【作者】 杨阳；

【导师】 钟惠波；

【作者基本信息】 北京理工大学 ， 理论经济学， 2015， 硕士

【摘要】 大量的互联网用户、互联网数字化的方式、方兴未艾的数据挖掘及人工智能等技术,使得通过互联网媒体信息分析投资者情绪,进而预测投资者行为的研究思路变得日益可行。但因为关于媒体和投资者行为关系的研究既涉及行为经济学、语言经济学等经济学新兴领域,也涉及心理学、计算机科学、语言学等交叉学科,对研究者及研究团队的复合能力要求较高,相关研究尚处于起步阶段。国际上的学者在该领域的研究所针对的文本信息主要以英文为主,并已开发了质性研究软件、分词技术等必要的研究工具。国内学者近年来在该领域也开始了零星的研究,但因为语言、媒体环境及市场环境等条件不同,基于汉语分词技术与情绪测量基础上的互联网新闻信息对资本市场的影响研究,还有很多技术难题需要解决。由此,该领域是一个既可以解决理论和技术难题,又有实践价值的研究课题。本文基于互联网新闻信息——投资者情绪——投资者行为——资产价格的研究逻辑,在文献综述的基础上,以上市公司的新闻情感倾向和股票价格作为研究对象,分析新闻的情感倾向是否影响了股票价格,以及影响的大小。论文选取和讯股票和新浪股票发布的股票新闻,上证180指数成份股作为研究样本,通过网络爬虫抓取网站新闻文本,经过去重、去噪处理后,运用文本分析软件量化分析新闻文本的情感倾向,考察新闻情感倾向是否对股票价格波动有显著的影响。文本分析是计算机学科近年来研究的重点问题,中文文本的情感量化分析有很多方法,如朴素贝叶斯、支持向量机、情感词典等,但是还没有像英文文本分析那样成熟的方法,本文的研究是一个全新的尝试,试图填补国内研究在这方面的空白。研究发现,新闻情感倾向能够部分解释股价的波动,新闻数量对股价的波动也具有一定的解释力。本文的研究分为六个部分:第一章为引言,论述本研究的研究背景、研究内容、研究意义等;第二章为文献综述,探讨国内外在这方面的基本研究状况;第三章为理论分析和模型设定;第四章为数据的搜集与整理;第五章为实证分析;最后一章为研究结论和建议。更多还原

【Abstract】 A great number of Internet users, the digitization of Internet, developing data mining technology and artificial intelligence etc., all of this make it possible for us to analyze investor sentiment with Internet media information and then forecast investor behavior. However, this research requests that researchers and research teams have strong comprehensive ability, because the research about media and investor is not only related to new disciplines like behavioral economics and linguistic economics, but also related to some interdisciplinary fields like psychology, computer science and linguistics. At present, relevant research is situated in the starting stage. Foreign scholars engaged in this research field mainly take English text message as research object, and they have developed some necessary research tools like qualitative research software and word segmentation technology. In recent years, domestic scholars have also been working on this research, but due to different language environment, media environment and market environment, there are lots of technological problems in the research of the effect of Internet coverage on capital market, which is based on Chinese segmentation technology and emotion measurement. Therefore, this research field in China is a research subject with practical value, and can solve theoretical and technological problems.Based on the research logic of Internet Coverage——Investor Sentiment——Investor Behavior——Asset Price and literature review, the coverage emotion tendency of listed companies and their stock prices are selected as the research objectives in this article. It will be analyzed whether the emotion tendency of Internet coverage affects the volatility of stock price.In this article, Internet coverage from the websites of Sina Stock and Hexun Stock and 180 stocks of Shanghai Stock Exchange 180 index are selected as research samples. The coverage texts obtained by web spider with duplicate removal and de-noising processing are quantitatively analyzed to measure their emotion tendency by use of text analysis software, following the empirical analysis of the effect of coverage emotion tendency on stock price. Text analysis is a key point problem of computer science, and there are lots of methods which quantize the emotion tendency of Chinese text, for example Na?ve Bayes, SVM(Support Vector Machine) and Emotion Dictionary, but there is still not a mature analysis method like English text analysis. This article is a brand new attempt to fill up domestic research blank in this field. This article draws the conclusion that coverage emotion tendency and coverage number can partly explain the volatility of stock price.This article is divided into six parts: the first chapter is the introduction which focuses on the research background, research content and research significance; the second chapter is literature review which presents the research situation abroad and at home; the third chapter includes theoretical analysis and econometric model; the fourth chapter is the collection and preprocessing of data; the fifth chapter presents empirical analysis; research conclusion and relevant advices are shown in the final chapter.更多还原