Incorporating HowNet-Based Semantic Relatedness Into Chinese Word Sense Disambiguation

摘要

This paper presents a semi-supervised learning method that incorporates sense knowledge into a Chinese word sense disambiguation (WSD) model. This research also effectively exploits HowNet-based semantic relatedness in order to leverage system performance. The proposed method includes Sense Colony task for improving context expansion and semantic relatedness calculating for sense feature representation. To incorporate sense knowledge into WSD, this paper employs the Semantic relatedness in a semi-supervised label propagation classifier. This research demonstrates state-of-the-art results on word sense disambiguation tasks.