A novel hybrid approach of recommending research resources in a university digital library based on demographic clustering
Paper ID : 1160-ICTCK (R1)
1مهرناز فتاحی *, 2مسعود نیازی, 3مهرداد جلالی
1مهندس کامپیوتر
2عضو هیات علمی گروه کامپیوتر دانشگاه آزاد اسلامی مشهد
3کمیته علمی
Despite the widespread use of Digital Libraries (DL), the problem of users accessing information relevant to their needs still remains. This could be due to lack of attention and limitations in software capabilities of Digital Libraries to retrieve information and address user issues in the search process. Recommender Systems (RS) could be used in Digital Libraries to aid users in finding and selecting relevant information and knowledge sources. This research attempts to assist users of a DL in retrieving their required resources by designing a recommendation system that uses academic demographic-based user clusters. Previous research have been carried out to recommend scientific resources in academic domain by combining various methods. But none of them apply user clusters alongside the content-based recommendation filtering. The proposed methods consist of three algorithms, content-based, collaborative filtering and clustering users based on their selected courses at the university. This method was evaluated using precision, recall and F-measure on the Knowledge Sharing Database for Ferdowsi University of Mashhad (PAD). Results show a significant improvement of 0.287 (118%) in the F-measure value, compared to the previous method used in this system.
Recommender System, Information Retrieval systems, Digital Libraries, Hybrid, Demographics, Clustering
Status : Paper Accepted