Ambiguous structure disambiguation based on semantic similarity computation
WU Ying-ze · 2006
In this paper we propose a novel method based the How Net,which is a Chinese-English bilingual knowledge-based dictionary,to disambiguate the ambiguous syntactic structure.The method uses the verb phrase + noun phrase + de/auxiliary word + noun phrase(VP+NP1+的/u+NP2) as sample to describe our approach for structure disambiguation.Our method is using HowNet as main knowledge resource.The strategy we presented is the modified similarity computation from the HowNet for structure disambiguation.The case study used in the experiment consists of 1000 examples which were extracted from the People's Daily.The experiment result shows that the disambiguation accuracy reached 83.7%,which demonstrate that the method which we used is effective for structure disambiguation.