Improved Semantic Similarity Method Based on HowNet for Text Clustering
Hongmei Nie, Jiaqing Zhou, Qi Guo, Zhiqi Huang · 2018
A new semantic similarity method based on HowNet is proposed in this paper. At the level of concept, the sememes are divided into several classes. In this paper, the ordering of the weights of the sememe classes is considered. At the same time, a function is reasonably set to make the weights change moderately, which solves the problem of excessive reliance on fixed weights. Moreover, for the similarity between two texts, the element matching method is proposed in this paper. Experimental results show that the proposed method makes the text clustering results more accurate and reasonable compared with the existing methods.