基于财经新闻的金融领域负面情感词典构建研究

Research on the Construction of Negative Sentiment Dictionary in Finance Based on Financial News

金融领域的情感倾向具有其领域特有性,通用情感词典无法准确判断其情感倾向。本文以道琼斯工业平均指数排名前30的公司的实时新闻信息为研究对象,设置人工标注规则,采用多种机器学习算法对已标注数据集进行训练,使用特征选择算法抽取特征词语作为负面情感词语,从而构建金融领域负面情感词典。本文以McDonald Financial Dictionary(麦当劳金融词典)为目标情感词典,通过直接和间接评测两种方式验证其准确性。研究结果表明:本文通过实验获得金融领域603个负面的领域情感词,其领域负面情感词典的覆盖率为82.5%,且对金融领域新闻信息情感识别的准确率达到92.8%,能够显著提高金融领域负面情感倾向分析的准确率。

The emotional tendencies in the field of economics have their own domain-specific characteristics,and general sentiment dictionaries cannot accurately judge their emotional tendencies.This paper takes the real-time news information of the top 30 companies in the Dow Jones Industrial Average as the research object,sets up manual labeling rules,uses a variety of machine learning algorithms to train the labeled data sets,and uses the feature selection algorithm to extract the characteristic words as negative emotion words,so as to construct the negative emotion dictionary in the field of finance.Taking the current classic McDonald Financial Dictionary as the target sentiment dictionary,the accuracy of the domain dictionary constructed based on dynamic news information in this study is verified and evaluated by direct and indirect evaluation.603 words have been successfully gotten in this research,and the results show that the coverage rate of the negative sentiment dictionary in the financial field constructed in this paper is 82.5%,and the accuracy of the recognition of financial news sentiment reaches 92.8%.The financial negative sentiment dictionary constructed in this paper can significantly improve the accuracy of negative sentiment tendency in finance.