Ontology Population from Web Product Information

Abstract

AbstractDue to the explosion of information on the Web, there is a need to structure Web data in order to make it accessible to both users and machines. E-commerce is one of the areas in which increasing data volume on the Web has serious consequences. This paper proposes a framework that populates tabular product information from Web shops in a product ontology. By formalizing product information in this way, one can make better product comparison or recommender applications on the Web. Our approach makes use of lexical and syntactic matching techniques for mapping properties and instantiating values. The performed evaluation shows that instantiating TVs and MP3 players from two popular Web shops, Best Buy and Newegg.com, results in an F1 score of 95.07% for property mapping and 76.60% for value instantiation.KeywordsProduct InformationRegular ExpressionProduct ClassLexical RepresentationBaseline ApproachThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.