Ontology-Based Approach for Neighborhood and Real Estate Recommendations

Abstract

Suggesting services or products to people is a task that should be handled by recommendation systems due to the important increase of information and the multitude of user criteria. In fact, when expressing wishes for a product, a user is influenced by his/her tastes or priorities. These influential characteristics tend to be challenging regarding their integration into recommendation systems, because interaction between the products/services and the user has to be captured through its preferences. Recommendation systems for neighborhood and real estate search are no exception, and to achieve reliable recommendation, we developed an ontology NAREO (Neighborhood And Real Estate Ontology) where environment characteristics related to user preferences are modeled with other geo-semantic descriptions. This ontology can be enriched by SWRL (Semantic Web Rule Language) rules that enhance the semantics of our knowledge base and allow reasoning process through built-ins. To illustrate a use case, we provide a basic set of predefined rules for the recommendation context. User preferences are managed through SPARQL queries taking into account the result of inferences.