SOM of Syntactic and semantic features based on Chinese sentences with multi-category words

Abstract

In this paper, SOM (self-organizing map) neural networks is introduced to Chinese multi-category words. Chinese multi-category words are those words which are of the same Chinese characters and different syntactic functions and meanings. If we only select characters of target sentences as the features of SOM, these multi-category words with the same characters will be mapped in same output nodes, however, brains neuroimaging of multi-category words with same characters and different functions will be mapped in different cortex area. Here we are used of the syntactic and semantic features to describe word sense of Chinese multi-category words. According to our experimental results, the syntactic and semantic features can distinguish effectively these multi-category words with same characters and different functions, and clustering result of SOM is distributed in different output nodes. It is coincident with human brains neuroimaging.