From Syntactic Structure To Semantic Relationship: Hypernym Extraction From Definitions By Recurrent Neural Networks Using The Part Of Speech Information

The hyponym-hypernym relation is an essential element in the semantic network. Identifying the hypernym from a definition is an important task in natural language processing and semantic analysis. While a public dictionary such as WordNet works for common words, its application in domain-specific scenarios is limited. Existing tools for hypernym extraction either rely on specific semantic patterns or focus on the word representation, which all demonstrate certain limitations. Here we propose a method by combining both the syntactic structure in definitions given by the word's part of speech, and the bidirectional gated recurrent unit network as the learning kernel. The output can be further tuned by including other features such as a word's centrality in the hypernym co-occurrence network. The method is tested in the corpus from Wikipedia featuring definition with high regularity, and the corpus from Stack-Overflow whose definition is usually irregular. It shows enhanced performance compared with other tools in both corpora. Taken together, our work not only provides a useful tool for hypernym extraction but also gives an example of utilizing syntactic structures to learn semantic relationships (Source code and data available at https://github.com/Res-Tan/Hypernym-Extraction).