A generic ontology for CBR-based recommender systems

The main idea behind the Case-Based Reasoning (CBR) paradigm is to solve new problems by reusing the records of previous similar experiences. Our objective is to design a domain-independent hybrid recommender engine that combines collaborative filtering and CBR in order to generate more accurate and personalized recommendations. Integrating CBR in our recommender system will allow adapting items or creating new ones when there exists no item that meets the needs or the preferences of the user. To ensure the relevancy of the recommendations, it is worthwhile using the general knowledge of the application domain during the different steps of the reasoning process. Since one of our main objectives is to implement a generic and domain-independent recommender system, we propose a reasoning approach based on a generic ontology that provides a top level description of the case structure and the knowledge requirements to execute the CBR steps using common recommendation formalism.