A Method of Mining Sentiment Word for Depression Patients Based on Word Frequency-Polarity Intensity

Abstract

Extracting sentiment words from the online comments of depression patients provides a necessary basis for the effective analysis of depression patients’ psychological tendency. How to build a domain-specific sentiment lexicon based on massive amounts of comments on the Web to analyse the sentiment tendencies of patients is a problem to solve. To solve this problem, this paper proposed a method of mining sentiment words based on word frequency-polarity intensity to construct Chinese depression sentiment lexicon. First, effective segment is done on the depression patients’ comment corpus by using bi-direction matching method and mutual information to select candidate seed word, and calculating word frequency-polarity intensity (IW) to select seed word. Then, the seed word set is expanded to construct the Chinese depression sentiment lexicon by calculating the semantic similarity between the basic Chinese sentiment lexicon and the seed word to get the sentiment word of depression field, and adding the sentiment word to the seed word set to get the Chinese depression sentiment lexicon. The experimental results show that the method proposed in this paper can accurately mine the sentiment words of depression.

Keywords

• sentiment analysis
• bi-direction matching method
• word frequency-polarity intensity
• semantic similarity
• depression sentiment lexicon