Skip to Main Content
Many real-world datasets can be considered as a linked collection of objects with multi-type relations, where each type of relations may play a distinct role. In this paper, we address the problem of link prediction in such multi-relational networks. While traditional link prediction methods are limited to single-type link prediction we attempt here to capture the correlations among the different relation types. For that, we use tensor formalization and formulate the link pattern prediction task as a tensor decomposition model which is solved by quasi-Newton optimization method. Extensive experiments on real-world multi-relational datasets demonstrate the accuracy and effectiveness of our model.