Distant Supervision for Relation Extraction Using Ontology Class Hierarchy-Based Features

摘要

Relation extraction is a key step in the problem of structuring natural language text. This paper demonstrates a multi-class classifier for relation extraction, constructed using the distant supervision approach, along with resources of the Semantic Web. In particular, the classifier uses a feature based on the class hierarchy of an ontology that, in conjunction with basic lexical features, improves accuracy and recall. The paper contains extensive experiments, using a corpus extracted from the Wikipedia and the DBpedia ontology, to demonstrate the usefulness of the new feature.