Unsupervised sentiment-bearing feature selection for document-level sentiment classification

Text sentiment classification aims to automatically clas-sify subjective documents into different sentiment-oriented categories (e.g. positive/negative). Given the high dimensionality of features describing documents, how to effectively select the most useful ones, referred to as sentiment-bearing features, with a lack of sentiment class labels is crucial for improving the classification performance. This paper proposes an unsu-pervised sentiment-bearing feature selection method (USFS), which incor-porates sentiment discriminant analysis (SDA) into sentiment strength cal-culation (SSC). SDA applies traditional linear discriminant analysis (LDA) in an unsupervised manner without losing local sentiment information be-tween documents. We use SSC to calculate the overall sentiment strength for each single feature based on its affinities with some sentiment priors. Experiments, performed using benchmark movie reviews, demonstrated the superior performance of USFS.