We demonstrate the value of combining ontological modeling with unsupervised algorithms to provide rich recommendations to users, modeling the semantics and relationships between cuisines, businesses and other properties that are of interest to users.

The goal behind this type of modeling is to bridge the gap between the type of data that is easy to enter, track, and maintain, with the types of information users know and would like to provide search systems. That is, by having a good ontology, and the ability to reason semantically, we can provider users richer recommendations. This would be possible even when we have only a very incomplete image of their preferences.

CS270 Paper