Link Prediction in Schema-Rich Heterogeneous Information Network

Abstract

Recent years have witnessed the boom of heterogeneous information network (HIN), which contains different types of nodes and relations. Many data mining tasks have been explored in this kind of network. Among them, link prediction is an important task to predict the potential links among nodes, which are required in many applications. The contemporary link prediction usually are based on simple HIN whose schema are bipartite or star-schema. In these HINs, the meta paths are predefined or can be enumerated. However, in many real networked data, it is hard to describe their network structure with simple schema. For example, the knowledge base with RDF format include tens of thousands types of objects and links. On this kind of schema-rich HIN, it is impossible to enumerate meta paths. In this paper, we study the link prediction in schema-rich HIN and propose a novel Link Prediction with automatic meta Paths method (LiPaP). The LiPaP designs an algorithm called Automatic Meta Path Generation (AMPG) to automatically extract meta paths from schema-rich HIN and a supervised method with likelihood function to learn weights of the extracted meta paths. Experiments on real knowledge database, Yago, validate that LiPaP is an effective, steady and efficient method.