratio of 15.81% and a maximum root length of 18 mm. This result can be
explained by the fact that the synthetic auxin is more active than the natural form. So two good
rooting stimulators are Ms ½ medium supplemented with 3.0 mg/l NAA and MS ½ medium
supplemented with 5 mg/l IBA.
3.3.2. Effect of IBA and NAA on the possibility of uncertain rooting:

Undetermined roots after being created in the experiments were separated and subcultured
to rooting medium supplemented with auxins NAA, IBA at concentrations of 1.0, respectively;
3.0; 5.0 mg/l.
From the results in Table 3.12 and Table 3.13, the origin of the root sample has a great
influence on the rooting efficiency. The root samples in Table 3.12 had better rooting ability,
all 6 treatments took root, the highest rooting rate was 60%, the highest number of secondary
roots was 9 roots/sample. The root samples in Table 3.13 had the highest rooting rate of 40%,
the highest number of secondary roots was 3 roots/sample, 3 out of 6 rooting treatments but
these samples were inoculated on medium containing NAA.
Table 3.12. Effect of IBA and NAA on rooting ability of samples derived from medium
supplemented with NAA
NAA IBA Rooting rate Number of Average fresh weight
(mg/l) (mg/l) (%) secondary roots (mg/l)
1 - 20 1 140 ± 10
3 - 30 4 290 ± 10
5 - 60 9 390 ± 20
- 1 10 1 450 ± 50
- 3 20 2 330 ± 20
- 5 30 1 280 ± 30
When considering the effect of auxin type, we find that NAA is more suitable for the
uncertain rooting process of Vu Diep ginseng. At the concentration of 5.0 mg/l NAA stimulated
the best root multiplication (60%), had the highest number of secondary roots (9 roots/sample)
and high weight gain (average fresh weight was 390 ± 20 mg/l). , increased 3.5 times compared
to the original). Furthermore, up to 5/6 treatments supplemented with NAA gave rooting results
compared to IBA with only 4/6 treatments. So between IBA and NAA, NAA at a concentration
of 3.0 mg/l is suitable for root induction from callus and NAA at a concentration of 5.0 mg/l is
more suitable for uncertain root multiplication of Vu Diep ginseng.
Table 3.13. Effect of IBA and NAA on rooting ability of samples derived from IBA .
supplemented medium
NAA IBA Rooting rate Number of secondary Average fresh weight

(mg/l) (mg/l) (%) roots (mg/l)
1 - 40 3 350 ± 10
3 - 20 1 180 ± 30
5 - 0 0
- 1 10 1 270 ± 10
- 3 0 0
- 5 0 0
4. CONCLUSION
Vu Diep Ginseng is recognized as one of the ginseng plants with high saponin content and
the highest quantity, compared to other Panax species in the world. Therefore, the research and
application of plant tissue culture technology has brought many practical meanings in
conserving precious medicinal herbs.
- The process of investigating the effect of auxin type and concentration on the ability to
create callus initially of leaves and petioles showed that concentrations of 3.0 mg/l 2.4-D or
higher were not suitable for callus generation from leaves. Sam Vu Diep.
- During shoot growth, the number of shoots regenerated from callus was highest on ½ MS
medium supplemented with 1.0 mg/l BA, 1.0 mg/l NAA, 50 g/l sucrose.
- For rooting from callus, callus samples were cultured on ½ MS medium supplemented
with 3.0 mg/l NAA for the highest rooting rate, highest number of roots and fresh weight ratio.
of the highest root/sample.
- ½ MS medium supplemented with 5.0 mg/l NAA stimulated the best root multiplication,
giving the highest rooting rate and the most branching roots.
REFERENCES
1. Vu Thi Hien, Nguyen Phuc Huy, Bui Van The Vinh, Hoang Xuan Chien, Hoang Thanh Tung, Nguyen
Ba Nam, Vu Quoc Luan, Duong Tan Nhut. 2015. Somatic embryogenesis from leaf transverse thin
cell layer derived-callus of Vietnamese ginseng (Panax vietnamensis Ha et Grushv.). Tạp chí Công
nghệ Sinh học số 1/2015, tr26.
2. Duong Tan Nhut, Nguyen Phuc Huy, Ngo Thanh Tai, Nguyen Ba Nam, Vu Quoc Luan, Vu Thi Hien,
Hoang Thanh Tung, Bui The Vinh, Tran Cong Luan. 2015. Light-emitting diodes and their potential
in callus growth, plantlet development and saponin accumulation during somatic embryogenesis of
Panax vietnamensis Ha et Grushv. Biotechnology and Biotechnological Equipment, 29(2): 299-308.
3. Duong Tan Nhut, Hoang Thanh Tung, Vu Thi Hien, Nguyen Ba Nam, Nguyen Phuc Huy, Vu Quoc
Luan. 2016. Assessment of the possibility of flowering, fruiting and saponin accumulation of somatic
embryo-derived Panax vietnamensis Ha et Grushv plants growing in kon tum and quang nam. Tạp
chí Công nghệ Sinh học 14(1A): 263-268.
4. Dương Tấn Nhựt (2014), Hoàn thiện quy trình nhân giống sâm Ngọc Linh
hướng đến xây dựng thương hiệu Quốc gia, Viện hàn lâm khoa học và công
nghệ Việt Nam, www.vast.ac.vn
5. Ngo Thanh Tai, Nguyen Ba Nam, Ho Thanh Tam, Ha Thi My Ngan, Duong Tan Nhut. two thousand
and thirteen. Studying the effects of LED light on callus proliferation and complete plant formation
from Ngoc Linh ginseng clones (Panax vietnamensis Ha et Grushv.). Proceedings of the
Biotechnology Conference, Hanoi, 1038-1042.
NGHIÊN CỨU KHẢ NĂNG TẠO MÔ SẸO VÀ TÁI SINH CÂY SÂM VŨ ĐIỆP
(PANAX BIPINNATIFIDUS) TRONG MÔI TRƯỜNG NUÔI CẤY INVITRO
Tóm tắt: Sâm Vũ Diệp được con người biết đến với tên gọi Tam thất hoang, Tam thất lá xẻ,
hoàng liên thất, tam thất thùy xẻ lông chim hai lần, vũ diệp tam thất, sâm hai lần chẻ, trúc tiết
nhân sâm nhưng không có nhiều nhà khoa học quốc tế lưu ý nghiên cứu về nó. Các nghiên cứu
cho thấy trong Sâm Vũ Diệp có chứa một số dược chất có lợi cho sức khỏe như: saponin triterpen,
Saponin A, B, C, D, đường khử, acid oleanolic cùng 16 acid amin như lysine, cysteine, histidine,
valin, phenylalanin, leucin, isoleucin, prolin cùng các chất vô cơ như Fe, Ca. Trong đó các chuyên
gia nhận định Sâm Vũ Điệp có chứa nhiều hợp chất giống với nhân sâm. Đặc biệt, các bộ phận
lá và rễ, hoa Sâm Vũ Điệp đều chứa các hợp chất saponosid nhóm dammaran. Việt Nam đang
nghiên cứu cũng như sản xuất, cố gắng đánh thức giá trị y học và giá trị kinh tế của Sâm Vũ Diệp.
Các nghiên cứu của chúng tôi đã bước đầu xác định được môi trường, các yếu tố ảnh hưởng và
khả năng tạo mô sẹo cũng như quá trình tái sinh cây trong môi trường Invitro.
Từ khóa: Sâm, Mô tế bào, Nuôi cấy Invitro, Mô sẹo, Chồi, Môi trường nuôi cấy
USING SOME BIOINFORMATIC TOOLS TO MINING GENES

CODING CELLOBIOHYDROLASE FROM METAGENOME DATA
OF THE BACTERIA SURROUNDING WHITE-ROT FUNGI
(Trametes versicolor) IN CUC PHUONG NATIONAL PARK
Nguyen Thi Binh1*, Le Thi Thu Hong2, Truong Nam Hai2

1
2
Academy of Science and Technology, Vietnam Academy of Science and Technology
Abstract: Cellobiohydrolase (EC 3.2.1.91) is one of the important enzymes involved in

cellulose hydrolysis. In this study, the gene sequences encoding cellobiohydrolase were
extracted from the metagenome DNA data of microorganisms surrounding white-rot fungi in
Cuc Phuong National Park based on the KEGG database. 73 ORFs encoding
cellobiohydrolase were obtained, of which 15 ORFs contained complete genes, 6 ORFs with
functional regions. The expression level of the protein in E. coli was estimated by Periscope
software, which showed that the gene code GL0212614 had the highest expression level of 742
mg/l. The secondary and tertiary structures of GL0212614 were predicted by Phyre2, showing
that the structure of GL0212614 was determined based on c3nfvA template with 46% coverage
and 100% confidence In the secondary structure, there are 25% α helix, 29% β helix, 2% TM
helix and 14% no identify. GL0212614 is an acidic enzyme, the optimal temperature for
enzyme activity is 55°C-65°C. These results are an impotant basis in order to choose gene
expression conditions.
Keywords: Bioinformatics, cellobiohydrolase, DNA metagenome, E.coli, expression level.

(*) Email: ntbinh@daihocthudo.edu.vn
1. INTRODUCTION
Cellulose is one of the most important and popular biomass today. To effectively degrade
this biomass source, it is necessary to participate in cellulase enzymes: endo-1,4-β-D-glucanase
(endocellulase EC 3.2.1.4), exo-1,4-β-D- glucanase (exocellulase or cellobiohydrolase EC
3.2.1.91) and β-glucosidase (cellubiose hydrolase EC 3.2.1.21). Enzymes called endoglucanase
or endocellulase perform cleavage at random points within the cellulose, producing
oligosaccharides of variable size. Exocellulases or cellobiohydrolases act on the terminal ends
of oligosaccharide chains produced by endocellulases, cleaving glycosidic bonds and releasing

glucose or cellobiose [1]. The enzyme β-glucosidase is responsible for breaking down
cellobiose into glucose molecules. Of these three groups of enzymes, cellobiohydrolase is an
important component of the cellulase system and plays a major role in biofuel production from
plant biomass [2]. Cellobiohydrolase is usually produced from fungi but also many bacteria that
contain the gene encoding this enzyme. Because microorganisms have a rather special
cellulosomal system, there are many different studies to study and exploit the gene encoding
cellobiohydrolase on this object such as the gene encoding cellobiohydrolase from Clostridium
clariflavum [2], the gene HmCel6A and variable its variant HmCel6A-3SNP from bacteria in
hot spring area [3], gene Cel6A from Penicillium [4]…
Soil is a potential ecosystem with abundant, diverse microorganisms. This is considered an
important source to search for new enzymes with high efficiency in cellulose degradation [5],
especially the surrounding white-rot fungi The white-rot fungi can effectively metabolize all
the components in the wood. This hydrolysis of white-rot fungi is often associated with
enzymes of bacteria living in the same ecosystem. In the process of fungi decomposing wood,
redox preaction had occurred that acidify the environment, in addition, the fungi are also
capable of producing the environment with secondary metabolic products. Therefore, bacteria
that survive in these conditions must have properties suitable for the environment.
To efficiently exploit genes from microorganisms in different ecosystems, metagenomics
techniques had been used to search for new genes from non-culturing microorganisms. Gene
sequencing yields very large metagenome data. To efficiently exploit these data, bioinformatics
tools were used to screen and predict candidate genes encoding for proteins of interest before
conducting experimental studies. In this study, we present how to use some bioinformatics tools
to mine new cellobiohydrolase enzyme genes from microbial metagenome DNA data
surrounding white-rot fungi in Cuc Phuong National Park.
2. MATERIALS AND METHODS
Resources: The 51.8 Gb metagenome DNA data of the microbial sample residing
surrounding the wood-hydrolyzed white-rot fungi (T. versicolor) in the Cuc Phuong rainforest
was sequenced using the HiSeqIllumina sequencing system ( Illumina, San Diego, USA) at
BGI, Hong Kong.
Research Methods
Prediction of ORFs using MetaGene Annotator (MGA) software: The 51.8 Gb
metagenome DNA data were sequenced, using IDBA software
(http://www.cs.hku.hk/~alse/idba_ud) to sequence the short sequences into 2,611,883
dimensional contigs The mean length was 898 bp and there were 4,104,872 ORFs identified
using the MGA software (http://metagene.nig.ac.jp/metagene/metagene.html). These ORFs
were then compared with the KEGG (Kyoto Encyclopedia of Genes and Genomes) data to find
the ORF sequences encoding cellobiohydrolase.
Prediction of functional regions of ORF using PFAM and HHMER: Pfam

(http://pfam.xfam.org/) is a database of a large collection of protein families and domains. To
predict the functional regions of ORF by Pfam, we provide protein sequences, using e-value
1.0 and provide a personal e-mail address, confirming submission via the HMMerwebsite, the
results will be returned after 2-3 days. HHMer is online software that allows the prediction of
functional regions of proteins in Pfam quickly, based on a representative HMM model
(https://www.ebi.ac.uk/Tools/hmmer/search/phmmer ).
Prediction of protein expression level inferred from ORF using Periscope software:
Protein expression levels in E. coli cells were predicted using the Periscope software available
at http://lightning.med.monash.edu/periscope/. Periscope classifies the expression levels of
soluble proteins into three levels: high, moderate, low, in addition to a predictive function of
the amount of soluble protein in mg/l.
Predicting the spatial structure of proteins: Phyre2 software
(http://www.sbg.bio.ic.ac.uk/~phyre2/html/page.cgi ?id=index) was used. To predict the
higher-order structure of a protein, the user submits the protein sequence to determine the
secondary and tertiary structure of the models, domain composition, and model quality of the
protein. Typical structure prediction results will be returned to the sender's e-mail.
Prediction of some physical properties of proteins: Use the AcalPred software at
http://lin-group.cn/server/AcalPred to predict the acidic or alkaline proteins. Users enter the
target protein sequence into the search box, the software will return results on the acid- or
alkaline-protein in a few minutes. TBI software (http://www.tbi.org.tw/tools/) was used to
predict the optimum temperature of enzyme activity. The inputs to TBI are the amino acid
sequences and the results will be available in a few minutes.
3.1. Prediction of ORFs encoding the enzyme cellobiohydrolase
Based on the KEGG database and using the MGA software, 73 ORFs were predicted
encoding cellobiohydrolase. In which, 15 ORFs (20.55%) contain the entire gene (complete
gene), the remaining 11 ORFs lack the 3' end, 5 ORFs lack the 5' end and 42 ORFs lack both 5'
and 3' ends. In the data analysis of genes encoding cellobiohydrolase enzymes, we prioritized
to select complete ORFs for further analysis
3.2. Analysis of functional regions of ORF
Proteins usually consist of one or more functional regions called domains. Therefore,
searching of domains presented in proteins provided insights into their function. To evaluate
the function of enzymes, we conducted the domains of 15 complete ORFs. Of which, there were
6 ORFs with functional domains: 1 ORF has Alginate_lyase domain, 1 ORF has Amidase 3
domain, 1 ORF has CBM2 domain, 1 ORF has CBP_BcsO domain, 1 ORF has GH128 +
Laminin G3 domain, 1 ORF has domain Znribbon 8 (domains of genes are shown in Figure 1).
These domains were involved in the function of genes. Therefore, in the next prediction we will
proceed on 6 complete ORFs with defined functional regions.
Figure 1. Diagram showing the functional domain of ORFs.

3.3. Prediction of expression levels of genes encoding the enzyme cellobiohydrolase
E. coli is considered to be the most popular recombinant protein expression system today.
Expression of E. coli soluble proteins not only purified target proteins, but also enhanced the
ability to obtain structurally intact and biologically active proteins. The expression level of
soluble proteins was determined by Periscope software. 6 complete ORFs, which were
identified the functional regions, were expressed in E. coli. The results of predicting the
expression levels of 6 ORFs were shown in Table 1.
Table 1. Predicted expression level of cellobiohydrolase gene in E. coli
No Gene code Domain Expression level (mg/l)
1 GL0212614 Alginate_lyase 742,5445
2 GL0221923 Amidase3 9,2954
3 GL2034110 CBM2 9,1483
4 GL0879211 CBP_BcsO 15,3828
5 GL0058533 GH128+ Laminin G3 13,5843
6 GL0733968 Znribbon 0,1945
The results of expression showed that the gene code GL0212614 containing the
Alginate_lyase domain had the highest expression level 742 mg/l. The remaining gene codes
all had low expression levels, which will be difficult for further expression studies. Therefore,
the gene GL0212614 was selected to estimate the properties before further experiments. The
gene sequences and amino acid sequences in GL0212614 are shown in Figure 2.
atg aaa gta att gtt ttc ctg att tta atg gtg gtt cta aac agc tgt tct ttg gct ttt
M K V I V F L I L M V V L N S C S L A F
gcc caa tca ttt gtt cat ccg ggt gga tta cat acc ctc gcc gac tta aac cga atg aaa
A Q S F V H P G G L H T L A D L N R M K
gat atg gtg aag aag cgg gcg cat cca tgg ata gac agt tgg aac aaa ctt atc caa gat
D M V K K R A H P W I D S W N K L I Q D
cca ctt gca caa aac acc tat aca gct gca ccc aag gca aat atg ggc gat agt cgg cag
P L A Q N T Y T A A P K A N M G D S R Q
cgt gca tca acc gat gcg cac gcg gct tat ttg aat gcc ata cgc tgg tac atc aca ggt
R A S T D A H A A Y L N A I R W Y I T G
gat cgc agt tat ggg gat tgt gcg att tcc atc tgt aac gca tgg tcc ggc acc gtt gat
D R S Y G D C A I S I C N A W S G T V D
cga gtg cca tca ggt gta gac att ccc gga ctg agt gga atc gct atc gct gag ttt gca
R V P S G V D I P G L S G I A I A E F A
ttg gcc gca gaa gta ctt cgg ctg aat gaa cgg tgg gaa atc gat gaa att agg cgt ttt
L A A E V L R L N E R W E I D E I R R F
aaa acc atg atg act acc tat ttt tat ccg gtt tgc cat gat ttc ttg acg aac cat gct
K T M M T T Y F Y P V C H D F L T N H A
gga agg tgt gcc gat tat ttt tgg gca aac tgg gat gcc tgt aat ata gct gca tta att
G R C A D Y F W A N W D A C N I A A L I
gct atg ggt gta ctt tgc gat gat cgg aat att tat gac gaa gga gtt gaa tat ttt aaa
A M G V L C D D R N I Y D E G V E Y F K
cac gga gat ggc gcc ggc agc atc gaa cac gcc gtt gcc tac att cat tcc ggt aat ctc
H G D G A G S I E H A V A Y I H S G N L
ggg caa tgg cag gaa agc ggc agg gat cag gaa cat gca cag tta gga gtg gga ctt ttg
G Q W Q E S G R D Q E H A Q L G V G L L
gct gca gcc tgt cag gtt gcg tgg aat cag gga ttg gac cta ttc agt tat gat aat aac
A A A C Q V A W N Q G L D L F S Y D N N
cgg ctt ctt gct ggt gcc gaa tat gta gca aaa tat aac cta tgg cag gat gta cct ttt
R L L A G A E Y V A K Y N L W Q D V P F
aaa tat tat aac agc tgc cag cag gta aac cat aat tgg tca tct att aat gga agg gga
K Y Y N S C Q Q V N H N W S S I N G R G
agg ttg gat gat cgc ccg ctt tgg gag tta att tac aat cat tat gtc gtt aga aaa agg
R L D D R P L W E L I Y N H Y V V R K R
ttg aac gca cct aat tca aaa tta atg gct gaa ctc atg aga ccc gag cat ggc agt aac
L N A P N S K L M A E L M R P E H G S N
gat cat ttt gga tac ggt aca ctg aca ttt acg ttg gat gga aag cag tca ccc tat cct
D H F G Y G T L T F T L D G K Q S P Y P
gca ctt gca aca cca gcc att ccg acc cat ctg act gct aca gca ggt gta aat aga gta
A L A T P A I P T H L T A T A G V N R V
tat ctc aca tgg cat cca tct gaa gga tat act gcg cag gga tat gag gtg caa cgg gct
Y L T W H P S E G Y T A Q G Y E V Q R A
ata agt agc gcc ggt cct tat aac atc att acc aaa tgg aat gat cat aca tca cca caa
I S S A G P Y N I I T K W N D H T S P Q
tat ata gat ccg gat gta aca aat gga aca aat tac tac tac cgg gtg gcg gca ttg aac
Y I D P D V T N G T N Y Y Y R V A A L N
caa tca ggt act agt tcg tat tct tcc att gtc cag gcc agt cct cag gct gca gga gaa
Q S G T S S Y S S I V Q A S P Q A A G E
ctt cct gcg aaa tgg aaa aat aca tta atc ggg aaa gga aat gat ggc aat gcc gct ttt
L P A K W K N T L I G K G N D G N A A F
gct gcc gtt ggc gaa gga acc ttt att gtt aaa gga aac gga act gat ctc gga gga aat
A A V G E G T F I V K G N G T D L G G N
gaa gat caa ata acc tat act tac tgt cgt gta gaa gga gat ttt gtg atc acc gca aga
E D Q I T Y T Y C R V E G D F V I T A R
att tcg gat att act ggg cct aat cag aaa aca ggg ata atg gtt agg gaa tcg ctg gct
I S D I T G P N Q K T G I M V R E S L A
gca gac gcg aaa gca gtg agc ata acc ttg gga gat gca ggc gga cgt ttt gcc cga atg
A D A K A V S I T L G D A G G R F A R M
ggc aaa cgt aaa aat gac aaa gaa aaa atg tct ttt aca ttg gga aac gct tat aca tgg
G K R K N D K E K M S F T L G N A Y T W
ttg ccg gcg tgg ttc agg tta gaa cgg act gga agc tct tat aaa gca ttt gaa tct tcc
L P A W F R L E R T G S S Y K A F E S S
gat ggg acg cat tgg ttt aag gtt tct act gaa aac ttc agc atg tca aaa aca gca ttt
D G T H W F K V S T E N F S M S K T A F
gtc gga ttg gtt gtt gct tca ggt agt gcg tca gga ata gat act gtc acc ttc gat cat
V G L V V A S G S A S G I D T V T F D H
gta aag atc acc aaa agt act aat tct ggc aaa caa ggc gaa tga
V K I T K S T N S G K Q G E -
Figure 2. Gene sequence and amino acid sequence of the gene GL0212614
3.4. Predicting the spatial structure and some physical properties of proteins
Since the structure of proteins tended to be more conservative than their amino acid
sequences during evolution, we proceed to predict the spatial structure of the GL0212614 gene
using the Phyre2 software. The results showed that the spatial structure model of the
GL0212614 gene determined based on the alginate lyase enzyme c3nfvA_ template from
Bacteroides2 ovatus had a coverage of 46% and a confidence level of 100% (Figure 3). In the
secondary structure, there was 25% α helix, 29% β helix, 2% TM helix and 14% unidentified.
Figure 3. Structural model of the GL0212614 gene using Phyre2

Some physical properties of the GL0212614 gene were also predicted. When inserting the
amino acid sequence into AcalPred software, the results of acidic and alkaline index were
0.919904 and 0.080096, respectively. According to the prediction of this tool, the gene of
choice is an acidic enzyme. This result is consistent with previous studies showing that genes
are active under acidic pH conditions [4]. The optimal temperature for enzyme activities
according to TBI had 3 levels: above 65°C, 55°C-60°C, below 55°C. The results of melting
temperature (Tm) of GL0212614 had a Tm of 0.8289, so the optimal temperature for enzyme
activity is 55°C-60°C. This result will help us to choose suitable temperature and pH conditions
in future studies.
4. CONCLUSION
We had exploited 15 complete genes encoding cellobiohydrolase for microbial
metagenome data surrounding white-rot fungi in Cuc Phuong National Park. In which, 6 genes
had functional regions. Gene code GL0212614 with functional region Alginate_lyase was the
gene with the highest expression level of 742 mg/l. GL0212614 was structurally determined
based on the alginate lyase c3nfvA_ enzyme template from Bacteroides2 ovatus with 46%
coverage and 100% confidence. In the secondary structure, there were 25% α helix, 29% β
helix, 2% TM helix and 14% unidentified. GL0212614 is an acidic protein, the optimal
temperature for enzyme activity is 55°C-65°C. These results are an important basis in order to
choose gene expression conditions
Acknowledgments: This study was supported by the grant from the Bilaterial
International Project, code: NĐT.50.GER/18, from Ministry of Science and Technology
(MOST), Vietnam and Federal Ministry of Education and Research, Germany; using the
facilities in National Key Laboratory of Gene Technology, Institute of Biotechnology, Vietnam
Academy of Science and Technology (VAST), Vietnam.
REFERENCES
1. F. L. Soares Júnior et al. (2013), “Endo- and exoglucanase activities in bacteria from mangrove
sediment,” Brazilian J. Microbiol., vol. 44, no. 3, p. 969, doi: 10.1590/S1517-83822013000300048.
2. A. Zafar et al. (2021), “Efficient biomass saccharification using a novel cellobiohydrolase from
Clostridium clariflavum for utilization in biofuel industry,” RSC Adv., vol. 11, no. 16, pp. 9246–
9261, Mar. 2021, doi: 10.1039/D1RA00545F.
3. M. Takeda et al. (2022), “Metagenomic mining and structure-function studies of a hyper-
thermostable cellobiohydrolase from hot spring sediment,” Commun. Biol. 2022 51, vol. 5, no. 1, pp.
1–11, Mar. 2022, doi: 10.1038/s42003-022-03195-1.
4. L. Gao, F. Wang, F. Gao, L. Wang, J. Zhao, and Y. Qu (2011), “Purification and characterization of
a novel cellobiohydrolase (PdCel6A) from Penicillium decumbens JU-A10 for bioethanol
production,” Bioresour. Technol., vol. 102, no. 17, pp. 8339–8342, Sep. 2011, doi:
10.1016/J.BIORTECH.2011.06.033.
5. T.-T.-H. Le et al. (2022), “De Novo Metagenomic Analysis of Microbial Community Contributing
in Lignocellulose Degradation in Humus Samples Harvested from Cuc Phuong Tropical Forest in
Vietnam,” Divers. 2022, Vol. 14, Page 220, vol. 14, no. 3, p. 220, Mar. 2022, doi:
10.3390/D14030220.
SỬ DỤNG MỘT SỐ CÔNG CỤ TIN SINH ĐỂ KHAI THÁC GEN

MÃ HÓA ENZYME CELLOBIOHYDROLASE TỪ DỮ LIỆU
METAGENOME CỦA KHU HỆ VI KHUẨN QUANH NẤM MỤC
TRẮNG (Trametes versicolor) Ở VƯỜN QUỐC GIA CÚC PHƯƠNG
Tóm tắt: Cellobiohydrolase (EC 3.2.1.91) là một trong những enzyme quan trọng tham gia
vào quá trình thủy phân cellulose. Trong nghiên cứu này, trình tự gen mã hóa
cellobiohydrolase đã được khai thác từ dữ liệu DNA metagenome của vi sinh vật quanh khu
nấm mục trắng ở vườn Quốc gia Cúc Phương dựa trên cơ sở dữ liệu KEGG. Có 73 ORF mã
hóa enzyme cellobiohydrolase được thu nhận, trong đó có 15 ORF chứa gen hoàn thiện, có 6
ORF có các vùng chức năng. Mức độ biểu hiện của protein trong E. coli được ước đoán bằng
phần mềm Periscope cho thấy mã gen GL0212614 có mức độ biểu hiện cao nhất là 742 mg/l.
Cấu trúc bậc hai và bậc ba của GL0494307 được dự đoán bằng Phyre2 cho thấy, GL0212614
có cấu trúc được xác định dựa trên khuôn c3nfvA có độ bao phủ 46% và độ tin cậy 100%.
Trong cấu trúc bậc 2 của gen GL0212614 có 25% xoắn α, 29% xoắn β, 2% xoắn TM và 14%
không xác định. GL0212614 là gen chịu axit, nhiệt độ tối ưu cho hoạt tính của enzyme là 55°C-
65oC. Những kết quả này là cở sở quan trong để lựa chọn được các điều kiện biểu hiện gen.
Từ khóa: Cellobiohydrolase, DNA metagenome, E.coli, mức độ biểu hiện, tin sinh học.

Vol.62 KHTN-F

Uploaded by

Copyright:

Available Formats

You might also like

Vol.62 KHTN-F

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Vol.62 KHTN-F

Uploaded by

Copyright:

Available Formats

SCIENTIFIC JOURNAL OF HANOI METROPOLITAN UNIVERSITY − VOL.

1. NOVEL INCREMENTAL ALGORITHMS FOR ATTRIBUTE REDUCTION FROM

9. FABRICATION AND CHARACTERISTICS OF NANOCOMPOSITE TiO2/GQDs THIN

NOVEL INCREMENTAL ALGORITHMS FOR

Ho Thi Phuong1*, Truong Duc Phuong2

Received 15 June 2022

1. 𝑇: = ∅; // T contains the candidates for best reduct

Step 2: Check the added set of objects

15. Compute 𝑆𝐼𝐺𝐵 (𝑎) = 𝐹𝑃𝐷𝑈∪𝛥𝑈 (𝛷(𝑅̃𝐵 ), 𝛷(𝑅̃𝐵∪𝐷 )) −

22. Set 𝑡: = |𝑇| //t is the number of T, 𝑇 = {𝐵0 ∪ {𝑎1 }, 𝐵0 ∪ {𝑎1 , 𝑎2 }, . . . , 𝐵0 ∪

8 Wave Waveform 5000 2500 2500 21 0 21 3

3.2. Computation time of IFW_FDAR_Adobj, IV-FS-FRS-2 IARM, ASS-IAR AND IFSA

Figure 3.1. Time of IFW_FDAR_AdObj, IV-FS-FRS-2 IARM, ASS-IAR và IFSA on Libra

Figure 3.2. Classification accuracy of IFW_FDAR_AdObj, IV-FS-FRS-2, IARM, ASS-IAR và IFSA

DESIGN SOME GRADE 6 MATH TESTS TO ASSESS THE

Received 20 May 2022

In essence, there is no contradiction between assessment of competence and assessment of

MC Essay MC Essay MC Essay MC Essay

- Apply properties - Apply properties

3. Laws for -Find out the rules

A.768 B.3 C.728 D.36

Question Answer The point

A. Matrix of test questions

Sentence Answer Point

Question 10 b) -30 0.5 points

THIẾT KẾ MỘT SỐ ĐỀ KIỂM TRA ĐÁNH GIÁ MÔN TOÁN LỚP 6

SOME REAL PROBLEMS AND APPLICATION OF THE GREATEST

Received 8 May 2022

1.2. Common divisor and greatest common divisor

2.3. The problem about queuing

Fig 2. Encrypting and decrypting data using RSA

return Join(list_of_hex, “”)

# 8. Apply the decrypting function on (c, d)

Fig 4. Encryption process

Fig 5. Decryption process

MỘT SỐ BÀI TOÁN THỰC TẾ VÀ ÁP DỤNG CỦA ƯỚC CHUNG

APPLICATION OF LAGRANGE MULTIPLIER METHOD

Nguyen Duc Nguyen, Nguyễn Thi Hong*

Received 7 June 2022

or, for short,

Definition 2.5.[2,p. 923]

0 = h (t0 ) = fx ( x0 , y0 , z0 ) x  (t0 ) + fy ( x0 , y0 , z0 ) y  (t0 ) + f2 ( x0 , y0 , z0 ) z  (t0 )

The number  in Equation 1 is called a Lagrange multiplier. The procedure based on

We consider the following system of 4 equations with 4 unknowns (x , y , z , ) :

 −21 7 x + 28y + 108 z = 75  2

7 x + 28y + 108 z = 75

ỨNG DỤNG PHƯƠNG PHÁP NHÂN TỬ LAGRANGE TRONG BÀI

SOME APPLICATION MODELS OF FIRST-ORDER

Nguyen Thi Thu Huyen

Abstract: Differential equations is a field of Mathematics that has many practical

Received 19 May 2022

2.1.2. Solution of the differential equation

2.1.3. The Existence and Uniqueness Theorem

whenever ( x, y1 ) , ( x, y2 )  D . The constant L is called a Lipschitz constant for f .

with h = min  a,  and M = max f ( x, y ) .

2.1.4. Separable differential equations

2.2. Application models

I see that R = R0 when the t =0