Self-organizing Map Analysis Consistent with Neuroimaging for Chinese Noun, Verb and Class-Ambiguous Word

摘要

In the paper we discussed the semantic distinction between Chinese noun, verb, and class-ambiguous word by using SOM (self-organizing map) neural networks. Comparing neuroimaging method with neural network method, our result shows neural network technique can be used to study lexical meaning, syntax relation and semantic description for the three kinds of words. After all, the response of human brain to Chinese lexical information is based mainly on conceptual and semantic attributes, seldom uses Chinese syntax and grammar features. Our experimental results are coincident with human brain's neuroimaging, our analysis will help to understand the role of feature description and relation of syntax and semantic features.