Web items recommendation using hybridized content-based and collaborative filtering techniques

Abstract

Recommender Systems are software agent developed to tackle the problem of  information overload by providing recommendations that assist individual users identify contents of interest by using the opinions of a community of users, similarities between items contents or the user’s preferences. The exponential growth of URLs is affecting the possibilities of search engine retrieving the right documents within the maze of web documents. Thus, recommender system has become a favourable alternative to solve the problems. This paper therefore adopts a hybridized technique that took advantages of content-based filtering to overcome the disadvantages of collaborative filtering to implement better recommender systems. From the results and experiments, it could be inferred that the adoption of hybrid recommender systems perform the best within the context of dataset size and accuracy. These factors are instrumental and essential in evaluation and designing of Recommender Systems

Keywords: Recommender systems, content based filtering, collaborative filtering, cold start problem