The Forcast System Of Stock Investment Behavior Based On Weibo Data

In the development process of stock market in different countries,there have been cases of large-scale soaring or plunging,of course,there are many accidental and inevitable factors.But for most investors,especially small individual investors,their investment behavior is often irrational,they choose to follow to chase sellblindly,eventually causing severe losses.Now Behavioral Finance came into being,which explains individual investors prone to herd mentality to follow trend,and laid a solid theoretical foundation for the study of investor sentiment.Whether Twitter abroad,or domestic Sina Weibo,social network has become more and more deeply into everyone's lives,study,and work.Now,we are accustomed to,but also willing to publish something on these platforms with their own interest,express the views of the outside world,our inner feelings and so on.For Internet analyst,This is equivalent to a wealth of valuable information.In the field of financial professionals,we hope to learn about the current market sentiment for the stock market,bullish or bearish sentiment,is very important to the behavior of investors.The modern theory of behavioral finance point out that psychological factors play an important role in investment decisions of investors and market volatility.So how can we measure investor sentiment has become the next problem.Therefore,this article is to design and implement the forecast system of stock investment behavior based on weibo data.It is mainly done through the use of natural language processing and other computer information technology,making weibo text sentiment analysis combined with the emotional dictionary in financial area,so that stock sentiment index will be calculated by a large amounts of data,based on this index,we can make some analysis and prediction onchanges of stocks,ultimately give the user a certain investment advice.The main work in this paper is outlined as follow:1.Given the calling frequency of the Weibo open platform API has some stringent restrictions,this system uses the Http Client client,simulating logining Sina Weibo,and collecting Weibo stock-related pages as the data source,and implements a parser to extract the effective field from the web page.2.The current sentiment analysis commonly used in two ways: a rule-based method based on emotional dictionary and statistical methods based on machine learning,taking into account the latter,it requires a lot of effort in training corpus annotationin the preparatory work,which is not suitable for the prototype system development,so this system uses the former method.Before the implementation system,we constructed a number of suitable financial emotional vocabularywith How Net,and we mapped the emotional expression to sentiment polarity,thuswe can introduce Weibo expression as the calculation factor in the text sentiment analysis.3.Due to the comprehensive survey of the overallstocks sentiment index,it needs all the stocks associated Weibo sentiment are summed up to a weighted mean,and the weight is the Weibo account's impact,we propose a computational Weibo account the influence of the method,which isa relatively objective assessment of its impact.4.the system uses B/S architecture based on Java EE framework,users can use the browser to access the system,and in the front-end interface,we use some data visualization plug-in,so that information is more readable and expressive which can enhance the user experience.