Abox Inference for Large Scale OWL-Lite Data

Abox inference is an important part in OWL data management. When involving large scale of instance data, it can not be supported by existing inference engines. In this paper, we propose efficient Abox inference algorithms for large scale OWL-Lite data. The algorithms can be divided into two categories: initial inference and incremental inference. Initial inference is used in situation where only raw data exists in storage system, and for this category we propose Rule Static Association Based (RSAB), Rule Dynamic Association Based (RDAB) and Rule Grouped-Sorted Based (RGSB) inference methods. Incremental inference algorithm is used in situation where large volume inference data exists in storage system, and for this category we extend the initial inference algorithm and propose Rule Pattern-Sharing Based (RPSB) method. At last, extensive experiments show that our methods are efficient in practice.