Sentiment analysis of Chinese micro-blog using semantic sentiment space model

Recently public, government, company and other forms of communities are extremely vocal about their opinions and perceptions on their appeals, policies and products on the micro-blog. Sentiment analysis on micro-blog data has attracted much attention for its application on opinion polarity, classification, summarization and query. However, the related state-of-the-art approaches for sentiment analysis are mainly focused on Twitter data. They don´t work well with Chinese micro-blog for word segmentation, feature word combination and feature extraction problems. In this paper, we propose to improve Chinese micro-blog sentiment analysis performance by 1) use sliding window feature combination detection and sentiment phrase dictionary combined method to address semantic recognition problems on metaphor, adversative, multiple negation and irony; 2)propose ten Chinese micro-blog features for sentiment analysis; 3)find most impactful feature combination for the sentiment classifier. Experimental results show that our semantic sentiment space model is helpful to Chinese micro-blog sentiment classification and our basic feature combination outperforms the traditional classification algorithm TD-IDF and KNN on standard measurement.