You are on page 1of 3

2019 IEEE International Conference on Computer Science and Educational Informatization (CSEI)

Research on the Construction of


Big Data Platform for College Education
1st
Zou Wei 2nd Chen Jianbing*
Key Laboratory of Ministry of Education for National Education Information Management Departmen
Informatization name of organization
Yunnan Normal University Yunnan Normal University
Kunming, China Kunming, China
237462254@qq.com  30690564@qq.com

Abstract—With the accumulation of data on mass information, make it better to carry out teaching work, and
education resources in colleges and universities, building a improve teaching quality. 
complete educational big data analysis platform will have a
positive impact on the school's education management model. B. Analysis of teachers' teaching ability.
In this paper, Yunnan Normal University is an example. Based In the past, because there was no support for a large
on the analysis of the functional requirements of the amount of data, it was all about teaching the teacher to
educational big data platform, a complete big data analysis follow the feeling. Now Yunnan Normal University can
platform architecture scheme is proposed, which can provide hang the teacher's teaching video online. Students can
reference for the planning and construction of the university
analyze the latitude of the online education course, the
education big data platform. 
online teaching course title of the online education platform,
Keywords—big data, Education platform, system the on-demand teacher course, the number of on-demand
architecture times and the frequency of revisiting. Find out where
students are not clear about the attractiveness of the live
I.RODUCTION course and help teachers improve their teaching and focus on
With the development of information technology and the teaching.
deepening of the educational informatization of colleges and C.Learning behavior analysis
universities, massive amounts of educational data have been
Through the collection of business data such as school
preserved in various ways. These massive educational data
business system, student management system, educational
are a valuable asset for schools. Value will have a positive
management system, file management system, library
and important impact on the management model and
management system, and card management system, the
education model of colleges and universities.
analysis model is constructed through the big data analysis
For example, through online behavior data and card data platform, which is aimed at students’ absenteeism, card
analysis, it is possible to accurately position poor students to swipe records, The library management book, the attendance
a certain extent; through the comprehensive analysis of of self-study in the morning and evening, and the student's
library system borrowing data and public class elective performance are analyzed to construct some models and
information, students can analyze their interests and hobbies; discover some characteristics of students' learning behavior.
Homework, final grades and student attendance data analysis Improve the management level of each department through
can provide a basis for the reform of traditional teaching the collaborative office system, thus improving the level of
models to a certain extent. teaching management and teaching quality. 
Structured and unstructured data for massive amounts of D.Student sentiment analysis
data needs to be stored and analyzed through a professional Student's public opinion analysis is the process of
platform. Therefore, building a complete set of university collecting relevant public opinion information according to
education big data platform will play an increasingly the needs of specific problems, and carrying out in-depth
important role in the informationization of college processing and analysis to obtain conclusions. Through the
education. collection and analysis of the unified identity authentication
system, the unified information portal, the online behavior
II.BIG DATA PLATFORM NEEDS ANALYSIS
management system, and the network operation and
The business system and security system data of Yunnan maintenance system data, Yunnan Normal University found
Normal University are collected through the big data analysi that the students of Yunnan Normal University were on the
s system. Finally, through the analysis of the big data platfor Internet forum, campus BBS, QQ, WeChat and other
m, the following aspects are demonstrated. platforms. communicate with. When an event occurs, many
A.Teaching quality assessment students can use the network to know the beginning and end
of the event, and then comment, or support or oppose, or
Big data analysis of data collected by existing business rational or sensible. When a certain viewpoint is widely
management systems, student management systems, library recognized by everyone, public opinion is likely to affect To
management systems, etc., through big data analysis from the development of the event. Through the analysis of these
the effects of teacher teaching, the use of multimedia paradoxes, we can understand the analysis of the publicity of
courseware, the number of students in class, and the the students in the university, and finally play a very
situation of absenteeism The platform is analyzed. It can important role in the ideological work and stability of the
provide decision support information for the teaching teaching staff of Yunnan Normal University. 
department, provide teachers with accurate feedback

978-1-7281-2308-0/19/$31.00 ©2019 IEEE 230 August 16-18, 2019•Kunming, China


E.A Analysis of woork-study neeeds Data acquisitio
A.D on layer
The Yunnaan Normal University
U sch
hool collects the The data acq quisition layerr is the found
dation of the bigb
traansaction recoords of the schhool card and analyzes
a it to send dataa analytics plaatform. Multipiple sets of co
ollection devices
wo ork-study infoormation to stuudents who may m have finanncial are deployed att multiple noodes of the school camppus
diffficulties. By tracking the student
s campu us card of Yunnnan work, and datta of the entiire school business system
netw m is
No ormal Univeersity, the students
s of Yunnan Noormal colllected throughh various data collection methods. After data
d
Un niversity weree counted in the morning g, mid-dinner and acqu uisition is completed, dataa preprocessin ng is perform med.
ev
vening, and thhe average coonsumption value of male and Datta preprocessing. Data prepprocessing is mainly dividded
femmale studentss was calculaated, and 20% % of the studdents into
o four steps: data normalizatation processinng, data filtering,
aft
fter the tie connsumption werre dynamically y analyzed. AAt the dataa merging and d acquisition annalysis. 
same time, the jjoint student management system confirrmed
B.Innfrastructure layer
l
thee need for innformation onn work-study assistance. A After
co
onfirming the sstudent inform mation through h the School WWork The infrastruucture layer mainly inclu udes distribuuted
Seervice Centerr, contact thhem. At the same time, the storrage, distribbuted compputing, data warehousiing,
co
ounseling stafff of the relevant
r depaartments willl be disttributed query, and metadataa storage. It iss the core partt of
co
onfirmed by thhe joint educaational manageement system m and the big data analy
ytics platform..
thee system officce system to further understand whetherr the 1) Distributed storage.Big Data platforrm uses HD DFS
stuudents are faciing family diffficulties and need
n help. tech
hnology on disstributed storaage to provide high-throughpput
F. Big data foreecast dataa access for applications oon large-scalee data sets. The
T
entiire Hadoop architecturee mainly im mplements the
The core oof big data is prediction. It applies data
undderlying suppoort for distribbuted storage through HDF FS,
alggorithms to massive amoounts of data to predict the
and
d implements program
p suppoort for distribu
uted parallel taask
lik
kelihood of things happpening. For Yunnan Noormal
proccessing throug
gh MR. 
Un niversity, the data of the sttudent manageement system m and
thee file manaagement systtem can bee analyzed. The 2) Distributed
D computing. Thhe platform uses u MapReduuce
ennrollment situaation, professiional situation
n and employm ment tech
hnology for distributed com mputing. The MRM frameworkk is
sittuation of the school are predicted by enrrolling studennts in run by a single JobTracker
J ruunning on thee main node anda
previous years. The various needs
n of studeents and the fuuture runn ning on each cluster from the node.Thee TaskTrackerr is
deevelopment off the school aree predicted an nd so on. commposed togeth her. The masster node is responsible for
scheeduling all thhe tasks thatt make up a job, which are
III.ARCHITEC
CTURE OF BIG DATA ANALY
YSIS PLATFORM
M
disttributed acrosss different slaave nodes. Thhe primary noode
Big data pplatform archhitecture is divided
d into data mon nitors their ex
xecution and re-executes previously
p faiiled
accquisition layeer, infrastructture layer, daata analysis laayer, task
ks. The slavee node is onnly responsiblle for the tasks
co
ore business laayer and platfoform display laayer. As showwn in assiigned by the master
m node. W When a Job isi submitted, the
Figure3.1 JobTracker will dispatch
d the cconfiguration information and
a
otheer informationn to the slavee node after receiving
r the job
j
andd configurationn informationn, and schedu ule the task and
a
mon nitor the execu
ution of the TaaskTracker.
3) Data
D warehouse. Mainly byy Client, ZooK
Keeper, HMastter,
HRegionServer,H HStore, HLogg data storaage, HFile data
d
storrage.
4) Distributed
D qu
uery.Big data pplatform uses Hive technoloogy
for distributed queries.
q Hive is based on n Hadoop's data
d
warrehousing tool maps structtured data files into a sinngle
dataabase table an
nd provides a ssimple sql queery that conveerts
sql statements intto MapReducee tasks.
5) Metadata sttorage. The big data an nalysis platfoorm
mettadata storagee mainly usees a relationaal database. The
T
mettadata storage mainly stores thee managem ment
con
nfiguration infformation off the big data analysis and a
anallyzes the report result storagge. 
C.D
Data analysis layer
l
The big data analysis systeem completes the construction
t data warehouse throuugh the data index modeling
of the
tech
hnology throu ugh the full understandin ng of the useer's
business data. Through thee construction of the data d
warrehouse, throuugh the visual component liibrary of the big
b
dataa analysis systtem, through th
the input formation The library
Figure3.1 System Architecture reallizes the inpu ut of data, reealizes the addition of daata,

231
numerical mapping, etc. through the field processing the collected protocol, target IP, device type, and other
assembly library, through time standardization, IP parameters to be collected, and generate responses through
geographic information mapping, etc.; realizes the data these parameters. The collection task periodically collects
through the record processing component of the visual data data on the target actively or continuously.
component library of the big data analysis platform. The
5) Knowledge base management. The big data analysis
data records are processed by filtering, sampling, merging,
platform provides the management of the knowledge base,
etc. At the same time, the big data analysis platform
and the model library built into the system can be called
provides a data set processing component, which can
during the data analysis process. Through the built-in visual
perform various processes such as merging, intersection, and
component data input component, the knowledge base data
union of the data sets. The processed data is analyzed by the
is imported, and finally the custom modeling and analysis is
correlation analysis, statistical analysis, data mining to
realized.
construct an analysis model and analysis tasks, and an
analysis process is defined by the flow control of the big 6) User authorization management. The big data analysis
data analysis platform. Through the process management, platform supports multi-role and multi-user systems. Each
the execution of the analysis task is realized, and the user can view or process the information that can be
management of the analysis node is realized through the task browsed within the scope of the user's own authority. The
scheduling management, and the analysis task is handed platform can add, delete, and modify the user. 
over to the plurality of analysis nodes and the analysis
The construction of big data platform for school
engine to perform the analysis task. Finally, an analysis
education and the application of big data are in the ascendant.
process of the big data processing and analysis system is
It is of great theoretical and practical significance to study
realized. 
and discuss the architecture of big data platform suitable for
D. Business Layer colleges and universities. Based on the actual situation of the
1) Modeling management. The multidimensional data school and engineering practice, this paper studies and
analysis capabilities of the platform are based on presents an intensive and integrated educational big data
multidimensional analysis techniques. Multi-dimensional architecture design. Due to the complexity of the current IT
analysis technology through the full understanding of system architecture and the diversity of application scenarios,
business data, first through the data index modeling the system architecture needs to consider the implementation
technology to complete the construction of the data complexity and management support in the actual
warehouse, and then based on the data warehouse based on architecture process.
statistical, correlation, mining and other analytical tools for
REFERENCE
the construction of data analysis models, data analysis tasks,
[1] Education Big Data Integration: Current Situation, Problems,
and then The output analysis results are performed by the Architecture and Implementation Strategies [J]. Li Zhen, Zhou
data analysis task. Each analytical model is described as an Dongyu, Liu Na. Library Science Research. 2017(20).
analytical process for a big data processing and analysis [2] The status quo and future development of learning analysis
system. These analysis processes can be performed on a research--Analysis of the 2017 International Conference on Learning
regular basis, and users can visually view the results of these Analysis and Knowledge[J]. Wu Yonghe, Li Ruochen, Wang Haonan.
Open Education Research. 2017(05).
analysis processes.
[3] Review of educational big data research [J]. Du Yumin, Fang
2) Component library management. Big data analytics Haiguang, Li Weiyang, Tongsaisai. China Education Informatization.
2016(19)
platform provides a rich library of visual components with
[4] Reviewing big data calculation from a system perspective [J]. Zheng
data-based input and output, processing component library Weimin. Big Data. 2015(01).
based on record field, visual component library based on [5] Research progress of parallel computing model in big data
data set, library based on various algorithms (such as data environment [J]. Pan Yi, Li Zhanhuai. Journal of East China Normal
clustering algorithm, data classification algorithm, etc.), University (Natural Science Edition). 2014(05).
component library based on process control, based on A [6] Access to education big data: acquisition and sharing of learning
visual component library for scripts. Analyze models and experience data based on xAPI specification [J]. Gu Xiaoqing, Zheng
Longwei, Jian Jing. Modern Distance Education Research. 2014(05).
analysis tasks for the user's business data through these
component libraries.
3) Task management. Analytic task management initiates an
instantiated analysis task by configuring relevant parameters
through a defined analysis template. After the task is
finished running, you can view the task execution status and
related results.
4) Collection management. The big data analysis platform
system can collect data and log information of related
application systems in the whole network, and normalize,
filter and merge the information to form a unified event
format. Data collection for each device supports the creation,
modification, and deletion of data acquisition sources in the
system. Each data acquisition source can select or configure

232

You might also like