Mining product features and opinions based on pattern matching

Acquiring available information from product reviews can not only instruct consumers to consume rationally, but also help companies to improve competitiveness and their products´ quality. A method based on pattern matching for mining features and opinions is proposed in this paper according to the characteristics of Chinese reviews. First, the reviews are segmented into fragments with relatively simple structures, and then different patterns are adopted to match fragments with different structures in order to mine the features and opinions in reviews. Finally a method based on features grouping was used to prune, which can keep both infrequent features and the comprehensiveness of mining results. Experimental results show that the method is effective.