LOD conversion system for generating large knowledge base from web contents

摘要

In recent years, the Linked Open Data (LOD) have been attracting attention in the world as the technology that can construct the semantic Web. The LOD are created based on the Resource Description Framework (RDF), and can link each of resources considering semantic relations. However, there are problems that the LOD do not have enough links between resources, and the LOD need a lot of time for creation. Therefore, this paper presents the new LOD conversion system that can convert the Web contents to the LOD. This system extracts keywords from sentences in the Web contents using DBpedia LOD, and generates the knowledge base. By experiments, the proposed system was confirmed that it can convert the target Web pages to the RDF data fully at 2.03 seconds per a page. In addition, the system realized to generate the knowledge base by estimating the concepts of the resources using the DBpedia LOD. The number of links in the LOD is increasing 2.0 times and more than non-estimated RDF data.