A Dynamic Web Mining Framework for E-Learning Recommendations using Rough Sets and Association Rule Mining

Abstract

wide web is a huge information source, broadly used for learning now-a-days due to flexibility of time, sharing of learning resources and infrastructure etc., Most of web based learning system lacks expert-learner interaction, assessment of user activities and learners are getting drowned by huge number of web pages in the learning web site and they find difficulties in choosing suitable materials relevant to their interest. This work attempts to engage e-learners at an early stage of learning by providing navigation recommendations. E-learning personalization is done by mining the web usage data like recent browsing histories of learners of similar interest. The proposed method uses upper approximation based rough set clustering and dynamic all k th order association rule mining using apriori for personalizing e-learners by providing learning shortcuts. The essence of combing association rule and clustering is that, using clustered access patterns can reduce the data set size for association rule mining task, and improves the recommendation accuracy.