Abstract
The e-learning system generates huge amount of data which contain hidden and valuable information and they are required to be explored for useful knowledge for decision making. Learner’s activity related data and all behavioral vis-a-vis navigational data are stored in the log files. Extracting knowledgeable information from these data by using Web Usage Mining technique is a very challenging and difficult task. Basically, there are three steps of Web Usage Mining i.e. preprocessing, pattern discovery and pattern analysis. This paper proposes a Dynamic Dependency Adaptive Model (DDAM) based on Bayesian Network. This model mines learner’s navigational accesses data and finds learner’s behavioral patterns which individualize each learner and provide personalized learning path to them according to their learning styles in the learning process. Result shows that learners effectively and efficiently access relevant information according to their learning style which is useful in enhancing their learning process. This model is learner centric but it also discovers patterns for decision making process for academicians and people at top management.
Access provided by Autonomous University of Puebla. Download conference paper PDF
Similar content being viewed by others
Keywords
1 Introduction
The term ‘‘learning styles’’ refers to the concept that individual differs in context of how they process information [1] and students’ preferred ways to learn [2]. There are many learning style models described in literature [3–6], which show that every individual prefers to learn in different learning style. According to above theories, Adaptive e-learning system can also strengthen above concept so that incorporating student learning style definitely enhances the learning process of learner. System that incorporate learning style can provide suggestion to students as well as instructor to optimize students’ learning path. This automatic detection system also overcome the drawbacks of the traditional detection method in e-learning system which is mainly based on questionnaire. In this paper, authors describe how learning process influenced by learning styles and which parameters affect to evaluate personalized learning path according to individual learning style because preferred mode of input varies from individual to individual. Rest of the paper is divided into four sections. Second section describes background and all related work done so far in this domain. Third section explains the concept of proposed model. Fourth section is an experimental section which gives the results and related discussion. Last section concludes the whole work and gives the future aspects.
2 Background and Related Works
2.1 Background
Adaptive personalized e-learning is able to support different learning paths and contents according to learner’s preferences so that it suits and fits into every individual learner’s diverse needs and backgrounds [7, 8]. Many literature surveys showed that many work in domain of e-commerce have been done, but not much work done in e-learning domain. As advancement in technology and development of new tools, the concept of e-learning is highly demanded in view of how to know our students preferences and enhancing their learning processes. It is a challenging task in recent era. In traditional e-learning system, only few experts’ opinions were responsible to provide learning paths and content in adaptive learning system so it was teacher centric. But here we have tried to provide learner centric adaptive system which enable to support learners with more self-control and efficient learning in the given e-learning environment. In context of adaptive e-learning system, Web Usages Mining is the application area of data mining techniques through which it discovers patterns from web data, targeted towards various applications of e-learning system. Figure 1 given below shows the different application of Web Usages Mining.
Web Usages Mining consists of mainly three components, preprocessing, and pattern discovery and pattern evaluation. This is show in flowchart given in Fig. 2.
There are many data mining techniques used in different processes of Web Usages Mining for example statistical analysis, association rule, clustering, classifications, sequential patterns and dependency model are used in pattern discovery.
2.2 Related Works
LS-AEHS is an adaptive e-learning system which incorporates learning style and show the effect of learning achievements of learners after adapting matching learning materials according to their learning style [9]. Increasing the effectiveness and efficiency of learning courses and learners’ satisfaction by adapting to prior knowledge is an important approach [10, 11].
The Web Watcher [12], Site Helper [13], Krishnapuram [14], and clustering work by Mobasher et al. [15] and Yan et al. [16], all they have focused on providing Web Site personalization based on web usage mining. Yan et al. [16] used web server log data. They found and analyzed clusters of users having similar access patterns. Web Usage Mining [17] is tool in the Internet community where data form online web is converted into meaningful knowledgeable utilities. There are many data mining techniques have been used to represent student models, such as rules [18], fuzzy logic [19], and case-based reasoning [20].
3 Proposed Approach for Detecting Learning Style
In the following subsections, about learning style, features of the patterns of navigational access behavioral data relevant to learning style and implementation details are presented.
3.1 Learning Style
Learning Style (LS) of a learner is a way how a learner collects, processes and organizes information [21]. This paper is based on Felder-Silverman learning style model. According to FSLSM, each learner has a preferred mode of learning style measured in four dimensions (active/reflective, sensing/intuitive, visual/verbal, sequential/global) [1, 5, 22]. The concept for providing adaptivity based on learning styles aims to enable LMSs to automatically generate personalized learning path according to the learner’s learning style.
3.2 Relevant Features of Behavior Pattern
There are many parameters or variables values generated by user navigational pattern stored in log file on server. We only analyses those variables which directly or indirectly relevant and correlated with corresponding learning style of FSLS. These variable are described in Table 1.
The threshold values for all given parameters are set according to the literature [23, 24] and by using some statistical functions. Prefix used before the variables name i.e. t, and freq, are used for time and frequency.
3.3 Implementation
Implementation of this approach basically uses multiple resources because not only one resource has all features, which gives the best result. Moodle is one of the best open source software for providing to create powerful, flexible and engaging online courses and experiences in learning management system [25]. The data gathered by Moodle LMS may require less amount of work in data pre-processing than data collected by other systems because it stored all the relevant and authenticate web usages data in database as well as in log file. Weka [26] is open source software that provides a collection of machine learning and data mining algorithms. Now DDAM model is formed based on Bayesian Network with the help of WEKA tool. Weka supports ARFF file format which gives the additional benefit for using Moodle platform.
A Bayesian Network (BN) is based on Bays theorem which gives the formula of calculating conditional probability and is composed of two components: qualitative for defining structure and quantitative part for quantify the network [27]. This Bayesian network gives the conditional probabilities of all dependent variables which represent the strength of the dependencies among nodes represented by variables used to find the corresponding learning characteristics for a particular learning style. Moodle does not have any visualization tool. Therefore, in this paper Gismo tool [28] used for visualization. Weka, Moodle and Gismos all three have many common features like they support ARFF file format, may be implemented in Java and they can work on same dataset. So, these are the main reasons of using WEKA, Moodle and Gismo tools in combination.
3.4 Dynamic Dependency Adaptive Model (DDAM)
Generally learners were not very attentive and responsive about to fill ILS questionnaire due to many reasons. They may be influenced by others which can lead to wrong information about their learning style [29]. Our proposed approach based on [30] is that the users’ preferences can change due to many reasons. Factors that affects are the type and quality of the learning objects of the course.. Therefore, the dynamic user modeling based on students’ behaviors in a course level or in a session level is strongly recommended.
Here, we only consider four dimensions of FSLS and eliminate organizational dimension because it is proved that induction is the natural human learning style. Experiments have also proved that most engineering students are inductive learners [1]. According to this, we collect all the relevant variables described above in Table 1 from LMS and set its threshold values in form of marginal probability distributions according to literature [23, 24]. By calculating the conditional probabilities of nodes which represent variables corresponding to the characteristics of particular learning style in reference to FSLS learning dimensions, Bayesian network is formed which shows the relationship among random variables. Arch provides the strength of relation among variables. Similarly based on this conditional probability distribution of parameters, we form all Bayesian Networks of all the four dimensions using relevant parameters which is shows in Fig. 3. For example for deciding input dimension, we set typ_Lo, t_t_Lo, no_visit_Lo and its threshold value as deciding factor. For example threshold values for chat, data access pattern and exam are given in Tables 2, 3, and 4.
4 Experiment and Result
Using our GyanDarshan e-learning Tutorial made-up on Moodle platform, it is shown in Figs. 4 and 5, 40 participants logging data who are registered in a course C++, used for experiment purpose. Now with the help of WEKA software applied with DDAM classifier which analyses the patterns and automatically detects learning. styles. The relevant variables are based on many activities data associated with topics of a course and of a particular session shown in Table 1. This also gives the learner’s navigational access and behavioral data. Gismo provides visualization tool for generating graphical representations that explores various learning aspects of students. It also shows the performance graph neatly.
Pre test and post test results show that if personalized learning path according to learners learning style is provided, they perform well in terms of their grade, time taken to acquire knowledge, understand learning process high precision value and get full satisfaction (Figs. 6, 7, and 8).
5 Conclusion and Future Works
This paper presents an approach to develop Dynamic Dependency Adaptive Model (DDAM) to integrate learning styles into Adaptive e-learning system to assess the effect of adapting educational system individualized to the student’s learning style and learning path. It evaluates its effectiveness in the learning process. This system is a learner centric instead of teacher centric which could increase the students’ autonomy. We are working on to obtain complete learner’s information’s and applying in descriptive data mining algorithms to obtain other hidden information of users so that students will get a personalized environment in near future. This model is based on the Felder Silverman learning theories but in future we will also explore other learning theories which may give other hidden and similar results which increases and strengthens of our research work and future prospects.
References
Jeremić, Z., Devedžić, V.: Design pattern its: student model implementation. In: Null, pp. 864–865. IEEE (2004)
Srivastava, J., Cooley, R., et al.: Web usage mining: discovery and applications of usage patterns from web data. ACM SIGKDD Explor. Newsl. 1(2), 12–23 (2000)
Kolb, A.Y., Kolb, D.A.: Learning styles and learning spaces: Enhancing experiential learning in higher education. Acad. Manage. Learn. Educ. 4(2), 193–212 (2005). [Curry, L.: An organization of learning styles theory and constructs (1983)]
Felder, R.M., Spurlin, J.: Applications, reliability and validity of the index of learning styles. Int. J. Eng. Educ. 21(1), 103–112 (2005)
Coffield, F., Moseley, D., et al.: Learning Styles and Pedagogy in Post 16 Learning: A Systematic and Critical Review. The Learning and Skills Research Centre, London (2004)
Pashler, H., McDaniel, M., et al.: Learning styles concepts and evidence. Psychol. Sci. Public Interest 9(3), 105–119 (2008)
Huang, S.L., Shiu, J.H.: A user-centric adaptive learning system for e-learning 2.0. J. Educ. Technol. Soc. 15(3), 214–225 (2012)
Joachims, T., Freitag, D., Mitchell, T.: Webwatcher: a tour guide for the World Wide Web. In: IJCAI, no. 1, pp. 770–777 (1997)
Boyle, C., Encarnacion, A.O.: Metadoc: an adaptive hypertext reading system. User Model. User-Adap. Interact. 4(1), 1–19 (1994)
Weibelzahl, S., Weber, G.: Advantages, opportunities and limits of empirical evaluations: evaluating adaptive systems. KI 16(3), 17–20 (2002)
Bra, P.D., Brusilovsky, P., Houben, G.J.: Adaptive hypermedia: from systems to framework. ACM Comput. Surv. 31(4), 1–6 (1999)
Ngu, D.S.W., Wu, X.: Sitehelper: A localized agent that helps incremental exploration of the World Wide Web. Comput. Netw. ISDN Syst. 29(8), 1249–1255 (1997)
Mobasher, B., Robert, C., Srivastav, J.: Automatic personalization based on web usage mining. Commun. ACM 43(8), 142–151 (2000)
Lee, C.H., Fu, Y.H.: Web usage mining based on clustering of browsing features. In: Eighth International Conference on Intelligent Systems Design and Applications, vol. 1, pp. 281–286. ISDA’08. IEEE (2008)
Yang, W., et al.: Mining social networks for targeted advertising. In: Proceedings of the 39th Annual Hawaii International Conference on System Sciences, HICSS’06, vol. 6. IEEE (2006)
Joshi, K.P., et al.: Warehousing and mining web logs. In Proceedings of the 2nd International Workshop on Web Information and Data Management, pp. 63–68. ACM (1999)
Graf, S., et al.: Providing adaptive courses in learning management systems with respect to learning styles. In: Richards, G. (ed.) Proceedings of the World Conference on E-Learning in Corporate, Government, Healthcare, and Higher Education (e-Learn), pp. 2576–2583. AACE Press, Chesapeake, VA (2007)
Xu, D., Wang, H., Su, K.: Intelligent student profiling with fuzzy models. In: Proceedings of the 35th Annual Hawaii International Conference on System Sciences, p. 8. HICSS. IEEE (2002)
Inés, C., et al.: de tutoría inteligente adaptativo considerando estilos de aprendizaje—Proceedings of IE2002, Vigo, España (2002)
Truong, H.M.: Integrating learning styles and adaptive e-learning system: current developments, problems and opportunities. Computers in Human Behavior (2015)
Garcia, P., et al.: Evaluating Bayesian networks’ precision for detecting students’ learning styles. Comput. Educ. 49(2007), 794–808 (2007)
Carmona, C., Castillo, G., Millán, E.: Designing a dynamic bayesian network for modeling students’ learning styles. In: Eighth IEEE International Conference on Advanced Learning Technologies, pp. 346–350. ICALT’08. IEEE (2008)
Popescu, E.: Diagnosing students’ learning style in an educational hypermedia system. Cognitive and Emotional Processes in Web-based Education: Integrating Human Factors and Personalization, Advances in Web-Based Learning Book Series, IGI Global, pp. 187–208 (2009)
Felder, R.M., Silverman, L.K.: Learning and teaching styles in engineering education. Eng. Educ. 78(7), 674–681 (1988)
Witten, I.H., Frank, E.: Data Mining: Practical Machine Learning Tools and Techniques. Morgan Kaufmann, San Francisco (2005)
Mazza, R., Milani, C.: Exploring usage analysis in learning systems: Gaining insights from visualisations. In: Communication Dans: The Workshop on Usage Analysis in Learning Systems, the Twelfth International Conference on Artificial Intelligence in Education, pp. 65–72. Amsterdam, The Netherlands (2005)
Felder, R.M., Soloman, B.A.: Index of learning style questionnaire (ILSQ). http://www.engr.ncsu.edu/learningstyles/ilsweb.html. Accessed 01-2008
Graf, S., Lan, C.H., et al.: Investigations about the effects and effectiveness of adaptivity for students with different learning styles. In: Ninth IEEE International Conference on Advanced Learning Technologies, pp. 415–419. ICALT 2009. IEEE (2009)
Popescu, E.: Dynamic adaptive hypermedia systems for e-learning. Doctoral dissertation, Université de Technologie de Compiègne (2008)
Graf, S., Kinshuk, Liu, T.C.: Supporting teachers in identifying students’ learning styles in learning management systems: an automatic student modelling approach. J. Educ. Technol. Soc. 12(4), 3–14 (2009)
Romero, C., et al.: Data mining in course management systems: Moodle case study and tutorial. Comput. Educ. 51(1), 368–384 (2008)
Rice, W.H:. Moodle E-learning Course Development: A Complete Guide to Successful Learning using Moodle. Packt Publishing, Birmingham (2006)
Samia, D., Abdelkrim, A.: An adaptive educational hypermedia system integrating learning styles: Model and experiment. In: 2012 International Conference on Education and e-Learning Innovations (ICEELI), pp. 1–6. IEEE (2012)
Blöchl, M., Rumetshofer, H., Wob, W.: Individualized e-learning systems enabled by a semantically determined adaptation of learning fragments. In: Proceedings of 14th International Workshop on Database and Expert Systems Applications, pp. 640–645. IEEE (2003)
Acknowledgments
We would like to show our gratitude towards the Prof. (Dr.) S.P. Lal Director, Birla Institute of Technology, Mesra, Ranchi (Patna) for sharing their pearls of wisdom with us during the course of this research.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2015 Springer International Publishing Switzerland
About this paper
Cite this paper
Sweta, S., Lal, K. (2015). Web Usages Mining in Automatic Detection of Learning Style in Personalized e-Learning System. In: Ravi, V., Panigrahi, B., Das, S., Suganthan, P. (eds) Proceedings of the Fifth International Conference on Fuzzy and Neuro Computing (FANCCO - 2015). Advances in Intelligent Systems and Computing, vol 415. Springer, Cham. https://doi.org/10.1007/978-3-319-27212-2_27
Download citation
DOI: https://doi.org/10.1007/978-3-319-27212-2_27
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-27211-5
Online ISBN: 978-3-319-27212-2
eBook Packages: Computer ScienceComputer Science (R0)