Recommender System for E-Learning through Content and Profile Based Approach

Abstract

E-learning environments have become a way of life. The aim of the recommender systems is to suggest an optimal set of modules that satisfy the needs of the user on a particular topic. Recommender systems research has used advances in learning styles, personal preferences of the users and tests of ability to suggest content. These suffer from two problems: gap in content matching and lack of context. The domain of information retrieval has shown us that the gap between perceived results and intended results is significant. This is due to complexity of the content. Hence the recommender systems need advanced mechanisms for content tagging and management as a part of their repertoire. The content management aspects need to look beyond the document management style or the data management aspects, and instead, focus on the content tagging for modeling. This is a significant challenge. The second key challenge is to leverage the advances in context based information retrieval. Context is a key contributor in narrowing down the domain of search process. Context can be leveraged and harnessed in recommender systems. When integrated with the research in learning styles and the tests of ability, the recommender systems can show significant results. The objective of this paper is to outline a pedagogical model for recommender systems that leverage the advances in query expansion and context awareness for narrowing down the learning objects that are closest to the user's query. This set of learning objects is now personalized using learning style research and the tests of aptitude for better match. The applicability of the method for a range of courses is tested and the results are bench marked against existing methods.