JP2005100221A - Investment judgement support information providing device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To quantitatively predict the fluctuation of a share price based on a document such as articles written about an enterprise. <P>SOLUTION: This investment judgement support information providing device 1 is provided with a vector database 21 in which statistical information acquired by statistically processing relations between articles and share price fluctuation is stored for each enterprise, an article collecting part 11 for acquiring article data related to a certain enterprise, an article analyzing part 13 for adding information for supporting the judgement of stock investment to the article data based on the articles shown by the article data to the article data based on the acquired article data and the statistical information stored in the vector database 21 and an information distributing part 15 for distributing the article data and the added information for supporting the judgement of the stock investment. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a technique for providing information for supporting investment judgment, and more particularly to a technique for adding investment judgment support information to a document such as a news article.

  Many investors make decisions about buying and selling stocks based on documents that describe companies, such as news articles. In particular, along with the fact that individuals can easily buy and sell stocks using the Internet, there are many individual investors who make investment decisions based on various news on the Internet and place trading orders. Therefore, news articles are thought to have an impact on stock prices.

  However, the decision on how to evaluate the content of a news article was based on individual investor experience and Kang. There are also news articles that are analyzed by analysts, but they are either qualitative or unclear.

  Accordingly, an object of the present invention is to quantitatively predict stock price fluctuations based on documents written about a company.

  Another object of the present invention is to provide investors with information for supporting investment decisions based on the expected result of stock price fluctuations based on documents written about a company.

  An investment decision support information providing device according to one embodiment of the present invention stores, for each company, means for storing statistical information obtained by statistically processing the relationship between articles and stock price fluctuations, and article data relating to a certain company. Based on the acquired article data and the statistical information stored in the storage means, information for supporting judgment of stock investment based on the article indicated by the article data is added to the article data. And means for outputting the article data and the information for supporting the determination of the added stock investment.

  The investment judgment support information display device according to one embodiment of the present invention comprises: means for receiving article data relating to a certain company; and information for supporting judgment of stock investment based on an article indicated by the article data; Means for displaying on the display screen side by side the information for supporting the judgment of the stock investment and the information for supporting the judgment of the stock investment is increased in the information for supporting the judgment of the stock investment. Information indicating whether it is a trend or a downward trend is included, and when it is an upward trend and when it is a downward trend, a display mode of information for supporting the judgment of the stock investment on the display screen Are different.

  An investment decision support information providing apparatus according to an embodiment of the present invention includes a unit for reading out document data including a document in which a company is written from a document data storage unit, and a plurality of past records written about the company. Means for associating and storing information indicating the contents of the document and information indicating stock price fluctuations of the company in each period including the time when each document was issued, and included in the read document data Means for analyzing the content of the document being documented, means for referring to the storage means based on the analysis result of the document, and determining similarity between the document to be analyzed and a plurality of past documents; Means for adding the information indicating the stock price fluctuation associated with the information indicating the content of the past document determined to be similar as a result of the similarity determination to the document data and outputting the information. .

  In a preferred embodiment, the information indicating the contents of the past document stored in the storage means is multidimensional vector data indicating the contents of the document, and the analyzing means includes the read document data. The means for generating a multidimensional vector indicating the content of the contained document and determining the similarity is generated by the multidimensional vector of a plurality of past documents stored in the storage means and the analyzing means. The degree of similarity with the multidimensional vector is determined.

  In a preferred embodiment, each of the multidimensional vectors of the plurality of past documents stored in the storage means is a centroid vector of multidimensional vectors based on the plurality of documents.

  In a preferred embodiment, the information indicating the stock price fluctuation is sensitivity to the stock price fluctuation in the period.

  In a preferred embodiment, a group of words having a predetermined relationship included in a plurality of past documents written about a company in the industry to which the company belongs, and each period including the time when each document is issued And word group storage means for storing information indicating stock price fluctuations of companies in the industry in association with each other, and when there is no similarity as a result of the similarity determination, it is included in the read document data Means for extracting a word group having the predetermined relationship from a document. Then, the output means refers to the word group stored in the word group storage means based on the extraction result, acquires information indicating the stock price fluctuation of the company in the industry, adds it to the document data, and outputs it To do.

  An investment decision support information providing apparatus according to an embodiment of the present invention is written for a means for reading out document data including a document in which a certain company is written from a document data storage means, and for a company in the industry to which the company belongs. In addition, a group of words having a predetermined relationship included in a plurality of past documents is associated with information indicating stock price fluctuations of companies in the industry in each period including the time when each document is issued. Means for storing, means for extracting a word group having the predetermined relationship from the document included in the read document data, and a word group stored in the storage means based on the extraction result Means for referring to, obtaining information indicating stock price fluctuations of the companies in the industry, and adding the information to the document data for output.

  In a preferred embodiment, the word group having a predetermined relationship stored in the storage means is a binary relationship or a ternary relationship obtained as a result of analyzing each dependency structure of the plurality of past documents. Is a group of words. The extraction unit analyzes the dependency structure of the document and extracts a word group having a binary or ternary relationship.

  Hereinafter, an embodiment to which the present invention is applied will be described with reference to the drawings.

  FIG. 1 shows the overall configuration of a system that acquires information about a company, adds information that supports investment decisions, and provides the information to a user. This system includes an investment support information providing device 1 that provides information for supporting investment judgment to information relating to a company, an information providing source system 5 that provides information to the investment support information providing device 1, and an investment support A user terminal 6 used by a user who receives information from the information providing apparatus 1 is connected via the network 2. A display device 61 is connected to the user terminal 6. There may be a plurality of information providing source systems 5 and user terminals 6, respectively.

  The investment support information providing device 1, the information providing source system 5, and the user terminal 6 are all configured by, for example, a general-purpose computer system, and individual components or functions in the devices 1, 5, and 6 described below are as follows. For example, it is realized by executing a computer program.

  In FIG. 1, an investment support information providing apparatus (hereinafter simply referred to as “information providing apparatus”) 1 analyzes contents written in a document data, for example, a document such as a news article, and the influence of the article contents on the stock price. Information to support investment decisions. Hereinafter, processing or functions performed by the information providing apparatus 1 will be specifically described.

  In FIG. 1, the information providing apparatus 1 includes an article collecting unit 11 that collects article data from the information providing source system 5, a company name determining unit 12 that determines which company the article is written on, and an article. An article analysis unit 13 that analyzes the contents of the information, a related information addition unit 14 that adds information related to the article to be distributed or link information to the related information, and an information distribution unit 15 that distributes information to the user terminal 6; Is provided. The article analysis unit 13 further includes a vector determination unit 131 that analyzes an article using a semantic vector and a dependency determination unit 132 that analyzes using a dependency relationship of documents.

  Moreover, the information provision apparatus 1 is further provided with the following various databases. That is, the information providing apparatus 1 includes a vector database 21, a dependency term database 22, an attention word database 23, an article database 31 for storing articles, a company database 32 storing company information, and each company Are classified according to type of business, a stock price database 34 storing stock prices and stock price charts, and a user database 35 storing information about users. The vector database 21 and the dependent term database 22 store information obtained by statistically processing the relationship between articles and stock price fluctuations.

  First, the procedure when the information providing apparatus 1 collects article data will be described using the flowchart shown in FIG.

<Article collection processing>
The article collection unit 11 acquires article data from the information providing source system 5 that provides various information about the company (S11). Article data may be acquired by receiving information automatically distributed by the information providing source system 5, or the article collecting unit 11 may access and acquire the information providing source system 5.

  This article data is, for example, data having a format as shown in FIG. That is, the article data 100 includes an article body 101, an article issue date and time 102, an article issuer 103, and a company name tag 104. The article body 101 is a document in which information about a company is written. The issuer 103 is the name or identification information of the news agency or newspaper company that issued the article. The company name tag 104 is a tag indicating which company the article body 101 is written in, and a company code is set therein. If the article text 101 describes a plurality of companies, the company name tag 104 includes the company code of each company.

  Further, depending on the article data 100, the company name tag 104 may not be attached. Therefore, the article collection unit 11 determines whether the acquired article data 100 has a company name tag 104 (S12).

  If the company name tag 104 is not attached, the company name determination unit 12 analyzes the content of the article body 101, extracts the company name registered in the company database 32, and sets the company code in the company name tag 104 (S13).

  Then, the article data 100 in which the company code is set in the company name tag 104 is stored in the article database 31 (S14).

  Here, the article database 31 has a data structure as shown in FIG. 4, for example. That is, the data item includes an article text 311, an issue date and time 312, an issuer 313, a company name tag 314, and a vector ID 315. At the time when the article data is stored, the vector ID 315 is not set. The setting of the vector ID 315 will be described later.

<Article analysis / delivery processing>
Referring to FIG. 1 again, the article analysis unit 13 sequentially analyzes each piece of article data according to the procedure shown in FIG. 5 with respect to the article data collected by the above processing. Predict the stock price trend. In addition to the prediction result, information related to the article is added and distributed.

  First, the article analysis unit 13 acquires one piece of article data from the article database 31 (S21). In the following processing, processing is performed on one piece of article data acquired here, and when the following processing is completed, processing is sequentially performed on the next article data.

  The vector determination unit 131 analyzes the content written in the article body 311 of the article data acquired here, and generates a multidimensional semantic vector (S22). Each element of the semantic vector is determined based on the appearance frequency for each keyword with respect to a predetermined keyword in the article text. For example, in order to generate a semantic vector, the vector determination unit 131 decomposes the document of the article body 311 by morphological analysis and extracts a predetermined keyword. Then, the appearance frequency of each keyword is adjusted according to the degree of bias of the corresponding keyword in the entire past article, and the value of each element of the vector is determined.

  Then, the vector determination unit 131 compares the semantic vector generated here with the vector stored in the vector database 21 (S23).

  Here, the vector database 21 stores semantic vectors generated from past article data. For example, as shown in FIG. 7, the vector database 21 stores a company code 211, a vector ID 212 for identifying a vector, a representative vector 213, and a sensitivity 214 in association with each other. The representative vector 213 is a vector representing a cluster composed of one or more semantic vectors, and indicates the approximate meaning of articles included in the cluster. In the storage area of the representative vector 213 in the vector database 21, the values of all elements constituting the representative vector 213 are stored. The company code 211 indicates a company referred to by an article included in the cluster. The sensitivity 214 is an index indicating the degree of fluctuation of the stock price. Here, the sensitivity 214 indicates the degree of influence that an article belonging to the cluster indicated by the representative vector 213 has on the stock price fluctuation of the company code 211 within a predetermined period after the article issuance. When the stock price rises, the sensitivity is positive (positive), and when the stock price falls, the sensitivity is negative (negative). Details of the method for generating the vector database 21 will be described later.

  Referring to FIG. 5 again, the vector comparison in step S23 is performed by the representative vector 213 corresponding to the company code 212 matching the company code included in the company name tag 314 of the processing target article data, and the vector generated in step S22. And the similarity is determined. Specifically, a cosine scale is obtained from the inner product of two vectors, and the similarity is determined based on the magnitude (S24).

  As a result of the similarity determination, when there is a representative vector 213 similar to the generated vector (S24: Yes), for example, when there is a representative vector 213 whose cosine scale is equal to or larger than a predetermined value, the representative vector 213 is associated. The sensitivity 214 that has been set is set as the sensitivity to the generated vector (S25). Thus, based on the past performance of the relationship between the article and the stock price fluctuation, when a certain article appears, it is predicted how much the stock price will rise or fall due to the article.

  When there are a plurality of representative vectors 213 similar to the generated vector, an average value of the sensitivity 214 corresponding to each representative vector 213 may be obtained and used as the sensitivity of the generated vector. Alternatively, the sensitivity 214 corresponding to the most similar representative vector 213 may be the sensitivity of the generated vector.

  The vector determination unit 131 stores the vector ID 212 of the representative vector 213 determined to be similar to the generated vector in the vector ID 315 of the article database 31.

  The related information adding unit 14 acquires and adds information related to the article data to be processed (S26). For example, the related information providing unit 14 searches for articles (for example, within one day, three days, or within one week) related to the company appearing in the article to be processed using the company tag 314 as a key, It is added to the article data to be processed. Alternatively, article data having a vector ID 315 having a representative vector similar to the representative vector indicated by the vector ID 315 of the article to be processed is searched, and the article data itself or the number of article data obtained by the search is used as the article data to be processed. It may be added. Alternatively, the current stock price and stock price chart of the company are acquired from the stock price database 34 and added.

  The information distribution unit 15 generates distribution data including article data to be processed, sensitivity obtained by the above processing, and related information, and user information including a distribution destination address and the like is registered in the user database 35 in advance. (S27). The distribution data 110 has a data structure as shown in FIG. 8, for example. That is, the distribution data 110 includes article data 111 (including the article text 112, the issue date 113, the issuer 114, and the company name tag 115), the sensitivity 116, and the related information 117.

  On the other hand, when the representative vector 213 similar to the generated vector is not found in step S24 (S24: No), the process moves to the flowchart shown in FIG. Then, when the sensitivity is determined by the processing of FIG. 6, the processing after step S26 of FIG. 5 is performed.

  The case where a similar representative vector is not found in step S24 means, for example, that there is no representative vector 213 whose cosine scale is equal to or greater than a predetermined value, and a company that is the target of the vector database 21 due to merger of companies. There is a case where there is no past article data about.

  The flowchart in FIG. 6 is processing performed by the reception determination unit 132. The dependency determination unit 132 analyzes the dependency structure of the article text, extracts feature words included therein, and determines the sensitivity of the stock price based on the extracted feature words.

  The dependency determination unit 132 performs dependency analysis on the article body 311 of the article data to be processed (S31). The dependency analysis is to analyze the dependency structure of a document. For example, the dependency determination unit 132 performs morphological analysis and syntax analysis on an article body, and extracts a word group having a binary relationship and a ternary relationship. (Extraction of ternary relationship may be omitted).

  For example, a dependency analysis of the example sentence “The price of domestic fruits and vegetables has fallen due to the expansion of imported fruits and vegetables and domestic farmers are less willing to plant.” "Expansion", "Domestic fruits and vegetables-price", "Domestic farmers-willingness to plant".

  Next, the reception determination unit 132 extracts a word group including the attention word registered in the attention word database 23 from the word group obtained here (S32). In the attention word database 23, words and phrases that are considered important in feature word extraction performed here, for example, predicates, are stored as attention words. The generation process of the attention word database 23 will be described later.

  If “Price”, “Expansion”, “Decrease”, “Decrease”, “Good”, “Slow”, etc. are registered in the attention word database 23, in the above example, “imported vegetables and fruits” and “domestic fruits and vegetables” “Price” is extracted.

  Next, with reference to the acceptance term database 22 and the industry database 33, the sensitivity to the extracted word group of the industry to which the target company belongs is obtained (S33). For example, the industry code is determined by referring to the industry database 33 using the company code included in the company name tag 314 of the article data to be processed as a key. Then, the sensitivity 223 is obtained from the acceptance term database 22 using the business type code and the word group extracted in step S32 as keys. The sensitivity obtained here is used as the sensitivity of the article to be processed.

  The dependency term database 22 stores dependency relationships extracted from past article data. For example, as shown in FIG. 9, the service term code 221, the service relationship 222, and the sensitivity 223 are stored in the service term database 22 in association with each other. Here, for an article, an industry code 221 indicating the industry of the company that is the subject of the article, and a word group in the dependency relationship 222 included in the article ("imported fruits and vegetables in the above example-expanded ”And“ Domestic fruit and vegetables—price ”) and the sensitivity 223 that the article gave to the stock price fluctuation of the company that is the subject of the article. Details of the method of generating the dependency term database 22 will be described later.

  Next, when the distribution data 110 is distributed to the user terminal 6 as a result of the processing of FIG. 5 and FIG. 6, each user terminal 6 receives it and displays it on the display device 61 provided in the user terminal 6. Let A display example is shown in FIG.

  In the example of FIG. 10A, an article body 63, a link 64 to information related to the article body 63, and information about a company written in the article 65 are displayed on the display screen 62 of the display device 61. ing. Here, when the related information link 64 is selected, the related information 117 included in the distribution data 110 is displayed. In the example of FIG. 10, a company code and a company name are displayed in the company information 65 based on the company name tag 115. The company information 65 may be displayed in different display modes, for example, in different colors depending on whether the sensitivity 116 is positive or negative. Thereby, the prediction result of the stock price fluctuation with respect to the article body 63 (when the sensitivity 116 is positive, it is an upward tendency, and when it is negative, the downward tendency) can be presented to the user at a glance. Further, the user (investor) sees this display and makes a decision to buy or sell the stock of the company with reference to this display.

  In the example of FIG. 10B, the article body 63 and the article body analysis result 66 are displayed on the display screen 62. The analysis result 66 includes a stock price tendency 661 indicating whether the stock price is rising or falling, a sensitivity 662 obtained from the article text 63, and the number of related articles 663 included in the related information 117. The stock price tendency 661 is positive if the sensitivity is positive, and negative if the sensitivity is negative. The related article number 663 is the number of related articles included in the related information 117. Information to be displayed in the analysis result 66 is determined by the user terminal 6 based on the distribution information 110.

  Next, processing for generating the vector database 21, the dependent term database 22, and the attention word database 23 will be described. The article data analysis apparatus 7 shown in FIG. 11 analyzes article data and analyzes stock price data. And based on the analysis and analysis result, the vector database 21, the reception term database 22, and the attention word database 23 are produced | generated.

  The article data analysis device 7 is configured by, for example, a general-purpose computer system, and individual components or functions in the article data analysis device 7 described below are realized by, for example, executing a computer program. Further, the article data analysis device 7 can be realized on the same computer machine as the information providing device 1.

  The article data analysis device 7 includes a fluctuation period extraction unit 71, a vector generation unit 72, a clustering unit 73, a representative vector calculation unit 74, a sensitivity calculation unit 75, an attention word collection unit 76, and a dependency analysis unit. 77, a dependency term extraction unit 78, an article database 31 in which past article data is stored, and a stock price database 34 in which information on past stock prices is stored.

<Extracting variable period and variable company>
The fluctuation period extraction unit 71 extracts a period in which the stock price fluctuates greatly for each company. For example, the fluctuation period extraction unit 71 refers to the stock price information stored in the stock price database 34, and extracts all companies (brands) whose stock price fluctuation range for a certain period is equal to or greater than a predetermined threshold. For example, compare the stock price at the beginning of the week with the stock price at the weekend, and the company code of the company (floating company) that fluctuated more than a predetermined threshold and the period (date of the week and weekend: fluctuation period) Extract. Here, the fixed period may be, for example, one week, two weeks, or one month. As the threshold value of the stock price fluctuation, for example, an increase or decrease of 5%, 10%, or 15% may be used.

<Representative vector generation>
The vector generation unit 72 analyzes the article data written about the variable company within the variable period extracted by the variable period extraction unit 71, and generates a vector indicating the semantic content of each article. For example, the vector generation unit 72 extracts from the article database 31 all the article data in which the company code extracted by the variable period extraction unit 71 is attached to the company name tag 314 and the issue date and time 312 is within the variable period. . Then, the vector generation unit 72 analyzes the article text 311 of each extracted article data and generates a multidimensional semantic vector. The semantic vector generation method is the same as that described in the vector determination unit 131.

  The clustering unit 73 clusters the plurality of semantic vectors generated by the vector generation unit 72 by company and by variation period. For example, for a certain company, a plurality of vectors generated from a plurality of article data within a certain variation period are classified into one or more clusters according to the degree of similarity.

  Here, when a significant cluster (a cluster having sufficient elements with respect to the number of analyzed article data) is obtained, the representative vector calculation unit 74 calculates a centroid vector of the cluster, Let it be a representative vector. When a plurality of significant clusters are obtained, a representative vector for each cluster is calculated.

  Each representative vector is stored in the vector database 21 together with the company code and the sensitivity calculated by the sensitivity calculation unit 75 described below. At this time, the vector ID 212 is determined to be unique.

<Sensitivity calculation>
The sensitivity calculation unit 75 calculates the sensitivity of the stock price of the variable company during the variable period based on the variable company and the variable period extracted by the variable period extraction unit 71. For example, the sensitivity calculation unit 75 acquires stock prices at the beginning and end of the fluctuation period, calculates a price / earnings ratio within the fluctuation period, and uses this as the sensitivity.

<Extracting terms from service>
The attention word collection unit 76 extracts attention words from the article text. For example, the attention word collection unit 76 extracts article data within the target period written about the target company from the article database 31, similarly to the vector generation unit 72. Then, the attention word collection unit 76 decomposes the article body 311 included in the article data into words, and extracts words having a high appearance frequency as attention words. For example, a predicate may be particularly noted as the attention word. Alternatively, the attention word may be extracted from predetermined words according to the appearance frequency. The extracted attention predicate is registered in the attention word database 23.

  The dependency analysis unit 77 analyzes the dependency structure of the article text. For example, similarly to the vector generation unit 72 and the attention word collection unit 76, the dependency analysis unit 77 extracts article data within the target period written about the target company from the article database 31. Then, a binary term and a ternary term word group are extracted from the article text 311 of the extracted article data by the same processing as the dependency determination unit 132.

  The dependency term extraction unit 78 extracts a word group including the attention word stored in the attention word database 23 from the word group of the binary relationship or the ternary relationship extracted by the dependency analysis unit 77. In addition, the dependent term extraction unit 78 acquires the industry code to which the target company belongs from the industry database 33, associates the extracted word group with this industry code, and further calculates the sensitivity calculated by the sensitivity calculation unit 75. Together with the dependent term database 22.

  The vector database 21, the dependent term database 22, and the attention word database 23 generated by the above processing are used in the information providing apparatus 1 described above. Therefore, the article data analysis device 7 provides the databases 21, 22, and 23 generated as described above to the information providing device 1. For example, as shown in FIG. 12, in a system in which the information providing source system 5, the article data analyzing device 7, and the information providing device 1 are connected via the network 2, the information providing device 1 and the article data analyzing device. 7 collects article data from the information providing source system 5, respectively. Then, the article data analysis device 7 can provide the above-described databases 21, 22, and 23 to the information providing device 1.

  Next, an automatic trading apparatus that generates and automatically executes stock trading orders based on distribution data 110 output by the information providing apparatus 1 will be described with reference to FIG.

  As shown in the figure, the automatic trading apparatus 8 is connected to an information providing apparatus 1 and a securities company system 9 via a network 2.

  The automatic trading apparatus 8 is configured by, for example, a general-purpose computer system, and individual components or functions in the automatic trading apparatus 8 described below are realized by executing, for example, a computer program. Further, the automatic trading apparatus 8 can also be realized on the same computer machine as the information providing apparatus 1.

  The automatic trading apparatus 8 includes a reliability determination unit 81, a trading order generation unit 82, an order execution unit 83, a trading condition database 84, a portfolio database 85, and an order database 86.

  The automatic trading apparatus 8 acquires the distribution data 110 generated by the information providing apparatus 1. Then, the reliability determination unit 81 determines the prediction of the stock price fluctuation based on the article data, that is, the reliability of the sensitivity 116 included in the distribution data 110. For example, when the related information 117 includes many related articles, the reliability of the sensitivity 116 is determined according to the number of articles.

  The buy / sell order generator 82 generates buy / sell order data based on the current portfolio stored in the portfolio database 85 for the stock of the company whose reliability is determined to be a certain level or higher by the reliability determination unit 81. . For example, if the propensity of the stock price is positive, it may be bought, and if it is negative, it may be sold. Furthermore, the trading quantity may be determined according to the sensitivity. Further, the sales order generation unit 82 refers to the sales condition database 84 in which the sales conditions such as the upper limit of the sales price and the number of times of sales per day are stored in advance. May be generated.

  The order execution unit 83 executes the sales order according to the sales order data generated by the sales order generation unit 82. For example, the automatic trading apparatus 8 transmits trading order data to the securities company system 9 and requests the securities company to trade. Then, the sales order transmitted to the securities company system 9 is stored in the order database 86.

  When the securities company system 9 receives the sales order data, the securities company system 9 performs sales based on the data.

  The above-described embodiments of the present invention are examples for explaining the present invention, and are not intended to limit the scope of the present invention only to those embodiments. Those skilled in the art can implement the present invention in various other modes without departing from the gist of the present invention.

It is a figure which shows the whole structure of the system which acquires the information regarding a company, adds the information which supports investment judgment, and provides to a user. It is a flowchart which shows the collection process procedure of article information. It is a figure which shows an example of the data structure of article data. It is a figure which shows an example of the data structure of an article database. It is a flowchart which shows the procedure of an article analysis / delivery process. It is a flowchart which shows the process sequence when there is no similar vector. It is a figure which shows an example of the data structure of a vector database. It is a figure which shows an example of the data structure of delivery data. It is a figure which shows an example of the data structure of an acceptance term database. It is a figure which shows the example of a display of the article and investment judgment assistance information in a user terminal. It is a figure which shows the whole structure of an article data analysis apparatus. It is a figure which shows the structure of the system which provides a database via a network. It is a figure which shows the system configuration for performing a trade order automatically.

Explanation of symbols

1: Investment support information providing device 2: Network 5: Information providing source system 6: User terminal 7: Article data analyzing device 8: Automatic trading device 9: Securities company system

Claims (17)

For each company, a means for storing statistical information obtained by statistically processing the relationship between articles and stock price fluctuations,
A way to get article data about a company,
Means for adding, to the article data, information for supporting judgment of stock investment based on the article indicated by the article data based on the acquired article data and statistical information stored in the storage means;
An investment decision support information providing device comprising: means for outputting the article data and the information for supporting the added stock investment decision. Means for receiving article data relating to a certain company and information for supporting judgment of stock investment based on an article indicated by the article data;
Means for displaying the received article data and information for supporting judgment of the stock investment side by side on a display screen;
The information for supporting the judgment of the stock investment includes information indicating whether the stock price of the company is on an upward trend or a downward trend. An investment determination support information display device, wherein a display mode of information for supporting determination of the stock investment on the display screen is different. Means for reading out document data including a document describing a company from the document data storage means;
Means for associating and storing information indicating the contents of a plurality of past documents written about the company and information indicating stock price fluctuations of the company in each period including the time when each document was issued; ,
Means for analyzing the content of a document included in the read document data;
Means for referring to the storage means based on the analysis result of the document and determining the similarity between the document to be analyzed and a plurality of past documents;
Investment judgment comprising: means for adding the information indicating the stock price fluctuation associated with the information indicating the content of the past document determined to be similar as a result of the similarity determination to the document data and outputting the information Support information providing device. The information indicating the content of the past document stored in the storage means is multidimensional vector data indicating the content of the document,
The analysis means generates a multidimensional vector indicating the content of the document included in the read document data,
The means for determining the degree of similarity determines the degree of similarity between a multidimensional vector of a plurality of past documents stored in the storage unit and the multidimensional vector generated by the analyzing unit. Investment decision support information providing device.   5. The investment determination support information providing apparatus according to claim 4, wherein each of the multidimensional vectors of the plurality of past documents stored in the storage means is a centroid vector of the multidimensional vectors based on the plurality of documents.   4. The investment decision support information providing apparatus according to claim 3, wherein the information indicating the stock price fluctuation is a sensitivity to the stock price fluctuation in the period. A group of words written in a plurality of past documents written about a company in the industry to which the company belongs and a list of companies in the industry in each period including the time when each document was issued Word group storage means for storing information indicating stock price fluctuations in association with each other;
Means for extracting a word group having the predetermined relationship from a document included in the read document data when there is no similarity as a result of the similarity determination;
The output means refers to the word group stored in the word group storage means based on the extraction result, acquires information indicating the stock price fluctuation of the company in the industry, and adds the information to the document data for output. Item 4. The investment decision support information providing device according to Item 3. Means for reading out document data including a document describing a company from the document data storage means;
A group of words written in a plurality of past documents written about a company in the industry to which the company belongs and a list of companies in the industry in each period including the time when each document was issued Means for associating and storing information indicating stock price fluctuations;
Means for extracting a word group having the predetermined relationship from a document included in the read document data;
Investment decision support comprising: means for referring to a word group stored in the storage means based on the extraction result, obtaining information indicating stock price fluctuations of companies in the industry, and adding the information to the document data for output. Information providing device. The word group having a predetermined relationship stored in the storage means is a word group having a binary relationship or a ternary relationship obtained as a result of analyzing each dependency structure of the plurality of past documents. ,
The investment determination support information providing apparatus according to claim 8, wherein the extraction unit analyzes a dependency structure of the document and extracts a word group having a binary or ternary relationship. For each company, an investment decision support information providing device having a storage means for storing statistical information obtained by statistically processing the relationship between articles and stock price fluctuations,
Get article data about a company,
Based on the acquired article data and the statistical information stored in the storage means, information for supporting judgment of stock investment based on the article indicated by the article data is added to the article data,
A method for providing investment judgment support information, wherein the article data and information for supporting judgment of the added stock investment are output. A computer program for realizing an investment decision support information providing apparatus when executed on a computer,
The investment determination support information providing device includes:
For each company, a means for storing statistical information obtained by statistically processing the relationship between articles and stock price fluctuations,
A way to get article data about a company,
Means for adding, to the article data, information for supporting judgment of stock investment based on the article indicated by the article data based on the acquired article data and statistical information stored in the storage means;
A computer program comprising: means for outputting the article data and information for supporting determination of the added stock investment. Computer
Receives article data relating to a company and information for supporting a stock investment decision based on an article indicated by the article data, including information indicating whether the stock price of the company is increasing or decreasing. And
When the stock price of the company is on an upward trend, the received article data and information for supporting the judgment of the stock investment in the first display mode are displayed side by side on a display screen,
When the stock price of the company is in a downward trend, the display of the investment determination support information that displays the received article data and information for supporting the determination of the stock investment in the second display mode side by side on the display screen Method. When executed on a computer,
Receives article data relating to a company and information for supporting a stock investment decision based on an article indicated by the article data, including information indicating whether the stock price of the company is increasing or decreasing. And
When the stock price of the company is on an upward trend, the received article data and information for supporting the judgment of the stock investment in the first display mode are displayed side by side on a display screen,
When the stock price of the company is on a downward trend, the received article data and the investment judgment support information for displaying on the display screen the information for supporting the judgment of the stock investment in the second display mode are displayed. Computer program to do. Computer
Reading out document data including a document describing a company from the document data storage means,
The information indicating the contents of a plurality of past documents written about the company and the information indicating the stock price fluctuation of the company in each period including the time when each document is issued are stored in association with each other.
Analyzing the content of the document included in the read document data;
Based on the analysis result of the document, refer to the storage means, determine the similarity between the document to be analyzed and a plurality of past documents,
Providing investment decision support information that outputs the information indicating the stock price fluctuation associated with the information indicating the contents of the past document determined to be similar as a result of the similarity determination, added to the document data and output Method. When executed on a computer,
Reading out document data including a document describing a company from the document data storage means,
The information indicating the contents of a plurality of past documents written about the company and the information indicating the stock price fluctuation of the company in each period including the time when each document is issued are associated with the storage means. Store and
Analyzing the content of the document included in the read document data;
Based on the analysis result of the document, refer to the storage means, determine the similarity between the document to be analyzed and a plurality of past documents,
Investment decision support information is provided for outputting, in addition to the document data, information indicating the stock price fluctuation associated with information indicating the content of a past document determined to be similar as a result of the similarity determination Computer program to do. Computer
Reading out document data including a document describing a company from the document data storage means,
A group of words written in a plurality of past documents written about a company in the industry to which the company belongs and a list of companies in the industry in each period including the time when each document was issued Corresponding information indicating stock price fluctuation is stored in the storage means,
Extracting a word group having the predetermined relationship from a document included in the read document data;
A method for providing investment decision support information that refers to a word group stored in a storage unit based on an extraction result, acquires information indicating a stock price fluctuation of a company in the industry, and adds the information to the document data for output. . When executed on a computer,
Read out document data including a document describing a company from the document data storage means,
A group of words written in a plurality of past documents written about a company in the industry to which the company belongs and a list of companies in the industry in each period including the time when each document was issued Corresponding information indicating stock price fluctuation is stored in the storage means,
Extracting a word group having the predetermined relationship from a document included in the read document data;
Provide investment decision support information that refers to the word group stored in the storage means based on the extraction result, obtains information indicating stock price fluctuations of companies in the industry, and outputs the information added to the document data Computer program for.