A FRAME LOGIC BASED APPROACH TO BUILD RECOMMENDATION SYSTEMS: AN APPLICATION TO E-COMMERCE DOMAIN.

This paper suggests a Frame Logic based approach to build recommendation systems. Frame Logic unifies the expressive inferential power of logic programming with some basic characteristics of object-oriented programming allowing the construction of complex inferential subsystems. In order to classify products on sale an extended ontology, in which categories of user interests are selected, has been implemented. The knowledge base used in the construction of user models (profiles) includes even user's hobbies. The realised recommendation system functionalities are inferred from this knowledge base. An inferential engine that, starting from the user profile knowledge (interests and hobbies), is able to derive and propose purchase recommendations to the user has been realised. The inferential system and the reference ontology have been implemented in Flora2. In this work a multi-tier software architecture is also presented. The architecture, designed adopting the Model-View-Controller project methodology, fully integrates Logic Programming (XSB Prolog and Flora2) related systems with the J2EE Application Server. [ABSTRACT FROM AUTHOR]