Building Chinese polarity lexicons with co-occurrence using non-negative matrix factorization

摘要

This paper looks inside sentiment lexicons acquisition, one of the most significant tasks of Sentiment Analysis. Aimed at the problems of the Chinese polarity lexicons only reflects the language knowledge, lack of pragmatic knowledge, we present a novel approach to building Chinese polarity lexicons with co-occurrence from real data by a semi-supervised learning algorithm. Firstly, a relation matrix was constructed between the emotion word and the evaluation object from corpora by PMI (Point-wise Mutual Information). Secondly, use NMF (Non-negative Matrix Factorization) down to its co-occurrence matrix between emotion words, and new matrix of emotion words and evaluation object. Finally, the two relation matrix combined with the feature of synonymy and morpheme, and Label Propagation (LP) algorithm was used to run the relation map to distinguish the polarity of the emotion words. Experimental results show that the proposed method improves the accuracy and recall.