Professional Documents
Culture Documents
net/publication/309088196
CITATIONS READS
8 861
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Marwah Alian on 13 December 2019.
ABSTRACT
Learners style is grouped into four types mainly; Visual, auditory, kinesthetic and
Read/Write. Each type of learners learns primarily through one of the main receiving
senses, visual, listening, or by doing. Learner style has an effect on the learning process
and learner’s achievement. It is better to select suitable learning tool for the learner
according to his learning style. In this work, a fuzzy model for predicting learner style
depending on characteristics of the learner is proposed. The system was tested on a
group of students and compared to their results from the online VARK questionnaire
which is a tool that is used to give the students information on how to maximize their
learning. The new proposed fuzzy inference system gave 48% similar classification
compared with the VARK.
1. INTRODUCTION
The differences among learners in the way they learn has been determined according to
some factors such as learning style ((Reid, 1987). There have been many definitions of
learning styles but according to (Zhang and Bonk, 2008) learning styles refer to “the
pattern or tendency of an individual’s learning behaviors and attitudes. These styles
influence how people learn and how they may be better taught”.
Learning style is forced by the learner through levels of learning from school to college.
Information is presented to learners in their kindergarten until third grade through
kinesthetic style. From fourth to eighth grade, new information demonstrated visually and
in grades 9 until college information is presented in lectures using auditory style (Girija,
2015).
It is important for a learner to understand his personal learning style; this will improve his
learning process, and assist him in communicating with others.
In adaptive eLearning systems a selected path of learning, that accommodates the
requirements and needs of the learner, is provided to the learner (Alian, 2011). Such
systems enhance the learners experience by providing a preferred and a suitable
environment for the learners and adapt to their needs (Shute and D. Zapata-Rivera, 2012).
Those needs can be determined based on different variables; one of those variables is
learning style that we are considering in this research.
Any learner has a preferred learning style, but may be using the three learning styles to
some degree and some learners may use more than one style in an equal degree. There is
a degree of uncertainty and ambiguity in determining the learning style. This has
motivated us to conduct this research, in which a Fuzzy system will be proposed and
introduced to measure the degree to which a learner belongs to with respect to the three
styles of learning.
This research is organized as follows; after the introduction, section 2 presents related
work. In section 3 a brief description for learning styles and their characteristics is
presented. The design of the proposed system is discussed in section 4. While
experimental results are presented in section 5. Finally, conclusion is given in section 6.
2. LITERATURE REVIEWS
Predicting learning styles for the learners support adaptive eLearning systems. Variety of
researches has been done in this area. For example, (Crockett et al. 2013) introduced a
model that predicts the learning style for a student by utilizing conversational intelligent
tutoring systems (CITS). In this work, the predicted learning style depends on the
behaviors that were extracted from a conversation with the learner through CITS
tutorials. The model uses a fuzzy rule based system and the rules were automatically
generated from tutorial dataset and the membership function perimeters were optimized
using a genetic algorithm. This model was tested on structured query language in the
environment of undergraduate teaching classes on two learning style dimensions, that are
perception and understanding (Crockett et al., 2013).
While (Huseyinov, 2011) proposed an adaptive strategy for predicting learning style.
The proposed system used fuzzy linguistic model as well as fuzzy multi-level
granulation. In this strategy, fuzzy if-then rules were used to map learning styles of the
learners to navigation information using natural language expressions. This strategy was
designed to be close to human reasoning. Therefore, it did provide a facility for
constructing a human computer interaction systems that has intelligent capabilities
(Huseyinov, 2011).
Learner style is important in an eLearning system, since it is one of the learner
characteristics that is used in adapting the eLearning system. For example, (Sbattella and
Tedesco, 2004) presented the tutoring module for virtual campus (TVM), in which the
behavior of a learner is traced and recorded in a virtual environment and generated a user
model according to the efficiency of learning strategies and learning attitude. Graphical
reports were generated from the extracted information. The system combined both
Bayesian networks and fuzzy rules to make the user model provide the learner with the
best suggested learning materials for the learner and produce reports for the teachers
about the behavior of the learners.
(Georgiou, 2011) presented Fuzzy Cognitive Map (FCM) using three layers and a
dynamic Hebbian rule in order to predict earning styles. In the proposed FCM model, the
weights among concepts were adjusted using additional information about learners such
as the Learning Ability factors. Two interconnected three layer FCM were demonstrated,
tested and compared after small changes on the weights among concepts (Georgiou,
2011).
In (Almohammadi and Hagras, 2013) an adaptive eLearning system was proposed based
on fuzzy logic. The proposed system relied on the characteristics and needs for the
learner in order to generate a fuzzy model which was used to enhance the delivered
knowledge to the learner. The system was able to learn the knowledge that was preferred
by the student and generated a suitable adaptive learning environment for the learner
based on his needs. In this system, rules were extracted based on gathered data about
learner’s characteristics.
In (Leite et. al., 2010) four learning styles were defined; Visual, Auditory, Kinesthetic
and Read/Write. The authors also have made a statistical database depending on the
responses of people through the online survey VARK (http://vark-learn.com/the-vark-
questionnaire/ accessed July 14, 2016).
In the previous work, trials to give the best learning environment to learners in class or
online learning were introduced and some of these focus on the behavior of the learner
through the learning process (Sbattella and Tedesco 2004). Some researcher use the
learning style as an input in the learner profile then they use it in their eLearning systems
to adapt to learner needs (Alian, 2011). On the other hand, some researchers have used
fuzzy logic in an attempt to enhance learning process or to produce more adaptability in
their systems (Huseyinov, 2011) (Almohammadi and Hagras, 2013). However, the
behavior of a learner is closely related to his learning style which is an important aspect
in the learning process.
In this research fuzzy logic will be used to design a system model for predicting the
learner style depending on characteristics of learners in order to determine to which
extent the learner is Visual, Auditory or Kinesthetic. Table1 summarizes the differences
between previous work and this research.
Table 1: Differences between our work and others work
Reference Features Methodology
Survey Fuzzy VARK
based based styles
TVM Sbattella and Tracing and recording
Tedesco 2004
× × behavior of a learner in
a virtual environment,
generating a user
model and producing
reports about rare
learner’s behavior.
EOS Alian, 2011 Depends on a given
× × attribute in the learner
profile about the
preferred learning
style.
CITS Crockett et al. Predicted learning style
2013
× × depends on the
collected behaviors
that were extracted
from a conversation
with the learner
through CITS tutorials.
Adaptive Huseyinov, Mapping learning
Predicting 2011
× × styles of the learners to
Strategy navigation information
using natural language
expressions.
VARK Leite, et al., Use a survey to
Model 2010
× determine learner style
depending on the
responses of the
learner.
FCM Georgiou, Uses Fuzzy Cognitive
2011
× × Map based on three
layers and utilizing
dynamic Hebbian rule
for predicting earning
styles.
New - Building fuzzy
Proposed membership functions
Model for characteristics of
(Ours) learners to predict
learner style VARK.
3. LEARNER STYLES
There are four different learning styles which are Auditory (A), Visual (V), kinesthetic
(K), and read/write (R) (Fleming and Mills, 1992).
VARK model (Fleming and Mills, 1992) was expanded from earlier conceptual methods
for example the VAK model that was introduced by Barbe and colleagues (1979) and the
VAKOG systems in neuro-linguistic programming (Fleming, 1995)
The auditory learner prefer to use tapes, involve into discussions, listen to another person
reading to him, interact with cd programs and enjoy the sound of words. Auditory
learners learn through listening to spoken lessons and discussing information (Clemons,
2004). While visual learners prefer to learn through graphs, videos, pictures, and
graphical representations but not words. Kinesthetic learners prefer to study and learn
using physical practice. They enjoy doing activities and need time to think but reading is
not their priority. Read/write learners prefer to use plain text and reading instructions for
any activity (Fleming, 2001) (Leite et. al., 2010) (Marcy, 2001). Table 2 presents the
differences among these 1learning styles.
It is important for teachers to keep their learning tools varied as possible in order to
achieve better learning for each learner. Statistical studies have shown that there are
differences in student achievement between those who’s teaching instructions had a great
match to their learning styles and those who’s teaching instructions was not (Dunn et al.,
1995).
Rule Based
Fuzzy Sets Engine
VDistraction
Speaking rate
ADistraction A
FIS
By Nature
Activity level K
Activity Enjoyment FIS
R
FIS
Using Instructions
Thinking Time: For this characteristic, three membership functions were used: Fast (x: 0,
15, 29), Moderate (x: 16, 24, 30, 43) and Slow (x: 31, 42 and >42) as shown in Figure4.
Visual: We used the linguistic values Low (x: 0, 4, 12) Moderate (x: 6, 10, 12, 15) and
High (x: 10, 16, >16) as shown in Figure 5.
Auditory: We used the linguistic values Low (x: 0, 5, 16), Moderate (x: 9, 13, 15.5, 21)
and High (x: 13, 22, >22) as shown in Figure 6. This input characteristic is measured
using the prolongation time in doing a task when the distractor is used for an auditory
and visual task based on the study by (Berti and Schroger, 2001) in which the authors
have measured the response time in auditory and visual conditions. The authors also
computed the response time prolongation in the auditory and visual task deviant stimuli
compared to duration in the standard trials (Berti and Schroger, 2001).
Speaking Rate is measured using the number of words spoken by a learner in a minute.
There are three membership functions used for this input; Slow (x: 0, 80,120), Medium
(x: 70, 120, 196), and Fast (x: 147, 220, >220) as shown in Figure 7.
By Nature is measured by how many times the learner going out per weak and
represented by two membership functions; quiet (x: 0, 1.5, 3.5) , outgoing (x: 1.5, 3,
>3.5) as shown in Figure 8.
Figure 8: By nature membership functions
Activity level: Three linguistic values are used for this input characteristic which are;
Mild (x: 1, 3, 6), Moderate (3, 5, 9), and Strenuous (x: 6, 9,13, 50) as shown in Figure 9.
This input characteristic is based on the study by (Godin and Shephard’s, 1985) which
used Leisure-time Exercise Questionnaire (LTEQ) to measure the number of times per
week a person practice a strenuous, moderate, and mild activity for 15 minutes in a week
(Raedeke , 2007).
Using verbal Instructions: This input characteristic as shown in Figure 10, uses two
membership functions; No Instruction (x: 0, 3, 5) and Verbal Instruction (x: 2.5, 5, 7 10).
This is based on the study by (Berry and Broadbent, 1990) that measures the time needed
for two groups to do a number of tasks, one group uses no instructions and the other uses
verbal instructions.
Figure 11: Linguistic values for the Enjoyment of physical Activity input
The output variables V,A ,R and K take three different membership function as shown in
Figure 12; Mild (x: 0, 4.5),Strong (x: 3,4.5 ,6), and High(x: 4.5, >5). The ranges are
based on the VARK system questionnaire and its statistics (Leite et al., 2010).
The proposed fuzzy system is based on the Mamdani model, where the computation unit
includes the clipping method (min), max for aggregation and the centroid technique was
used for the defuzzification step. The system was implemented using MATLAB.
Figure 13 demonstrates the surface for each of the learning styles in the propose system.
Figure 13.a demonstrates the Visual surface where the relation is between speaking rate
and read likeness. As can be seen from Figure 13.a the output visual learning style is
effected by the inputs speaking rate and read likeness. The higher the speaking rate and
read likeness the higher is to classify the learning rate as visual. While Figure13.b
represents the relation between auditory distraction and thinking time for the auditory
surface. In Figure13.c, Kinesthetic surface is presented using the relation between
activity enjoyment and activity level. Also, the surface for Read/Write presenting the
relation between read likeness and using instructions as in Figure 13.d.
a. visual surface b. Auditory Surface
Figure 13: Learner styles surface used in the proposed fuzzy system
Table 6 shows that we have multimodality learning styles for the majority of the tested
students. Many of the students have two-learning style modes; they have two preferred
styles of learning. Some tested students have tri-learning style modes; they prefer three
styles of learning in a close degree of preference. Few of the tested students have one
preferred learning style compared to the number of students with bi-modes or tri-modes.
The proposed fuzzy inference system gives the same classified learning style as the
VARK for 16 students from 33 (shown with * in Table 6).
Figure 15 shows the preferred percentage for learning styles among the students in the
dataset for each learning style. For the proposed fuzzy inference system, the Visual
learning style was preferred 45% among students while the Auditory was highly
preferred with 70% and the Kinesthetic was preferred with 64%. On the other hand, for
the scores from the VARK Figure 15 shows that the Visual learning style was highly
preferred with a percentage of 36% while the Auditory learning style was highly
preferred with a percentage of 76% and the Kinesthetic with a percentage of 82%. In
both models, the Read/write learning style was the least preferred percentage with 18%.
0.90
0.80
0.70
0.60 Visual
0.50
Auditory
0.40
Read/Write
0.30
Kinesthetic
0.20
0.10
0.00
Fuzzy VARK
The number of learning style scores that were matched with the VARK are shown in
Figure 16. For example, cases 2, 4 and 20 had two matches with VARK since the
learners have a bi-modal preferred learning styles while cases 22, 23, 24, 25 and 26 had
three matches with VARK scores since these learners have tri-mode learning styles as it
is demonstrated in Table 6.
Some cases did not give a match with VARK scores such as case 9, 11, 13 and 14 but the
characteristics of these students according to how they learn gave similarity to the
predicted learner style that was determined by the proposed fuzzy system.
According to (Fleming, 2006) for many learners, solving the questions of VARK
repeatedly over time is a valuable exercise, even though the scores may vary. While
other learners have a good knowledge about their learning style or the way they learn
and they do not need any help from any questionnaire.
Fuzzy Matches with VARK
4
# of styles match
3
2
1
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33
case #
6. CONCLUSION
In this research a fuzzy inference system for predicting preferred learner style was
proposed based on the features used for the VARK questionnaire. The system was tested
on a group of students (males and females). The input data was collected from our own
questionnaire that was designed specially to get the inputs for the fuzzy inference
system. The experiment was performed in two phases and the output scores of the
proposed system was tested and compared with the VARK scores. The experiment
shows that 48% of the test cases by the proposed fuzzy inference system gave the same
classification as the VARK system for learning styles. So, we conclude that the new
proposed fuzzy system is a competitor approach to predict the rate of learner’s styles.
For future work, an interactive system would be developed to dynamically select the
linguistic values for the input variables for the fuzzy inference system so that the system
can maximize or minimize the learning styles that a user has.
REFERENCES
Berry, D. C., and Broadbent, D. E. (1990). The role of instruction and verbalization in
improving performance on complex search tasks. Behavior & Information
Technology, 9, 175-190.
Barbe, Walter Burke; Swassing, Raymond H.; Milone, Michael N. (1979). Teaching
through modality strengths: concepts and practices. Columbus, Ohio: Zaner-Bloser.
C. Girija Navaneedhan, (2015),Visual, Auditory and Kinesthetic Approach to Enhance
the Information Processing Ability in Teaching Learning Teaching Chemistry,
International Educational E-Journal, Volume-IV, and Issue-I, Jan-Feb-Mar2015, pp:
61-66.
Dunn, R., Griggs, S. A., Olson, J., & Beasly, M., (1995). A meta-analytic validation of
the Dunn and Dunn model of learning style preferences. Journal of Educational
Research, 88, 353-362.
D. A. Georgiou, S. Botsios, V. Mitropoulou, M. Papaioannou, C. Schizas, G. Tsoulouhas,
(2011), learning style recognition based on adjustable three-layer fuzzy cognitive
map, JAISCR, 2011, Vol.1, No.4, pp333-347.
Fleming, N. D. and Mills, C. (1992), Not Another Inventory, Rather a Catalyst for
Reflection From To Improve the Academy, Vol. 11, 1992. Page 137
Fleming, N.D; (1995), I'm different; not dumb. Modes of presentation (VARK) in the
tertiary classroom, in Zelmer,A., (ed.) Research and Development in Higher
Education, Proceedings of the 1995 Annual Conference of the Higher Education and
Research Development Society of Australasia (HERDSA),HERDSA, Volume 18, pp.
308 - 313
Fleming, N. (2001). How do I learn best: A student’s guide to improved learning: VARK
- visual, aural, read/write, kinaesthetic.
Fleming, N., and Baume, D. (2006) Learning Styles Again: VARKing up the right tree!
Educational Developments, SEDA Ltd, Issue 7.4, Nov. 2006, pp.4-7
Godin, G., & Shephard, R. J. (1985). A simple method to assess exercise behavior in the
community. Canadian Journal of Applied Sport Science, 10, 141–146
Ilham N. Huseyinov, (2011), Fuzzy Linguistic Modelling Cognitive / Learning Styles for
Adaptation through Multi-level Granulation, Human Computer Interaction Users and
Applications, Volume 6764 of the series Lecture Notes in Computer Science, 2011,
pp: 39-47
Ke Zhang, Curtis J. Bonk, (2008), “Addressing diverse learner preferences and
intelligences with emerging technologies: Matching models to online opportunities”,
Canadian Journal of Learning and Technology, V34 (2) Spring, 2008.
K. Crockett, A. Latham, D. Mclean; J.O'Shea, (2013),A fuzzy model for predicting
learning styles using behavioral cues in a conversational intelligent tutoring system,
IEEE International Conference on Fuzzy Systems (FUZZ), 2013, pp: 1 – 8
Khalid Almohammadi, Hani Hagras, (2013), An Adaptive Fuzzy Logic Based System for
Improved Knowledge Delivery within Intelligent ELearning Platforms. IEEE
International Conference on Fuzzy Systems · July 2013, pp. 1-8, DOI:
10.1109/FUZZ-IEEE.2013.6622350
Leite, W. L., Svinicki, M. & Shi, Y. (2010). Attempted Validation of the Scores of the
VARK: Learning Styles Inventory with Multi-trait-Multimethod Confirmatory Factor
Analysis Models. Educational and Psychological Measurement. 70, 323-339.
L. Sbattella; R. Tedesco, (2004), Profiling and tutoring users in virtual campus, ITHET
2004. Proceedings of the Fifth International Conference on Information Technology
Based Higher Education and Training, (2004). pp. 256 – 262.
Marcy V., MMSc, PA-C, 2001, Adult Learning Styles: How the VARK Learning,
Perspective on Physician Assistant Education, Vol. 12, No. 2. Pp.117-120.
Marwah Alian, (2011) Formalization and Implementation of Eliminating and Optimizing
Selection, (EOS) approach, Education and Information Technologies: Volume 16,
Issue 1, pp. 89-103. DOI: 10.1007/s10639-009-9113-0,”Special Issue: Online
Learning Technologies", Springer.2011.
Reid, J. M.,(1987), The learning style preferences of ESL Students. TESOL Quarterly,
21(1), 1987, pp.87-111.
Raedeke T., (2007), The relationship between enjoyment and affective responses to
exercise. Journal of Applied Sport Psychology.2007; 19:105–115. doi:
10.1080/10413200601113638.
Stefan Berti, Erich Schroger, (2001), A comparison of auditory and visual distraction
effects: behavioral and event-related indices, Research report, Cognitive Brain
Research, 10 (2001). pp. 265–273
Stephanie A. Clemons, (2004), Developing On-Line Courses for Visual/Kinesthetic
Learners: A Case Study, International Journal of Instructional Technology and
Distance Learning, vol 1, No. 11, pp. 51-62
Sean P Mullen, Erin A Olson, Siobhan M Phillips, Amanda N Szabo, Thomas R
Wójcicki, Emily L Mailey, Neha P Gothe, Jason T Fanning, Arthur F Kramer, and
Edward McAuley, (2011), Measuring enjoyment of physical activity in older adults:
invariance of the physical activity enjoyment scale (paces) across groups and time,
The International Journal of Behavioral Nutrition and Physical Activity. (2011)8:
103.
V. J. Shute and D. Zapata-Rivera, (2012), Adaptive educational systems, Adaptive
technologies for training and education, pp. 7-27, 2012.