Coordinate relationship extraction on sentence level in Chinese corpus

An important problem in text mining is the automatic extraction of relations. The study of extracting semantic relation between event semantic chunks from sentences is rarely involved in the current research. This paper presents a metric theory on identifying semantic relation based on matching degree of the deictic word and the similarity of event semantic chunk, and discusses the calculation of the similarities of concept and event semantic tree. Finally, a novel method based on ontology is proposed to extract coordinate relation from sentences. This method explores the effect of extracting coordinate relation for each kind of deictic word of coordinate relation and none of deictic word. This paper provides a contrast with the experimental comparison between Ontology-based, Naïve Bayesian (NB) and Support Vector Machine (SVM). The overall performance of Ontology-based method is tested by selecting randomly 100 sentences of coordinate relation and 600 sentences of non-coordinate relation from data corpus. This Experimental data shows that the method is effective.