Emerging Pragmatic Patterns in Large-Scale

With the development of the Linked Data, an increasing number of RDF data sets are published in many application domains. To understand the underlying meaning and characteristics of large RDF data, and to reuse popular domain terms when publishing data, captur- ing emerging pragmatic patterns is critical. In this paper, we propose the notion of term co-instantiation graph (TIG) and a method to build a TIG for a given RDF dataset. We also describe a clustering-based approach to distill a set of pragmatic patterns from a TIG, which reveal the prag- matic custom of highly-correlated terms. Through extensive experiments on a real big dataset containing 21 M RDF documents, we analyze the macroscopic structure of the term co-instantiation graph and pragmatic patterns from the complex network point of view, and demonstrate our approach can not only give an elaborated ontology partitioning from the pragmatic perspective to ease the ontology reuse, but also provide a new way to explore the Linked Data.