JP2011204199A - System for providing information of stock price fluctuation event, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system that automatically detects a stock price fluctuation event in the past.SOLUTION: A system searches a dictionary data storage unit 22 that registers a correspondence relation between a specific expression character string and an abstraction tag, an article data storage unit 20 that stores article data, and a rule data storage unit 24 that stores extraction rules specifying that the character string should be extracted for each combination of the abstraction tags included in the article data or each combination of the abstraction tag and a specific character string where the article data storage unit 20 is retrieved by using a stock name and an event type which are transmitted from a client terminal 34 as keywords and the related article data are resolved into morphemes, a corresponding abstraction tag is given to a character string that expresses at least a stock name, an event type, and an event occurred time among the morphemes, the extraction rule is applied to each article data, stock price fluctuation event information composed of the stock name, the event type, the event occurrence time is created, and a search result screen showing the stock price fluctuating event information is transmitted to the client terminal 34.

Description

  The present invention relates to a stock price fluctuation event information providing system and program, and more particularly, to a technology for searching past events that affect stock prices for each brand and presenting the time of occurrence and stock price information at that time.

  In the event of an event that adversely affects stock prices, such as product recalls or significant downward revisions to corporate performance, investors need to determine how far the stock price will drop in order to purchase the stock at the lowest price. . Conversely, if an event that has a positive impact on the stock price occurs, such as the launch of a revolutionary new product or a public offering of a subsidiary, the investor sells the stock at the ceiling price, so how far the stock price will increase in the future? It is necessary to identify.

In this way, as a method to divide the bottom price and ceiling price when a relatively large stock price fluctuation event occurs, it is often determined whether the current value is appropriate or not by referring to the stock price movement when a similar event in the past has occurred. It is practiced by investors.
For example, when a large-scale recall of an automobile manufacturer occurred, the same manufacturer had caused the same recall problem 10 years ago, when the stock price fell by up to 30% compared to the latest peak. Referring to the example shown, the investor speculates, “The current value, which has fallen by only 20% compared to the latest peak, is not the bottom, and may drop by about 10%.” Can do. Alternatively, when the current value has already dropped by 35% compared to the latest peak, it can be recognized that “the current value is too low and may soon increase by about 5%”.

Events and Stock Price / Stock and Brand Features Internet URL: http://www1394.hp.infoseek.co.jp/money/kabumeigara.html Search Date: March 20, 2010 Stock Price Event Analysis DP Final Version Internet URL: http://www.jftc.go.jp/cprc/DP/CPDP-12-J.pdf Search Date: March 20, 2010

  In this way, when a stock price fluctuation event occurs, referencing similar past cases is extremely effective in determining the timing of buying and selling, so investors can use the “Yamada Car Recall” on the Internet search site. In this way, a search expression in which the brand name and the event type are connected with the AND condition is input, the occurrence time of the same type event in the past is examined, and stock price information at that time is obtained.

  However, the reliability of information written on Web pages on the Internet is a mix of cobblestones and is not organized in an orderly manner, so investors need to continue to search hard until the intended information is obtained.

  Moreover, even if the past occurrence time of similar events can be detected successfully, it takes a lot of trouble to individually acquire the stock price fluctuation data at that time and superimpose it on the current price movement.

  In addition, since it is often difficult to estimate the scale (impact = impact level) from the description of the Web page, if an event of the same type has occurred several times in the past, There is also a situation where it is wondering whether to refer to price movements.

  Furthermore, the impact of a particular event on stock prices is basically common to all countries, so if you can refer to similar projects in other countries, you can expect a wider range of judgment materials and more appropriate conclusions. In fact, there was a problem that only cases in the English-speaking world could be referred to in addition to the cases in their own country.

The present invention has been devised to overcome such a current situation, and has as its first object the realization of a system in which a user can detect a specific type of stock price fluctuation event in the past very easily.
A second object of the present invention is to realize a system that automatically acquires stock price fluctuation data before and after the detected past event occurrence date and enables comparison with the current price movement.
A third object of the present invention is to realize a system in which a user can select an event corresponding to the current event size by providing a mechanism for determining the same event size in the past.
Further, the present invention provides a system that can widely refer to stock price fluctuation events occurring in foreign countries by extracting events from information described in a language other than the user's native language. The purpose is.

  In order to achieve the above object, the stock price fluctuation event information providing system according to claim 1 stores dictionary data defining a correspondence relationship between a specific expression character string and an abstract tag indicating the type of the character string. Dictionary data storage means to be stored, article data storage means to store digitized article data, and combinations of abstract tags included in article data, or combinations of abstract tags and specific character strings, Extraction rule storage means for storing an extraction rule that defines a character string to be extracted as a constituent element of stock price fluctuation event information, and when a stock price fluctuation event search request is transmitted from a client terminal, at least a brand and an event A search screen provided with a type designation field is transmitted to prompt the designation of the brand and event type, and from the client terminal, the brand and the event When the specified information of the event type is transmitted, the article data storage means is searched using the brand and event type as a key, and the related article data is extracted, and the extracted article data is decomposed into morphemes. Morpheme analyzing means for referring to the dictionary data, abstract means for assigning a corresponding abstract tag to a character string representing at least a brand, an event type, and a time in each morpheme, and each article data By applying the above extraction rule, event information generating means for generating stock price fluctuation event information by extracting at least a character string representing a stock, a character string representing an event type, and a character string representing an event occurrence time, and the stock price fluctuation A search result screen in which event information is described is generated and transmitted to the client terminal.

  Further, the stock price fluctuation event information providing system according to claim 2 is the system of claim 1, and the dictionary data storage means, article data storage means, and extraction rule storage means include data related to a plurality of languages. The article data extraction means converts the brand and event type transmitted from the client terminal into corresponding words in each language, and then related articles from the article data described in the corresponding language using these as keys. A process of extracting data is executed, the morpheme analyzing means executes a process of decomposing each related article data into morphemes corresponding to a corresponding language, and the abstracting means adds an abstract tag corresponding to each language. The event information generation means executes processing to be assigned to each morpheme and applies the extraction rule for the corresponding language to each article data to change the stock price fluctuation event. It is characterized by executing a process of extracting information.

  The stock price fluctuation event information providing system according to claim 3 is the system according to claim 1 or 2, further storing a scale determination rule defining a correspondence relationship between the number of article data and the scale of the stock price fluctuation event. Storage means, and means for counting the number of articles when there are multiple article data related to the same issue and stock price fluctuation event. Means for specifying the scale of the fluctuation event, and the search result screen includes the scale of each stock price fluctuation event.

  The stock price fluctuation event information providing system according to claim 4 is the system according to claim 1 or 2, further comprising a scale determination that defines a correspondence relationship between a temporal distribution range of similar article data and a scale of a stock price fluctuation event. The interval between the reporting date of the first article data and the reporting date of the last article data when there is a plurality of article data related to the same brand and stock price fluctuation event, and the storage means for storing the rules And means for identifying the scale of the stock price fluctuation event by applying this interval to the scale determination rule, and the search result screen includes the scale of each stock price fluctuation event. It is characterized by.

  The stock price fluctuation event information providing system according to claim 5 is the system according to claim 1 or 2, and further, the dictionary data storage means includes an expression character string representing the scale of the stock price fluctuation event and its abstraction. When the correspondence relationship with the tag is registered and the abstraction means includes an expression character string representing the scale of the stock price fluctuation event in the article data, the processing of assigning the corresponding abstract tag is performed. And the event information generating means executes a process of extracting a character string to which an abstract tag representing the scale is attached according to the extraction rule as a constituent element of the event information. It is characterized by the scale of the event.

  In the stock price fluctuation event information providing system according to claim 6, a stock performance fluctuation event search request is received from a company performance data storage means for storing performance forecast data and performance data for each company's settlement period, and a client terminal. When it is sent, a search screen with at least a brand and event type designation field is sent, and a means for prompting designation of the brand and event type, and a downward revision of corporate performance as a brand and event type from the client terminal or When information specifying upward revision is transmitted, means for extracting the performance forecast data and performance data for each fiscal period related to the stock from the above-mentioned company performance data storage means, the performance data for each fiscal period and the forecast for the current period If the difference from the data is calculated and the actual data exceeds the expected data, at least the When stock price fluctuation event information including the closing event type and closing date as the event occurrence time is generated, and if the actual data is lower than the expected data, at least the stock, the downward revision event type, and the event occurrence time The event information generating means for generating the stock price fluctuation event information including the closing period of the stock price and the means for generating the search result screen in which the stock price fluctuation event information is described and transmitting it to the client terminal are provided.

  The stock price fluctuation event information providing system according to claim 7 is the system according to claim 6, and the company performance data storage means stores data in a plurality of countries described in a plurality of languages. It is characterized by that.

  The stock price fluctuation event information providing system according to claim 8 is the system according to claim 6 or 7, further defining a correspondence relationship between a range of difference between the actual data and the forecast data and a scale of the stock price fluctuation event. The scale of the downward revision event or the upward revision event by applying the difference between the forecast data and the forecast data for each fiscal year to the scale judgment rule, and the scale judgment rule storage means for storing the scale judgment rule. And the scale of each stock price fluctuation event is written on the search result screen.

  The stock price fluctuation event information providing system according to claim 9 is the system according to claims 1 to 8, further comprising stock price history data storage means for storing stock price history data of each brand, and a specified from a client terminal. Means for extracting stock price history data within a predetermined range from the stock price history data storage means based on the occurrence time of the stock price fluctuation event when a stock price fluctuation chart display request relating to a stock price fluctuation event is transmitted; Means for generating a stock chart including a first graph representing a time-series change in stock prices drawn based on stock price history data; means for transmitting a stock chart display screen on which the stock chart is described to the client terminal; A re-display request for a stock price chart that specifies the time of a stock price fluctuation event to be compared is sent from the client terminal A stock price history data in a predetermined range based on the occurrence time of the stock price fluctuation event from the stock price history data storage means, and a time series of stock prices drawn based on the extracted stock price history data Means for generating a stock chart in which the second graph representing the change and the first graph are arranged in the same graph area so that the event occurrence times of the two graphs overlap, and a stock chart describing the stock chart And a means for transmitting the display screen to the client terminal.

  The stock price fluctuation event information providing program according to claim 10, wherein the computer stores dictionary data in which dictionary data defining a correspondence relationship between a specific expression character string and an abstract tag indicating its type is stored. Article data storage means for storing digitized article data, a combination of abstract tags included in article data, or a component of stock price fluctuation event information for each combination of an abstract tag and a specific character string Extraction rule storage means for storing an extraction rule that defines a character string to be extracted as a stock price fluctuation event search request is sent from a client terminal, at least a brand and event type designation field is provided A means of prompting the designation of the brand and event type by sending a search screen. When the fixed information is transmitted, the article data storage means is searched using the brand and event type as a key, the article data extracting means for extracting the related article data, and the morpheme analysis for decomposing the extracted article data into morphemes Means, referring to the above dictionary data, abstracting means for assigning a corresponding abstract tag to a character string representing at least a brand, an event type, and a time in each morpheme, and applying the above extraction rule to each article data The event information generating means for generating stock price fluctuation event information by extracting at least a character string representing a stock, a character string representing an event type, a character string representing an event occurrence time, and the stock price fluctuation event information are described. It is characterized by generating a search result screen and functioning as means for transmitting to the client terminal.

  According to the stock price fluctuation event information providing system according to claim 1 and the stock price fluctuation event information providing program according to claim 10, a past stock price fluctuation event that matches a brand and an event type specified by a user is digitized. Since it is automatically extracted from the article data and its occurrence time is presented, the user can refer to past similar events very easily.

  According to the stock price fluctuation event information providing system according to claim 2, since the stock price fluctuation event is also extracted from the article data described in a plurality of languages, the user can widely refer to similar events that have occurred in foreign countries. Become.

  According to the stock price fluctuation event information providing system according to claim 3, the scale of the event is determined according to the amount of article data related to the same stock price fluctuation event, and is presented to the user as an attribute of the stock price fluctuation event. The user can accurately refer to past events corresponding to the scale of the same type of event that is currently occurring.

  According to the stock price fluctuation event information providing system according to claim 4, the scale of the event is determined according to the length of the temporal distribution range (span) of the article data related to the same stock price fluctuation event, and the stock price fluctuation event Since it is presented to the user as an attribute, the user can accurately refer to a past event corresponding to the scale of the same kind of event currently occurring.

  According to the stock price fluctuation event information providing system according to claim 5, the scale of the event is determined based on a specific description in the article data related to the stock price fluctuation event, and is presented to the user as an attribute of the stock price fluctuation event. Therefore, the user can accurately refer to past events corresponding to the scale of the same type of event that is currently occurring.

  According to the stock price fluctuation event information providing system according to claim 6, based on the company performance data that is quantitative data, the performance downward correction event or the performance upward correction event related to the stock designated by the user is automatically extracted. Since the occurrence time is presented, the user can refer to past performance correction events very easily.

  According to the stock price fluctuation event information providing system according to claim 7, since the business performance downward revision event and the business performance upward revision event are generated from the corporate performance data described in a plurality of languages, the user can perform past business performance in a foreign country. It is possible to refer to correction events widely.

  According to the stock price fluctuation event information providing system according to claim 8, the scale of the performance correction event is determined according to the degree of divergence between the forecast data related to the company performance and the performance data, and is presented to the user as an attribute of the stock price fluctuation event. Therefore, the user can accurately refer to past events corresponding to the scale of the same type of event that is currently occurring.

  According to the stock price fluctuation event information providing system according to claim 9, the stock price history data related to the past stock price fluctuation event selected by the user is automatically extracted, and a stock price fluctuation graph based on the stock price history data is presented. A stock price fluctuation graph relating to the comparison target event (for example, the current event) designated by is generated and a stock price fluctuation chart in which both graphs are superimposed is presented to the user. Validity and future price movements can be estimated.

  FIG. 1 is a block diagram showing a functional configuration of a stock price fluctuation event information providing system 10 according to the present invention. A web server 12, a qualitative event generating unit 14, a quantitative event generating unit 16, a chart generating unit 18, The article data storage unit 20, the dictionary data storage unit 22, the rule data storage unit 24, the company attribute data storage unit 26, the company performance data storage unit 28, and the stock price history data storage unit 30 are included.

The qualitative event generation unit 14, the quantitative event generation unit 16, and the chart generation unit 18 are realized by the CPU of the computer executing necessary processes according to the OS and application programs.
The article data storage unit 20, dictionary data storage unit 22, rule data storage unit 24, company attribute data storage unit 26, company performance data storage unit 28, and stock price history data storage unit 30 are provided in the hard disk of the computer. ing.

The web server 12 is connected to a client terminal 34 operated by a user via the Internet 32, and a qualitative event generation unit 14, a quantitative event generation unit 16, a chart generation unit according to a request transmitted from each client terminal 34 18 has a function of requesting processing to 18 and transmitting various screens (Html files) describing the processing results to the client terminal 34 (details will be described later).
The client terminal 34 is a PC or the like equipped with a web browser program.

The article data storage unit 20 stores newspaper article data and magazine article data that have been converted into text for the past several decades.
This article data is not limited to Japanese articles published in Japan, but is published in foreign languages such as English, Chinese, German, French, Korean, etc. Is widely available.

In the dictionary data storage unit 22, a large number of various types of dictionary data defining the correspondence between a specific expression character string and an abstract character string indicating its type are registered (details will be described later). This dictionary data is not limited to a Japanese dictionary, and dictionary data for major foreign languages such as English, Chinese, German, French and Korean are prepared.
In the dictionary data storage unit 22, a large number of correspondence relationships between character strings representing event types and synonyms of the character strings are registered for each language unit (details will be described later).

The rule data storage unit 24 stores qualitative event extraction rules, anaphora analysis rules, event scale determination rules, and quantitative event extraction rules (details will be described later).
In each rule, the correspondence between application conditions and processing contents is described in a predetermined programming language.
A large number of qualitative event extraction rules, anaphora analysis rules, event scale determination rules, and quantitative event extraction rules are also prepared for each national language.

The company attribute data storage unit 26 stores attribute information (official name, abbreviation, listed market, brand code, business type, etc.) relating to the brand company of the stock (details will be described later).
A lot of this attribute information is also stored for each country.

The company performance data storage unit 28 stores past company performance data (predicted values of sales, operating profit, ordinary profit, etc. and actual values thereof) announced by each brand company.
Many company performance data are also stored for each country.

The stock price history data storage unit 30 stores daily stock price data for each brand for the past several decades.
This stock price data is not limited to stock price data in Japan, but stock price data at exchanges in each country is prepared.

  The article data, the company attribute data, the company performance data, and the stock price history data may be acquired from an external database server as needed instead of being held inside the system 10.

Next, a processing procedure in the stock price fluctuation event information providing system 10 will be described with reference to the flowcharts of FIGS.
First, when the Web server 12 receives a request for a stock price fluctuation event search from the client terminal 34 (S10), a search screen is transmitted to the client terminal 34 (S12).

FIG. 4 shows the stock price fluctuation event search screen 40, in which an event type, a business type, a brand, and a scale are provided as search condition input items.
On the other hand, the user first expands an event type pull-down menu and selects an arbitrary event type. As the event type, a plurality of terms representing types of stock price fluctuation events are prepared in advance, such as recall, hit product, food poisoning, downward revision of performance, upward revision of performance, and public offering of subsidiaries. Here, it is assumed that “recall” is selected.

Next, the user expands the industry pull-down menu and selects an arbitrary industry. Here, it is assumed that “automobile” is selected.
Next, the user expands a brand pull-down menu and selects an arbitrary brand. In this case, since the user selects “automobile” as the business type, only the brands having the business type attribute of the car are listed in the pull-down menu. Here, it is assumed that “Yamada Motor Corporation” is selected.
Next, the user expands a scale pull-down menu and selects an arbitrary scale (large, medium, small, no designation). Here, it is assumed that “not specified (= all scales)” is selected.

When the user completes the search condition setting and clicks the search button 42, the set search condition is transmitted to the Web server 12.
Receiving this (S14), the Web server 12 determines whether the event type selected by the user is a qualitative event or a quantitative event (S16).
In FIG. 4, since “recall”, which is a qualitative event, is selected as the event type, the Web server 12 passes a search condition to the qualitative event generating unit 14 and requests a process for generating qualitative event information.
Hereinafter, the procedure of processing by the qualitative event generation unit 14 will be described with reference to the flowchart of FIG.

First, the qualitative event generation unit 14 filters article data that matches the current search condition from a large amount of article data stored in the article data storage unit 20 (S18-01).
Since the search conditions this time are “event type: recall”, “industry: automobile”, “brand: Yamada automobile”, the qualitative event generation unit 14 appropriately stores these keywords and stores them in the article data storage unit 20. Article data that matches the conditions is retrieved.

As described above, the article data storage unit 20 includes article data in each language. As shown in FIG. 6, the dictionary data storage unit 22 stores definition data in which a plurality of synonyms are associated for each language such as Japanese, English, Chinese, etc., for the event type “recall”. .
Therefore, the qualitative event generation unit 14 extracts Japanese synonyms as OR conditions such as “Recall OR Product Recovery OR Free Recovery OR…” when extracting the corresponding article data from the article data group described in Japanese. The search expression combined with is used.
In addition, the qualitative event generation unit 14 combines Chinese synonyms with OR conditions such as “recovery OR complete OR…” when extracting the corresponding article data from the article data group described in Chinese. Search formula.
The same applies to other event types.

Although not shown in the figure, the dictionary data storage unit 22 also registers a plurality of words corresponding to the automobile manufacturing and sales business for each country language for “automobile” which is an industry.
Also, although not shown in the figure, the dictionary data storage unit 22 also stores a plurality of patterns of notation examples in each language for the brand “Yamada Motor”.

  In short, the qualitative event generation unit 14 expands the search conditions set by the user into corresponding synonyms for each language unit, generates a search expression that connects them with OR conditions, and uses this to generate article data in each language. In order to improve the efficiency of subsequent processing, a rough sieving is performed between the target event article and other general articles.

Next, the qualitative event generation unit 14 decomposes each extracted article data into morpheme units and identifies parts of speech (S18-02).
Next, the qualitative event generation unit 14 refers to the company name dictionary, event dictionary, time expression dictionary, employee expression dictionary, amount expression dictionary, etc. stored in the dictionary data storage unit 22, and the corresponding abstract tag for the corresponding morpheme (Meta tag) is assigned (S18-03).

  For example, as shown in FIG. 7 (a), there is a character string “Yamada Automobile” in the sentence No.001 of an event article X (Article ID: 2002 ...), and “Yamada Automobile” in the company name dictionary. "Is registered as the brand company name, the qualitative event generation unit 14 associates the <company name> abstract tag with the character string" Yamada Motor ".

  In addition, since the character string “15th of this month” exists in the same sentence and the registered example of “this month * day (* is a wild card)” exists in the time expression dictionary, the qualitative event generation unit 14 Associates the <period> abstract tag with the character string “15th of this month”.

  In addition, since the “recall” character string exists in the same sentence and the registered example of “recall” exists in the event dictionary, the qualitative event generation unit 14 < Associate event> abstract tag.

  In addition, since the character string “PR” is present in the sentence No. 0002 of the event article X, and there is a registered example of “PR” in the employee expression dictionary, the qualitative event generation unit 14 Associates the <employee> abstract tag to the string "PR spokesperson".

Next, the qualitative event generation unit 14 executes an anaphora analysis process in which, when a predetermined anaphor is included in the article, this is replaced with a corresponding antecedent (S18-04).
For example, in the sentence No. 0002 of the event article X, there is a character string “Company”, and this “Company” is defined as an anaphor in the anaphoric analysis rule stored in the rule data storage unit 24. In accordance with this anaphora analysis rule, the event generation unit 14 recognizes “Yamada Automobile”, which is a character string with the <company name> abstract tag included in the latest preceding sentence, as an antecedent and executes a replacement process. To do. As a result, “Company” is replaced with “Yamada Motor Corporation <Company Name>”.

Next, the qualitative event generation unit 14 specifies the company name (brand) that is the subject of the event from the event article according to the qualitative event extraction rules stored in the rule data storage unit 24 (S18-05).
For example, as an extraction rule, the case subject “GA” or the auxiliary particle “HA” is connected immediately after the character string with the abstract tag <company name> → the character string with <company name> attached to the event subject Therefore, the qualitative event generation unit 14 recognizes “Yamada Motor” in the event article X as an event subject.

In addition, as an extraction rule, “<company name><employee>”,<event> and “announcement” character strings exist in the same sentence → character string with <company name> specified as the subject of the event "Is stored in the rule data storage unit 24," Yamada Motor "is recognized as the event subject from the following sentence α.
(Sentence α) Yamada Motor Corporation <company name> bulletin officer <employee> announced on 15th of this month <time> that the new Orion will be recalled <event>.

Next, the qualitative event generation unit 14 specifies the event occurrence date from the event article according to the extraction rules stored in the rule data storage unit 24 (S18-06).
For example, as the extraction rule “character string with abstract tag of <time> → specified as event occurrence date” is stored, the qualitative event generation unit 14 follows this from the event article X to “ ”Is recognized as the event occurrence date.
At this time, the qualitative event generation unit 14 refers to the reporting date of the event article X, March 16, 2002, and replaces the character string “current month” with a specific “March 2002”.
If there is no character string representing the time in the article, it is recognized as “event occurrence date = report date of the article” for convenience.

Next, the qualitative event generation unit 14 generates article unit event information from each event article (S18-07).
As shown in FIG. 7B, the article-unit event information includes data items such as article ID, group ID, event type, event subject, country of occurrence, and date of occurrence. However, since the group ID item is filled with a value in the next processing, it is blank at this point.

Next, the qualitative event generating unit 14 executes a grouping process for bringing together event articles related to the same event (S18-08).
For example, event articles whose event type, event subject, and date of occurrence are completely the same are assigned to the same group ID item as belonging to the same group.
Even if there is a deviation within a predetermined range (for example, within two days) on the day of occurrence, if the event type and the event subject match, the same group is recognized and the same group ID is assigned.
On the other hand, even if the event type and the date of occurrence match, if the event subject does not match, the comparison target articles are not recognized as belonging to the same group.

Next, the qualitative event generation unit 14 counts the number of event articles belonging to each event group, and applies the scale determination rule stored in the rule data storage unit 24 to this, thereby the scale of the event (impact = impact) Force) is determined (S18-09).
For example, the scale determination rule stipulates that “recall scale criteria: less than 10 articles → small, 10 ≦ number of articles <30 → medium, 30 ≦ number of articles → large”, and the number of articles belonging to the article group. If there are 35 cases, it is determined that “scale: large”.

As described above, instead of determining the size based on the number of related articles, it is also possible to determine the scale based on the length of the span of the reporting date between related articles. The fact that the media has been reported for a long period of time represents a high level of interest in the event and the severity of the impact.
For example, in the group of articles with the same group ID, the number of days between the oldest report date and the latest report date is counted. If it is less than 10 days, “scale: medium” and if it is 10 days or more, “scale: large” is determined.
For this purpose, a dedicated rule for determining the scale according to the span of the article is registered in the rule data storage unit 24 in advance.

Alternatively, the size can be determined based on a specific description in the article.
For example, if there is a sentence in the article that says “the amount of loss due to this recall <event> is estimated to be 20 billion yen <amount>”, the <amount> abstract tag is assigned from here. In addition, “20 billion yen” is taken out and applied to the following scale determination rule stored in the rule data storage unit 24 to recognize “scale = large”.
■ Recall scale judgment rule: Amount <100 million yen → Small, 100 million yen ≤ Amount <500 million yen → Medium, 500 million yen ≤ Amount → Large

As this premise, the qualitative event generation unit 14 refers to the money amount expression dictionary in the dictionary data storage unit 22 and assigns an abstract tag of <money amount> to a predetermined money amount expression.
For example, when “100 million yen”, “10 million yen”, and “million yen” are defined as money expression in the money expression dictionary, the qualitative event generation unit 14 places these character strings and immediately before them. The calculated arithmetic numbers or Chinese numerals are recognized as character strings representing money amounts, and an abstract tag of <money amount> is given to them.

Next, the qualitative event generation unit 14 generates qualitative event information based on the processing result (S18-10).
FIG. 7C shows an example of qualitative event information, which includes data items of event ID, event type, brand, country of occurrence, date of occurrence, and scale.

Finally, the qualitative event generation unit 14 extracts information that matches the search condition from a large number of generated qualitative event information (S18-11).
For example, when the user has specified “event scale: large” as a search condition, the qualitative event generation unit 14 extracts only qualitative event information in which “large” is set in the scale data item, and “ Discards qualitative event information for which "Medium" or "Small" is set.

Receiving the qualitative event information from the qualitative event generating unit 14, the Web server 12 generates a search result screen and transmits it to the client terminal 34 (S22 in FIG. 2).
FIG. 8 shows the search result screen 44, in which two pieces of event information are displayed in a list.
When the user checks the check box for event information No. 2 and clicks the detailed browsing button 46, a detailed browsing request for the event information is transmitted to the Web server 12.

Receiving this (S24), the Web server 12 generates a detailed screen of the event information and transmits it to the client terminal 34 (S26).
Although not shown in the figure, this detailed screen lists the title, news source, and news report date of the article data that is the basis for generating the event information.

  Each article data is displayed in a clickable manner and is linked to the corresponding article data stored in the article data storage unit 20, so when the user clicks one of these, the article is sent via the web server 12. The display screen is transmitted to the client terminal 34.

  Next, when the user clicks the chart button 47 with the No. 2 event information check box checked in the search result screen 44 of FIG. 8, a stock price chart display request for the event information is sent to the Web server 12. Is done.

Receiving this (S28 in FIG. 3), the Web server 12 passes the brand selected by the user, the country of occurrence of the past event, and the date of occurrence to the chart generation unit 18 and requests the generation process of the stock price chart.
On the other hand, the chart generation unit 18 extracts stock price information (for example, 30 days before and after the occurrence date) of the corresponding issue in the relevant country from the stock price history data storage unit 30 before and after the past event occurrence date, A stock price chart including the plotted graph is generated (S30).

Receiving the stock price chart from the chart generator 18, the Web server 12 generates a stock price chart display screen and transmits it to the client terminal 34 (S32).
FIG. 9 shows the stock price chart display screen 50. In the graph area in which the stock price index is set on the vertical axis and the time is set on the horizontal axis, the stock prices before and after the past event occurrence date selected by the user are displayed. A line graph 51 showing the transition is drawn with a one-dot chain line.

  Here, the user sets “March 20, 2010” as the current event occurrence date in the current event occurrence date designation column 52, sets “Japan” in the current event occurrence country designation column 53, and then displays the button 54 again. When is clicked, a stock price chart redisplay request specifying the current event occurrence date and country is transmitted to the Web server 12.

Receiving this (S34), the Web server 12 passes the brand selected by the user, the date of occurrence of the current event, and the country of occurrence to the chart generation unit 18, and requests the generation process of the stock price chart.
In response to this, the chart generation unit 18 extracts the stock price information of the corresponding stock before and after the current event occurrence date in the corresponding country from the stock price history data storage unit 30, and generates a stock price chart to which a graph in which the stock prices are plotted in time series is added. (S36).

Receiving the stock price chart from the chart generator 18, the Web server 12 generates a stock price chart display screen and transmits it to the client terminal 34 (S38).
FIG. 10 shows this stock price chart display screen. In the same graph area, a one-dot chain line line graph 51 showing the transition of stock prices before and after the past event occurrence date and the transition of stock prices before and after the current event occurrence date are shown. A broken line graph 55 is drawn.

As shown in the figure, the past stock price graph 51 and the current stock price graph 55 are arranged so that the event occurrence dates overlap, so the user can see at a glance how the stock price has changed due to the occurrence of the same kind of event. It becomes possible to compare.
And if there is a divergence between the bottom price of the stock price at the past event and the current stock price, the user says, “Because the current stock price is too low compared to the past case, It can be determined that there is a possibility that a value return will occur.

When the user selects “downward revision of achievement” or “upward revision of achievement” on the stock price fluctuation event search screen 40 in FIG. 4, the Web server 12 determines that the quantitative event has been selected ( 2 (S16 in FIG. 2), the search condition (business type, brand, scale) is passed to the quantitative event generation unit 16 to request generation processing of quantitative event information.
Hereinafter, the procedure of processing by the quantitative event generation unit 16 will be described with reference to the flowchart of FIG.

First, the quantitative event generation unit 16 refers to the company performance data storage unit 28, and extracts company performance data of a business type or brand that matches the search condition (S20-01).
Next, the quantitative event generation unit 16 analyzes the extracted company performance data and extracts an event that matches the search condition (S20-02).

For example, when the designated event type is “downward revision of business performance” and the designated brand is “Yamada Motor”, the quantitative event generation unit 16 stores the corporate performance data related to Yamada Motor Company in the corporate performance data storage unit 28. After extracting from the above, the difference between the actual value and the expected value of the performance item is calculated for each accounting period.
If there is a negative deviation in which the difference exceeds a preset threshold value, the quantitative event generation unit 16 recognizes an event of “downward revision of performance” for the settlement period.

  FIG. 12 (a) shows an example of this corporate performance data. In the past settlement period of Yamada Motor, the sales performance is only 18 trillion yen compared with the sales forecast of 20 trillion yen, minus 2 between them. Since there is a trillion yen divergence, the quantitative event generation unit 16 extracts events for downward revision of business results.

In the figure, only sales and operating profit are listed as indices representing corporate performance, but other indices (ordinary profit, net profit, etc.) can also be used.
Further, when the actual value exceeds the predetermined threshold value exceeding the expected value, the quantitative event generation unit 16 extracts the business performance upward revision event.

  Next, the quantitative event generation unit 16 specifies “May 31, 1998”, which is the settlement date of the business performance data, as the event occurrence date (S20-03).

Next, the quantitative event generation unit 16 determines the scale of the event (large, medium, small) by applying the above difference to the quantitative event size determination rule stored in the rule data storage unit 24 (S20-04). ).
In this scale determination rule, for example, for each range of the ratio of the actual value to the expected value, “Less than 5% → Small, 5% to less than 10% → Middle, 10% or more → Large” A small value is set.

Finally, the quantitative event generation unit 16 generates quantitative event information (S20-05).
FIG. 12 (b) shows an example of this quantitative event information, which includes data items of event ID, event type, brand, country of occurrence, date of occurrence, and scale.

The web server 12 that has received the quantitative event information from the quantitative event generation unit 16 generates a search result screen and transmits it to the client terminal 34 as described above (S22 in FIG. 2).
Further, when a detailed browsing request for the qualitative event information is transmitted from the client terminal 34, a detailed screen in which the settlement data is described in detail is transmitted from the Web server 12 to the client terminal 34 (S24, S26).

  Further, when a stock price chart display request is transmitted from the client terminal 34, the chart generation unit 18 extracts the stock price history data before and after the closing date from the stock price history data storage unit 30, generates a stock price chart, and generates a stock price chart from the Web server 12. A chart display screen is transmitted to the client terminal 34 (S28 to S32 in FIG. 3).

  Further, when a redisplay request with the designation of the current event occurrence date and the country of occurrence is transmitted from the client terminal 34, the chart generation unit 18 displays the stock price data before and after the current event occurrence date related to the corresponding stock in the stock price history data storage unit A stock price chart is extracted from 30 and the past and current stock price fluctuation graphs are superimposed, and a stock price chart display screen is transmitted from the Web server 12 to the client terminal 34 (S34-38).

In the above, an example in which past stock price fluctuation events are searched by brand is explained. However, in the stock price fluctuation event search screen 40 of FIG. Thus, it is possible to comprehensively acquire past stock price fluctuation events related to a plurality of issues. In this case, the search result screen 44 displays a list of stock price fluctuation events of multiple brands belonging to the same industry.
Depending on the type of business, even if the brands are different, the same price movement may occur when a specific type of event occurs, and the user can estimate the future price movement of the specific brand from a wider viewpoint.

It is a block diagram which shows the function structure of the stock price fluctuation event information provision system which concerns on this invention. It is a flowchart which shows the process sequence in a stock price fluctuation event information provision system. It is a flowchart which shows the process sequence in a stock price fluctuation event information provision system. It is a figure which shows a stock price fluctuation event search screen. It is a flowchart which shows the process sequence which concerns on extraction of a qualitative event. It is a conceptual diagram which shows a mode that several synonyms are defined per language with respect to the character string showing an event type. It is explanatory drawing which shows a mode that qualitative event information is extracted from an event article. It is a figure which shows a search result screen. It is a figure which shows the stock price chart display screen on which the fluctuation graph of the stock price concerning a past event was drawn. It is a figure which shows the stock price chart display screen on which the fluctuation graph of the stock price concerning a past event and a present event was drawn. It is a flowchart which shows the process sequence which concerns on extraction of a fixed quantity event. It is explanatory drawing which shows a mode that quantitative event information is extracted from company performance data.

10 Stock price fluctuation event information provision system
12 Web server
14 Qualitative event generator
16 Quantitative event generator
18 Chart generator
20 Article data storage
22 Dictionary data storage
24 Rule data storage
26 Company attribute data storage
28 Corporate performance data storage
30 Stock history data storage
32 Internet
34 Client terminal
40 Stock price fluctuation event search screen
42 Search button
44 Search results screen
46 View details button
47 Chart button
50 Stock chart display screen
51 Historical stock price line graph
52 Current event occurrence date designation column
53 Current event country designation field
54 Refresh button
55 Line graph of current stock price

Claims (10)

Dictionary data storage means for storing dictionary data defining a correspondence relationship between a specific expression character string and an abstraction tag indicating its type;
Article data storage means for storing digitized article data;
Extraction that stores extraction rules that specify character strings to be extracted as constituent elements of stock price fluctuation event information for each combination of abstract tags included in article data, or a combination of abstract tags and specific character strings Rule storage means;
When a request for a stock price fluctuation event search is transmitted from a client terminal, a search screen including at least a brand and event type designation field is transmitted, and means for prompting designation of the brand and event type;
Article data extraction means for searching the article data storage means using the brand and event type as a key and extracting related article data when designation information of the brand and event type is transmitted from the client terminal,
Morphological analysis means for decomposing the extracted article data into morphemes;
An abstraction unit that refers to the dictionary data and adds a corresponding abstract tag to a character string representing at least a brand, an event type, and a time in each morpheme,
An event information generating means for generating stock price fluctuation event information by extracting at least a character string representing a brand, a character string representing an event type, and a character string representing an event occurrence time by applying the extraction rule to each article data; ,
A means for generating a search result screen in which the stock price fluctuation event information is described and transmitting it to the client terminal;
Stock price fluctuation event information provision system with The dictionary data storage means, article data storage means, and extraction rule storage means store data relating to a plurality of languages,
The article data extraction means performs a process of extracting related article data from article data described in a language corresponding to a key after converting the brand and event type transmitted from the client terminal into the corresponding language in each language. Run,
The morpheme analysis means executes a process of decomposing each related article data into morphemes corresponding to a corresponding language,
The abstraction means executes a process of assigning an abstract tag corresponding to each language to each morpheme,
2. The stock price fluctuation event information provision according to claim 1, wherein the event information generation means executes processing for extracting stock price fluctuation event information by applying an extraction rule for a corresponding language to each article data. system. Storage means for storing a scale determination rule that defines the correspondence between the number of article data and the scale of stock price fluctuation events;
Means for counting the number of articles when there are multiple article data related to the same issue and stock price fluctuation event;
By applying this number of articles to the above-mentioned size determination rule, a means for identifying the size of the stock price fluctuation event is provided,
3. The stock price fluctuation event information providing system according to claim 1, wherein the scale of each stock price fluctuation event is written together on the search result screen. Storage means for storing a size determination rule that defines a correspondence relationship between the temporal distribution range of similar article data and the size of a stock price fluctuation event;
Means for calculating an interval between the reporting date of the first article data and the reporting date of the last article data when there are a plurality of article data relating to the same brand and stock price fluctuation event;
Means for identifying the scale of the stock price fluctuation event by applying this interval to the scale determination rule,
3. The stock price fluctuation event information providing system according to claim 1, wherein the scale of each stock price fluctuation event is written together on the search result screen. In the dictionary data storage means, the correspondence between the expression character string representing the scale of the stock price fluctuation event and the abstract tag is registered,
The abstraction means executes a process of assigning a corresponding abstract tag when an expression character string representing the scale of the stock price fluctuation event exists in the article data,
The event information generation means executes a process of extracting a character string provided with an abstract tag representing a scale according to the extraction rule as a component of event information,
3. The stock price fluctuation event information providing system according to claim 1, wherein the scale of each stock price fluctuation event is written together on the search result screen. Company performance data storage means for storing performance forecast data and performance data for each company's fiscal period,
When a request for a stock price fluctuation event search is transmitted from a client terminal, a search screen including at least a brand and event type designation field is transmitted, and means for prompting designation of the brand and event type;
When information specifying downward revision or upward revision of corporate performance is sent from the client terminal as the stock and event type, the forecast data and performance data for each fiscal period related to the stock are sent from the corporate performance data storage means. Means for extracting;
Calculate the difference between the actual data for each fiscal period and the forecast data for the current period. If the actual data exceeds the expected data, include at least the stock, the event type for upward revision of the performance, and the fiscal period as the event occurrence period. Event information that generates stock price fluctuation event information and generates stock price fluctuation event information that includes at least the stock, the event type for downward revision of the business results, and the closing date as the event occurrence time if the actual data is lower than the expected data Generating means;
A means for generating a search result screen in which the stock price fluctuation event information is described and transmitting it to the client terminal;
Stock price fluctuation event information provision system with   7. The stock price fluctuation event information providing system according to claim 6, wherein said company performance data storage means stores data in a plurality of countries described in a plurality of languages. A scale determination rule storage means for storing a scale determination rule that defines the correspondence between the range of the difference between the actual data and the forecast data and the scale of the stock price fluctuation event;
By applying the difference between the performance forecast data for each fiscal period and the performance data to this scale determination rule, the scale determination rule has a means for identifying the scale of the downward revision event or the upward revision event,
8. The stock price fluctuation event information providing system according to claim 6, wherein the scale of each stock price fluctuation event is written together on the search result screen. Stock price history data storage means for storing stock price history data of each brand;
When a request for displaying a stock price fluctuation chart relating to a specific stock price fluctuation event is transmitted from the client terminal, stock price history data in a predetermined range is extracted from the stock price history data storage means based on the occurrence time of the stock price fluctuation event. Means,
Means for generating a stock price chart including a first graph representing a time-series change in stock price drawn based on the extracted stock price history data;
Means for transmitting a stock price chart display screen in which the stock price chart is described to the client terminal;
When a re-display request for a stock price chart that designates the time of a stock price fluctuation event to be compared is transmitted from the client terminal, the stock price history data storage means has a predetermined range based on the occurrence time of the stock price fluctuation event. Means for extracting stock price history data;
A stock chart in which a second graph representing a time-series change in stock price drawn based on the extracted stock price history data and the first graph are arranged in the same graph area so that their event occurrence times overlap. Means for generating
Means for transmitting a stock price chart display screen in which the stock price chart is described to the client terminal;
The stock price fluctuation event information providing system according to any one of claims 1 to 8, further comprising: Computer
Dictionary data storage means for storing dictionary data defining a correspondence relationship between a specific expression character string and an abstract tag indicating the type,
Article data storage means for storing digitized article data,
Extraction that stores extraction rules that specify character strings to be extracted as constituent elements of stock price fluctuation event information for each combination of abstract tags included in article data, or a combination of abstract tags and specific character strings Rule storage means,
When a request for a stock price fluctuation event search is sent from the client terminal, a means for sending a search screen provided with at least a brand and event type designation field and prompting designation of the brand and event type,
Article data extraction means for searching the article data storage means by using the brand and event type as a key and extracting related article data when designation information of the brand and event type is transmitted from the client terminal,
Morphological analysis means for decomposing the extracted article data into morphemes,
An abstraction unit that refers to the dictionary data and attaches a corresponding abstract tag to a character string representing at least a brand, an event type, and a time in each morpheme,
Event information generation means for generating stock price fluctuation event information by extracting at least a character string representing a stock, a character string representing an event type, and a character string representing an event occurrence time by applying the above extraction rule to each article data;
Means for generating a search result screen in which the stock price fluctuation event information is described and transmitting it to the client terminal;
Stock price fluctuation event information provision program to function as.

Images