Improved algorithm for tag-based collaborative filtering

Abstract

Important aspect in the modern e-learning
 systems is selecting the most adequate learning materials
 based on learners’ requirements, needs and knowledge
 goals. Recommender systems based on collaborative
 filtering contribute to overcoming the information
 overload in personalized learning environments. That’s
 why there is imminent need of using systems that have the
 capability to detect the learners’ needs and to recommend
 them the most adequate learning context. In recent years,
 it is common practice to use tags in the process of filtering
 the most useful learning materials.Through the tagging,
 learners can mark or highlight some learning materials
 and can contribute to organizing and retrieving useful
 learning materials.
 Our previous researches were focused on tag-based
 collaborative filtering and learning style determination,
 the factors that affect the tag-based collaborative filtering,
 in order to suggest useful learning material in adequate
 format.
 In this paper, we propose a new tag-based collaborative
 algorithm that takes in consideration the factors that
 affect the tag-based collaborative filtering in order to
 develop more efficient and accurate algorithm, and
 suggest the learning materials based on posted tags rating
 and students rating.
 The developed system was implemented at the Faculty of
 Law – Bitola, and the evaluation results are shown in this
 paper.