Non-hierarchical Relation Extraction of Chinese Text Based on Scalable Corpus

As for ontology construction from Chinese text, the non-hierarchical relation extraction is harder than the concept extraction and its extraction effect is still not satisfactory. In this paper, we put forward a scalable corpus model, which uses Tongyici Cilin and word2vec to calculate terms’ similarity and add the qualified candidate terms to the corpora. In this way we can expand the scalable corpus while extracting non-hierarchical relations. In turn, the scalable corpus that has been expanded with the new terms will facilitate the non-hierarchical relation extraction further. We carry out the experiment with Chinese texts in the domain of Computer, whose results show that with expansion of the corpus, the extraction effect will be better and better.