User Reviews contains a large number of product features and user's opinions towards these features. This paper proposed an approach to extract product features, which is based on Apriori algorithm, and using PMI with the seed set and co-occurrence degree with opinion words to filter features. And then an approach to group product features based on K-means algorithm is proposed, in which sharing words, lexical similarity and opinion words are chosen as the tokens to represent the association of product features. With the Chinese reviews of restaurants from the Internet, experimental results demonstrate the validity of the proposed method．