基于支持向量机的上市公司午间公告新闻自动阅读与决策支持系统

An Automated Midday Bulletin News of Listed Companies Read and Decision Support System Based on Support Vector Machine

随着互联网和信息技术的不断发展,投资者获得相关信息的渠道日益丰富,方式也愈加便捷。互联网的膨胀带来了海量的非结构化数据,如新闻、微博等等,如何利用这些信息从而进一步为投资者提供决策支持成为近年来的研究热点。本文从午间公告新闻类型的角度出发,通过提取关键词与K-Means聚类得到初步的新闻类型,然后利用支持向量机进行新闻的分类预测。最后,我们从事件研究的角度出发探讨了新闻类型对当天下午股票价格的影响。

With the development of the Internet and information technology,investors could obtain the relevant information more abundantly and more conveniently.The enlargement of the Internet brings a large number of unstructured data,like news,microblog and so on.How to use this kind of information so as to provide support about investors' decision is hot in the research area.Existed research mainly focused on switching the large number of unstructured data in disclosures,news and social network into specific value so as to assist the forecast of the time series of stock price.In this paper,we start with dividing the news into different types,using keyword extraction method and K-Means.Having got the key words vectors and news types,support vector machine to is utilized train a classifier to predict the news type.At last,the impact of the midday bulletin news on the intraday stock price is discussed in the terms of event study.