Research on HowNet-based Chinese Word Lexical Semantic Similarity Measurement
Lin Peng, Yuejie Zhang, Tao Zhang · 2009
HowNet-based Chinese Word Lexical Semantic Similarity Measurement (WLSSM) pattern is proposed in this paper, in which the organization structure of HowNet is used to extract abundant semantic information. This paper utilizes the grammatical rules of Knowledge Database Mark-up Language (KDML) and adopts both maximum matching in each level of sememe and depth information of sememes for WLSSM. Compared with some classic measurements, the proposed method can distinguish word pairs in different semantic similarity levels, and the computation results are more consistent with human being's subjective feeling.