Ontology Enrichment: A Tool to Empower Hybrid Recommender Systems

Tremendous growth of online available information creates the need of Recommender Systems (RS). Different recommendation techniques are used based on need and purpose. Each technique is having its own advantages and disadvantages. Hybrid approach is used to overcome the limitations of different approaches for Recommendations. Approach of hybridization also varies based on the desired results. We have developed a hybrid recommender system using mixed strategy of hybridization. Designed model suppresses the limitations of individual techniques. Our designed portal recommends news based on content based approach as well collaborative approach. The system is empowered by making use of semantically enriched ontology as a knowledge base. The ontology plays a central role because it is used to model the knowledge of the domain, to make product profile as well as user profiles and the recommendation process. This empowerment helps in handling issues in recommender system. We have tested ontology enrichment for recommendations in news domain. For recommendations RSS feed news items are first pre-classified into categories and entity types without ambiguity making use semantic knowledge of ontology like synonymy, hyponymy and meronymy information. Semantics in recommendations helps in handling the issue of over specialization, diversity, sparsity, cold start problem for new item, accuracy, adaptivity and missing information in recommendations. The semantic approach also makes it possible for user to query the annotated ontology to retrieve desired information. User has also been given the choice of news source to improve personalization in recommendations. Moreover, Users are also provided with information popular among all the users using collaborative approach. Incorporating semantic knowledge into hybrid recommendation approach increases the flexibility and quality of the personalized recommendations, making it usefulness for end user.