An improved sentiment analysis algorithm for Chinese news

In recent years, news sentiment analysis is a hotspot in the field of natural language processing, and it is also a challenging problem. Methods based on semantic direction almost only consider polarity of emotion words of every sentence in news. We present an improved news sentiment analysis method. It divides news sentiment analysis into title sentiment analysis and text sentiment analysis. For the title, we use our rule set to process. For the text, we use an algorithm of subjective sentences recognition and an algorithm of subject word recognition to analyze the sentiment of Chinese news. We call the proposed method is Improved Sentiment Analysis (ISA). Experimental data shows that the proposed method improves the accuracy of news sentiment analysis.