Extracting Implicit Features Based on Association Rules

Product reviews in the network shopping platform provide references to customs' purchase decision. However, existing researches on opinion objects mainly focus on explicit features, and few of scholars take implicit features into consideration. In this paper, based on Chinese online comments data preprocessing. We proposed a Fuzzy C-means algorithm based on Simulated Annealing (SA-FCM) to cluster the explicit comment sentences into 9 classes. And put each class of comment sentences into a document set. Then association rules between opinion words and opinion objects in every document set are mined and build an association rules table among classes, opinion targets and opinion words. The implicit features are discovered according to the opinion words in the association rule table. Finally, the implicit features excavate method proposed in this paper can effectively improve the accuracy of the extraction effect through an experiment verification.