A movie recommender system using ontology based semantic similarity measure

Abstract

Continuous growth in information available on the Internet overwhelms the users during navigation. This information overload may result in users’ dissatisfaction which is undesirable. Users’ satisfaction is very important aspect in every domain. Recommender systems play a vital role in dealing with information overload problems. The recommender systems filter the huge information on the Internet to generate limited and personalized information to users. This helps in increasing users' satisfaction by retaining his/her interests during navigation. Pure Web usage data based recommender systems have been used from last few years. However, they lag in precise recommendations because of absence of domain knowledge. Further, the similarity measures play a vital role in recommendation process and hence affect the performance of the recommender systems. The performance of recommender systems can be enhanced through integration of domain knowledge with usage data. This paper presents an approach to movie recommender system that integrates domain knowledge with usage data. The ontology is used to represent domain knowledge. The proposed approach is based on a new ontology based semantic similarity measure. The experimental results prove that the recommendations’ quality andaccuracy of prediction can be enhanced through integration of ontological domain knowledge with Web usage data.