A Hybrid Approach for Personalized Recommender System Using Weighted Term Frequency Inverse Document Frequency

Abstract

Recommender systems are gaining a great popularity with the emergence of eCommerce and social media on the internet. These recommender systems enable users’ access products or services that they would otherwise not be aware of due to the wealth of information on the internet. Two traditional methods used to develop recommender systems are content-based and collaborative filtering. While both methods have their strengths, they also have weaknesses; such as limited content analysis, overspecialization and new user problem in content-based filtering, scalability, data scarcity and new item problem in collaborative filtering. These weaknesses leads to poor recommendation quality, but some of them can be overcome by combining two or more recommender methods to form a hybrid recommender system. This thesis deals with issues related to the design and evaluation of a hybrid approach for personalized recommender system that combines content-based and collaborative filtering methods to improve the precision of a recommendation. The content based and collaborative filtering methods use weighted Term Frequency Inverse Document Frequency to compute similarities among users and items. Experiments done using Movie Lens dataset shows that the hybrid approach for personalized recommender system overcomes the challenge of recommendation precision experienced in pure recommender systems.