Design of a Neurofuzzy­based Model for Active and Collaborative Online Learning

An e-learning model that adapts instructional content to individual learning differences and creates new assessment parameters such as studylevel, reviewstudylevel, collabotivelevel, assignments and finalexam for a focused and improved study performance is proposed in this paper. The model was adopted from Honey and Mumford model learning style questionnaire with four classes (activist, pragmatist, theorist and reflector) which was used to determine the learners’ learning preference and matched with the appropriate content presentation. Fuzzy c-means clustering technique was used to analyze the learners’ responses stored in the learner profile database to obtain the degree at which learners belong to each of the four classes. The highest degree of membership value was selected among the four classes as the most acceptable pattern of learning for individual learner. The result of the analysis showed classification accuracy in the ratio of 96%:4% on the fifty learners’ data collected. Furthermore, the learners’ responses from assessment parameters as they gainfully engaged with learning process were obtained in form of studylevel, reveiewstudylevel, collabotivelevel, assignments and finalexam and stored in leaner profile database. The values of the assessment parameters of each learner were fed into neurofuzzy-based network where fuzzy logic technique was performed on a four-layer network. The stages are fuzzification of the input variables, realization of fuzzy relations and rule evaluation, fuzzy aggregation of the rule outputs and defuzzification. These were used to determine the learners learning capability. The results obtained from the implementation of this model showed that the varied learning rate of 0.05, 0.10, 0.15 and 0.20 has the best network classification performance of 97% at 0.05. The tools used for the implementation of this model include MySQL, PHP, HTML and Java.