基于情感分析技术的股票研究报告分类

基于情感分析技术的股票研究报告分类 Stock Research Reports Classification Based on Sentiment Analysis 彭敏 PENG Min 1 2 first-author 汪清 WANG Qing 1 黄济民 HUANG Jimin 1 周李 ZHOU Li 1 胡鑫汇 HU Xinhui 1 武汉大学计算机学院 武汉大学计算机学院 School of Computer,Wuhan University School of Computer,Wuhan University 武汉大学深圳研究院 武汉大学深圳研究院 Shenzhen Institute of Wuhan University Shenzhen Institute of Wuhan University 基于情感文本分析技术对股票研究报告中的投资建议进行分类.提取股票研究报告中的"组合特征";采用改进的卡方统计方法进行特征提取,并通过支持向量机（SVM）和朴素贝叶斯算法进行分类,验证分类效果;探讨了权重计算、特征维度和样本数量对分类效果的影响.基于东方财富网上采集的14 000篇股票研究报告的实验表明,通过提取"组合特征"、部分特征维度以及对训练样本重采样,可以取得较好的分类效果. The stock research report is the important professional investment advice in stock areas. Based on web information extraction and retrieval,automatic analysis of the investment advice in massive stock reports will make the significant impact on investors’ behaviors. In this paper,we propose a classification strategy for the stock research report based on sentiment analysis methods. Firstly,we extract the integrated features in the stock research report. Secondly,we leverage feature selection with the improved CHI statistical methods,and classify the stock research reports through the SVM and Naive Bayes classifiers. Finally,we evaluate the classification result considering feature weight,feature dimension and sampling number.Based on 14000 research reports collected from the www. eastmoney. com,the experimental results show that,the strategy of integrated features selection,dimension reduction as well as training resampling,can achieve higher performance. 情感分析 特征提取 支持向量机 朴素贝叶斯 不平衡数据集 股票研究报告 sentiment analysis feature selection SVM（support vector machine） naive Bayes imbalanced data stock research report 10.14188/j.1671-8836.2015.02.004 TP391.1 国家自然科学基金资助项目（61472291,61303115） 2015-02-01 2021-03-31 2 61