Predicting stock return via social media has attracted a great deal of attention in the finance and information management disciplines. However,most of these efforts focus on English social media and there have been very few studies on stock predicting via Chinese social media. Extracting information from Chinese social media for predicting stock return has great significance. In this paper, text feature set for Chinese social media analysis is proposed and feature extraction technology is used to identify stakeholders and topics in Chinese social media. Further more,we construct stock regression models to analyze the relationships between stakeholder groups/topics and stock return. Finally, message board of China Railway Group Limited on Eastmoney website is chosen as our experimental platform. The experimental result tested the validity and practicability of our proposed method.%利用社会媒体信息预测股票收益已经成为近年来金融和信息管理等领域的研究热点.然而,现有的研究大多是在英文社会媒体环境下,如何获取中文社会媒体信息,并用其预测股票收益具有重要意义.本文提出了适合中文社会媒体分析的文本特征集,利用特征提取等技术,抽取中文社会媒体上的干系人和话题.进一步,通过构建股票收益率的回归模型,研究中文社会媒体上干系人和话题的活动状况对股票收益率的影响.最后,以东方财富网的中国中铁吧为实验平台进行实验研究,验证了所提出方法的有效性和实用性.
展开▼
