Improving recommendations by assigning weights to semantic concepts

Recommender systems are needed due to information overload in the field of libraries. These systems are different than search engines by the way that profiles of users are learned and accordingly recommendations are provided to the users. Agent based framework for developing such recommender system is proposed. The experiment shown cosine similarity is better measure to find similarity distance between vectors of user profiles and library resources than Euclidean, Manhattan, and Chebyshev measures. The recommendation results are further improved by assigning weights to the semantic concepts. This paper also describes about weight assignment to improve the recommendations.