Research on Extraction for Feature Words from Chinese Product Comments

This paper firstly uses Baidu Baike and co-occurrence proportion of adjacent words after sentence segmentation to identify new words to decreases the effect on recognition of feature because of segmentation errors. We design part of speech sequence patterns to obtain candidate feature word set from Chinese product comments, then utilize a series of effective statistical technique and natural language processing technique to filter candidate feature words. The experiment results show our approaches are promising.