Overcoming Weaknesses of Current Personalization Techniques by Semantic Web Technologies ⋆ Yolanda Blanco-Fernández, José J. Pazos-Arias, Alberto Gil-Solla, Manuel Ramos-Cabrer, Martín López-Nores ETSE de Telecomunicación, University of Vigo, 36310, Spain {yolanda,jose,agil,mramos,mlnores}@det.uvigo.es Abstract. Nowadays, users are exposed to an overwhelming amount of informa- tion in several application domains. Recommender systems fight such an overload by selecting the products which are more appealing to each user, according to his personal preferences or needs. Current personalization techniques are based on more or less sophisticated syntactic methods which miss a lot of knowledge during the elaboration of the recommendations. In this paper, we propose an ap- proach that effectively overcomes the drawbacks of the existing personalization techniques by resorting to reasoning mechanisms inspired in Semantic Web tech- nologies. Such a reasoning provides recommender system with extra knowledge about the user’s preferences, thus favoring more accurate personalization pro- cesses. 1 Introduction Nowadays, users are exposed to an overload of information in numerous application domains (e.g. WWW, e-commerce, Digital TV). In this scenario, it is necessary to de- velop recommender systems which select automatically products interesting for each user, according to his/her preferences or needs (modeled in a personal profile). In order to predict the relevance of a product for a given user, some personalization techniques establish simple comparisons between the main attributes of this product, and those defined in his/her profile. In contrast with these strategies, other approaches dismiss these content descriptions and only consider the levels of interest contained in the user’s profile. All the existing personalization techniques have a common draw- back, due to the fact that the selection of the recommendations is based on syntactic mechanisms which miss huge amounts of knowledge about the user’s preferences. Such knowledge is related to the semantics of the user’s interests and of the products avail- able in the recommender system. This limitation reduces the quality of the offered sug- gestions, and causes weaknesses in the personalization techniques adopted by existing recommender systems. In this paper, the aim is to fight these syntactic limitations, by taking advantage the experience gained in the Semantic Web field. As stated in [4], this initiative permits to discover semantic relationships among resources annotated by metadata, which are ⋆ Work supported by the Spanish Ministry of Education and Science Project TSI2004-03677.