JP2019016263A - Calculation device, calculation method, calculation model, and model - Google Patents

Abstract

To calculate credibility in consideration of a relationship between business operators.SOLUTION: A calculation device includes an acquisition unit and a calculation unit. The acquisition unit obtains first scores indicating evaluation values of business operators. The calculation unit calculates a second score indicating the credibility of a first business operator on the basis of the first score of a first business operator and the first score of a second business operator having a transaction relationship with the first business operator out of the first scores that are obtained by the acquisition unit. For example, the acquisition unit obtains a first score that is calculated on the basis of information on the financial conditions of business operators.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a calculation device, a calculation method, a calculation program, and a model.

  Generally, when a financial institution such as a bank determines a loan destination, the credit quality of a business operator that is the loan destination is determined. In other words, a financial institution investigates the financial status of a business operator and measures the creditworthiness of the business operator to determine whether or not to make a loan to the business operator and how much the loan amount is set. Determine information about. Information such as the creditworthiness of a business may be provided by an external organization that specializes in collecting and analyzing corporate information rather than being determined by the financial institution itself.

  Here, as an example of a technique for determining the creditworthiness of a business operator, a technique for calculating a fund trace using transfer data is known. According to this technology, it is possible to graph the flow of funds of a business operator without depending on financial statements or information on major business partners from external organizations. There is also known a technique for calculating the creditworthiness of a business operator based not only on financial information but also on various information based on user behavior.

Japanese Patent Laying-Open No. 2015-88037 Japanese Patent No. 5960887

  However, in the above-described conventional technology, it is not always possible to calculate the degree of trust taking into account the relationship between the operators. Specifically, the prior art merely creates a graph reflecting business-to-business transactions by calculating funds trace using transfer data. For this reason, in the prior art, it is not always possible to calculate the degree of credit for determining whether or not the loan is useful for financial institutions.

  The present application has been made in view of the above, and an object thereof is to provide a calculation device, a calculation method, a calculation program, and a model capable of calculating a degree of reliability in consideration of the relationship between operators.

  The calculation device according to the present application includes an acquisition unit that acquires a first score indicating an evaluation value of each operator, and the first score of the first operator out of the first scores acquired by the acquisition unit, A calculation unit that calculates a second score indicating the creditworthiness of the first operator based on the first score of the second operator having a business relationship with the first operator. .

  According to one aspect of the embodiment, there is an effect that it is possible to calculate the degree of trust that takes into account the relationship between operators.

FIG. 1 is a diagram illustrating an example of a calculation process according to the embodiment. FIG. 2 is a diagram illustrating a configuration example of a calculation system according to the embodiment. FIG. 3 is a diagram illustrating a configuration example of the calculation apparatus according to the embodiment. FIG. 4 is a diagram illustrating an example of a financial table according to the embodiment. FIG. 5 is a diagram illustrating an example of a stock price exchange table according to the embodiment. FIG. 6 is a diagram illustrating an example of a search data table according to the embodiment. FIG. 7 is a diagram illustrating an example of a transaction information storage unit according to the embodiment. FIG. 8 is a diagram illustrating an example of a model storage unit according to the embodiment. FIG. 9 is a flowchart (1) illustrating a processing procedure according to the embodiment. FIG. 10 is a flowchart (2) illustrating a processing procedure according to the embodiment. FIG. 11 is a hardware configuration diagram illustrating an example of a computer that realizes the function of the calculation device.

  Hereinafter, a mode for carrying out a calculation device, a calculation method, a calculation program, and a model according to the present application (hereinafter referred to as “embodiment”) will be described in detail with reference to the drawings. Note that the calculation device, the calculation method, the calculation program, and the model according to the present application are not limited by this embodiment. In the following embodiments, the same portions are denoted by the same reference numerals, and redundant description is omitted.

[1. Example of calculation process)
First, an example of calculation processing according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of a calculation process according to the embodiment. In FIG. 1, an example of processing in which the calculation device 100 according to the embodiment calculates the creditworthiness of a predetermined operator will be described. In the embodiment, the creditworthiness of the business operator is obtained by quantifying the credit of the business operator based on, for example, the business condition of the business operator or the business stability of the business operator. In addition, a company is not necessarily limited to a company, but is a concept that includes a wide range of persons who run businesses, such as individual business owners and small corporations.

  A calculation device 100 illustrated in FIG. 1 is a server device that calculates a score indicating the creditworthiness of a business operator. In the embodiment, an operator for which a score indicating credit is calculated by the calculation device 100 is referred to as “first operator”, and an operator having a business relationship with the first operator is referred to as “second operator”. Written as “person”. Further, in the embodiment, the score used when the calculation device 100 calculates the credit rating, the score indicating the evaluation value of each business operator is expressed as “first score”, and is calculated using the first score. A score indicating the creditworthiness of the first operator, which is finally calculated by the calculation device 100, is referred to as a “second score”. That is, the calculation apparatus 100 according to the embodiment acquires a first score indicating an evaluation value of each business operator, and among the acquired first score, the first score of the first business operator and the first business operator And a second score indicating the creditworthiness of the first operator based on the first score of the second operator having a business relationship with the first score.

  The business server 20 shown in FIG. 1 is a server device used by each business. The content distribution server 40 shown in FIG. 1 is a server device that distributes content such as websites to general users (not shown in FIG. 1). The calculation device 100, the operator server 20, and the content distribution server 40 are connected to be communicable via a network N (eg, the Internet, see FIG. 2) not shown in FIG.

  For example, the calculation device 100 calculates a credit rating (second score) as an index used by the financial institution to measure the credit of the business operator, and provides the calculated credit rating to the financial institution. In general, a financial institution independently investigates a business operator and determines the creditworthiness as a loan destination. For example, a financial institution determines creditworthiness as a loan destination in cooperation with an external organization such as a research company that specially investigates the financial status of a business operator.

  However, in the above method, there are cases where the financial institution cannot determine the credit of the borrower from various angles. For example, a research company or a financial institution can obtain financial status for a relatively large company that is listed, but it does not have a financial status for a small unlisted company or a venture company that has not yet been established. The situation cannot always be acquired. For this reason, research firms and financial institutions sometimes estimate the creditworthiness of unlisted small companies or venture companies that are not established relatively low. However, some of these businesses are also mixed with some businesses that are expected to steadily improve their business by dealing with large companies. However, financial institutions may not be able to accurately evaluate these small businesses, or may not recognize them as small customers because they are unaware of the presence of small businesses. In addition, small businesses may not be properly evaluated and may not be able to receive loans from financial institutions.

  Therefore, the calculation device 100 according to the embodiment determines the creditworthiness of the first operator by using the processing described below, so that it is not limited by the financial status or scale of the first operator itself. From the above, the creditworthiness of the first operator is calculated. Thereby, the calculation apparatus 100 can bring about the situation where the lending destination of a financial institution can be expanded, or a 1st provider can receive a loan actively from a financial institution.

  Specifically, the calculation device 100 has a predetermined relationship with the first business operator as well as information on the first business operator as to the first business operator that is the target of creditworthiness determination as a loan destination. Calculation processing is performed using the information of the business operator. More specifically, the calculation device 100 calculates the credit rating of the first operator based on information related to the business relationship between the first operator and the second operator. As a result, it is difficult for the calculation device 100 to calculate the creditworthiness based on the financial information of the first operator itself, because the first operator is not listed or is a small business operator such as a venture company. Even in this case, the creditworthiness of the first operator can be calculated with high accuracy. Hereinafter, an example of a calculation process performed by the calculation apparatus 100 will be described along a flow with reference to FIG.

  First, the calculation device 100 performs a process of calculating the first score of each operator based on the information of each operator. In the embodiment, the first score is a score indicating the evaluation value of each business operator calculated without taking into account the business operator's score. For example, the first score is calculated based on the financial status of each business operator. The calculation apparatus 100 acquires information (hereinafter referred to as “evaluation information”) related to the evaluation of each business from the content distribution server 40 in order to calculate the first score of each business (step S11).

  The content distribution server 40 is assumed to be a server that holds various information regarding each business operator. Further, the content distribution server 40 may include a server owned by a financial institution, a server of a specialized business operator that conducts business for evaluating a company, and the like. The content distribution server 40 may include, for example, a web server that provides a search service.

  The calculation apparatus 100 acquires evaluation information from the content distribution server 40, and calculates the first score of the business operator (step S12). Note that the calculation device 100 may calculate the first score based on various conventional techniques. As an example, the calculation device 100 may calculate the first score using any known company evaluation model. For example, the Alt value (Edward Altman) Z value and the like are widely known as a company evaluation model for obtaining a score indicating the probability of bankruptcy of an operator based on financial information. The calculation apparatus 100 may employ a score indicating the bankruptcy probability of the business operator as an example of the first score using such an existing model. Or the calculation apparatus 100 may calculate the 1st score of each provider based on the user action (the action etc. which searched the provider) with respect to each provider.

  Below, an example of the model which the calculation apparatus 100 produces | generates for the 1st score calculation which concerns on embodiment is shown. For example, the calculation device 100 generates a model for calculating the first score of the business operator by regression analysis using predetermined correct answer data. For example, for listed businesses, there are cases where numerical values obtained by evaluating businesses based on the financial information of businesses, etc. are published by external organizations (organizations that investigate businesses and evaluate businesses). The calculation apparatus 100 uses the evaluation value as correct answer data (objective variable), and performs regression using each piece of information such as financial status such as financial data and currency exchange data, or search data that is an example of user behavior as an element (explanatory variable). Perform analysis. And the calculation apparatus 100 calculates the 1st score of each provider using the model derived recursively. For example, the calculation device 100 generates a regression equation such as the following equation (1).

Y = ω ₁ · x ₁ + ω ₂ · x ₂ + ω ₃ · x ₃ + ... + ω _n · x _n ... (1)

In the above equation (1), for example, “Y” is input with an evaluation value calculated by an external organization or correct data preset by the administrator of the calculation device 100. In addition, in “x _{1 to} x _n ”, for example, a variable (for example, financial status, Stock prices, exchange rates, information obtained by quantifying the number of searches for business operators, and the like are input (respectively, n indicates an arbitrary number). By performing learning using the above equation (1) (that is, calculating a sufficient number of samples recursively), the calculation apparatus 100 optimizes the value of “ω _n ” corresponding to each item. That is, for example, when calculating the evaluation value of the first operator, the calculation device 100 has a weight value “ω _n ” such as how much the feature such as “stock price” affects the score. Ask for. Then, the calculation device 100 sets the obtained weight value and generates a model. The calculation apparatus 100 can obtain the first score of the business operator by inputting the information of the business operator as a target to the generated model. That is, the calculation device 100 generates the following first score output model.

First score = ω ₁ · x ₁ + ω ₂ · x ₂ + ω ₃ · x ₃ + ... + ω _n · x _n (2)

The above equation (2) represents the first score output model. In the above equation (2), specific numerical values obtained by performing regression analysis on the equation (1) are substituted into the respective weight values ω. Therefore, the calculation apparatus 100 calculates the first score of the first operator by inputting the evaluation information of the first operator into “x ₁ ” to “x _n ” in the above formula (2). Can do.

  The model described above is an example, and the model generated by the calculation device 100 is not limited to the above example. In other words, the calculation device 100 can appropriately select a model different from the above example by combining known methods as long as it is a model that receives evaluation information and outputs a value indicating the evaluation value (first score) of the operator. It may be generated. For example, in the above example, an example is shown in which a model is generated by regression analysis, but the calculation apparatus 100 may generate a model by other statistical processing.

  In the example of FIG. 1, it is assumed that the calculation device 100 calculates the first score of all the operators to be processed. Subsequently, the calculation device 100 acquires transaction information from the business operator server 20 (step S13).

  In the embodiment, the transaction information indicates information related to the transaction of the first business operator, which is the second score calculation target. Specifically, the transaction information includes information on the second business operator having a business relationship with the first business operator, the transaction amount with each second business operator, the number of transactions, the transaction frequency, and the transaction. It is the number of years. Further, the transaction information may include information such as the transition of the transaction amount with each second operator, the transition of the number of transactions, the transition of the transaction frequency, and the like.

  The calculation device 100 acquires transaction information from the business server 20, and generates a model for calculating the second score based on the acquired transaction information (step S14). Specifically, the calculation device 100 calculates the second score of the first company based on the first score of the first company and the first score of the second company having a business relationship with the first company. Generate a model that outputs

  For example, the calculation device 100 adds a feature calculated based on the first score of the second operator having a business relationship with the second operator to the model generated to output the first score. Update the generated model. For example, the calculation apparatus 100 obtains a regression equation for performing model update processing based on the following equation (3). In order to simplify the explanation, in the following formula (3), it is assumed that only the transaction amount (payment / withdrawal amount) of the transaction information in a predetermined period among the transaction information is used as a feature.

Y = ω ₁ · x ₁ +... + Ω _n · x _n + ω _IN · x _IN + ω _OUT · x _OUT (3)

As described above, the expression (3) further includes the elements “ω _IN · x _IN ” and “ω _OUT · x _OUT ” compared to the expression (1). “X _IN ” indicates a feature related to payment from the second business operator to the first business operator. “Ω _IN ” indicates a weight value of the feature related to the deposit. “X _OUT ” indicates a feature related to withdrawal from the first business operator to the second business operator. “Ω _OUT ” indicates a weight value of a feature related to withdrawal.

“X _IN ” is obtained, for example, by the relationship between the amount of money deposited by the second business operator to the first business operator and the first score of the second business operator who has made the payment. For example, FIG. 1 shows an example when the calculation device 100 generates a regression equation based on the transaction information of the business operator A who is the first business operator. As shown in FIG. 1, business operator A has a business relationship with business operator B or business operator C. In this case, when viewed from the business operator A, the business operator B and the business operator C correspond to the second business operator.

  In the example of FIG. 1, it is assumed that the first score of the business operator B is “90” and the first score of the business operator C is “40”. Further, the deposit amount from the business B to the business A is “8 million yen” in a predetermined period (for example, one year). The deposit amount from the business operator C to the business operator A in the predetermined period is “4 million yen”. Further, the amount of money paid out by the business operator A to the business operator B in the predetermined period is “3 million yen”. The amount of payment to the operator C by the operator A for a predetermined period is “7 million yen”.

The calculation apparatus 100 generates a new regression equation by adding a feature indicating the relationship between the information regarding the transaction amount and the first score of each second operator. For example, the calculation device 100 calculates a weighted average of each numerical value. Specifically, the calculation apparatus 100 calculates the feature related to the deposit of the operator A by dividing the sum of the values obtained by multiplying the first score of the second operator and the deposit amount by the total deposit amount. . In the example of FIG. 1, the calculation device 100 calculates a numerical value “about 73” as the feature related to the deposit of the business operator A. Such a numerical value corresponds to “x _IN ” in the regression equation. Similarly, the calculation apparatus 100 calculates the feature regarding the withdrawal of the business operator A by dividing the sum of the values obtained by multiplying the first score of the second business operator and the withdrawal amount by the total of the withdrawal amount. In the example of FIG. 1, the calculation device 100 calculates a numerical value “55” as the feature related to the withdrawal of the business operator A. Note that the above calculation method is an example, and the calculation device 100 may employ any method as long as a numerical value is obtained based on the relationship between the first score and the amount of money.

In this way, the calculation apparatus 100 updates the model for outputting the first score using the regression equation (3) including the calculated “x _IN ” and “x _OUT ”. That is, the calculation apparatus 100 updates the model based on the regression equation (3) to which a new feature has been added, and obtains a model for obtaining a new score (denoted as “provisional score” for explanation). Generate.

For example, the calculation device 100 performs the calculation as described above for each business operator as well as the business operator A, and generates a regression equation (3) corresponding to each business operator. That is, the calculation apparatus 100 generates a large number of regression equations (samples) corresponding to each business operator, and updates the model based on the generated regression equations. Specifically, the calculation apparatus 100 generates a regression equation like the following equation (4) to which the values of the weight values “ω _IN ” and “ω _OUT ” are added.

Provisional score = ω ₁ · x ₁ + ... + ω _n · x _n + ω _IN · x _IN + ω _OUT · x _OUT · · · (4)

Thus, in the above formula (4), “ω _IN · x _IN ” and “ω _OUT · x _OUT ” are added as elements. Thereby, the calculation apparatus 100 can obtain | require the provisional score which is a score which considered the business relationship of the provider A instead of the 1st score calculated | required by the information of the provider A itself. That is, the calculation device 100 can calculate the provisional score in such a manner that the evaluation value (first score) of the other party of the transaction affects the first business operator. For this reason, for example, even a small-scale first business operator has many transactions with a large company that is assumed to have a relatively high first score, or a second business operator with a stable first business operator. The first business operator who trades has a high provisional score.

  By the way, in the above processing, the first score of the operator A is updated to the provisional score, and accordingly, the first score of the operator B is updated to the provisional score, and the first score of the operator C is changed to the provisional score. Updated to That is, in the calculation of the provisional score of each business operator, since the first scores updated with each other influence each other, an error may occur depending on the calculation order. For this reason, the calculation apparatus 100 repeats the update of the model until the updated provisional score falls below a predetermined threshold with the provisional score obtained by the model before the update. In other words, the calculation device 100 repeats the model update until the provisional score blur converges. Specifically, the calculation device 100 repeats the update until the provisional scores of all the operators are “| updated score−pre-update score | ≦ predetermined threshold (for example, 1)”.

  Then, the calculation apparatus 100 repeats updating the model until the above condition is satisfied, and finally generates a model like the following formula (5).

Second score = ω ₁ · x ₁ +... + Ω _n · x _n + ω _IN · x _IN + ω _OUT · x _OUT (5)

The above equation (5) represents the second score output model. In the above equation (5), specific numerical values obtained by regression analysis are substituted for each weight value ω. For this reason, the calculation apparatus 100 inputs the evaluation information of the first operator into “x ₁ ” to “x _n ” in the above formula (5), and further, the transaction information of the first operator is “x _IN ”. Or by inputting into “x _OUT ”, the second score of the first operator can be output. In this way, the calculation device 100 calculates the second score (that is, the credit rating) of the first operator (step S15).

  As described above, the calculation device 100 according to the embodiment acquires the first score indicating the evaluation value of each business operator. And the calculation apparatus 100 is based on the 1st score of a 1st provider among the acquired 1st scores, and the 1st score of the 2nd provider who has a business relationship with the said 1st provider. A second score indicating the creditworthiness of one operator is calculated.

  As described above, the calculation apparatus 100 according to the embodiment uses not only the financial information of the first operator itself but also the first score (evaluation value) of the second operator that has a business relationship with the first operator. Then, a second score indicating the creditworthiness of the first operator is calculated. Thereby, the calculation apparatus 100 can calculate the 2nd score of the 1st provider with whom the 2nd provider with higher evaluation is related relatively high. In other words, the calculation device 100 can calculate the degree of credit taking into account the relationship between the operators, not the evaluation of the first operator alone. Specifically, even if the first business operator does not have a high evaluation value (first score) required from the information of the business operator alone because the sales and scale are small, the business operator actually has business with a large company. In the case of a business operator who has increased transactions with a reliable company, the calculation device 100 can calculate a high degree of credit (second score) of the first business operator. .

  Hereinafter, the calculation apparatus 100 that performs the calculation process as described above and the calculation system 1 including the calculation apparatus 100 will be described in more detail.

[2. Configuration of calculation system]
Next, the configuration of the calculation system 1 according to the embodiment will be described with reference to FIG. FIG. 2 is a diagram illustrating a configuration example of the calculation system 1 according to the embodiment. As illustrated in FIG. 2, the calculation system 1 includes a user terminal 10, an operator server 20, a content distribution server 40, and a calculation device 100. Each device included in the calculation system 1 is connected to be communicable by wire or wirelessly via a network N (for example, the Internet) which is a communication network. Note that the number of devices included in the calculation system 1 shown in FIG. 2 is not limited to the illustrated one. For example, the calculation system 1 may include a plurality of user terminals 10.

  The user terminal 10 is an information processing apparatus used by a user who browses content such as a web page. For example, the user terminal 10 is a desktop PC (Personal Computer), a notebook PC, a mobile phone such as a smartphone, a tablet terminal, or a PDA (Personal Digital Assistant). For example, the user terminal 10 acquires a web page from the content distribution server 40 according to an operation by the user, and displays the acquired web page.

  Further, the user terminal 10 transmits predetermined information to the content distribution server 40. For example, the user terminal 10 transmits the name of the business operator as a query to the content distribution server 40 in the search service. In addition, the user terminal 10 may transmit a review related to a product provided by the operator to the content distribution server 40 at a review site or the like.

  The business server 20 is a server device used by the business. For example, the company server 20 holds information related to transactions between companies. The business server 20 provides information related to transactions between business operators in response to a request from the calculation device 100. Specifically, the business entity server 20 provides account information and the like to be held to the calculation device 100, thereby providing the calculation device 100 with transaction information included in the account information.

  The content distribution server 40 is a web server that distributes web pages included in a website that provides various services to the user terminal 10. For example, the content distribution server 40 is a portal site, news site, auction site, weather forecast site, shopping site, finance (stock price) site, route search site, map providing site, travel site, restaurant introduction site, SNS (Social Networking). Service) A web page including various information related to a service site, a web blog, and the like is distributed to the user terminal 10.

  The various data distributed from the content distribution server 40 to the user terminal 10 is actually an HTML (HyperText Markup Language) file or image that forms a web page, but in the following, from the content distribution server 40 to the user terminal 10 may be described as content.

  In addition, the content distribution server 40 stores user behavior on the network N. Here, the user behavior on the network N means information transmitted from the user terminal 10 in accordance with a user operation when using services provided from various websites. For example, the user behavior on the network N is input of a search query on a search site, purchase behavior on a shopping site, posting of a review from a user on a product evaluation site, and the like. Further, the user behavior includes, for example, message exchange on an SNS (Social Networking Service) site, follow-up behavior with respect to others, and the like.

  Further, according to the acquisition request from the calculation device 100, the content distribution server 40 uses the above-described information on the user behavior on the network N as an evaluation information of the business operator, and an index value (for example, stock price or exchange rate) that affects the business. May be transmitted to the calculation apparatus 100.

  As described above, the calculation device 100 acquires the first score indicating the evaluation value of the operator, and among the acquired first scores, the first score of the first operator and the transaction with the first operator. This is a server device that calculates a second score indicating the creditworthiness of the first operator based on the first score of the second operator having the relationship.

[3. Configuration of calculation device]
Next, the configuration of the calculation apparatus 100 according to the embodiment will be described with reference to FIG. FIG. 3 is a diagram illustrating a configuration example of the calculation apparatus 100 according to the embodiment. As illustrated in FIG. 3, the calculation device 100 includes a communication unit 110, a storage unit 120, and a control unit 130. The calculation apparatus 100 includes an input unit (for example, a keyboard and a mouse) that receives various operations from an administrator who uses the calculation apparatus 100, and an output unit (for example, a liquid crystal display) for outputting various types of information. You may have.

(About the communication unit 110)
The communication unit 110 is realized by, for example, a NIC (Network Interface Card). The communication unit 110 is connected to the network N in a wired or wireless manner, and transmits / receives information to / from the user terminal 10, the provider server 20, and the content distribution server 40 via the network N.

(About the storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. The storage unit 120 according to the embodiment includes an evaluation information storage unit 121, a transaction information storage unit 125, and a model storage unit 126. Hereinafter, each storage unit will be described in order.

(About the evaluation information storage unit 121)
The evaluation information storage unit 121 stores information related to the evaluation of the operator. The evaluation information storage unit 121 includes a financial table 122, a stock exchange table 123, and a search data table 124 as data tables for storing information. Hereinafter, each data table will be described in order.

(About financial table 122)
The financial table 122 stores information related to the business operator's finances. Here, FIG. 4 shows an example of the financial table 122 according to the embodiment. FIG. 4 is a diagram illustrating an example of the financial table 122 according to the embodiment. As illustrated in FIG. 4, the financial table 122 includes items such as “business” and “financial information”. The “financial information” has small items such as “total assets”, “total liabilities”, “sales”, “operating income”, and “market capitalization”.

  “Operator” indicates the name of the operator. “Financial information” indicates information related to the financial status of the operator. For example, the financial information is updated on the basis of data or the like published from the business operator every quarter. In the example of FIG. 4, “total assets”, “total liabilities”, “sales”, “operating profit”, and “market capitalization” are illustrated as small items of financial information. The stored financial information is not limited to the example shown in FIG. For example, the financial information may include various types of information such as the capital adequacy ratio of the operator, the collateral capacity, and the settlement status.

  In the example shown in FIG. 4, items such as “total assets” are represented by the concept “aaa”. Actually, however, each item has specific information such as an amount corresponding to the total assets. Is memorized.

  That is, in FIG. 4, as an example of information stored in the financial table 122, the financial information of “operator A” has a total asset of “aaa”, a total liability of “bbb”, and a sales amount of “ccc”. ”, Operating profit is“ ddd ”, and market capitalization is“ eeee ”.

(About Stock Price Exchange Table 123)
The stock price exchange table 123 stores information on stock prices and exchange rates. Here, FIG. 5 shows an example of the stock exchange table 123 according to the embodiment. FIG. 5 is a diagram illustrating an example of the stock price exchange table 123 according to the embodiment. As shown in FIG. 5, the stock price exchange table 123 includes items such as “business operator”, “industry”, “stock price”, “stock price transition information”, and “exchange-linked information”.

  “Operator” corresponds to the same item shown in FIG. “Industry” indicates the industry to which the business operator belongs. In the example shown in FIG. 5, the “industry” item is represented by a concept such as “B01”. However, in reality, the “industry” item includes specific items such as the steel industry, the home appliance industry, and the communication industry. Information indicating a specific industry is stored.

  “Stock price” indicates the stock price of the operator. “Stock price transition information” indicates the transition of the stock price. In the example shown in FIG. 5, the items “stock price” and “stock price transition information” are represented by concepts such as “E01” and “F01”. The specific numerical value of is stored. In addition to specific stock price figures, the “stock price trend information” item includes information such as how much the stock price has increased or decreased during a given period, and differences in movement with the market (the overall market is As the price drops, transition information such as a rise in the stock of the operator is stored. Note that when the company is a private company or the like and the stock price cannot be acquired or there is no stock price information, the items of “stock price” and “stock price transition information” are blank.

  In “Forex-linked information”, information on the transition of performance linked to the movement of the exchange is stored. In the example shown in FIG. 5, the item “Foreign Exchange Linked Information” is represented by a concept such as “G01”. Specific numerical values, such as how the business will affect the business and the industry as a whole, and whether the business operated by the business, such as whether or not it is a business (industry) where the strong yen is more advantageous Information indicating a general interlocking relationship is stored.

  That is, in FIG. 5, as an example of information stored in the stock price exchange table 123, “business A” has “B01” as the industry to which it belongs, “E01” as the stock price, and “F01” as the stock price transition information. Yes, the exchange-linked information is “G01”.

(Regarding the search data table 124)
The search data table 124 stores information related to a search history in which a user searches for a business. Here, FIG. 6 shows an example of the search data table 124 according to the embodiment. FIG. 6 is a diagram illustrating an example of the search data table 124 according to the embodiment. As shown in FIG. 6, the search data table 124 includes items such as “business”, “total period”, “number of searches”, “degree of increase”, and “target word”.

  “Operator” corresponds to the same item shown in FIG. “Aggregation period” indicates a period in which information related to searches performed by the user is aggregated. In the example of FIG. 6, the aggregation period is set to one week, but the aggregation period may be a different period.

  “Number of searches” indicates, for example, the number of times that a searcher has been searched by a search engine on a predetermined search site. The search query counted as the number of searches is not limited to the name of the business operator itself, and may include the name of a product provided by the business operator, the name of the business operator of the business operator, and the like.

  The “degree of increase” indicates an increase / decrease value of the number of searches with respect to the number of searches in the immediately preceding aggregation period. The “target word” indicates a word that is counted as the number of searches related to the operator when the described word is transmitted as a search query. For example, when the name of a product provided by a business is higher than the name of the business, the user may perform a search action using the product name. In this case, even if a search is performed using the product name as a search query, if the product name is set as the target word, the number of searches for the business provider that provides the product is counted. The target word may be set artificially by an application of the administrator of the calculation apparatus 100 or a business operator, or may be automatically set by analyzing a link of a website based on a search result, for example. Good. Specifically, when the user searches for “product P01”, when a relatively large number of websites related to “company A” are displayed as search results, “product P01” is displayed together with company A. "Is automatically set as the target word.

  That is, in FIG. 6, as an example of information stored in the search data table 124, “Business A” is “2000 times” during the aggregation period “June 26, 2017 to July 2, 2017”. The search indicates that the degree of increase from the previous number of searches is “1000 times”. Further, it is shown that “business operator A”, “product P01”, “manager Q01”, etc. are set for the business operator A as target words in the search.

  Note that information other than the information illustrated in FIG. 6 may be appropriately stored in the search data table 124. For example, the search data table 124 stores search ranking information based on the number of searches of each business operator in the same industry, search ranking information based on the increase in the number of searches and the rate of increase in the number of searches, and the like. May be. For example, a business with a higher search ranking is evaluated as a business with a strong business or a high degree of attention from the user.

(About transaction information storage unit 125)
The transaction information storage unit 125 stores information related to the business of the business operator. Here, FIG. 7 shows an example of the transaction information storage unit 125 according to the embodiment. FIG. 7 is a diagram illustrating an example of the transaction information storage unit 125 according to the embodiment. As illustrated in FIG. 7, the transaction information storage unit 125 includes items such as “business” and “transaction history”. The “transaction history” has small items such as “transaction date”, “customer”, “transaction type”, and “amount”.

  “Operator” corresponds to the same item shown in FIG. The “transaction history” indicates a history related to the business of the operator. “Transaction date” indicates the date on which the transaction was performed. “Business partner” indicates a business partner. “Transaction type” indicates the type of transaction such as deposit or withdrawal. “Amount” indicates the amount of the transaction. In the example shown in FIG. 7, the item “amount” is represented by a concept such as “C01”. Actually, a specific amount of money that is traded is stored in the item “amount”.

  That is, in FIG. 7, as an example of the information stored in the transaction information storage unit 125, the transaction history of “operator A” includes, as an example, the transaction date “June 5, 2017”. “B” indicates that there was “payment” of the amount “C01”.

(About the model storage unit 126)
The model storage unit 126 stores information related to the model generated by the calculation device 100. Here, FIG. 8 illustrates an example of the model storage unit 126 according to the embodiment. FIG. 8 is a diagram illustrating an example of the model storage unit 126 according to the embodiment. As illustrated in FIG. 8, the model storage unit 126 includes items such as “model ID”, “update date”, and “industry”.

  “Model ID” indicates identification information for identifying a model. “Update date” indicates the date on which the model was updated. “Industry” corresponds to the same item shown in FIG.

  That is, in FIG. 8, as an example of information stored in the model storage unit 126, the model identified by the model ID “M01” has an update date “July 1, 2017” and corresponds to the industry “B01”. It shows that it is a model to do.

Although illustration in FIG. 8 is omitted, the information stored in the model storage unit 126 includes specific model data (such as a weight value set in ω _i ) constituting the model. Further, the model stored in the model storage unit 126 corresponds to the second score output model as shown in the equation (5) shown in FIG. 1, but the model storage unit 126 learns the second score output model. Learning data, a first score used in the process until the second score output model is generated, and the like may be stored.

  In addition, the model which concerns on embodiment is the 1st score of the 2nd provider who has business relations with the 1st provider among the 1st scores which showed the evaluation value of the provider, the 2nd provider, and the 1st The input layer, the output layer, and any layer from the input layer to the output layer to which the value indicating the relationship with the information on deposits and withdrawals with the business operator is input belongs to a layer other than the output layer A first element, and a second element whose value is calculated based on the weight of the first element and the first element. In addition, the model performs the first operation on the information input to the input layer by using each element belonging to each layer other than the output layer as a first element, based on the first element and the weight of the first element. The computer is caused to function so as to output the value of the second score indicating the creditworthiness of the business operator from the output layer. In the example shown in the embodiment, the calculation device 100 does not perform learning of a multilayer structure such as deep learning, and therefore, the calculation device 100 is any layer from the input layer to the output layer. The “layers other than the output layer” and “each layer other than the output layer” indicate the input layer.

For example, when the model according to the embodiment is realized by the regression model represented by the above-described equations (2) and (5), the first element included in the model is input data (x _i) such as x ₁ or x ₂ (I is an arbitrary number)), and the weight of the first element corresponds to the coefficient ω _i corresponding to x _i . In general, the regression model can be regarded as a simple perceptron having an input layer and an output layer, but when the model is regarded as a simple perceptron, the first element corresponds to any node of the input layer, and the second The element can be regarded as a node included in the output layer.

  Furthermore, when each model is realized by a neural network having one or a plurality of intermediate layers such as DNN (Deep Neural Network), the first element included in each model is any of the input layer or the intermediate layer The second element corresponds to a node to which a value is transmitted from a node corresponding to the first element, that is, a node in the next stage, and the weight of the first element is the first element and A weight to be considered for a value transmitted from a corresponding node to a node corresponding to the second element, that is, a connection coefficient.

  The calculation device 100 calculates the second score using a model having an arbitrary structure such as the above-described regression model or DNN. More specifically, the calculation device 100 adds a regression equation in which the relationship between the first score of the second operator having a transaction with the first operator and the transaction amount with the second operator is added as an element. By recursively learning, the process of updating the score of the first operator is repeated. Then, the calculation device 100 considers that the score has converged when the difference between the pre-update score and the post-update score of each business operator reaches a predetermined threshold, and learning is completed (a model is generated). ).

(About the control unit 130)
Returning to FIG. 3, the description will be continued. The control unit 130 is a controller, for example, various programs (an example of a calculation program) stored in a storage device inside the calculation device 100 by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like. This is realized by executing the RAM as a work area. The control unit 130 is a controller, and is realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

  In addition, the control unit 130 receives the first score of the second operator input to the input layer of the model by information processing according to the model stored in the storage unit 120 (this value is a “provisional score”). In some cases) and a value indicating the relationship between the second business operator and the information relating to the deposit and withdrawal between the first business operator and a coefficient of the model (that is, a weight value corresponding to the feature learned by the model) An operation is performed, and a value indicating the reliability of the first operator, that is, a second score is output from the output layer of the model.

  As illustrated in FIG. 3, the control unit 130 according to the embodiment includes an acquisition unit 131, a first generation unit 132, a first calculation unit 133, a second generation unit 134, and a second calculation unit 135. Then, the functions and operations of information processing described below are realized or executed. Note that the internal configuration of the control unit 130 is not limited to the configuration illustrated in FIG. 3, and may be another configuration as long as the information processing described below is performed. Further, the connection relationship between the processing units included in the control unit 130 is not limited to the connection relationship illustrated in FIG. 3, and may be another connection relationship.

(About the acquisition unit 131)
The acquisition part 131 acquires the 1st score which showed the evaluation value of each provider. For example, the acquisition unit 131 acquires the first score of each business operator calculated by the first calculation unit 133 described later.

  For example, the acquisition unit 131 acquires evaluation information for each business from the content distribution server 40. For example, the acquisition unit 131 acquires information regarding the financial status of each business operator. Then, the acquisition unit 131 sends the acquired evaluation information to the first generation unit 132. A first generation unit 132 and a first calculation unit 133, which will be described later, calculate the first score of each business operator based on the evaluation information acquired by the acquisition unit 131. That is, the acquisition unit 131 acquires a first score calculated based on information regarding the financial status of each business operator.

  Note that the acquisition unit 131 does not necessarily need to acquire the first score calculated by the first calculation unit 133. For example, when there is an external organization that evaluates the business operator, the acquisition unit 131 may acquire a first score calculated in advance based on the evaluation value indicated by the external organization.

  Further, the acquisition unit 131 may acquire transaction information of each business operator from the business operator server 20. Further, the acquisition unit 131 may acquire not only information such as financial status but also various information as evaluation information. For example, the acquisition unit 131 acquires stock price information related to a business as an example of evaluation information. For example, the acquisition unit 131 acquires stock price information from the content distribution server 40 having information on a stock market on which a business is listed. Moreover, the acquisition part 131 may acquire the stock price information of the company related to the company as the stock price information about the company. For example, the acquisition unit 131 acquires stock price information of a business entity that has a capital relationship with the first business entity to be processed.

  In addition, the acquisition unit 131 acquires exchange information in the economic zone to which the business operator belongs. An economic zone to which a business belongs is, for example, a business that operates in yen, and corresponds to exchange information regarding the relationship between the country in which the yen circulates and other countries. Further, the acquisition unit 131 acquires exchange rate transition information indicating whether or not there is a possibility that the business performance of the business operator may change in relation to the exchange information. For example, the acquisition unit 131 receives in advance settings of business operators and industries whose business performance increases due to a strong yen, or business operators whose business results from a weak yen and industry, etc., from an administrator of the calculation apparatus 100. And the acquisition part 131 acquires the industry and business type to which the 1st provider and 2nd provider made into a process belong based on the set information.

  Moreover, the acquisition part 131 may acquire the search data regarding a provider as evaluation information. For example, the acquisition unit 131 acquires the type of search query transmitted from the user terminal 10, the number of transmissions, and the like from the content distribution server 40 that provides the search service.

  In the example of FIG. 1, the acquisition unit 131 has acquired an example of acquiring various types of information from the content distribution server 40, but the information acquisition destination is not limited to the content distribution server 40. For example, the acquisition unit 131 may acquire evaluation information from the user terminal 10. The acquisition unit 131 may acquire evaluation information not only from the user terminal 10 and the content distribution server 40 but also from a predetermined external server. For example, the acquisition unit 131 may acquire information from a cloud server or the like that accumulates information related to the user terminal 10 or the content distribution server 40.

  The acquisition unit 131 stores the acquired information in the storage unit 120 as appropriate. For example, the acquisition unit 131 stores the acquired evaluation information in the evaluation information storage unit 121.

(About the first generation unit 132)
The 1st production | generation part 132 produces | generates the model for calculating a 1st score of a provider based on the evaluation information of each provider. Specifically, the first generation unit 132 generates a regression equation such as the above equation (1) shown in FIG. 1 and performs regression analysis on the generated regression equation to output a first score. Generate a model.

  Note that the first generation unit 132 may use various known methods when generating the model. For example, the first generation unit 132 normalizes the value indicating the financial status among the acquired evaluation information among the operators, normalizes the stock price information between the operators, or normalizes the exchange information between the operators. Or And the 1st production | generation part 132 produces | generates a regression type for every provider based on the evaluation information of each provider.

(About the first calculator 133)
The first calculation unit 133 calculates the first score of each business operator using the model generated by the first generation unit 132. Specifically, the first calculation unit 133 inputs evaluation information of the first business operator to be processed into the model generated by the first generation unit 132, and obtains a first score corresponding to the first business operator. By outputting, the first score is calculated.

For example, the first calculator 133 calculates the first score of the first operator based on the stock price information regarding the first operator. For example, the first calculation unit 133 determines whether “x _i ” in the model is based on the absolute price of the stock price related to the first business operator, the relative price in the industry, or whether the stock price is increasing. ”Is adjusted. Thereby, the 1st calculation part 133 can compute a 1st score, ie, an evaluation value comparatively high about the 1st business operator whose stock price is more favorable than another business operator, for example.

The first calculation unit 133 calculates the first score corresponding to the financial status of each business operator by using the above-described method for the financial status, exchange rate, search data, and the like. Can do. As such a method, for example, a known calculation method may be appropriately used. For example, the first calculation unit 133 may perform the first score calculation process by digitizing the evaluation information or the number of searches based on various known methods. For example, the first calculation unit 133 may use a numerical value normalized based on the relationship with the entire statistical data for the number of searches. For example, the first calculation unit 133 may receive, from the administrator of the calculation apparatus 100 in advance, a setting that adds a value when the stock price increases. In this case, the first calculation unit 133 inputs the model based on the event acquired by the acquisition unit 131 (such as a rise in the stock price of the second operator) and a numerical value added when the event occurs. Adjust the information ("x _i ").

  Note that the first calculation unit 133 may calculate the first score based on the evaluation value of the operator acquired from a predetermined external organization without using the model generated by the first generation unit 132. For example, the first calculation unit 133 may treat the evaluation value of the business acquired from a predetermined external organization as it is as the first score. In other words, the calculation device 100 is a value obtained by an arbitrary target based on the information of the operator itself as the first score, and can be a value that can compare and show the evaluation values of the operators. Any value may be adopted.

(About the second generation unit 134)
The 2nd generation part 134 is related to the 1st score of the 2nd company among the 1st scores of each company acquired by acquisition part 131, and the deposit / withdrawal of the 2nd company and the 1st company. A model is generated based on the relationship with information. The generation of the model by the second generation unit 134 includes the meaning of updating the model generated by the first generation unit 132.

  For example, the second generation unit 134 inputs a relatively high value as a value indicating the relationship between the first score of the second business operator and the information regarding the deposit / withdrawal between the second business operator and the first business operator. The model is learned so that a coefficient (weight value) for each element is set so that the score increases when the value is input and the score decreases when a relatively low value is input.

  The second generation unit 134 generates a model using, for example, a value indicating the relationship between the first score of the second business operator and the deposit / withdrawal between the second business operator and the first business operator. Specifically, as shown in FIG. 1, the second generation unit 134 weights the first score of the second operator for the first operator, and the amount of money received or paid for each second operator. A model is generated using the averaged value.

  Moreover, the 2nd production | generation part 134 is not only the amount of money of deposit / withdrawal in the transaction between a 1st provider and a 2nd provider as information regarding deposits / withdrawals, but between a 1st provider and a 2nd provider. A model may be generated based on the number of transactions, the transaction frequency, and the like. In addition, the second generation unit 134 may generate a model using at least one of the transition of the amount of deposit / withdrawal of the transaction in the predetermined period, the transition of the number of transactions in the predetermined period, or the transition of the transaction frequency. Good. That is, the second generation unit 134 may generate a model using a value obtained by using not only the amount of transaction amount but also the number of transactions as an index value.

  Further, the second generation unit 134 learns a regression equation in which the provisional score, which is a provisional score output by the model, is replaced with the value of the first score used when the provisional score is output. , Update the model. As shown in FIG. 1, the model generated by the second generation unit 134 includes the first score (or provisional score) as a feature, and therefore, when the first score of any operator is updated The first score of other operators is also affected. For this reason, the 2nd production | generation part 134 updates the 1st score (provisional score) used for the last calculation about each provider, and repeats model learning (for example, regression analysis by a regression equation).

  Then, the second generation unit 134 repeats updating the model until the provisional score output by the updated model and the provisional score output by the model before the update are within a predetermined threshold. Then, when the provisional score output by the updated model and the provisional score output by the model before the update are within a predetermined threshold, the second generation unit 134 displays the updated model, It is determined as a model for outputting the second score of the operator.

  The second generation unit 134 stores the generated model in the model storage unit 126. Note that the second generation unit 134 may generate models for various categories such as for each industry to which the business belongs and for each business scale of the business.

  The second generation unit 134 may generate a model using any learning algorithm, not limited to the machine learning based on the regression equation as described above. For example, the second generation unit 134 may generate a model using a learning algorithm such as a neural network, a support vector machine, clustering, and reinforcement learning. As an example, when the second generation unit 134 generates a model using a neural network, the model according to the embodiment includes an input layer including one or more neurons, an intermediate layer including one or more neurons, and one or more And an output layer including neurons. When the second calculation unit 135 performs a calculation process using GAN (Generative Adversarial Networks), the model generated by the second generation unit 134 may be a model that forms part of the GAN.

(About the second calculator 135)
Of the first score acquired by the acquisition unit 131, the second calculation unit 135 uses the first score of the first operator and the first score of the second operator who has a business relationship with the first operator. Based on this, a second score indicating the creditworthiness of the first operator is calculated.

  For example, the second calculator 135 calculates the second score of the first operator using the model generated by the second generator 134. Specifically, the second calculator 135 determines that the provisional score output by the model updated by the second generator 134 and the provisional score output by the model before update are within a predetermined threshold. In addition, the second score of the first operator is calculated using the model.

  For example, the second calculation unit 135 may use the value output from the model as it is as the second score, or a predetermined calculation process (for rounding the number to facilitate comparison between the first operators). The value after the calculation or the like may be used as the second score.

[4. Processing procedure)
Next, a processing procedure performed by the calculation apparatus 100 according to the embodiment will be described with reference to FIGS. 9 and 10. First, with reference to FIG. 9, a processing procedure until the first score is stored in the calculation processing procedure according to the embodiment will be described. FIG. 9 is a flowchart (1) illustrating a processing procedure according to the embodiment.

  As shown in FIG. 9, the calculation apparatus 100 acquires evaluation information from the content distribution server 40 (step S101). The calculation device 100 generates a model for calculating the first score based on the acquired evaluation information (step S102).

  And the calculation apparatus 100 calculates the 1st score of each provider using the produced | generated model (step S103). The calculation device 100 stores the calculated first score in the storage unit 120 (step S104).

  Next, with reference to FIG. 10, the procedure of the process until the second score is calculated among the procedures of the providing process according to the embodiment will be described. FIG. 10 is a flowchart (2) illustrating a processing procedure according to the embodiment.

  As illustrated in FIG. 10, the calculation device 100 acquires transaction information from the provider server 20 included in each provider (Step S <b> 201). The calculation device 100 calculates a value indicating the relationship between the operators based on the acquired transaction information and the first score of each second operator (step S202). As an example, the calculation device 100 A value indicating the relationship between the operators is calculated based on the first score of the second operator and the deposit / withdrawal amount of the transaction to the first operator.

  And the calculation apparatus 100 updates the model produced | generated for calculation of a 1st score by using the value which shows the relationship between providers as a feature (step S203). Subsequently, the calculation apparatus 100 calculates the score (provisional score) of each business operator using the updated model (step S204).

  At this time, the calculation apparatus 100 determines whether the score before update and the score after update are within a predetermined threshold (step S205). When the score before update and the score after update are not within the predetermined threshold (step S205; No), the calculation apparatus 100 further updates the model based on the output value (new provisional score) (step S206). ). Furthermore, the calculation device 100 calculates a score using the updated model (step S204), and determines whether the score before the update and the score after the update are within a predetermined threshold (step S204). S205) is repeated.

  When the score before the update and the score after the update are within a predetermined threshold (step S205; Yes), the calculation device 100 stores the updated model in the storage unit 120 (step S207). The finally updated model is a model for outputting the second score. Using the model, the calculation device 100 calculates the second score of the first operator that is the processing target (step S208).

[5. (Modification)
The calculation system 1 according to the above-described embodiment may be implemented in various different forms other than the above-described embodiment. Therefore, in the following, another embodiment of each device included in the calculation system 1 will be described.

[5-1. (Calculation including seasonal factors)
In the above-described embodiment, an example in which the calculation device 100 calculates the second score based on the transaction amounts of the first company and the second company has been described. Here, the calculation device 100 may calculate the second score based on information other than the transaction amount of the first business operator and the second business operator.

  For example, the calculation device 100 may generate a model based on seasonal factors in the transaction between the first business operator and the second business operator. For example, in transactions between business operators, the transaction volume may increase or decrease from season to season. For this reason, the calculation apparatus 100 may be able to calculate an appropriate credit (second score) at the timing when the credit of the first operator is calculated, for example, by incorporating the seasonal factor into the model. is there. For example, by using the second score in consideration of seasonal factors, a financial institution or the like can measure the timing of financing to the first business operator.

  For example, the calculation apparatus 100 uses the transaction amount as a transaction information based on the ratio of the transaction amount such as every month out of the annual transaction amount between the first business operator and any second business operator. Get information such as. Specifically, in the calculation device 100, the transaction between the first business operator and any second business operator is relatively increased in the summer (for example, the July-September period), and in the winter (for example, 1-3 months). Information that it decreases relatively in the period.

  Then, for example, the calculation device 100 sets an index value indicating the degree of the seasonal factor (hereinafter referred to as “seasonal factor index”) by an arbitrary method. For example, the calculation device 100 sets a seasonal factor index between the first business operator and an arbitrary second business operator based on the ratio of the annual transaction amount. Specifically, the calculation device 100 sets the seasonal factor index of the second operator D among the second operators to “20” and the seasonal factor index of the second operator E among the second operators as “80”. It is assumed that it has been set. When the first score of the second operator D is “90” and the first score of the second operator E is “40”, the calculation device 100 includes the first operator and the second operator. A value indicating the relationship with the above is calculated as (20 * 90 + 80 * 40) / (20 + 80) = 50.

  Thus, the calculation apparatus 100 acquires information on the seasonal factors of each business operator, and includes the first score of the second business operator and the information on the seasonal factors of the second business operator and the first business operator. A model may be generated based on the relationship. That is, the calculation apparatus 100 can set a value indicating the relationship between the operators using not only the transaction amount but also various information. And the calculation apparatus 100 can quantify an operator's creditworthiness from a more diversified viewpoint by calculating an operator's 2nd score using such various information.

[5-2. Example of evaluation information)
In the above embodiment, as an example of the evaluation information, the financial status of the business operator and the search behavior etc. among the user behavior are exemplified. However, the example of the evaluation information is not limited to this. For example, if the content distribution server 40 is a server that handles news data, the calculation device 100 may acquire news data as an example of evaluation information. And when the news data positive with respect to the business is acquired, the calculation device 100 determines that the business of the business is favorable and calculates the first score of the business high. A simple model may be generated. Alternatively, when negative news data is acquired for a business operator, the calculation device 100 determines that the business operator of the business operator is out of order and calculates the first score of the business operator low. A simple model may be generated.

  Alternatively, if the content distribution server 40 is a server that distributes advertisements, the calculation device 100 may acquire data related to advertisements as an example of evaluation information. For example, the calculation apparatus 100 acquires information such as the number of advertisements put out by the business operator and the advertisement fee spent on the advertisement. Then, the calculation device 100 determines that the operator of the operator is more favorable as the number of advertisements issued by the operator, the advertising fee paid by the operator, and the like are higher. A model that calculates a high value may be generated. Alternatively, the calculation device 100 determines that the operator of the operator is inferior as the number of advertisements issued by the operator, the advertisement fee paid by the operator, and the like are lower. A model that calculates a low value may be generated.

  Further, the calculation device 100 may use any information that can be acquired from the content distribution server 40 as the evaluation information. For example, the calculation apparatus 100 displays the content of text data posted by the user on the review site (for example, whether the review is favorable to the business operator), the content of text data posted by the user on the SNS, and the like. You may acquire as evaluation information.

[5-3. Each device of the calculation system]
Each device included in the calculation system 1 may be realized by various modifications. For example, the calculation device 100 may also serve as processing executed by the content distribution server 40. Moreover, the calculation apparatus 100 may perform a process of providing the second score to the web page provided by the content distribution server 40. Moreover, the calculation apparatus 100 may acquire transaction information from an external server other than the provider server 20.

[5-4. (External organization data)
In the said embodiment, in determining the creditworthiness of a 1st provider, the example which calculates the 1st score which is an index value regarding evaluation of a 2nd provider was shown. Here, the calculation apparatus 100 may use data of an external organization for the first score of the second operator. That is, the calculation apparatus 100 acquires a predetermined evaluation value when an external organization evaluates the second operator as the first score, and based on the acquired predetermined evaluation value, the first score of the first operator May be calculated.

  For example, the calculation device 100 uses an evaluation value published by an external organization that evaluates a business for listed companies or the like as the first score of the second business. That is, the calculation apparatus 100 may perform calculation processing using evaluation values such as publicly disclosed corporate value data and corporate asset data. As described above, the calculation apparatus 100 can provide objective reliability to the reliability of data and can simplify the calculation process by adopting the data of the external organization.

[5-5. Second operator]
In the said embodiment, the business operator who has dealings with the 1st business operator was handled as a 2nd business enterprise, and the example which performs the calculation process which concerns on embodiment was shown. Here, the calculation apparatus 100 may handle only a business that satisfies a predetermined condition among business operators that have a transaction with the first business operator as the second business operator.

  For example, the calculation device 100 may handle only a predetermined number of upper operators as the second operator in the annual transaction amount for the first operator. In addition, the calculation apparatus 100 may exclude an operator having a capital relationship with the first operator (that is, a parent company or a subsidiary) from the second operator.

  Moreover, the calculation apparatus 100 may handle a business operator that does not have a transaction with the first business operator as the second business operator. For example, the calculation apparatus 100 may handle businesses co-occurring in a portfolio formed by the same user as a first business operator and a second business company in a finance site or the like. In other words, the calculation device 100 determines that the operators having the relationship that the same user purchases the stock is likely to be the operators having some relationship, and thus the first operator and the second operator It may be handled as a relationship.

  Note that the calculation device 100 may indicate, for example, a vector notation as to which business operator the first business operator has a relationship with. For example, the calculation apparatus 100 sets a vector corresponding to “1” as a dimension corresponding to an operator having a relationship with the first operator and “0” as a dimension corresponding to an operator having no relationship with the first operator. The value having the relationship between the first business operator and the second business operator may be calculated.

[5-6. convergence〕
In the above-described embodiment, an example is shown in which the calculation is repeated until the provisional score calculated by the updated model and the provisional score calculated by the model before the update are within a predetermined threshold. Here, the calculation apparatus 100 may set the number of updates in advance, and may determine that the model generation is completed after the set number of updates has been performed. By such processing, the calculation apparatus 100 can prevent in advance the occurrence of a regression equation that does not converge the score semipermanently.

[6. Others]
In addition, among the processes described in the above embodiment, all or part of the processes described as being automatically performed can be performed manually, or the processes described as being performed manually can be performed. All or a part can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the document and drawings can be arbitrarily changed unless otherwise specified.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured.

  For example, the evaluation information storage unit 121, the transaction information storage unit 125, and the model storage unit 126 illustrated in FIG. 3 may be stored in an external storage server or the like without being stored in the calculation device 100. In this case, the calculation device 100 acquires evaluation information, transaction information, and the like by accessing the storage server.

  In addition, for example, the calculation device 100 described above is a front-end server that mainly executes exchanges with external devices such as obtaining evaluation information from the content distribution server 40 and obtaining transaction information from the business server 20. And the back-end server side that executes mainly the process of generating a model and calculating the second score.

[7. Hardware configuration)
Further, the calculation device 100, the user terminal 10, the provider server 20, the content distribution server 40, and the like according to the above-described embodiment are realized by a computer 1000 having a configuration as shown in FIG. Hereinafter, the calculation apparatus 100 will be described as an example. FIG. 11 is a hardware configuration diagram illustrating an example of a computer 1000 that implements the functions of the calculation apparatus 100. The computer 1000 includes a CPU 1100, a RAM 1200, a ROM (Read Only Memory) 1300, an HDD (Hard Disk Drive) 1400, a communication interface (I / F) 1500, an input / output interface (I / F) 1600, and a media interface (I / F). ) 1700.

  The CPU 1100 operates based on a program stored in the ROM 1300 or the HDD 1400 and controls each unit. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 is started up, a program depending on the hardware of the computer 1000, and the like.

  The HDD 1400 stores a program executed by the CPU 1100, data used by the program, and the like. The communication interface 1500 receives data from other devices via the communication network 500 (corresponding to the network N shown in FIG. 2) and sends the data to the CPU 1100, and other data generated by the CPU 1100 via the communication network 500 To the device.

  The CPU 1100 controls an output device such as a display and a printer and an input device such as a keyboard and a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. Further, the CPU 1100 outputs the data generated via the input / output interface 1600 to the output device.

  The media interface 1700 reads a program or data stored in the recording medium 1800 and provides it to the CPU 1100 via the RAM 1200. The CPU 1100 loads such a program from the recording medium 1800 onto the RAM 1200 via the media interface 1700, and executes the loaded program. The recording medium 1800 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase change rewritable disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. Etc.

  For example, when the computer 1000 functions as the calculation device 100, the CPU 1100 of the computer 1000 implements the function of the control unit 130 by executing a program loaded on the RAM 1200. In addition, each data in the storage unit 120 is stored in the HDD 1400. The CPU 1100 of the computer 1000 reads these programs from the recording medium 1800 and executes them, but as another example, these programs may be acquired from other devices via the communication network 500.

[8. effect〕
As described above, the calculation device 100 according to the embodiment includes the acquisition unit 131 and the second calculation unit 135 (an example of a calculation unit). The acquisition part 131 acquires the 1st score which showed the evaluation value of each provider. Of the first score acquired by the acquisition unit 131, the second calculation unit 135 uses the first score of the first operator and the first score of the second operator who has a business relationship with the first operator. Based on this, a second score indicating the creditworthiness of the first operator is calculated.

  As described above, the calculation apparatus 100 according to the embodiment calculates the credit rating of the first operator based on the information related to the second operator having a business relationship with the first operator. As a result, the calculation device 100 calculates the second score of the first operator that is associated with the higher-evaluated second operator relatively high, or the evaluation value of the second operator that is the supplier is high or low. It is possible to calculate the second score of the first operator that fluctuates according to the above. In other words, the calculation device 100 can calculate the degree of credit taking into account the relationship between the operators, not the evaluation of the first operator alone.

  In addition, the calculation device 100 according to the embodiment, based on the first score of the first operator and the first score of the second operator who has a business relationship with the first operator, A second generation unit 134 (an example of a generation unit) that generates a model that outputs the second score is further included. The second calculator 135 calculates the second score of the first operator using the model generated by the second generator 134.

  As described above, the calculation apparatus 100 according to the embodiment generates a model that uses not only the first score of the first operator but also the first score of the second operator, and uses the model to generate the second score. Calculate the score. Thereby, the calculation apparatus 100 can calculate the 2nd score of a 1st provider accurately, considering the evaluation value of the 2nd provider who is a supplier.

  Moreover, the acquisition part 131 acquires the 1st score calculated based on the information regarding the financial condition of each provider. The second generation unit 134 generates a model using the first score acquired by the acquisition unit 131.

  Thus, the calculation apparatus 100 according to the embodiment generates a model using the first score based on the financial status of the first operator. Thereby, the calculation apparatus 100 can generate a model reflecting the current financial status of the first operator.

  Moreover, the 2nd production | generation part 134 produces | generates a model using the value which shows the relationship between the 1st score of a 2nd provider, and the deposit / withdrawal of the said 2nd provider and a 1st provider.

  As described above, the calculation apparatus 100 according to the embodiment generates a model having information on the deposit / withdrawal of the second business operator, which is a business partner, as a feature. As a result, the calculation device 100 calculates the credit quality in accordance with the actual situation of the transaction between the business operators, such as calculating a high credit quality for the first business operator who has transactions with the second business operator with high credit. Can do.

  Moreover, the 2nd production | generation part 134 is the information regarding deposit / withdrawal as the amount of deposit / withdrawal in the transaction between the 1st provider and the 2nd provider, the number of transactions, the transaction frequency, the amount of deposit / withdrawal of the transaction in the predetermined period A model is generated using at least one of the following transitions, transitions in the number of transactions in a predetermined period, and transitions in transaction frequency.

  As described above, the calculation apparatus 100 according to the embodiment may generate a model using various types of information indicating the actual transaction status such as the amount of deposit and withdrawal and the number of transactions. Thereby, the calculation apparatus 100 can calculate the credit rating according to the actual situation of the transaction.

  In addition, the second generation unit 134 learns a regression equation in which the provisional score, which is a provisional score output by the model, is replaced with the value of the first score used in the model when the provisional score is output. To update the model.

  As described above, the calculation device 100 according to the embodiment updates the model by recursively using the score output from the generated model. Thereby, the calculation apparatus 100 can improve the precision of the numerical value to output in the model which outputs a score in consideration of the relationship between a 1st provider and a 2nd provider.

  The second generation unit 134 repeats updating the model until the provisional score output by the updated model and the provisional score output by the model before the update are within a predetermined threshold. When the provisional score output by the updated model and the provisional score output by the model before the update are within a predetermined threshold, the second calculation unit 135 uses the model to perform the first project The second score of the person is calculated.

  As described above, the calculation device 100 according to the embodiment calculates the second score using the model generated by repeating the update until the score output from the model converges. Thereby, the calculation apparatus 100 can calculate an appropriate second score in which fluctuation due to factors such as the calculation order is suppressed.

  Moreover, the acquisition part 131 acquires the 1st score calculated based on the evaluation value shown from the external organization which evaluates a provider. The second generation unit 134 generates a model based on the relationship between the first score of the second business operator and the information regarding the deposit / withdrawal between the second business operator and the first business operator.

  Thus, the calculation apparatus 100 according to the embodiment may use the first score calculated by the external organization for the first score. Thereby, since the calculation apparatus 100 can omit the process of calculating the first score, it is possible to reduce the load of information processing or to speed up the process.

  Moreover, the acquisition part 131 acquires the information regarding a seasonal factor of each provider. The second generation unit 134 generates a model based on the relationship between the first score of the second business operator and the information regarding the seasonal factors between the second business operator and the first business operator.

  As described above, the calculation apparatus 100 according to the embodiment may generate a model using seasonal factors regardless of the amount of deposit and withdrawal. As a result, the calculation device 100 can calculate the creditworthiness according to the season so that the financial institution can make an active loan to an operator whose business performance improves according to the season.

  As described above, some of the embodiments of the present application have been described in detail with reference to the drawings. However, these are merely examples, and various modifications, including the aspects described in the disclosure section of the invention, based on the knowledge of those skilled in the art, It is possible to implement the present invention in other forms with improvements.

  The above-described calculation device 100 may be realized by a plurality of server computers, and depending on the function, the configuration may be realized by calling an external platform or the like with an API (Application Programming Interface) or network computing. Can be changed flexibly.

  Further, “section (module, unit)” described in the claims can be read as “means”, “circuit”, and the like. For example, the acquisition unit can be read as acquisition means or an acquisition circuit.

DESCRIPTION OF SYMBOLS 1 Calculation system 10 User terminal 20 Provider server 40 Content delivery server 100 Calculation apparatus 110 Communication part 120 Storage part 121 Evaluation information storage part 122 Financial table 123 Stock price exchange table 124 Search data table 125 Transaction information storage part 126 Model storage part 130 Control Unit 131 acquisition unit 132 first generation unit 133 first calculation unit 134 second generation unit 135 second calculation unit

Claims (12)

An acquisition unit for acquiring a first score indicating an evaluation value of each operator;
Of the first scores acquired by the acquisition unit, the first business based on the first score of the first business operator and the first score of the second business operator having a business relationship with the first business operator. A calculation unit for calculating a second score indicating the creditworthiness of the person,
A calculation device comprising: Based on the first score of the first operator and the first score of the second operator having a business relationship with the first operator, a model for outputting the second score of the first operator is generated. Generator,
Further comprising
The calculation unit includes:
Using the model generated by the generating unit, the second score of the first operator is calculated.
The calculation apparatus according to claim 1. The acquisition unit
Obtaining the first score calculated based on information on the financial status of each operator;
The generator is
Generating the model using the first score acquired by the acquisition unit;
The calculation apparatus according to claim 2, wherein: The generator is
Using the first score of the second operator and a value indicating the deposit / withdrawal relationship between the second operator and the first operator, the model is generated.
The calculation apparatus according to claim 2 or 3, wherein The generator is
As information on the deposit / withdrawal, the amount of deposit / withdrawal, the number of transactions, the transaction frequency, the transition of the amount of deposit / withdrawal of transactions during a predetermined period, the number of transactions The model is generated using at least one of transition of transaction frequency or transition of transaction frequency.
The calculation apparatus according to claim 4. The generator is
By learning a regression equation in which the provisional score, which is a provisional score output by the model, is replaced with the value of the first score used in the model when the provisional score is output, Update,
The calculation device according to any one of claims 2 to 5, wherein The generator is
Repeatedly updating the model until the provisional score output by the updated model and the provisional score output by the model before the update are within a predetermined threshold,
The calculation unit includes:
When the provisional score output by the updated model and the provisional score output by the model before the update are within a predetermined threshold, the second score of the first operator is calculated using the model. calculate,
The calculation apparatus according to claim 6. The acquisition unit
Obtain a first score that is calculated based on the evaluation value indicated by the external organization that evaluates the operator,
The generator is
Generating the model based on the relationship between the first score of the second operator and the information on the deposit and withdrawal of the second operator and the first operator;
The calculation device according to any one of claims 2 to 7. The acquisition unit
Obtain information on the seasonal factors of each business operator,
The generator is
Based on the relationship between the first score of the second operator and the information on the seasonal factors of the second operator and the first operator, the model is generated.
The calculation device according to any one of claims 2 to 8. A calculation method executed by a computer,
An acquisition step of acquiring a first score indicating an evaluation value of each operator;
Of the first scores acquired in the acquisition step, the first business based on the first score of the first business operator and the first score of the second business operator having a business relationship with the first business operator. A calculation step of calculating a second score indicating the creditworthiness of the person,
The calculation method characterized by including. An acquisition procedure for acquiring a first score indicating an evaluation value of each operator;
Of the first scores acquired by the acquisition procedure, the first business based on the first score of the first business operator and the first score of the second business operator having a business relationship with the first business operator. A calculation procedure for calculating a second score indicating the creditworthiness of the person,
A calculation program for causing a computer to execute. Of the first score indicating the evaluation value of each business operator, the first score of the second business operator having a business relationship with the first business operator and the deposit / withdrawal between the second business operator and the first business operator An input layer to which a value indicating the relationship between information and information is input;
The output layer,
A first element belonging to any layer from the input layer to the output layer other than the output layer;
A second element whose value is calculated based on the first element and a weight of the first element;
The information input to the input layer is calculated based on the first element and the weight of the first element, with each element belonging to each layer other than the output layer as the first element. To output the value of the second score indicating the creditworthiness of one operator from the output layer,
A model for making a computer function.

Images