A hybrid web recommender system based on Q-learning

Abstract

Different efforts have been made to address the problem of information overload on the Internet. Recommender systems aim at directing users through this information space, toward the resources that best meet their needs and interests. Web Content Recommendation has been an active application area for Information Filtering, Web Mining and Machine Learning research. Recent studies show that combining the conceptual and usage information can improve the quality of web recommendations. In this paper we exploit this idea to enhance a reinforcement learning framework, primarily devised for web recommendations based on web usage data. A hybrid web recommendation method is proposed by making use of the conceptual relationships among web resources to derive a novel model of the problem, enriched with semantic knowledge about the usage behavior. With our hybrid model for the web page recommendation problem we show the apt and flexibility of the reinforcement learning framework in the web recommendation domain, and demonstrate how it can be extended in order to incorporate various sources of information. We evaluate our method under different settings and show how this method can improve the overall quality of web recommendations.