Graph Compression Strategies for Instance-Focused Semantic Mining

Download PDF
Open Webpage

Xiaowei Jiang, Xiang Zhang, Feifei Gao, Chunan Pu, Peng Wang

Published: 2013, Last Modified: 06 Oct 2025CSWS 2013EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Semantic mining is a research area that sprung up in the last decade. With the explosively growth of Linked Data, instance-focused Semantic Mining technologies now face the challenge of mining efficiency. In our observation, graph compression strategies can effectively reduce the redundant or dependent structures in Linked Data, thus can help to improve mining efficiency. In this paper, we first describe Typed Object Graph as a generic data model for instance-focused Semantic Mining; and then we propose two graph compression strategies for Linked Data: Equivalent Compression and Dependent Compression, each of which is demonstrated in specific mining scenarios. Experiments on real Linked Data show that graph compression strategies in Semantic Mining is feasible and effective for reducing the volume of Linked Data to improve mining efficiency.