The Design And Implementation Of A Stock Market Hotspot Analysis System Based On Text Analysis

The stock market in the western developed countries has been developing for more than 200 years,and the Chinese stock market has only 20 years of history.Due to its immaturity,the Chinese stock market is still going through ups and downs under the influence of information or policies,so it is hard to understand the running rules by purely using conventional methods.Although,there is a great deal of stock analysis software in the domestic market,most just displays various data to users in the forms of graphs and curves,and it only focused on the analysis of fundamental and technical perspectives,the market lacks an analyzing system,which aims to analyze the phenomenon of subject speculation in Chinese stock market.Subject speculation in Chinese stock market is reflected as surrounding various hotspot speculations.Therefore,this thesis focuses on the analysis of hotspot concepts in the stock market based on human behaviors,and based on the design for text analysis techniques to realize a Stock Market Hotspot Analysis System.The above Stock Market Hotspot Analysis System has four major functions,1)Financial and economic information collection,it collects the original data,which the system needs to analyze.The collective data includes data for daily trends of the stock,financial information data,and the existed hotspot concept data.2)Hotspot identification and storage,process the data extracted from financial and economic information collecting module,which includes generating training model identified by hotspot concepts,applying training model to collect hotspot concepts,establishing heterogeneous network interconnection between hotspot concepts and stocks.3)Hotspot Popular Degree Index calculation,which relies on hotspot concepts and related stock data generated by hotspot identification and storage module.The way to interconnect hotspot concepts and stock is called heterogeneous network structure,calculated results from the time sequences of stock popularity can be analyzed associated with stock prices or stock-related industry price index.4)Data display.It is used for displaying hotspot concepts and stock popularity degrees,and using HTML to provide data visualization analysis,which makes users easier to grasp the moving trends of the stock market.The stock analysis software in the market base in fundamental and technical perspectives,only a few contents are displayed in informational perspectives.However,hotspot analysis system for the stock market work regarding to informational perspectives to analyze stocks,which compensates the drawbacks in working in fundamental and technical perspectives effectively.Hotspot concept identification is the main point for this thesis.Hotspot concept is a key word,which denotes the good or bad news of a stock,which is also a consensus between investors in the stock market,and has an enormous advertising effect.In order to get the latest hotspot concept of the market,the author raised a hot concept identification method in the section of system requirement analysis,and mentioned three important steps in the hotspot concept identification,which are boundary templates identification,nominated entities identification,disambiguation by a search engine.This thesis illustrates the algorithm design and procedure in detail in system design section,and,finally,provides the crucial part of the algorithm code realization in system realization section.The calculation of Hotspot Popular Degree Index is another crucial point discussed in this thesis.Hotspot Popular Degree Index is a weighted value for a period of financial information for a stock which calculated by the hotspot concept,it is used to represent how popular the stock is in public opinions within current period.This thesis learns the PR value concept from Google search engine and the voting idea of Page Rank algorithm,and noted the Stock Market Hotspot Mining algorithm based on heterogeneous network,called SMHM,which used to calculate the popular degree of a stock depending on its contribution degree.Currently,Stock Market Hotspot Analysis System has launched and been operated smoothly for 8 months.In the past,the average period for artificial identification on one hotspot concept was 4 days,after using this system,it can generate 2 hotspots from 7270 news on each day.This,therefore,makes up the shortcomings of artificial identification,which are slow response,long identification cycle and narrow coverage.In the meanwhile,the system calculates the Hotspot Popular Degree Index of 2830 stocks per day,which shows the distribution of hotspots in the stock market effectively,and it helps users to analyze stock market in terms of information perspectives.