Using web crawling technology crawls investors’ comments of SANY stock(Stock Code: 600031) and Fujian Expressway stock(Stock Code: 600033) from February 11, 2015 to August 16, 2017. Then using semi-supervised machine learning method construct investor sentiment index. Moreover, collecting the daily closing stock price and trading volume data from Qianlong software explore the relationship between investor sentiment and stock market volatility based on VAR model and Granger Test Method. The results show that the rate of return and trading volume have a two-way Granger causality, while negative emotion and the rate of return have a one-way Granger causality. Furthermore, with the impulse response function and variance decomposition, the results show that trading volume has significant effects on rate of return and negative emotions of investors have significant negative effects on rate of return and trading volume.