Sentiment Analysis of Social Networks' Comments to Predict Stock Return

Financial intelligence has become a research hotspot in recent years with the development of behavioral finance which introduces the social emotion and behavior factors in the decision-making. The data mining technology is widely used in the research on financial intelligence. This paper collected the investors' comments from Social Network Sites (SNS) by crawler technology and segmented each piece of comment into words by Chinese text processing technology to build a financial sentiment lexicon. Applying the sentiment lexicon, a sentiment computing model based on SO-PMI algorithm was designed to compute the sentiment indices of the investors. Finally, the paper made an empirical analysis through linear regression between the return of the stock and its investors' sentiment index. The result proved that the sentiment indices based on the investors' comments are better to measure the investors' sentiment and can be used to predict the stock return.