Relation Extraction and Validation Algorithm
Abstract
Ontological concept Relation extraction is a difficult research problem. In this paper, we aim to extract multi-type relations from the text analyses and the existent relations (in the concept hierarchy). Our approach combines a verb centered method, lexical analyses, syntactic and statistic ones. It is based on a rich contextual modeling that strengthens the term co-occurrence selection, a lexical analysis, a use of the existent relations in the concept hierarchy and a stepping between the extracted relations to validate them and facilitate the evaluation made by the domain experts.