Applying Multicriteria Algorithms to Restaurant Recommendation

In this paper we propose two novel multicriteria recommendation algorithms and present a comparison with other recommendation approaches in the gastronomic domain. The motivation comes from the fact that traditional single criterion approaches consider that two users share the same taste when they provide similar global ratings on the experienced items. However, these users could agree on global ratings while having completely different priorities on item attributes and different preferences on attribute values. Multicriteria recommenders seem to be a promising solution for this problem as they aggregate user ratings on several item components in order to generate more accurate recommendations. Experiments conducted on Santiago(e)Tapas, a real gastronomic contest where customers evaluate different aspects of several restaurants, demonstrate that one of our algorithms, Support Distance Weighting, outperforms other multi-criteria and single-criterion algorithms in terms of prediction precision.