Semantics-enhanced Recommendation Strategies and Time-Driven Filtering: A Must in Personalized E-commerce

Yolanda Blanco-Fernandez
University of Vigo
Spain

Martin Lopez-Nores
University of Vigo
Spain

Jose J. Pazos-Arias
University of Vigo
Spain

Abstract:
In order to ensure high quality personalization services, e-commerce recommender systems must adapt the suggestions as the personal preferences and needs of each user evolve over time. In order to cope with this adaptation process, we argue that it is necessary to develop recommendation strategies that consider the changes happened in the users’ preferences, the kind of products bought in the past, and the time elapsed from the purchase. This way, the proposed approach will improve the quality and accuracy of the recommendations, preventing suggestions of items that are not currently useful for the users. Realizing this view will require research in knowledge representation, user modeling techniques, reasoning mechanisms over an ontology, and filtering procedures, among others.