An Intelligent System for Rumor Recognition and Rumor Sentiment Judgment

The healthy operation of the stock markets depends on the accuracy, timeliness, and openness of information. However, rumors interfere with investors’ decision-making, influencing their sentiments. More notably, with the development of Internet technologies, rumors based on digitized platforms have more serious interference with the stock market. Not only undermining information transparency, but also imposing huge risks to the stock market. Therefore, we propose an intelligent system based on the CNN algorithm, which can simultaneously recognize and judge the sentiment of the rumor information in digitized platforms. Our results found: 1) CNN-based algorithm is more accurate in classifying massive short-text information and rumor categorization performance is better than the other algorithm; 2) CNN-based algorithm has obvious advantages in rumor sentiments judgment with accuracy of experimental results higher than both other algorithm and traditional financial method; 3) rumor sentiment judgment results have been validated proving they are appropriate for econometric models.