A Feature Representation Algorithm Based on Emotional Tendency for Texts Clustering
Bin Xue · Computer Development & Applications · 2013
Firstly,feature and evaluation word are extracted form product review texts based on domain ontology.On that basis,making use of the feature evaluation word emotional tendency with the sentence emotional tendency to carry out the feature representation gets the text feature matrix to use KMeans clustering,realize the evaluation object emotional clustering.In order to validate the effectiveness of the proposed method,in the real car comment text data on the experiment,the experiment results show that the feature of emotional tendency,than clustering results feature representation using Boolean and LDA,the clustering result have increased significantly in the purity and F-measure.