JP2021144356A - Illegal financial transaction detection program - Google Patents

Abstract

To provide an illegal financial transaction detection program for automatically and highly accurately detecting an illegal financial transaction.SOLUTION: An illegal financial transaction detection program for detecting an illegal financial transaction causes a computer to execute: an information acquisition step for acquiring difference pattern information on a transaction history in a bank account of a new detection object person to newly detect an illegal transaction; and a determination step for referring to three or more levels of degree of association between reference difference pattern information on a transaction history in a bank account of a past detection object person acquired in advance and a possibility of illegal transaction of the past detection object person and for determining a possibility of illegal transaction of the new detection object person on the basis of the difference pattern information acquired in the information acquisition step.SELECTED DRAWING: Figure 5

Description

The present invention relates to a fraudulent financial transaction detection program suitable for detecting fraudulent financial transactions.

In recent years, the so-called money laundering method has become widespread, in which illegally acquired funds are pretended to be funds obtained by legitimate means. In this money laundering, in order to obscure the source of funds, so to speak, dirty funds obtained by crimes such as tax evasion and window dressing are used to transfer money from time to time using a fictitious or financial institution account in the name of another person. Repeat, buy stocks and bonds, make large donations, etc.

Such money laundering methods have become particularly sophisticated in recent years, and there is a problem that it is difficult to detect them accurately only from the transaction history in a normal bank account. In addition, it takes a very long time to manually examine such transaction history one by one, and there is a problem that the burden of labor becomes excessive.

Therefore, the present invention has been devised in view of the above-mentioned problems, and an object thereof is to automatically and highly accurately detect fraudulent financial transactions such as the above-mentioned money laundering. It is to provide a possible fraudulent financial transaction detection program.

In order to solve the above-mentioned problems, the fraudulent financial transaction detection program according to the present invention is a fraudulent financial transaction detection program that detects fraudulent financial transactions. The information acquisition step of extracting the history over two or more terms and acquiring the difference pattern information including the pattern of the difference in the transaction history between the extracted terms, and the transaction history in the bank account of the past detection target person acquired in advance. Extract over two or more terms, and check the degree of association between the reference difference pattern information including the pattern of the difference in transaction history between the extracted terms and the possibility of fraudulent transactions of the person to be detected in the past. It is characterized in that the computer is made to execute the determination step of determining the possibility of fraudulent transaction of the new detection target person based on the difference pattern information acquired in the information acquisition step with reference to the reference.

Even if you do not have any special skills or experience, it is possible to automatically and accurately detect fraudulent financial transactions such as money laundering.

It is a block diagram which shows the whole structure of the system to which this invention is applied. It is a figure which shows the specific configuration example of a search device. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention. It is a figure for demonstrating the operation of this invention.

Hereinafter, the fraudulent financial transaction detection program to which the present invention is applied will be described in detail with reference to the drawings.

The first embodiment FIG . 1 is a block diagram showing an overall configuration of a fraudulent financial transaction detection system 1 in which a fraudulent financial transaction detection program to which the present invention is applied is implemented. The fraudulent financial transaction detection system 1 includes an information acquisition unit 9, a search device 2 connected to the information acquisition unit 9, and a database 3 connected to the search device 2.

The information acquisition unit 9 is a device for a person who uses this system to input various commands and information, and specifically, is composed of a keyboard, buttons, a touch panel, a mouse, a switch, and the like. The information acquisition unit 9 is not limited to a device for inputting text information, and may be configured by a device such as a microphone that can detect voice and convert it into text information. Further, the information acquisition unit 9 may be configured as an image pickup device capable of capturing an image of a camera or the like. The information acquisition unit 9 may be configured by a scanner having a function of recognizing a character string from a paper-based document. Further, the information acquisition unit 9 may be integrated with the discrimination device 2 described later. The information acquisition unit 9 outputs the detected information to the determination device 2. Further, the information acquisition unit 9 may be configured by means for specifying the position information by scanning the map information. Further, the information acquisition unit 9 may be composed of an illuminance sensor for measuring a temperature sensor, a humidity sensor, and a wind direction sensor. Further, the information acquisition unit 9 may be composed of a communication interface for acquiring data on the weather from the Japan Meteorological Agency or a private weather forecast company. Further, the information acquisition unit 9 may be composed of a body sensor that is attached to the body to detect body data, and this body sensor detects, for example, body temperature, heart rate, blood pressure, number of steps, walking speed, and acceleration. It may be composed of a sensor for the purpose. Further, the body sensor may acquire biological data of animals as well as humans. Further, the information acquisition unit 9 may be configured as a device that acquires information such as drawings by scanning or reading it from a database. In addition to these, the information acquisition unit 9 may be configured by an odor sensor that detects an odor or scent.

The database 3 stores various information necessary for detecting fraudulent financial transactions. As information necessary for detecting fraudulent financial transactions, the transaction history in the bank account of the person to be detected in the past is extracted over two or more terms, and for reference including the pattern of the difference in transaction history between the extracted terms. In each term when acquiring difference pattern information, reference geographical information regarding the geographical status of the area to which the past detection target person belongs, reference account opening information regarding the account opening status of the past detection target person, and difference pattern information. Reference market information regarding market conditions, reference external environment information regarding the external environment of the industry of the past detection target, reference credit information regarding the claims held by the past detection target, and trading behavior of the past detection target Reference trading behavior information, reference financial information regarding the finances of past detection target persons, reference pension insurance information regarding pensions and insurance paid by past detection target persons, reference attribute information regarding attributes of past detection target persons, A data set with the possibility of fraudulent transactions of the person to be detected in the past is stored.

That is, in the database 3, in addition to such reference difference pattern information, reference geographical information, reference account opening information, reference market condition information, reference external environment information, reference credit information, and reference trading behavior Any one or more of the information, the reference financial information, the reference annuity insurance information, and the reference attribute information and the possibility of fraudulent transactions of the detection target person determined in the past are stored in association with each other. By the way, the possibility of fraudulent transactions of the detection target person judged in the past is not limited to individuals and corporations that were actually judged whether or not they actually correspond to fraudulent transactions in the past, but a fictitious persona is set. , The possibility of fraudulent transactions of the detection target person determined by the financial institution or the like may be included in the learning data.

The search device 2 is composed of, for example, an electronic device such as a personal computer (PC), but is embodied in any other electronic device such as a mobile phone, a smartphone, a tablet terminal, a wearable terminal, etc., in addition to the PC. It may be converted. The user can obtain a search solution by the search device 2.

FIG. 2 shows a specific configuration example of the search device 2. The search device 2 performs wired communication or wireless communication with a control unit 24 for controlling the entire search device 2 and an operation unit 25 for inputting various control commands via operation buttons, a keyboard, or the like. A communication unit 26 for the purpose, an estimation unit 27 for making various judgments, and a storage unit 28 for storing a program for performing a search to be executed represented by a hard disk or the like are connected to the internal bus 21, respectively. .. Further, a display unit 23 as a monitor that actually displays information is connected to the internal bus 21.

The control unit 24 is a so-called central control unit for controlling each component mounted in the search device 2 by transmitting a control signal via the internal bus 21. Further, the control unit 24 transmits various control commands via the internal bus 21 in response to the operation via the operation unit 25.

The operation unit 25 is embodied by a keyboard or a touch panel, and an execution command for executing a program is input from the user. When the execution command is input by the user, the operation unit 25 notifies the control unit 24 of the execution command. Upon receiving this notification, the control unit 24, including the estimation unit 27, executes a desired processing operation in cooperation with each component. The operation unit 25 may be embodied as the information acquisition unit 9 described above.

The estimation unit 27 estimates the search solution. The estimation unit 27 reads out various information stored in the storage unit 28 and various information stored in the database 3 as necessary information when executing the estimation operation. The estimation unit 27 may be controlled by artificial intelligence. This artificial intelligence may be based on any well-known artificial intelligence technique.

The display unit 23 is composed of a graphic controller that creates a display image based on the control by the control unit 24. The display unit 23 is realized by, for example, a liquid crystal display (LCD) or the like.

When the storage unit 28 is composed of a hard disk, predetermined information is written to each address based on the control by the control unit 24, and is read out as needed. Further, the storage unit 28 stores a program for executing the present invention. This program is read and executed by the control unit 24.

The operation in the fraudulent financial transaction detection system 1 having the above-described configuration will be described.

In the fraudulent financial transaction detection system 1, for example, as shown in FIG. 3, it is premised that the degree of association between the reference difference pattern information and the possibility of fraudulent transaction is set and acquired in advance. .. The reference difference pattern information extracts the transaction history in the bank account of the past detection target person acquired in advance over two or more terms, and indicates the difference in the transaction history between the extracted terms. This past detection target person can be any person who has been actually investigated by a financial institution, etc. in the past to see if it corresponds to a fraudulent transaction, and as a result, a judgment has been made regarding the possibility of fraudulent transaction. .. This past detection target person includes both corporations and individuals.

The transaction history referred to here includes the amount of transaction and the sense of scale. This transaction history can be obtained from deposits and withdrawals to and from bank accounts and settlement data at financial institutions. Deposit and withdrawal information can be obtained from these. In addition, transaction history can be obtained from quotations, invoices, and invoices. This transaction history may be extracted from account deposit / withdrawal details, accounting data, WEB sites, financial statements, and book information.

The transaction history may be broadly classified into any of the remittance amount, the remittance country, and the remittance currency in the bank account of the detection target person. Further, this transaction history may be formed by a combination of any two or more of the remittance amount, the remittance country, and the remittance currency.

Further, this reference difference pattern information includes a transaction history including time-series changes in transactions in the bank account of the past detection target person. As shown in FIG. 4, when the horizontal axis is time and the vertical axis is amount, the amount of the transaction amount (including both remittance and deposit) and its time-series frequency are acquired, and this is used as reference difference pattern information. May be good. The horizontal axis is separated by terms T1 to T5 having the same time length. Term T1 is time t1 to t2, term T2 is time t2 to t3, term T3 is time t3 to t4, and term T4 is time t4 to. t5 and term T5 consist of times t5 to t6. At this time, as shown in FIG. 4, information regarding the currency of the transaction here may be included (unmarked currency indicates yen). It may also include information about the sending country as an alternative to currency. The information about the sending country may also include information on whether or not the country has had a transaction in the past.

As shown in FIG. 4, the terms T4 and T5 have a significantly higher transaction density than the previous terms T1 to T3, a large remittance amount, and a high degree of currency diversity. In this way, the number of remittances, the amount of remittance, the currency and the variety of remittance countries in each period are acquired as the above-mentioned transaction history.

In the present invention, transaction history in at least two or more terms is acquired. For example, it is assumed that the transaction history of the term T1 and the term T4 is acquired. At this time, when one of the terms is used as a reference, the pattern of difference between the other term and the other term is extracted. For example, when the term T1 is used as a reference, the number of remittances, the average transaction amount per transaction, the total transaction amount, and the type of currency of the other term T4 are increasing. The reference difference pattern information is obtained by assigning such a difference pattern to a preset type. In this case, it is applied to the category consisting of the category of "number of remittances, average transaction amount per transaction, total transaction amount, and currency type increase".

Further, for example, it is assumed that the transaction history of the term T4 and the term T5 is acquired. At this time, when the term T4 is used as a reference, the other term T5 has a decrease in the number of remittances, a total transaction amount, and a type of currency, but an increase in the average transaction amount per transaction. In this case, it is applied to the category of "the number of remittances, the total amount of transactions, and the types of currencies are decreasing, but the average transaction amount per transaction is increasing".

In the present invention, the difference in the transaction history between two or more terms may be extracted once for one past detection target person, and applied to one reference difference pattern information for one past detection target person. .. Further, the difference in the transaction history between two or more terms may be extracted twice or more for one past detection target person, and applied to two or more reference difference pattern information for one past detection target person. In the latter case, in the above example, two reference difference pattern information between the term T1 and the term T4 and two reference difference pattern information between the term T4 and the term T5 are obtained. The obtained two reference difference pattern information is linked to one of the past detection target persons. The same applies to the case where one reference difference pattern information is assigned with three or more terms.

Further, when detecting this difference pattern, the difference between the transaction history and the transaction history in two or more terms is considered in the coordinate system when the horizontal axis is time (time) and the vertical axis is the transaction amount. A value may be obtained and the tendency of this difference value may be applied to each of the above-mentioned patterns, or the difference value itself may be used as reference difference pattern information.

This difference value may be the remittance country or currency instead of the transaction amount on the vertical axis, and in this case, the difference value is ± 0 if it is the same currency (remittance country), and different currencies (remittance countries). ), It may be displayed as plus or minus between them.

The possibility of fraudulent transactions may be ranked as "normal", "need attention", "possible fraudulent transactions", "obvious fraudulent transactions", etc., or displayed as a score, 1000 points. Is the most secure, in other words, it is likely to be normal, 0 points is the least secure, in other words, the possibility of fraudulent transactions is high. The safest score may be any score other than 1000 points. Further, if the possibility of this fraudulent transaction is represented by two values of whether or not it was a fraudulent transaction, the two values may be learned as a data set of the output solution. This potential for fraudulent transactions is pre-collected as a dataset to and from reference variance pattern information. As for the possibility of this fraudulent transaction, for example, if there is a record of a judgment made by a financial institution before, it may be used as it is. Further, the possibility of fraudulent transaction may be calculated from the number of cases recognized as fraudulent transaction and the number of cases not recognized as fraudulent transaction with the same or similar reference difference pattern information. By the way, the possibility of this fraudulent transaction is attached to the company or individual who provided the difference pattern information for reference. For this reason, in the past examination, by acquiring the possibility of fraudulent transactions attached to the company or individual who provided the difference pattern information for reference and the company or individual, these data sets were obtained. Can be obtained. I will collect many such cases.

That is, through this reference difference pattern information and the data set of the possibility of fraudulent transactions, it is possible to know whether or not various data generated in the reference difference pattern information are determined as fraudulent transactions, or the possibility thereof. That is, the data described in the reference difference pattern information and the possibility of fraudulent transactions are a data set. Therefore, by collecting the reference difference pattern information and the data set of the possibility of fraudulent transactions, how the possibility of fraudulent transactions was judged in the examination after what kind of transaction history was in the past. It becomes possible to know.

In the example of FIG. 3, it is assumed that the input data is, for example, reference difference pattern information P01 to P03. The reference difference pattern information as such input data is linked to the output. In this output, the possibility of fraudulent transactions as an output solution is displayed.

The reference difference pattern information is associated with each other through three or more levels of association with respect to the possibility of fraudulent transactions as this output solution. The reference difference pattern information is arranged on the left side through this degree of association, and the possibility of fraudulent transactions is arranged on the right side through this degree of association. The degree of association indicates the degree of which fraudulent transaction is more likely to be related to the reference difference pattern information arranged on the left side. In other words, this degree of association is an index showing what kind of fraudulent transaction possibility each reference difference pattern information is likely to be associated with, and is the most probable fraudulent transaction from the reference difference pattern information. It shows the accuracy in selecting the possibility. In the example of FIG. 3, w13 to w19 are shown as the degree of association. These w13 to w19 are shown in 10 stages as shown in Table 1 below, and the closer to 10 points, the higher the degree of relevance of each combination as an intermediate node to the possibility of fraudulent transactions as output. On the contrary, the closer to one point, the lower the degree of relevance of each combination as an intermediate node to the possibility of fraudulent transactions as an output.

The search device 2 acquires in advance the degree of association w13 to w19 of three or more stages shown in FIG. That is, the search device 2 accumulates past data as to whether the reference difference pattern information or the possibility of fraudulent transaction in that case is adopted in determining the actual search solution, and stores these. By analyzing and analyzing, the degree of association shown in FIG. 3 is created.

For example, it is assumed that a certain reference difference pattern information is determined to be "apparent fraudulent transaction". In such a situation, it is assumed that the reference difference pattern information of a similar pattern is similarly determined to be "apparent fraudulent transaction". In such a case, the degree of association of "apparent fraudulent transactions" becomes stronger. On the other hand, in the exact same pattern (classification) of the difference pattern information for reference, it is assumed that many are judged to be "normal" and few are judged to be "clear fraudulent transactions". In such a case, the degree of association of "normal" becomes stronger, and the degree of association of "obvious fraudulent transaction" becomes lower.

This analysis and analysis may be performed by artificial intelligence. In such a case, for example, in the case of reference difference pattern information P01, analysis is performed from the data of the determination result of the possibility of fraudulent transactions in the past. This may be extracted from the data of the determination result of the possibility of fraudulent transactions in the past fraudulent transaction examinations stored in, for example, financial institutions, government agencies, the National Tax Agency, and the like. In the case of the difference pattern information P01 for reference, when there are many cases of "obvious fraudulent transaction", the degree of association leading to this "obvious fraudulent transaction" is set higher, and there are many cases of "normal". Is set to a higher degree of association that leads to this "normal". For example, in the example of the difference pattern information P01 for reference, it is linked to "possible fraudulent transaction" and "obvious fraudulent transaction", but the association of w13 leading to "possible fraudulent transaction" from the previous case. The degree is set to 7 points, and the degree of association of w14 that leads to "apparent fraudulent transactions" is set to 2 points.

Further, the degree of association shown in FIG. 3 may be composed of the nodes of the neural network in artificial intelligence. That is, the weighting coefficient for the output of the node of this neural network corresponds to the above-mentioned degree of association. Further, the network is not limited to a neural network, and may be composed of all decision-making factors constituting artificial intelligence.

In such a case, as shown in FIG. 5, reference difference pattern information is input as input data, the possibility of each fraudulent transaction is output as output data, and at least one or more hidden between the input node and the output node. Layers may be provided for machine learning. On the contrary, the possibility of fraudulent transactions may be input and the difference pattern information for reference may be output.

Such a degree of association is what is called learned data in artificial intelligence. After creating such learned data, in actually determining the possibility of fraudulent transactions from now on, the possibility of fraudulent transactions will be predicted using the above-mentioned learned data. In such a case, the latest difference pattern information of the new detection target person who actually newly determines the possibility of fraudulent transaction is acquired. This difference pattern information is composed of the same type of data as the above-mentioned reference difference pattern information, and the acquisition method thereof is the same as that of the reference difference pattern information. The new detection target person referred to here may be specified in advance at the time of detection, but is not limited to this, and all the persons included in a certain population are included in the new detection target person to make a judgment. You may do so. In such a case as well, the determination described below will be similarly executed for each new detection target person included in the population.

The newly acquired difference pattern information is input by the information acquisition unit 9 described above. The information acquisition unit 9 may acquire such difference pattern information as electronic data.

Based on the difference pattern information newly acquired in this way, the possibility of fraudulent transactions that are likely to be actually determined is searched for the difference pattern information. In such a case, the degree of association shown in FIG. 3 (Table 1) acquired in advance is referred to. For example, when the newly acquired difference pattern information is the same as or similar to P02, "normal" is associated with w15 and "obvious fraudulent transaction" is associated with association degree w16 through the association degree. There is. In such a case, "normal" with the highest degree of association is selected as the optimum solution. However, it is not essential to select the one with the highest degree of association as the optimum solution, and the “clear fraudulent transaction” that has the lowest degree of association but the association itself is recognized may be selected as the optimum solution. In addition to this, it goes without saying that an output solution to which the arrows are not connected may be selected, and as long as it is based on the degree of association, it may be selected in any other priority.

In this way, it is possible to search for the possibility of fraudulent transactions to be determined from the newly acquired difference pattern information and display it to the user (consultant, bank employee, customer, government agency, etc.). By looking at the search result, the user can grasp the possibility of the searched fraudulent transaction. By the way, in the process of outputting the possibility of fraudulent transaction, in addition to simply displaying the possibility of fraudulent transaction searched, a specific warning or alarm is given based on the possibility of fraudulent transaction. May be generated.

In the example of FIG. 6, it is assumed that the input data is, for example, reference difference pattern information P01 to P03 and reference geographical information P14 to 17. The intermediate node shown in FIG. 6 is a combination of reference geographical information and reference difference pattern information as such input data. Each intermediate node is further linked to the output. In this output, the possibility of each fraudulent transaction as an output solution is displayed.

In the example of FIG. 6, it is premised that a combination of the reference difference pattern information and the reference geographical information is formed. The reference geographical information is information on the area to which the past detection target actually belongs, and may be designated and classified by country, region, or municipality. In addition to the transaction status, it may be possible to determine the possibility of fraudulent transactions with higher accuracy by determining from which region such transactions are occurring, so these are used as explanatory variables. It is an addition.

In the example of FIG. 6, it is assumed that the input data is, for example, reference difference pattern information P01 to P03 and reference geographical information P14 to 17. The intermediate node shown in FIG. 6 is a combination of reference geographical information and reference difference pattern information as such input data. Each intermediate node is further linked to the output. In this output, the possibility of fraudulent transactions as an output solution is displayed.

Each combination (intermediate node) of the reference difference pattern information and the reference geographical information is associated with each other through three or more levels of association with respect to the possibility of fraudulent transactions as this output solution. The reference difference pattern information and the reference geographical information are arranged on the left side through this degree of association, and the possibility of fraudulent transactions is arranged on the right side through this degree of association. The degree of association indicates the degree of possibility of each fraudulent transaction and the degree of relevance to the reference difference pattern information and the reference geographical information arranged on the left side. In other words, this degree of association is an index showing what kind of fraudulent transaction possibility each reference difference pattern information and reference geographical information is likely to be associated with, and is referred to as reference difference pattern information. It shows the accuracy in selecting the most probable possibility of fraudulent transactions from the reference geographical information. In addition to the difference pattern information, the possibility of fraudulent transactions may change depending on the actual geographical situation, and it is possible to make a judgment based on this. Therefore, by combining these reference difference pattern information and reference geographical information, the optimum possibility of fraudulent transactions will be searched for.

In the example of FIG. 6, w13 to w22 are shown as the degree of association. As shown in Table 1, these w13 to w22 are shown in 10 stages, and the closer to 10 points, the higher the degree of relevance of each combination as an intermediate node to the output, and conversely, 1 point. The closer they are, the less relevant each combination as an intermediate node is to the output.

The search device 2 acquires in advance the degree of association w13 to w22 of three or more stages shown in FIG. That is, the search device 2 accumulates past data as to which of the reference difference pattern information, the reference geographical information, and the possibility of fraudulent transaction in that case was suitable for determining the actual search solution. By analyzing and analyzing these, the degree of association shown in FIG. 6 is created.

This analysis and analysis may be performed by artificial intelligence. In such a case, for example, in the case of the reference difference pattern information P01 and the reference geographical information P16, the possibility of fraudulent transactions is analyzed from the past data. When there are many cases where the possibility of fraudulent transactions is "possible fraudulent transactions", the degree of association that leads to this "possible fraudulent transactions" is set higher, and there are many cases of "normal". When there are few cases of "obvious fraudulent transactions", the degree of association leading to "normal" is set high, and the degree of association leading to "obvious fraudulent transactions" is set low. For example, in the example of the intermediate node 61a, it is linked to the output of "possible fraudulent transaction" and "normal", but from the previous case, the degree of association of w13 that leads to "possible fraudulent transaction" is 7 points. In addition, the degree of association of w14 that leads to "normal" is set to 2 points.

Further, the degree of association shown in FIG. 6 may be composed of the nodes of the neural network in artificial intelligence. That is, the weighting coefficient for the output of the node of this neural network corresponds to the above-mentioned degree of association. Further, the network is not limited to a neural network, and may be composed of all decision-making factors constituting artificial intelligence.

In the example of the degree of association shown in FIG. 6, the node 61b is a node of the combination of the reference difference pattern information P01 and the reference geographical information P14, and the degree of association of "obvious fraudulent transaction" is w15, " The degree of association of "normal" is w16. The node 61c is a node that is a combination of the reference geographical information P15 and P17 with respect to the reference difference pattern information P02, and the “normal” association degree is w17 and the “attention required” association degree is w18. ing.

Such a degree of association is what is called learned data in artificial intelligence. After creating such learned data, the above-mentioned learned data will be used in the actual search for determining the possibility of fraudulent transactions. In such a case, geographical information is acquired in addition to the difference pattern information from the company that newly determines the possibility of fraudulent transactions. This geographical information corresponds to the above-mentioned reference geographical information, and is input in units of, for example, a country, a region, and a city, ward, town, or village.

Based on the newly acquired difference pattern information and geographical information in this way, the optimum possibility of fraudulent transactions is searched for. In such a case, the degree of association shown in FIG. 6 (Table 1) acquired in advance is referred to. For example, when the newly acquired difference pattern information is the same as or similar to P02 and the geographical information is the same as or similar to P17, the node 61d is set via the degree of association. The node 61d is associated with a "substantial failure" w19 and a "need attention" with a degree of association w20. In such a case, the “substantial bankruptcy” with the highest degree of association is selected as the optimum solution. However, it is not essential to select the one with the highest degree of association as the optimum solution, and it is also possible to select the “point of caution” as the optimum solution, which has the lowest degree of association but the association itself is recognized. In addition to this, it goes without saying that an output solution to which the arrows are not connected may be selected, and as long as it is based on the degree of association, it may be selected in any other priority.

In addition, Table 2 below shows examples of the degrees of association w1 to w12 extending from the input.

The intermediate node 61 may be selected based on the degree of association w1 to w12 extending from this input. That is, the larger the degree of association w1 to w12, the heavier the weighting in the selection of the intermediate node 61 may be. However, the degrees of association w1 to w12 may all have the same value, and the weights in the selection of the intermediate node 61 may all be the same.

FIG. 7 shows an example in which the combination of the above-mentioned reference difference pattern information, the reference account opening information, and the possibility of fraudulent transactions with respect to the combination are set to three or more levels of association.

The reference account opening information includes all the information submitted or registered when the person to be detected in the past opened a bank account. The reference account opening information includes information on personal identification as a corporation (registration certificate, etc.), personal identification as an individual account opening applicant, as well as information such as transaction purpose and business content to be submitted when opening an account. Is done. In addition, the face image of the identification card with a face photo for identity verification may be included in the reference account opening information.

The face image is composed of a face photograph or a digital image taken through a camera, and may be either a still image or a moving image. Each such face image is analyzed. The face image may be automatically discriminated based on the feature amount of the analyzed image and converted into data by using image analysis and, if necessary, deep learning technology. For example, it is possible to extract features that affect the credit aspect, such as a part that seems to be unconfident by lying and a part that contains malicious intent through a face image.

Various items such as transaction purpose and business content described in the documents submitted at the time of opening an account may be extracted as they are if they are submitted as text information, or a character recognition function such as OCR may be used. You may want to read this through.

The extracted transaction purpose and business content may be categorized by a predetermined pattern (for example, the transaction purpose is foreign exchange remittance for trade, buying and selling of intellectual property, etc.).

In the example of FIG. 7, it is assumed that the input data is, for example, reference difference pattern information P01 to P03 and reference account opening information P18 to 21. The intermediate node shown in FIG. 7 is a combination of the reference difference pattern information and the reference account opening information as such input data. Each intermediate node is further linked to the output. In this output, the possibility of fraudulent transactions as an output solution is displayed.

Each combination (intermediate node) of the reference difference pattern information and the reference account opening information is associated with each other through three or more levels of association with respect to the possibility of fraudulent transactions as this output solution. The reference difference pattern information and the reference account opening information are arranged on the left side through this degree of association, and the possibility of fraudulent transactions is arranged on the right side through this degree of association. The degree of association indicates the degree of high relevance to the possibility of fraudulent transactions with respect to the reference difference pattern information and the reference account opening information arranged on the left side. In other words, this degree of association is an index indicating what kind of fraudulent transaction possibility each reference difference pattern information and reference account opening information is likely to be associated with, and is a reference difference pattern information and reference. It shows the accuracy in selecting the most probable possibility of fraudulent transactions from the account opening information. In addition to the difference pattern information, the possibility of fraudulent transactions changes depending on what kind of account opening is. For example, if a transaction that clearly deviates from the original purpose with respect to the transaction purpose or business content at the time of opening an account is reflected as difference pattern information, there is a high possibility that it corresponds to a fraudulent transaction. Therefore, by combining these reference difference pattern information and reference account opening information, the optimum possibility of fraudulent transactions will be searched for.

The search device 2 acquires in advance the degree of association w13 to w22 of three or more stages shown in FIG. 7. That is, the search device 2 accumulates past data as to which of the reference difference pattern information, the reference account opening information, and the possibility of fraudulent transaction in that case was suitable for determining the actual search solution. By analyzing and analyzing these, the degree of association shown in FIG. 7 is created.

This analysis and analysis may be performed by artificial intelligence. In such a case, for example, in the case of the reference difference pattern information P01 and the reference account opening information P20, the possibility of fraudulent transactions is analyzed from past data. For example, in the example of the intermediate node 61a, it is linked to the output of "possible fraudulent transaction" and "normal", but from the previous case, the degree of association of w13 that leads to "possible fraudulent transaction" is 7 points. In addition, the degree of association of w14 that leads to "normal" is set to 2 points.

Further, the degree of association shown in FIG. 7 may be composed of the nodes of the neural network in artificial intelligence. That is, the weighting coefficient for the output of the node of this neural network corresponds to the above-mentioned degree of association. Further, the network is not limited to a neural network, and may be composed of all decision-making factors constituting artificial intelligence.

In the example of the degree of association shown in FIG. 7, the node 61b is a node of the combination of the reference difference pattern information P01 and the reference account opening information P18, and the degree of association of "clear fraudulent transaction" is w15, " The degree of association of "normal" is w16. The node 61c is a node that is a combination of the reference account opening information P19 and P21 with respect to the reference difference pattern information P02, and the "normal" association degree is w17 and the "attention required" association degree is w18. ing.

Such a degree of association is what is called learned data in artificial intelligence. After creating such trained data, when actually giving advice from now on, the above-mentioned trained data will be used. In such a case, in addition to the difference pattern information described above, the account opening information of the new detection target person is acquired. The account opening information corresponds to the reference account opening information.

Based on the newly acquired difference pattern information and account opening information in this way, the possibility of fraudulent transactions is searched for. In such a case, the degree of association shown in FIG. 7 (Table 1) acquired in advance is referred to. For example, when the newly acquired difference pattern information is the same as or similar to P02 and the account opening information is P21, the node 61d is associated via the degree of association. The node 61d is associated with "obvious fraudulent transaction" by w19 and "need attention" by the degree of association w20. In such a case, the “obvious fraudulent transaction” with the highest degree of association is selected as the optimal solution. However, it is not essential to select the one with the highest degree of association as the optimum solution, and it is also possible to select the “point of caution” as the optimum solution, which has the lowest degree of association but the association itself is recognized. In addition to this, it goes without saying that an output solution to which the arrows are not connected may be selected, and as long as it is based on the degree of association, it may be selected in any other priority.

FIG. 8 shows an example in which a combination of the above-mentioned reference difference pattern information, reference market condition information, and the possibility of fraudulent transactions with respect to the combination are set to three or more levels of association.

Reference market information is various information related to market conditions. The market conditions referred to here may target any range including one company, the entire industry including the company, the entire Japan, and the entire world. Examples of this reference market information include interest rates, foreign exchange, stock prices of each stock, crude oil, futures, precious metals, and price movements of Bitcoin. This reference market information may be displayed as a time-series chart, a line graph, or the like for these targets. Further, information such as Bollinger band, MACD, moving average line and the like may be attached. In addition, this market information may include fundamental indicators of each brand's company, and in addition to annual sales, profits, and costs, PER (price-earnings ratio), PBR (price-to-book value ratio), ROE ( Indicators such as return on equity) may be included. For foreign exchange, information such as a chart showing price movements between currencies, Bollinger Bands, MACD, and moving average lines may be attached. The reference market information may categorize the market potential itself. That is, the reference market information P26 to P29 in FIG. 8 may be information classified by type, and may be classified by dividing according to whether or not the growth rate of the stock price is 0% or more per year. Further, it may be categorized by a pattern (for example, a pattern of whether the growth rate of the stock price increases rapidly or gradually) or the like.

This reference market condition information relates to the market condition corresponding to each term when obtaining the above-mentioned reference difference pattern information. If this reference difference pattern information is shown in a time-series trend as shown in Fig. 4, this reference market information also has interest rates, exchange rates, and stock price movements of each stock on the time axis along this line. May be indicated.

In the example of FIG. 8, it is assumed that the input data is, for example, reference difference pattern information P01 to P03 and reference market condition information P18 to 21. The intermediate node shown in FIG. 8 is a combination of reference market condition information and reference difference pattern information as such input data. Each intermediate node is further linked to the output. In this output, the possibility of fraudulent transactions as an output solution is displayed.

Each combination (intermediate node) of the reference difference pattern information and the reference market condition information is associated with each other through three or more levels of association with the possibility of fraudulent transactions as this output solution. The reference difference pattern information and the reference market information are arranged on the left side through this degree of association, and the possibility of fraudulent transactions is arranged on the right side through this degree of association. The degree of association indicates the degree of high possibility of fraudulent transactions and the degree of relevance to the reference difference pattern information and the reference market condition information arranged on the left side. In other words, this degree of association is an index showing what kind of fraudulent transaction possibility each reference difference pattern information and reference market condition information is likely to be associated with, and is a reference difference pattern information and reference. It shows the accuracy in selecting the most probable possibility of fraudulent transactions from market information. In addition to the difference pattern information, the possibility of fraudulent transactions changes depending on market conditions. For example, even though the market is currently appreciating the yen or stocks are falling sharply, transactions that clearly deviate from the original transaction pattern under that situation are reflected as difference pattern information. If so, there is a high possibility that it corresponds to a fraudulent transaction. Therefore, by combining these reference difference pattern information and reference market condition information, the possibility of optimal fraudulent transactions will be searched for.

The search device 2 acquires in advance the degree of association w13 to w22 of three or more stages shown in FIG. That is, the search device 2 accumulates past data as to which is more suitable for the reference difference pattern information, the reference market condition information, and the possibility of fraudulent transaction in that case in determining the actual search solution. By analyzing and analyzing these, the degree of association shown in FIG. 8 is created.

In the example of the degree of association shown in FIG. 8, the node 61b is a node of the combination of the reference difference pattern information P01 and the reference market condition information P18, and the degree of association of "clear fraudulent transaction" is w15, "normal". The degree of association is w16. The node 61c is a node that is a combination of the reference market condition information P19 and P21 with respect to the reference difference pattern information P02. There is.

Such a degree of association is what is called learned data in artificial intelligence. After creating such trained data, when actually giving advice from now on, the above-mentioned trained data will be used. In such a case, in addition to the above-mentioned difference pattern information, the market condition information from the time when the difference pattern information is acquired to the point retroactive to a predetermined period is acquired. The market information corresponds to the reference market information.

Based on the newly acquired difference pattern information and market condition information in this way, the possibility of fraudulent transactions is searched for. In such a case, the degree of association shown in FIG. 8 (Table 1) acquired in advance is referred to. For example, when the newly acquired difference pattern information is the same as or similar to P02 and the market condition information is P21, the node 61d is associated via the degree of association. The node 61d is associated with "obvious fraudulent transaction" by w19 and "need attention" by the degree of association w20. In such a case, the “obvious fraudulent transaction” with the highest degree of association is selected as the optimal solution. However, it is not essential to select the one with the highest degree of association as the optimum solution, and it is also possible to select the “point of caution” as the optimum solution, which has the lowest degree of association but the association itself is recognized. In addition to this, it goes without saying that an output solution to which the arrows are not connected may be selected, and as long as it is based on the degree of association, it may be selected in any other priority.

FIG. 9 shows an example in which the combination of the above-mentioned reference difference pattern information, the reference external environment information, and the possibility of fraudulent transactions with respect to the combination are set to three or more levels of association.

The reference external environment information is all information about the external environment of the industry of the past detection target person. The external environmental information referred to here is economic data (GDP, employment statistics, mining and industry production index, capital investment, labor force survey, consumer price index, Japan Bank Short View, etc.), household data (household consumption status survey, household data, one week). Average working hours, savings amount statistical data, annual income statistical data, etc.), real estate data (office vacancy rate, unit price per tsubo, rent market, land price, vacant house data, etc.), natural environment data (disaster data, temperature data, precipitation, etc.) It is typified by quantity data, wind direction data, humidity data, etc.). External environmental information includes not only part or all of these data, but also information on all external environments such as politics, economy, society, technological evolution, epidemics, trends, epidemics, and natural disasters. good. Further, the reference external environment information may be such external environment information defined by text information, and these patterns (for example, whether the growth rate of GDP increases rapidly or gradually, etc.) It may be categorized by (pattern) or the like.

In the example of FIG. 9, it is assumed that the input data is, for example, reference difference pattern information P01 to P03 and reference external environment information P18 to 21. The intermediate node shown in FIG. 9 is a combination of the reference difference pattern information and the reference external environment information as such input data. Each intermediate node is further linked to the output. In this output, the possibility of fraudulent transactions as an output solution is displayed.

Each combination (intermediate node) of the reference difference pattern information and the reference external environment information is associated with each other through three or more levels of association with the possibility of fraudulent transactions as this output solution. The reference difference pattern information and the reference external environment information are arranged on the left side through this degree of association, and the possibility of fraudulent transactions is arranged on the right side through this degree of association. The degree of association indicates the degree of high relevance to the possibility of fraudulent transactions with respect to the reference difference pattern information and the reference external environment information arranged on the left side. In other words, this degree of association is an index indicating what kind of fraudulent transaction possibility each reference difference pattern information and reference external environment information is likely to be associated with, and is a reference difference pattern information and reference. It shows the accuracy in selecting the most probable possibility of fraudulent transactions from the external environment information. In addition to the difference pattern information, the possibility of fraudulent transactions changes depending on market conditions. For example, in the external environment, when the consumer price index is rapidly declining, but the transaction that clearly deviates from the original transaction pattern under that situation is reflected as the difference pattern information. Is more likely to fall under fraudulent transactions. Therefore, the optimum possibility of fraudulent transactions is searched for by combining the reference difference pattern information and the reference external environment information.

The search device 2 acquires in advance the degree of association w13 to w22 of three or more stages shown in FIG. That is, the search device 2 accumulates past data as to which is more suitable for the reference difference pattern information, the reference external environment information, and the possibility of fraudulent transaction in that case in determining the actual search solution. By analyzing and analyzing these, the degree of association shown in FIG. 9 is created.

In the example of the degree of association shown in FIG. 9, the node 61b is a node of the combination of the reference difference pattern information P01 and the reference external environment information P18, and the degree of association of "clear fraudulent transaction" is w15, " The degree of association of "normal" is w16. The node 61c is a node that is a combination of the reference external environment information P19 and P21 with respect to the reference difference pattern information P02. ing.

Such a degree of association is what is called learned data in artificial intelligence. After creating such trained data, when actually giving advice from now on, the above-mentioned trained data will be used. In such a case, in addition to the difference pattern information described above, the external environment information from the time when the difference pattern information is acquired to the point retroactive to a predetermined period is acquired. The external environment information corresponds to the reference external environment information.

Based on the newly acquired difference pattern information and external environment information in this way, the possibility of fraudulent transactions is searched for. In such a case, the degree of association shown in FIG. 9 (Table 1) acquired in advance is referred to. For example, when the newly acquired difference pattern information is the same as or similar to P02 and the external environment information is P21, the node 61d is associated via the degree of association. The node 61d is associated with "obvious fraudulent transaction" by w19 and "need attention" by the degree of association w20. In such a case, the “obvious fraudulent transaction” with the highest degree of association is selected as the optimal solution. However, it is not essential to select the one with the highest degree of association as the optimum solution, and it is also possible to select the “point of caution” as the optimum solution, which has the lowest degree of association but the association itself is recognized. In addition to this, it goes without saying that an output solution to which the arrows are not connected may be selected, and as long as it is based on the degree of association, it may be selected in any other priority.

In the present invention, as an alternative to the reference external environment information shown in FIG. 9, the reference information shown below is learned by being associated with the degree of association by a combination of reference difference pattern information. May be good.

As an example of the reference information, the reference claim information may be used. This reference claim information is information on the claims held by the person to be detected in the past. In addition to stock certificates and futures, this claim includes all claims such as intellectual property rights such as patent rights, rights related to leasing, and loan refund claim rights to have the loaned money returned. .. This claim information includes the validity period of the claim and all the contents of the claim. When acquiring this reference bond information, stock certificates, futures, corporate bonds and government bonds may be acquired directly from the database in which the contents of the stock certificates and government bonds actually held are stored. Further, when the content of the claim is described in the contract, the description may be read by using the above-mentioned OCR or the like, or may be acquired as text information.

In such a case, there are three stages: a combination of reference difference pattern information and reference claim information regarding the claims held by the past detection target person, and the possibility of fraudulent transactions of the past detection target person. Obtain the above degree of association in advance. At the time of actual judgment, the claim information regarding the claim held by the new detection target person is acquired. This claim information corresponds to the above-mentioned reference claim information, and the acquisition method is also the same. Then, with reference to the degree of association, the possibility of fraudulent transaction of the newly detected target person is determined in the same manner as described above based on the newly acquired difference pattern information and the credit information.

Further, as an example of other reference information, reference trading behavior information may be used. This trading behavior information includes all information related to trading behavior performed by the person to be detected in the past. Examples of this reference trading behavior information include information on transactions conducted through the EC site, information on the purchase and sale of stock certificates and futures, corporate bonds and government bonds described above, and include at least the amount of the object of the transaction. These reference trading behavior information can be obtained from accounting data, WEB sites, financial statements, and the like. In addition, if a transaction history has been negotiated with a trading partner in the past and the transaction history holds a paper document or electronic data of the document, reference difference pattern information may be extracted from this transaction history document. .. Further, if the individual is a self-employed person, the difference pattern information for reference may be extracted from the main business partner information described in the site, SNS, etc. launched for the self-employed person.

By determining fraudulent transactions in combination with such reference trading behavior information, it is possible to suppress the case of money laundering by remittance by using an online settlement service and mediating legal commercial transactions.

In addition, this reference trading behavior information may include a business partner. If the transaction is biased to a specific company from the past deposit history by comparing with the difference pattern information for reference, or if the field of the business partner company is different from the past detection target person in the first place, it corresponds to fraudulent transaction. In some cases, the probability of doing so can be increased, and the determination accuracy can be improved.

For reference trading behavior information, these purchasing information may be extracted from the books. Further, these purchase information may be obtained from a receipt or the like. The reference trading behavior information may include sales on the EC (electronic commerce) site, purchase history, and purchase history imported from cloud accounting information. In addition, the reference trading behavior information may statistically indicate what is actually purchased and at what ratio in the purchases that have been made so far. For example, one individual spends a quarter of the annual money on housing rent, one quarter for food, another quarter for hobby golf, and a quarter. The breakdown of purchase targets, such as buying clothing and miscellaneous goods, may be shown as a ratio. When applying this reference consumption behavior information to the above-mentioned reference information, for example, a certain reference information was the most purchased last month, with 15% for clothes, 13% for food expenses, and dating. The cost is 10%, ..., and other reference information is 25% for rent, 20% for food, 14% for automobiles, etc. Is.

In such a case, the degree of association between the combination of the reference difference pattern information and the reference trading behavior information of the past detection target person and the possibility of fraudulent transaction of the past detection target person is determined. Obtain it in advance. At the time of actual determination, the trading behavior information regarding the trading behavior of the new detection target person is acquired. This trading behavior information corresponds to the above-mentioned reference trading behavior information, and the acquisition method is also the same. Then, after referring to the degree of association, the possibility of fraudulent transaction of the newly detected target person is determined in the same manner as described above based on the newly acquired difference pattern information and the trading behavior information.

Further, as an example of other reference information, reference financial information may be used. The reference financial information is data related to all financial statements including the financial statements of the person to be detected in the past. Debt, equity capital ratio Shareholders' equity, total capital, gearing ratio, interest-bearing debt, equity capital, ordinary income margin, ordinary income, sales, total capital ordinary income ratio, ordinary income, total capital, net income, ordinary income Profit growth rate, current term ordinary income, previous term ordinary income, sales, repayment ability, debt repayment years, interest-bearing debt, operating funds, cash flow, operating income, depreciation expenses, etc. In addition to the figures in these financial statements, the reference financial information may include whether or not the debt is insolvent, and how much borrowing from officers and inventories (inventory) are. These reference financial information may be for one year or several years.

By combining such financial information with the difference pattern information to determine fraudulent transactions, the accuracy of the determination can be further improved.

In such a case, the combination of the reference difference pattern information and the reference financial information of the past detection target person and the possibility of fraudulent transaction of the past detection target person are linked in advance to three or more levels. Get it. At the time of actual judgment, financial information regarding the finance of the new detection target person is acquired. This financial information corresponds to the above-mentioned reference financial information, and the acquisition method is also the same. Then, with reference to the degree of association, the possibility of fraudulent transactions of the newly detected target person is determined in the same manner as described above based on the newly acquired difference pattern information and financial information.

Further, as an example of other reference information, reference pension insurance information may be used. Reference annuity insurance information includes all information about annuities and insurance actually paid by past detection targets. Since there is also a method of money laundering through pensions and insurance, adding this to the explanatory variables improves the judgment accuracy.

The pension mentioned here includes not only those that are obliged to pay such as welfare pension and national pension, but also pensions that can be accumulated by one's own will. In addition to health insurance, insurance also includes all types of insurance that you can contract on your own initiative.

In such a case, the combination of the reference difference pattern information and the reference pension insurance information of the past detection target person and the possibility of fraudulent transactions of the past detection target person are linked to each other at three or more levels. Obtain it in advance. At the time of actual judgment, pension insurance information related to the pension and insurance of the newly detected person is acquired. This pension insurance information corresponds to the above-mentioned reference pension insurance information, and the acquisition method is also the same. Then, with reference to the degree of association, the possibility of fraudulent transactions of the newly detected person is determined in the same manner as described above based on the newly acquired difference pattern information and the annuity insurance information.

Further, as an example of other reference information, reference attribute information may be used. The reference attribute information is all information related to the attributes of the past detection target person. The reference attribute information includes the age and health status, annual income, date of birth, birthplace, and family environment of the person to be detected (whether or not the wife is pregnant, the family is pursuing a disability or illness, and whether or not they live with their grandparents. , Whether the care recipient is at home, whether or not they are affected by the disaster, whether or not they are DV or abused, whether or not they are divorced, whether or not they are divorced, whether or not they are divorced, whether or not they have household income, whether or not they have livelihood protection, whether or not they have wives and children, and their age structure, etc. , Child's school attendance status), and if the child is in school, the school expenses and expected school expenses in the future, whether the individual is currently in employment, working, place of employment, years of continuous employment, etc. are also included. Regarding the state of health, whether or not the individual is in perfect health or has some kind of congenital disorder and the degree of the disorder, and whether or not there is an acquired disorder after birth and the degree of the disorder. Also, whether or not you have some kind of illness after birth, whether or not the illness continues at present and its degree, allergic condition, inflammatory condition, injury condition, pre-existing condition, medication status, etc. , Contains all information that indicates health status. The health condition included in the reference attribute information may be derived from medical data itself such as heart rate, pulse rate, blood data, electrocardiogram data, and X-ray image. All of these may be obtained through declarations and documents submitted by each household, medical certificates of doctors, and the like. When applying this reference attribute information to the above-mentioned reference information, for example, one reference information is age 46 years old, birthplace: Shizuoka, annual income of ¥ 100,000, 〇 × working at a company, family environment: with wife. The eldest son, the eldest daughter's family of four, the eldest son is a high school student, the eldest daughter is a junior high school student, health condition: I have had surgery for gastric ulcer in the past, other reference information is age 38 years old, birthplace: Tokyo, annual income 100,000 yen , ▲ 〇 Worked at a company, Family environment: Family of three, wife and eldest daughter, eldest daughter is elementary school student, health condition: good, etc.

In such a case, the combination of the reference difference pattern information and the reference attribute information of the past detection target person and the possibility of fraudulent transaction of the past detection target person are linked in advance to three or more levels. Get it. At the time of actual determination, the attribute information regarding the attribute of the new detection target person is acquired. This attribute information corresponds to the above-mentioned reference attribute information, and the acquisition method is also the same. Then, with reference to the degree of association, the possibility of fraudulent transaction of the newly detected target person is determined in the same manner as described above based on the newly acquired difference pattern information and attribute information.

In addition to the difference pattern information, if you want to acquire any two or more of geographical information, account opening information, market condition information, external environment information, credit information, trading behavior information, financial information, pension insurance information, and attribute information. , Learning data consisting of three or more levels of association between the combination of two or more reference information according to the two or more information to be acquired with the reference difference pattern information and the possibility of fraudulent transactions for the combination. By creating, it is possible to search for a solution to the possibility of fraudulent transactions in the same way.

In addition to the difference pattern information, in addition to any one or more of geographical information, account opening information, market condition information, external environment information, credit information, trading behavior information, financial information, pension insurance information, attribute information, and other parameters. By creating learning data consisting of three or more levels of association between the combination with and the possibility of fraudulent transactions with respect to the combination, it is possible to similarly search for a solution to the possibility of fraudulent transactions.

Further, as shown in FIG. 10, the present invention determines the possibility of each fraudulent transaction based on the degree of association of two or more types of information, the reference information U and the reference information V. It is assumed that the reference information U is the reference difference pattern information, and the reference information V is other reference information. Alternatively, it is assumed that the reference information U is other reference information and the reference information V is the reference difference pattern information.

At this time, as shown in FIG. 10, the output obtained for the reference information U may be used as the input data as it is, and may be associated with the output via the intermediate node 61 in combination with the reference information V. For example, for the reference information U, after the output solution is output, this may be used as an input as it is, and the output may be searched by using the degree of association with other reference information V.
Further, the present invention is not limited to the above-described embodiment, and for example, as shown in FIG. 11, search for reference difference pattern information from the transaction history of each term T through machine learning by artificial intelligence. It may be. FIG. 11 shows a case where one reference difference pattern information is assigned from the transaction history of three terms. Data sets of each type of reference pattern information for the combination of transaction history of each term T are collected. Let me learn this. Then, by using this learned model and inputting the transaction history of two or more terms actually acquired from the detection target person, the reference difference pattern information matching the transaction history may be output. That is, a learned model associated with each other through three or more levels of association with reference difference pattern information for a combination of transaction histories of each term is constructed in advance, and based on each transaction history of two or more terms. , Find the difference pattern information for reference by referring to this trained model. Not only when the difference pattern information for reference is obtained, but also when the difference pattern information is obtained, the same procedure may be performed.

In the above-mentioned degree of association, the degree of association is expressed by a 10-step evaluation, but it is not limited to this, and it may be expressed by a degree of association of 3 or more levels, and conversely, it may be expressed by 3 or more levels. For example, 100 steps or 1000 steps may be used. On the other hand, this degree of association does not include those expressed in two stages, that is, whether or not they are related to each other, either 1 or 0.

According to the present invention having the above-described configuration, anyone can easily search for the possibility of fraudulent transactions of a company considering financing without any special skill or experience. Further, according to the present invention, it is possible to determine the search solution with higher accuracy than that performed by a human being. Further, by configuring the above-mentioned degree of association with artificial intelligence (neural network or the like), it is possible to further improve the discrimination accuracy by learning this.

Since there are many cases where the input data and the output data described above do not exist exactly the same in the process of learning, the input data and the output data may be classified by type. That is, the information P01, P02, ... P15, 16, ... That constitute the input data are classified according to the criteria classified in advance on the system side or the user side according to the content of the information, and the classified inputs. A data set may be created between the data and the output data and trained.

Further, according to the present invention, there is a feature that an optimum solution search is performed through the degree of association set in three or more stages. The degree of association can be described by, for example, a numerical value from 0 to 100% in addition to the above-mentioned 10 steps, but the degree of association is not limited to this, and any step can be described as long as it can be described by a numerical value of 3 or more steps. It may be configured.

Candidates for the possibility of search solutions by determining the possibility of fraudulent transactions with higher credibility and lower misidentification based on the degree of association expressed by such numerical values of three or more levels. It is also possible to search and display in descending order of the degree of association under the situation where there are a plurality of possible situations.

In addition to this, according to the present invention, it is possible to judge without overlooking the discrimination result of the output having an extremely low degree of association such as 1%. It warns the user that even a judgment result with an extremely low degree of association is connected as a slight sign and may be useful as the judgment result once every tens or hundreds of times. be able to.

Further, according to the present invention, there is an advantage that the search policy can be determined by the method of setting the threshold value by performing the search based on the degree of association of three or more stages. If the threshold value is lowered, even if the above-mentioned degree of association is 1%, it can be picked up without omission, but it is unlikely that a more appropriate discrimination result can be detected favorably, and a lot of noise may be picked up. be. On the other hand, if the threshold value is raised, there is a high possibility that the optimum search solution can be detected with a high probability. Sometimes the solution is overlooked. It is possible to decide which one to prioritize based on the ideas of the user side and the system side, but it is possible to increase the degree of freedom in selecting the points to be emphasized.

Further, in the present invention, the above-mentioned degree of association may be updated. This update may reflect information provided via a public communication network such as the Internet. In addition to market information, when event information, external environmental information, household information, real estate information, expert opinion information, and natural environment information knowledge, information, and data are acquired, the degree of association is increased or decreased accordingly. Let me.

In other words, this update corresponds to learning in the sense of artificial intelligence. It can be said that it is a learning act because it acquires new data and reflects it in the learned data.

In addition, this update of the degree of association is not based on the information that can be obtained from the public communication network, but is also updated by the system side or the user side based on the contents of research data, treatises, conference presentations, newspaper articles, books, etc. by experts. It may be updated artificially or automatically. Artificial intelligence may be utilized in these update processes.

Further, the process of first creating the trained model and the above-mentioned update may use not only supervised learning but also unsupervised learning, deep learning, reinforcement learning, and the like. In the case of unsupervised learning, instead of reading and learning the data set of input data and output data, information corresponding to the input data is read and trained, and the degree of association related to the output data is self-formed from there. You may let it.

1 Fraudulent financial transaction detection system 2 Search device 21 Internal bus 23 Display unit 24 Control unit 25 Operation unit 26 Communication unit 27 Estimate unit 28 Storage unit 61 node

Claims (12)

In a fraudulent financial transaction detection program that detects fraudulent financial transactions
With the information acquisition step of extracting the transaction history in the bank account of the newly detected person who newly detects fraudulent transactions over two or more terms and acquiring the difference pattern information including the pattern of the difference in the transaction history between the extracted terms. ,
Transaction history in the bank account of the past detection target person acquired in advance is extracted over two or more terms, reference difference pattern information including the pattern of the transaction history difference between the extracted terms, and the past detection target person. Based on the difference pattern information acquired in the above information acquisition step, the determination step for determining the possibility of fraudulent transaction of the new detection target person is performed by referring to the degree of association with the possibility of fraudulent transaction. A fraudulent financial transaction detection program characterized by having a bank execute. In the above information acquisition step, the transaction history of any one or more of the remittance amount, the remittance country, and the remittance currency in the bank account of the new detection target person is extracted.
In the determination step, among the transaction histories of any one or more of the remittance amount, the remittance country, and the remittance currency in the bank account of the past detection target person, the reference including the same type of transaction history in forming the difference pattern information. The fraudulent financial transaction detection program according to claim 1, wherein the above-mentioned degree of association is formed with the difference pattern information for use. In the above information acquisition step, geographical information regarding the geographical situation of the area to which the new detection target belongs is acquired.
In the determination step, the combination having the reference difference pattern information and the reference geographical information regarding the geographical situation of the area to which the past detection target belongs, and the possibility of fraudulent transactions of the past detection target person. A claim characterized by determining the possibility of fraudulent transactions of the new detection target person based on the difference pattern information and the geographical information acquired in the above information acquisition step by referring to the degree of association of three or more levels with. The fraudulent financial transaction detection program described in item 1 or 2. In the above information acquisition step, the account opening information regarding the account opening status of the new detection target person is acquired, and the information is acquired.
In the determination step, the combination having the reference difference pattern information and the reference account opening information regarding the account opening status of the past detection target person and the possibility of fraudulent transactions of the past detection target person are 3 Claim 1 or claim 1, characterized in that the possibility of fraudulent transactions of the new detection target person is determined based on the difference pattern information acquired in the information acquisition step and the account opening information with reference to the degree of association of the stages or higher. Fraudulent financial transaction detection program described in item 2. In the above information acquisition step, market information regarding market conditions in each of the above terms is acquired.
In the determination step, the degree of association between the combination of the reference difference pattern information and the reference market information regarding the market conditions in each of the terms and the possibility of fraudulent transactions of the past detection target person is determined. The fraudulent financial transaction according to claim 1 or 2, characterized in that the possibility of fraudulent transaction of the new detection target person is determined based on the difference pattern information acquired in the above information acquisition step and the market condition information. Detection program. In the above information acquisition step, the external environment information regarding the external environment of the industry of the new detection target person is acquired.
In the determination step, the combination of the reference difference pattern information and the reference external environment information regarding the external environment of the industry of the past detection target person and the possibility of fraudulent transactions of the past detection target person are determined. Claim 1 is characterized in that the possibility of fraudulent transactions of the newly detected target person is determined based on the difference pattern information acquired in the above information acquisition step and the external environment information by referring to the degree of association of three or more stages. Or the fraudulent financial transaction detection program described in item 2. In the above information acquisition step, the claim information regarding the claims held by the new detection target person is acquired.
In the determination step, the combination of the reference difference pattern information and the reference claim information regarding the claims held by the past detection target person and the possibility of fraudulent transactions of the past detection target person are determined. Claim 1 or claim 1, characterized in that the possibility of fraudulent transaction of the newly detected target person is determined based on the difference pattern information and the claim information acquired in the above information acquisition step with reference to the degree of association of three or more stages. Fraudulent financial transaction detection program described in item 2. In the above information acquisition step, the trading behavior information regarding the trading behavior of the new detection target person is acquired, and the information is acquired.
In the determination step, there are three stages: a combination having the reference difference pattern information and the reference trading behavior information regarding the trading behavior of the past detection target person, and the possibility of fraudulent transactions of the past detection target person. Claim 1 or 2 characterized in that the possibility of fraudulent transaction of the new detection target person is determined based on the difference pattern information and the trading behavior information acquired in the above information acquisition step with reference to the above degree of association. Fraudulent financial transaction detection program described in the section. In the above information acquisition step, the financial information regarding the finance of the new detection target person is acquired, and the information is acquired.
In the determination step, there are three or more stages of the combination having the reference difference pattern information and the reference financial information regarding the finance of the past detection target person, and the possibility of fraudulent transactions of the past detection target person. The invention according to claim 1 or 2, wherein the possibility of fraudulent transaction of the new detection target person is determined based on the difference pattern information and the financial information acquired in the information acquisition step with reference to the degree of association. Fraudulent financial transaction detection program. In the above information acquisition step, the annuity insurance information regarding the annuity and insurance paid by the newly detected person is acquired.
In the determination step, the combination of the reference difference pattern information and the reference annuity insurance information regarding the annuity and insurance paid by the past detection target person and the possibility of fraudulent transactions of the past detection target person are determined. Claim 1 characterized in that the possibility of fraudulent transactions of the newly detected person is determined based on the difference pattern information acquired in the above information acquisition step and the annuity insurance information by referring to the degree of association of three or more stages. Or the fraudulent financial transaction detection program described in item 2. In the above information acquisition step, attribute information related to the attributes of the new detection target person is acquired.
In the determination step, there are three or more stages of the combination having the reference difference pattern information and the reference attribute information regarding the attributes of the past detection target person, and the possibility of fraudulent transactions of the past detection target person. The invention according to claim 1 or 2, wherein the possibility of fraudulent transaction of the new detection target person is determined based on the difference pattern information and the attribute information acquired in the information acquisition step with reference to the degree of association. Fraudulent financial transaction detection program. The fraudulent financial transaction detection program according to any one of claims 1 to 11, wherein in the determination step, the degree of association corresponding to the weighting coefficient of each output of the node of the neural network in artificial intelligence is used. ..

Images