Mining Link Patterns in Linked Data

摘要

As the explosive growth of online linked data, an emerging problem is what and how we can learn from these data. An important knowledge we can obtain is the link patterns among objects, which are helpful for characterizing, analyzing and understanding of linked data. In this paper, we present a novel approach of mining link patterns. A Typed Object Graph is proposed as the data model, and a gSpan-based algorithm is proposed for pattern mining. A type determination policy is introduced in cases of multi-types and a data clustering algorithm is proposed to improve scalability. Time performance and mining results are discussed by experiments.