A Preference-Learning Framework for Modeling Relational Data

Ivano Lauriola, Mirko Polato, Guglielmo Faggioli, Fabio Aiolli

Abstract: Nowadays large scale Knowledge Bases (KBs) represent very important resources when it comes to develop expert systems. However, despite their huge sizes, KBs often suffer from incompleteness. Recently, much effort has been devoted in developing learning models to reduce the aforementioned issue. In this work, we show how relational learning tasks, such as link prediction, can be cast into a preference learning tasks. In particular, we propose a preference learning method, called REC-PLM, for learning low-dimensional representations of entities and relations in a KB. Being highly parallelizable, REC-PLM is a powerful resource to deal with high-dimensional modern KBs. Experiments against state-of-the-art methods on a large scale KB show the potential of the proposed approach.