Semi-automatic Knowledge Graph Construction by Relation Pattern Extraction

Yingju Xia, Zhongguang Zheng, Yao Meng, Jun Sun


Knowledge graphs represent information in the form of entities and relation-ships between them. A knowledge graph consists of multi-relational data, having entities as nodes and relations as edges. The relation indicates a relationship between these two entities. Relation extraction is the key step to construct a knowledge graph. Conventional relation extraction methods usually need large scale labeled samples for each website. It’s difficult to deal with the large number of relations and the various representations of each relation. This paper proposed a novel semi-automatic method that builds knowledge graph by extracting relation patterns and finding new relations. The proposed method models the relation pattern as a tag sequence and learns the pattern similarity metric using the existing relation instances. The pattern similarity is adopted to extract new patterns for existing relations. The new relations are detected by using the pattern similarity and clustering technique. The experimental results on large scale web pages show the effectiveness and efficiency of the proposed method.


Semantic web, relation extraction, knowledge graph

Full Text: PDF