利用社交媒体情感分析的短期股价趋势预测方法

Short-Term Stock Price Movement Prediction Model Based on Social Media Sentiment Analysis

本文旨在利用社交媒体中的情感信息来提升股价涨跌预测性能.与以往粗粒度地使用文本中的情感信息不同,将与某公司特定话题相关的细粒度情感信息引入预测模型中,并提出一个用于短期股价预测的全新特征——“话题情感”,该特征同时抽取话题和情感信息,并协同利用二者来预测股价涨跌.此外,以往的测试数据集中交易日数量非常少或者仅包含单支股票的数据,本文方法构建了包含众多股票的长时间跨度数据集,并在此数据集上验证了细粒度情感分析对股价涨跌预测的良好效用.

To improve the capability of stock price movement prediction,the sentiment information in social media was utilized.Different from the coarse-grained utilization of all the sentiment information in text,the fine-grained sentiment information correlated with the specific topic of a company was taken into consideration in this study.The topic and its corresponding sentiment information were co-extracted and simultaneously used to predict the stock price movement.Moreover,to verify the efficiency of the fine-grained sentiment information,a stock dataset,consisting of more trading dates and stocks,was utilized.It is superior to previous research with only few trading dates or few stocks.