基于信息可信度判别股市涨跌预测方法(CCIR2014)(FINAL).pdfVIP

基于信息可信度判别的股市涨跌预测方法 钱忠 邹博伟 李培峰 朱巧明 苏州大学计算机科学与技术学院 苏州 215006 E-mail:qianzhongqz@163.com, zoubowei@, pfli@, qmzhu@ 摘 要：信息可信度判别是信息抽取研究中的关键任务。该任务通过判断信息的可信度，将事实信息与不 可靠信息相分离，以确保所抽取信息的准确性。本文针对新浪财经频道关于沪深两市的评论文本进行了标 注，并采用基于词法和句法特征的最大熵分类器识别不确定或否定线索词及其作用范围，并以此判断评论 的可信程度。本文提出基于信息可信度的股市涨跌预测方法，该方法考虑了信息可靠程度对预测结果的影 响。实验结果表明，本文提出的信息可信度判别算法能够提高股市涨跌预测系统的性能。 关键词：信息可信度；最大熵分类器；股市涨跌；不确定；否定 The Prediction Method of Rise or Fall in Stock Markets Based on the Discriminat ion of Information Credibility Qian Zhong Zou Bowei Li Peifeng Zhu Qiaoming School of Computer Science and Technology，Soochow University，Suzhou 215006 E-mail:qianzhongqz@163.com, zoubowei@, pfli@, qmzhu@ Abstract: The discrimination of information credibility is a key task in the research of information extraction. This task separates factual information and unreliable information by judging the credibility of the information to insure the accuracy of extracted information. This paper annotates the comment texts about Shanghai and Shenzhen stock market from Sina finance website, and then discriminates the credibility of comments through recognizing negative and speculative cues and their scopes by the maximum entropy classifier based on lexical and syntactic features. This paper proposes a method of predicting rise or fall in stock markets based on information credibility, and this method has considered the influence of information credibility for predictive result. The experimental results show that the information credibility discriminating algorithm proposed by this paper can improve the performance of the prediction system of rise or fall in stock market. Keywords: Information Credibility; Maximum Entropy Classifier; Rise or Fall in Stock Market; Speculation; Negation 1