JP2023129229A - Data processing device, data processing method, and program - Google Patents

Abstract

To determine whether or not voucher data is authentic.SOLUTION: A data processing device 1 comprises: a storage section 22 that stores, in association with a plurality of issue sources of vouchers, a plurality of pieces of determination data indicating at least any feature of a predetermined character or image included in authorized voucher data; an acquisition section 231 that acquires voucher data; an identification section 232 that identifies an issue source of the voucher data on the basis of at least any of a character or image included in the acquired voucher data acquired by the acquisition section 231; a determination section 233 that determines whether or not the acquired voucher data is authentic data by comparing at least any feature of the predetermined character or image included in the determination data stored in the storage section 22 in association with the issue source identified by the identification section 232 with at least any feature of the predetermined character or image included in the acquired voucher data; and an output section 234 that outputs a determination result of the determination section 233.SELECTED DRAWING: Figure 2

Description

The present invention relates to a data processing device, a data processing method, and a program for processing billing data.

BACKGROUND ART Conventionally, the authenticity of the issuer of document data such as voucher data has been determined (for example, see Patent Document 1). Patent Document 1 describes that the authenticity of form data is verified by verifying an electronic signature embedded in the form data.

Japanese Patent Application Publication No. 2007-206964

The technique described in Patent Document 1 has a problem in that if an electronic signature is not embedded in the voucher data, it cannot be determined whether the voucher data is authentic.

Therefore, the present invention has been made in view of these points, and an object thereof is to provide an information output device, an information output method, and a program that can determine whether or not evidence data is genuine. do.

A data processing device according to a first aspect of the present invention stores a plurality of pieces of determination data indicating characteristics of at least one of predetermined characters or images included in regular voucher data in association with a plurality of issuers of vouchers. an acquisition unit that acquires the voucher data; and an identification unit that identifies the issuer of the voucher data based on at least one of characters or images included in the acquired voucher data acquired by the acquisition unit; the characteristics of at least one of the predetermined characters or images included in the determination data stored in the storage unit in association with the issuer identified by the identification unit; and the predetermined characteristics included in the acquired voucher data. a determination unit that determines whether the acquired evidence data is genuine data by comparing the characteristics of at least one of characters or images; and an output unit that outputs the determination result of the determination unit. Be prepared.

The determination unit is configured to determine the position of the predetermined character or image included in the determination data stored in the storage unit in association with the issuer identified by the identification unit, and the predetermined character or image included in the acquired voucher data. It may be determined whether or not the acquired evidence data is genuine data by comparing the position of the character or image. The acquisition unit acquires an electronic invoice including a markup language for specifying a data structure as the evidence data, and the determination unit stores it in the storage unit in association with the issuer specified by the identification unit. By comparing the predetermined characters included in the determined determination data and the predetermined characters included in the acquired voucher data, it is determined whether the acquired voucher data is genuine data. Good too.

The determination unit is configured to distinguish between the predetermined characters included in the determination data stored in the storage unit in association with the issuer identified by the identification unit and the predetermined characters included in the acquired voucher data. The degree of coincidence may be specified, and based on the specified degree of coincidence, it may be determined whether the acquired evidence data is genuine data. The determination unit uses the voucher data and information indicating the issuer of the voucher data as input data, and outputs information indicating whether the voucher data is authentic issued by the publisher. The acquired voucher data acquired by the acquisition unit and the information indicating the issuer identified by the identification unit are input into the trained machine learning model, and the voucher data output by the machine learning model It may be determined whether the acquired voucher data is genuine data based on information indicating whether the acquired voucher data is genuine or not issued by the source.

The data processing device generates learning data including the voucher data associated with a label indicating that the voucher data was acquired from a genuine publisher and the voucher data associated with a label indicating that the voucher data was acquired from a fake publisher. The image forming apparatus may further include a generation unit that generates the machine learning model by performing machine learning. The data processing device further includes a reception unit that receives a user's input as to whether or not the determination result is correct, and the generation unit receives a user input indicating that the determination result that the acquired voucher data is genuine is correct. When input is accepted, the machine learning model performs machine learning using new learning data obtained by associating the corresponding evidence data with a label indicating that it is genuine and adding the corresponding evidence data to the learning data. may be generated.

A data processing device according to a second aspect of the present invention stores a plurality of pieces of determination data indicating characteristics of at least one of predetermined characters or images included in regular voucher data in association with a plurality of issuers of vouchers. an acquisition unit that acquires evidence data; a characteristic of at least one of the predetermined characters or images included in each of the plurality of determination data stored in the storage unit; an identifying unit that identifies the determination data that shows the most similar feature to the acquired voucher data by comparing the characteristics of at least one of the predetermined characters or images included in the acquired voucher data; a determining unit that determines whether a publisher identified based on at least one of characters or images included in the evidence data matches the publisher corresponding to the determination data specified by the identifying unit; , and an output section that outputs the determination result of the determination section.

A data processing method according to a third aspect of the present invention includes the step of acquiring evidence data executed by a computer, and the step of acquiring evidence data based on at least one of characters or images included in the acquired acquired voucher data. the step of identifying the issuer, and referring to a storage unit that stores a plurality of determination data indicating characteristics of at least one of a predetermined character or an image included in the regular voucher data in association with a plurality of issuers of the voucher; and the characteristics of at least one of the predetermined characters or images included in the determination data stored in the storage unit in association with the specified issuer, and the predetermined characters or images included in the acquired voucher data. The method includes a step of determining whether the acquired evidence data is genuine data by comparing at least one feature of the image, and a step of outputting the determination result.

A data processing method according to a fourth aspect of the present invention includes the step of acquiring voucher data executed by a computer; With reference to a storage unit that stores a plurality of pieces of determination data indicating any one of the characteristics, at least one of the characteristics of the predetermined character or image included in each of the plurality of pieces of determination data stored in the storage unit and identifying the determination data that is most similar to the acquired voucher data by comparing the characteristics of at least one of the predetermined characters or images included in the acquired acquired voucher data; determining whether a publisher identified based on at least one of characters or images included in the data matches the publisher corresponding to the identified determination data; and outputting a determination result. and a step.

A program according to a fifth aspect of the present invention includes the steps of: acquiring evidence data; and specifying the issuer of the voucher data based on at least one of characters or images included in the acquired acquired voucher data. and a step of identifying the information by referring to a storage unit that stores a plurality of determination data indicating at least one of characteristics of predetermined characters or images included in the regular voucher data in association with a plurality of issuers of the vouchers. characteristics of at least one of the predetermined characters or images included in the determination data stored in the storage unit in association with the issuer; and at least one of the predetermined characters or images included in the acquired voucher data. A step of determining whether or not the acquired evidence data is genuine data by comparing the characteristics with the above characteristics, and a step of outputting the determination result are executed.

A program according to a sixth aspect of the present invention includes the steps of: acquiring voucher data; and determining characteristics of at least one of predetermined characters or images included in regular voucher data in association with a plurality of issuers of vouchers; The characteristics of at least one of the predetermined characters or images included in each of the plurality of determination data stored in the storage section and the acquired identifying the determination data that is most similar to the determination data by comparing the characteristics of at least one of the predetermined characters or images included in the acquired voucher data; a step of determining whether a publisher specified based on at least one of characters or images matches the publisher corresponding to the identified determination data; and a step of outputting a determination result. Let it run.

According to the present invention, it is possible to determine whether or not evidence data is genuine.

FIG. 1 is a diagram for explaining an overview of a data processing system according to a first embodiment. The configuration of a data processing device is shown. An example of a determination data table is shown. An example of feature extraction of acquired voucher data is shown. Another example of the determination data is shown. Another example of feature extraction of evidence data is shown. 3 is a flowchart illustrating a processing procedure for determining the authenticity of the issuer of evidence data by the data processing device. It is a flowchart which shows the processing procedure of the authenticity determination of evidence data by the data processing apparatus 2 of a modification. child 7 shows the configuration of a data processing device according to a second embodiment. 3 is a flowchart illustrating a processing procedure for determining the authenticity of the issuer of evidence data by the data processing device.

[Overview of data processing system S]
FIG. 1 is a diagram for explaining an overview of a data processing system S according to the first embodiment. The data processing system S includes an external device 1, a data processing device 2, and a person in charge terminal 3. The data processing system S may include a plurality of person-in-charge terminals 3.

The external device 1 is, for example, a computer installed in a facility of a business partner such as a business partner, a computer of an employee of the business who uses the data processing device 2, or an access point of a network through which electronic invoices are sent and received. . External device 1 communicates with data processing device 2 via a network. The external device 1 transmits evidence data such as bill data to the data processing device 2. The evidence data is, for example, bill data and receipt data. The evidence data may be an electronic invoice or electronic data such as PDF (Portable Document Format) data. The evidence data may be obtained by reading a paper bill using OCR (Optical Character Recognition). The person in charge terminal 3 may function as the external device 1.

The person in charge terminal 3 is, for example, a computer used by a person in charge of an accounting department who uses the data processing device 2. The person in charge terminal 3 communicates with the data processing device 2 via the network. Note that the person in charge terminal 3 may transmit and receive data to and from the external device 1 via the network.

The data processing device 2 communicates with the external device 1 and the person in charge terminal 3 via the network N. The data processing device 2 is, for example, a server connected to a local network of a business operator, but may also be a cloud server. The data processing device 2 acquires evidence data (hereinafter also referred to as "obtained evidence data") from the external device 1 ((1) in FIG. 1). If the person in charge terminal 3 receives the evidence data from the external device 1, the data processing device 2 may acquire this evidence data from the person in charge terminal 3.

The data processing device 2 identifies the issuer of the acquired voucher data based on characters, images, etc. (hereinafter referred to as "characters, etc.") indicating the company included in the acquired acquired voucher data ((2 in FIG. 1). )). The data processing device 2 stores a determination data table including a plurality of determination data indicating characteristics such as predetermined characters included in the regular voucher data in association with a plurality of issuers of the vouchers. The data processing device 2 refers to the stored determination data table and identifies determination data stored in association with the identified issuer.

The data processing device 2 compares the characteristics of predetermined characters included in the identified determination data with the corresponding features of predetermined characters included in the acquired voucher data to determine whether the issuer of the acquired voucher data is genuine. ((3) in FIG. 1). The data processing device 2 outputs the determination result of whether the obtained voucher data is genuine to the person in charge terminal 3 ((4) in FIG. 1).

In this way, the data processing device 2 outputs the determination result of whether or not the request data is genuine based on the characteristics of at least one of the characters and images included in the acquired voucher data, so that the issuer can be identified. It is possible to prevent a user from falling victim to fraud when false evidence data is received.

[Configuration of data processing device 2]
FIG. 2 shows the configuration of the data processing device 2. As shown in FIG. The data processing device 2 includes a communication section 21, a storage section 22, and a control section 23. The control unit 23 includes an acquisition unit 231, a identification unit 232, a determination unit 233, and an output unit 234.

The communication unit 21 is an interface for communicating with the person in charge terminal 3. The storage unit 22 includes storage media such as a ROM (Read Only Memory), a RAM (Random Access Memory), and a hard disk. The storage unit 22 stores programs executed by the control unit 23. The storage unit 22 stores issuer information that associates the issuer of the voucher data with an image such as a logo indicating the issuer of the voucher data.

The storage unit 22 includes a determination data table in which information indicating a plurality of issuers of vouchers is associated with a plurality of determination data indicating characteristics of at least one of predetermined characters or images included in regular voucher data. is memorized. The regular voucher data is voucher data that is known in advance to have been issued by a regular publisher. The determination data indicates, for example, the position of a character string indicating the issuer as a feature of a predetermined character or image included in the regular evidence data. The position of the character string indicating the publisher includes, for example, the distance from the left edge of the paper to the position of the character string indicating the publisher, and the distance from the top edge of the paper to the position of the character string indicating the publisher, and is expressed in millimeters. expressed in units.

FIG. 3 shows an example of a determination data table. FIG. 4 shows an example of feature extraction of acquired voucher data. In the judgment data table shown in FIG. 3, the issuer of the regular voucher data, the first judgment data indicating the distance from the left edge of the page of the position of the issuer name included in the regular voucher data, and the issuer of the regular voucher data. It is associated with second determination data indicating the distance from the top edge of the page of the publisher's name.

The distance from the left edge of the page of the publisher's name is, for example, D1 shown in FIG. 4. In the example shown in Figure 4, the left end of the character string indicating the publisher name is the position of the publisher name, but the position of the publisher name may be the right end of the character string indicating the publisher name, and the character string It may be in the center of Similarly, the distance of the publisher's name from the top of the page is, for example, D2 shown in FIG. 4. In the example shown in Figure 4, the upper end of the character string indicating the publisher name is the position of the publisher name, but the position of the publisher name may be the lower end of the character string indicating the publisher name. It may be in the center of

In the example in the first row from the top of Figure 3, the distance from the left edge of the page of the character string indicating the publisher's name in the official document data issued by the publisher "AB Shoji" is 201 mm, indicating the publisher's name. The distance from the top of the paper to the position of the character string is 60 mm. In the example in the second row from the top of FIG. 3, the distance from the left edge of the page of the position of the character string indicating the publisher's name in the official document data issued by the publisher "CD Manufacturing" is 170 mm, and the distance from the left edge of the paper is 170 mm. The distance from the top of the page to the position of the character string indicating the name is 30 mm.

FIG. 4 shows an example of extracting characteristics of voucher data. In FIG. 4, the company name "AB Shoji", which is the publisher's name, is indicated by a dashed frame line. In the example of FIG. 4, the distance D1 from the left edge of the page of the character string "AB Shoji" indicating the company name of the issuer included in the acquired voucher data is extracted as a feature by the determining unit 233, which will be described later. The distance D2 from the top edge of the page of the character string "AB Shoji" indicating the issuer included in the acquired voucher data is extracted as a feature by the determining unit 233.

FIG. 5 shows another example of the determination data. FIG. 6 shows another example of feature extraction of evidence data. In the determination data shown in FIG. 5, the issuer of the regular evidence data, the first determination data indicating the distance from the left edge of the page of the position of the logo image indicating the issuer included in the regular evidence data, and the issuer of the regular evidence data. It is associated with second determination data indicating the distance from the top edge of the page of the position of the logo image indicating the included publisher.

The distance of the position of the logo image indicating the publisher from the left edge of the page is, for example, D1' shown in FIG. 6. In the example shown in FIG. 6, the left end of the logo image showing the publisher is the position of the logo image, but the position of the logo image showing the publisher may be the right end of the logo image showing the publisher. It may be in the center. Similarly, the distance of the position of the logo image from the top of the page is, for example, D2' shown in FIG. In the example shown in FIG. 6, the top edge of the logo image indicating the publisher is the position of the logo image, but the position of the logo image may be the bottom edge of the logo image indicating the publisher, or the center of the logo image. It's okay.

In the example in the first row from the top of Figure 5, the distance from the left edge of the page of the position of the logo image indicating the publisher in the official document data issued by the publisher "AB Shoji" is 20 mm, indicating the publisher. The distance of the logo image from the top of the page is 16 mm. In the example in the second row from the top of FIG. 5, the distance from the left edge of the page of the position of the logo image indicating the issue name in the official document data issued by the publisher "CD Manufacturing" is 17 mm, which indicates that the issuer is The distance of the logo image shown from the top of the page is 30 mm.

FIG. 6 shows another example of extracting features of acquired voucher data. As shown in the example of FIG. 6, the distance D1' from the left edge of the paper of the position of the logo image in the shape of "AB" of the publisher "AB Shoji" included in the acquired voucher data is determined as a characteristic by the determination unit 233, which will be described later. Extracted. The distance D2' from the top of the page of the position of the logo image shaped like "AB" of the publisher "AB Shoji" included in the acquired voucher data is extracted as a feature by the determination unit 233.

The control unit 23 in FIG. 2 is, for example, a CPU (Central Processing Unit). The control unit 23 functions as an acquisition unit 231, a identification unit 232, a determination unit 233, and an output unit 234 by executing a program stored in the storage unit 22.

The acquisition unit 231 communicates with the external device 1 or the person in charge terminal 3 via the communication unit 21 . The acquisition unit 231 acquires evidence data via the external device 1. The acquisition unit 231 may acquire an electronic invoice including a markup language for specifying a data structure as evidence data. For example, the acquisition unit 231 may acquire an electronic invoice written in XML format as evidence data. The acquisition unit 231 may acquire the evidence data from the person in charge terminal 3 when the person in charge terminal 3 receives the evidence data from the external device 1 . The acquisition unit 231 outputs the acquired evidence data to the identification unit 232.

[Identification of publisher]
The identification unit 232 identifies the issuer of the evidence data based on at least one of the characters and images included in the acquired evidence data acquired by the acquisition unit 231. For example, the specifying unit 232 specifies, as the issuer of the voucher data, a company name that does not have a keyword such as "Ochu" indicating the destination among the company names included in the acquired voucher data. When the evidence data is an electronic invoice, the identification unit 232 identifies a character string with a tag indicating the issuer as the issuer name.

The identifying unit 232 may identify the issuer of the voucher data based on a logo image indicating the issuer of the voucher data. First, the specifying unit 232 reads from the storage unit 22 issuer information that associates the issuer of the voucher data with a logo image indicating the issuer of the voucher data. The identifying unit 232 refers to the read publisher information and identifies the publisher stored in association with the logo image whose similarity to the logo image included in the acquired voucher data is equal to or higher than the threshold value. Good too. The threshold value is, for example, a value determined as a value necessary to specify the issuer with the same accuracy as when specifying the issuer of the voucher data based on the company name included in the acquired voucher data. The identifying unit 232 outputs information indicating the identified issuer to the determining unit 233.

[Verification of authenticity of issuer of acquired voucher data]
The determination unit 233 determines whether the acquired evidence data is authentic. First, the determination unit 233 extracts the characteristics of at least one of a predetermined character or image included in the acquired voucher data. For example, the determination unit 233 extracts the position, size, font, etc. of a predetermined character or image included in the acquired voucher data as a feature.

The determination unit 233 refers to the determination data table stored in the storage unit 22 and identifies the determination data stored in association with the publisher identified by the identification unit 232. This determination data includes characteristics such as the position, size, or font of a predetermined character or image extracted from the regular evidence data. The determination unit 233 determines whether the acquired voucher data is genuine by comparing features such as the position, size, or font of a predetermined character or image extracted from the acquired voucher data with the corresponding features included in the specified determination data. Determine whether the data is valid.

In the example of FIG. 4, as described above, the determination unit 233 extracts the distance D1 from the left edge of the page of the position of the character string "AB Shoji" indicating the company name of the issuer included in the acquired voucher data as a feature. The determination unit 233 extracts as a feature the distance D2 from the top of the page of the position of the character string "AB Shoji" indicating the issuer included in the acquired voucher data.

The determination unit 233 refers to the determination data table (see FIG. 3) stored in the storage unit 22 and determines the first determination data table stored in association with the publisher “AB Shoji” identified by the identification unit 232. It is specified that the distance from the left edge of the page of the position of the issuer's name included in the regular evidence data indicated by the data is 201 mm.

The determination unit 233 refers to the determination data table stored in the storage unit 22 and determines whether the genuine evidence is indicated by the second determination data stored in association with the issuer “AB Shoji” identified by the identification unit 232. It is specified that the distance from the top edge of the page of the publisher name included in the data is 160 mm. The determination unit 233 determines whether the difference between the extracted distance D1 and the distance (201 mm) indicated by the specified first determination data is less than or equal to a threshold value. The determination unit 233 determines whether the difference between the extracted distance D2 and the distance (160 millimeters) indicated by the specified second determination data is less than or equal to a threshold value.

The determination unit 233 determines that the difference between the extracted distance D1 and the distance indicated by the identified first determination data is less than or equal to the threshold value, and that the difference between the extracted distance D2 and the distance indicated by the identified second determination data is less than or equal to the threshold value. If the difference is less than or equal to the threshold, it is determined that the acquired voucher data is genuine voucher data issued by the issuer "AB Shoji". On the other hand, if the difference between the extracted distance D1 and the distance indicated by the identified first determination data exceeds the threshold, or the difference between the extracted distance D2 and the distance indicated by the identified second determination data If the difference exceeds a threshold value, it is determined that the acquired voucher data is not genuine voucher data issued by the issuer "AB Shoji".

As shown in the example of FIG. 6, the determination unit 233 determines the distance D1' from the left edge of the paper of the position of the logo image shaped like "AB" of the publisher "AB Shoji" included in the acquired voucher data, as described above. It may be extracted as a feature. The determination unit 233 may extract as a feature the distance D2' from the top of the page of the position of the logo image shaped like "AB" of the publisher "AB Shoji" included in the acquired voucher data. Regarding the method of determining the authenticity of the publisher of the acquired voucher data using the distance D1' and the distance D2' from the left edge and top edge of the page of the position of the logo image extracted by the determination unit 233, This method is the same as the method of determining the authenticity of the issuer of the acquired voucher data using the distance D1 and distance D2 from the left end and top end of the page, so the explanation will be omitted.

Further, the determination unit 233 is not limited to the example in which the authenticity of the issuer is determined based on the position of the character string indicating the issuer included in the acquired voucher data. The determination unit 233 may determine the authenticity of the issuer based on the position of a character string that corresponds one-to-one with the issuer, such as a character string indicating the name of a payee account.

[Verification of authenticity of electronic invoice issuer]
If the evidence data acquired by the acquisition unit 231 is an electronic invoice, the determination unit 233 determines whether the document data acquired by the acquisition unit 231 is an electronic invoice or not, the determination unit 233 determines whether or not the document data acquired by the acquisition unit 231 is an electronic invoice. By comparing the obtained document data with predetermined characters included in the obtained document data, it is determined whether the obtained document data is genuine data or not.

First, the determination unit 233 extracts an XML text indicating tag information or layout information corresponding to a character string indicating an issuer or a logo image in the acquired voucher data. The determination unit 233 may extract XML text indicating tag information or layout information corresponding to a character string indicating a transfer destination account in the acquired voucher data.

The determination unit 233 refers to the determination data table stored in the storage unit 22 and identifies the determination data stored in association with the publisher identified by the identification unit 232. This determination data includes a character string indicating the publisher, or an XML text indicating tag information or layout information corresponding to the logo image, as a predetermined character feature. The determination unit 233 compares the XML text indicating the tag information or layout information included in the identified determination data with the extracted XML text indicating the tag information or layout information, thereby determining whether the issuer of the acquired evidence data is genuine. Determine whether or not.

More specifically, the determination unit 233 determines that the issuer of the acquired evidence data is genuine when the XML text indicating tag information etc. included in the identified determination data matches the extracted XML text indicating tag information etc. It is determined that The determination unit 233 determines that the issuer of the acquired evidence data is not genuine when the XML text indicating tag information etc. included in the identified determination data does not match the extracted XML text indicating the corresponding tag information etc. It is determined that

The determination unit 233 identifies the degree of matching between predetermined characters included in the determination data stored in the storage unit 22 in association with the issuer identified by the identification unit 232 and predetermined characters included in the acquired voucher data. . The determination unit 233 may determine whether the obtained evidence data is genuine evidence data based on the specified degree of matching.

For example, when the determination unit 233 extracts as predetermined characters a plurality of tag information or a character string indicating layout information corresponding to a character string indicating the issuer in the acquired voucher data, the determination unit 233 Among the character strings indicating tag information or layout information, the proportion of character strings that match any one of the plurality of character strings indicating tag information or layout information included in the specified determination data is specified as a degree of matching. The determination unit 233 may determine that the acquired evidence data is genuine evidence data when the identified degree of matching is equal to or greater than a threshold value. The determination unit 233 may determine that the acquired evidence data is not genuine evidence data when the specified degree of matching is less than a threshold value. The threshold value is specified by the user, for example.

[Output of judgment results]
The output unit 234 communicates with the person in charge terminal 3 via the communication unit 21 . The output unit 234 outputs the determination result of the determination unit 233 as to whether the issuer of the acquired voucher data is authentic. For example, the output unit 234 causes the issuer identified by the identifying unit 232 and the determination result of the determining unit 233 to be displayed on the display of the person in charge terminal 3.

[Processing procedure for determining authenticity of evidence data by data processing device 2]
FIG. 7 is a flowchart showing a processing procedure for determining the authenticity of the issuer of evidence data by the data processing device 2. This processing procedure starts, for example, when the acquisition unit 231 acquires evidence data from the external device 1. First, the acquisition unit 231 determines whether the acquired evidence data is an electronic invoice (S101). When the acquisition unit 231 determines that the acquired voucher data is not an electronic invoice (NO in S101), the identification unit 232 identifies the acquired voucher data based on the characters included in the acquired voucher data acquired by the acquisition unit 231. The publisher is identified (S102).

The determination unit 233 extracts the position of a predetermined character included in the acquired voucher data as a feature. The determination unit 233 refers to the determination data table stored in the storage unit 22 and identifies the determination data stored in association with the publisher identified by the identification unit 232. The determination unit 233 determines whether the deviation between the position of a predetermined character extracted from the acquired voucher data and the position of a predetermined character included in the specified determination data is less than or equal to a threshold value (S103).

If the difference between the position of a predetermined character extracted from the acquired voucher data and the position of a predetermined character included in the specified determination data is equal to or less than a threshold value (YES in S103), the determination unit 233 determines whether the acquired voucher data It is determined that the publisher is genuine (S104). The output unit 234 outputs the determination result of the determination unit 233 to the person in charge terminal 3 (S105), and ends the process.

When the acquisition unit 231 determines that the acquired voucher data is an electronic invoice in the determination at S101 (YES at S101), the acquisition unit 231 acquires the XML text included in the acquired voucher data (S106). The identification unit 232 identifies the issuer of the acquired evidence data based on the acquired XML text (S107). The determination unit 233 extracts, as a predetermined character feature, an XML text indicating tag information or layout information corresponding to a character string indicating an issuer in the acquired voucher data.

The determination unit 233 refers to the determination data table stored in the storage unit 22 and identifies the determination data stored in association with the publisher identified by the identification unit 232. The determining unit 233 determines whether a character string indicating tag information or layout information included in the specified determination data matches a character string indicating extracted tag information or layout information (S108). If the character string indicating the tag information or layout information included in the identified determination data does not match the extracted character string indicating the tag information or layout information (NO in S108), the determination unit 233 determines whether the acquired evidence data It is determined that the publisher is not genuine (S109), and the process proceeds to S105.

If the difference between the position of the predetermined character extracted from the acquired evidence data and the position of the predetermined character included in the specified determination data is larger than the threshold value in the determination of S103 (NO in S103), the determination unit 233 determines that It is determined that the issuer of the acquired voucher data is not genuine (S109), and the process proceeds to S105. In the determination of S108, the determination unit 233 determines that the issuer of the acquired voucher data is genuine if the tag information, etc. included in the identified determination data matches the extracted tag information, etc. (YES in S108). It is determined that (S110), and the process proceeds to S105.

[Effects of the data processing device 2 of the first embodiment]
In the data processing device 2 of the first embodiment, the output unit 234 determines whether the issuer of the billing data is authentic based on the characteristics of at least one of characters and images included in the acquired voucher data. Output the results. For this reason, the output unit 234 can prevent the business entity that received the voucher data from falling victim to fraud when the acquisition unit 231 acquires voucher data with a false origin.

<Modified example>
In the first embodiment, an example has been described in which, after the identification unit 232 identifies the issuer of the evidence data, the determination unit 233 determines whether or not the identified issuer of the evidence data is authentic. The present invention is not limited to this. For example, the identifying unit 232 may identify regular voucher data having the most similar features to the features extracted from the acquired voucher data, and identify the issuer of the identified regular voucher data as the issuer of the acquired voucher data.

Similar to the first embodiment, the identification unit 232 of this modification extracts the characteristics of at least one of a predetermined character or image included in the acquired evidence data acquired by the acquisition unit 231. The specifying unit 232 refers to the determination data table stored in the storage unit 22 and identifies at least one of the predetermined characters or images included in each of the plurality of determination data stored in the determination data table of the storage unit 22. By comparing any of the features with the features extracted from the acquired voucher data, determination data indicating the most similar feature to the acquired voucher data is identified.

The identifying unit 232 refers to the determination data table stored in the storage unit 22 and identifies the first issuer corresponding to the determination data that includes the characteristics most similar to the acquired voucher data. Similar to the first embodiment, the identifying unit 232 identifies the second issuer of the acquired voucher data based on at least one of a character string such as the publisher's name or an image included in the acquired voucher data.

The determining unit 233 determines whether the first issuer identified by the identifying unit 232 and the second issuer match. The determining unit 233 determines that the issuer of the acquired voucher data is authentic when the first issuer and the second issuer match. If the first issuer and the second issuer do not match, the determining unit 233 determines that the issuer of the acquired voucher data is not authentic.

FIG. 8 is a flowchart showing a processing procedure for determining the authenticity of evidence data by the data processing device 2 of this modification. This processing procedure starts, for example, when the acquisition unit 231 acquires the acquired evidence data from the external device 1. First, the acquisition unit 231 determines whether the acquired evidence data is an electronic invoice (S201). When the acquisition unit 231 determines that the acquired voucher data is not an electronic invoice (NO in S201), the identifying unit 232 determines the position of a predetermined character included in the acquired voucher data acquired by the acquisition unit 231 as a feature. Extract.

The specifying unit 232 compares the position of a predetermined character included in each of the plurality of pieces of determination data stored in the storage unit 22 with the position of a predetermined character extracted from the acquired voucher data, thereby identifying the acquired voucher data. Determination data indicating the position of the most similar predetermined character is specified (S202). The identifying unit 232 refers to the determination data table stored in the storage unit 22 and identifies the first issuer that is stored in association with the determination data that includes the characteristics most similar to the acquired voucher data.

Similar to the first embodiment, the identifying unit 232 identifies the second issuer of the acquired voucher data based on at least one of the characters and images included in the acquired voucher data. The determining unit 233 determines whether the second publisher specified by the identifying unit 232 matches the first publisher stored in association with the determination data (S203). If the second issuer identified by the identifying unit 232 matches the first issuer (YES in S203), the determining unit 233 determines that the issuer of the acquired voucher data is authentic (S204). The output unit 234 outputs the determination result of the determination unit 233 to the person in charge terminal 3 (S205), and ends the process.

When the acquisition unit 231 determines that the acquired voucher data is an electronic invoice in the determination of S201 (YES in S201), the acquisition unit 231 acquires the XML text included in the acquired voucher data (S206). The specifying unit 232 extracts, as a predetermined character feature, an XML text indicating a plurality of tag information or layout information corresponding to a character string indicating an issuer in the acquired voucher data. The specifying unit 232 determines which tag information, etc. has the highest degree of matching with the tag information or layout information specified by the specifying unit 232, among the plurality of determination data stored in the determination data table of the storage unit 22. (S207).

The identifying unit 232 refers to the determination data table stored in the storage unit 22 and identifies the first issuer stored in association with determination data indicating tag information etc. with the highest degree of matching. Similar to the first embodiment, the identifying unit 232 identifies the second issuer of the acquired voucher data based on the XML text acquired from the acquired voucher data, and proceeds to determination in S203.

If the second issuer identified by the identifying unit 232 does not match the first issuer in the determination of S203 (NO in S203), the determination unit 233 determines that the issuer of the acquired voucher data is not authentic (S208). ), the process proceeds to S205.

<Second embodiment>
In the second embodiment, an example will be described in which the authenticity of acquired voucher data is determined using a trained machine learning model. FIG. 9 shows the configuration of a data processing device 300 according to the second embodiment. The data processing device 300 of the second embodiment is the same as that in FIG. 4 except that the control unit 23 includes a determination unit 301, a generation unit 302, and a reception unit 303. Functional blocks similar to those in FIG. 4 are designated by the same reference numerals as in FIG. 4, and description thereof will be omitted.

The determination unit 301 is trained to use voucher data and information indicating the issuer of this voucher data as input data, and to output information indicating whether the voucher data is authentic issued by the issuer as output data. This machine learning model is used to determine whether the acquired evidence data is authentic. The evidence data may be an electronic invoice. First, the determination unit 301 reads this learned machine learning model from the storage unit 22. The determining unit 301 inputs the acquired voucher data acquired by the acquiring unit 231 and information indicating the issuer identified by the specifying unit 232 to the read machine learning model, and determines whether the acquired voucher data is genuine issued by this publisher. Obtain the information output by the machine learning model that indicates whether the The determination unit 301 determines whether the acquired evidence data is genuine data based on the information output by this machine learning model.

[Processing procedure during machine learning model learning]
Hereinafter, processing during learning of the machine learning model stored in the storage unit 22 will be described. The generation unit 302 uses machine learning that uses voucher data and information indicating the issuer of this voucher data as input data, and uses information indicating whether the voucher data is authentic issued by the issuer as output data. Generate the model. The generation unit 302 performs machine learning on learning data including evidence data associated with a label indicating that the information was obtained from a genuine publisher and evidence data associated with a label indicating that the information was obtained from a fake publisher. By doing so, a machine learning model is generated.

Evidence data associated with a label obtained from a fake publisher includes, for example, the publisher's name, the logo image representing the publisher, or the location of the remittance account name included in the voucher data, but the location of the label obtained from the genuine publisher is A plurality of pieces of evidence data are used that are deviated by a threshold value or more from the position of the issuer name, etc. of the piece of evidence data associated with the label indicating the fact. The threshold value is determined by an expert in the field, for example, depending on the accuracy required for determining the authenticity of the issuer of evidence data using a trained machine learning model.

Evidence data associated with a label obtained from a fake publisher includes evidence data associated with a label that indicates that the size or font of the publisher name included in this evidence data is obtained from a genuine publisher. A plurality of pieces of evidence data may be used that are different in size or font of the publisher's name. The evidence data may be an electronic invoice.

Further, in the data processing device of the second embodiment, after the determination unit 301 determines whether the acquired voucher data is genuine data, the determination unit 301 determines whether the issuer of the acquired voucher data is genuine. The computer may also include a reception unit 303 that accepts a user's input as to whether or not the determination result indicating whether or not the determination result is correct. When the reception unit 303 receives a user input indicating that the information indicating whether the issuer is authentic or not by the determination unit 301 is correct, the generation unit 302 generates a new information corresponding to the determination result by the determination unit 301. Create training data.

When the reception unit 303 receives a user input indicating that the determination result by the determination unit 301 that the voucher data is genuine is correct, the generation unit 302 determines that the target acquired voucher data determined by the determination unit 301 is genuine. The acquired evidence data is added to the learning data in association with a label indicating that there is a certain fact. When the reception unit 303 receives a user input indicating that the determination result by the determination unit 301 that the voucher data is not authentic is correct, the generation unit 302 generates a message indicating that the target acquired voucher data determined by the determination unit 301 is not genuine. The acquired evidence data is added to the learning data in association with the label indicating the .

When the generation unit 302 receives a user input indicating that the determination result of the determination unit 301 that the voucher data is genuine is incorrect, the generation unit 302 indicates that the target acquired voucher data determined by the determination unit 301 is not genuine. The acquired evidence data is added to the learning data in association with the label. When the reception unit 303 receives a user input indicating that the determination result by the determination unit 301 that the voucher data is not genuine is incorrect, the generation unit 302 determines whether the target acquired voucher data determined by the determination unit 301 is genuine. The acquired evidence data is added to the learning data in association with a label indicating that there is a certain fact.

The generation unit 302 may generate a machine learning model by performing machine learning on new learning data after adding the evidence data whose authenticity has been determined by the user in this way. In this way, the generation unit 302 can further improve the accuracy of determining the authenticity of the issuer of evidence data using the trained machine learning model. The generation unit 302 causes the storage unit 22 to store the generated trained machine learning model.

[Processing procedure for determining authenticity of evidence data by data processing device 300]
FIG. 10 is a flowchart showing a processing procedure for determining the authenticity of the issuer of evidence data by the data processing device 300. This processing procedure starts, for example, in a state where the data processing device 300 is communicating with the external device 1. First, the acquisition unit 231 acquires evidence data from the external device 1 (S301). The identifying unit 232 identifies the issuer of the acquired voucher data (S302). When the acquired voucher data is an electronic invoice, the identification unit 232 identifies the issuer of the acquired voucher data based on the XML text included in the acquired voucher data.

The determination unit 233 has been trained to use voucher data and information indicating the issuer of this voucher data as input data, and to output data as information indicating whether the voucher data is authentic issued by the issuer. The machine learning model is read from the storage unit 22. The determining unit 233 inputs the acquired voucher data acquired by the acquiring unit 231 and the information indicating the publisher specified by the specifying unit 232 to the read machine learning model (S303), and determines whether the acquired voucher data is issued by this publisher. Obtain information output by a machine learning model that indicates whether the item is genuine or not. The determination unit 301 determines whether the acquired evidence data is genuine data based on the information output by this machine learning model (S304). The output unit 234 outputs the determination result by the determination unit 233 as to whether the acquired voucher data is genuine data to the person in charge terminal 3 (S305), and ends the process.

[Effects of the data processing device 2 of the second embodiment]
In the data processing device 2 of the second embodiment, the output unit 234 improves the accuracy of determining the authenticity of the issuer of acquired voucher data using the trained machine learning model, so It is possible to prevent users from falling victim to fraud.

Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes can be made within the scope of the gist. be. All or part of the device can be configured by being functionally or physically distributed and integrated in arbitrary units. In addition, new embodiments created by arbitrary combinations of multiple embodiments are also included in the embodiments of the present invention. The effects of the new embodiment resulting from the combination have the effects of the original embodiment.

1 External device 2 Data processing device 3 Person in charge terminal 21 Communication section 22 Storage section 23 Control section 231 Acquisition section 232 Specification section 233 Judgment section 234 Output section 300 Data processing device 301 Judgment section 302 Generation section 303 Reception section

Claims (12)

a storage unit that stores a plurality of determination data indicating characteristics of at least one of predetermined characters or images included in the regular voucher data in association with a plurality of issuers of the voucher;
an acquisition unit that acquires evidence data;
an identification unit that identifies the issuer of the voucher data based on at least one of characters or images included in the acquired voucher data acquired by the acquisition unit;
the characteristics of at least one of the predetermined characters or images included in the determination data stored in the storage unit in association with the issuer identified by the identification unit; and the predetermined characteristics included in the acquired voucher data. a determination unit that determines whether the acquired evidence data is genuine data by comparing the characteristics of at least one of characters or images;
an output unit that outputs the determination result of the determination unit;
A data processing device comprising: The determination unit is configured to determine the position of the predetermined character or image included in the determination data stored in the storage unit in association with the issuer identified by the identification unit, and the predetermined character or image included in the acquired voucher data. determining whether the acquired evidence data is genuine data by comparing the position of the character or image of
The data processing device according to claim 1. The acquisition unit acquires an electronic invoice including a markup language for specifying a data structure as the voucher data,
The determination unit is configured to identify the predetermined characters included in the determination data stored in the storage unit in association with the issuer identified by the identification unit and the predetermined characters included in the acquired voucher data. determining whether the acquired evidence data is genuine data by comparing;
The data processing device according to claim 1 or 2. The determination unit is configured to distinguish between the predetermined characters included in the determination data stored in the storage unit in association with the issuer identified by the identification unit and the predetermined characters included in the acquired voucher data. identifying a degree of coincidence, and determining whether the acquired evidence data is genuine data based on the identified degree of coincidence;
The data processing device according to claim 3. The determination unit uses the voucher data and information indicating the issuer of the voucher data as input data, and outputs information indicating whether the voucher data is authentic issued by the publisher. The acquired voucher data acquired by the acquisition unit and the information indicating the issuer identified by the identification unit are input into the trained machine learning model, and the voucher data output by the machine learning model Determining whether the acquired voucher data is genuine data based on information indicating whether or not it is genuine issued by the source;
The data processing device according to claim 1. By performing machine learning on learning data including the evidence data associated with a label indicating that the information was obtained from a genuine publisher and the evidence data associated with a label indicating that the information was obtained from a fake publisher, further comprising a generation unit that generates the machine learning model;
The data processing device according to claim 5. further comprising a reception unit that receives a user's input as to whether or not the determination result is correct;
When the generation unit receives a user input indicating that the determination result that the acquired voucher data is genuine is correct, the generating unit generates the voucher data in association with a label indicating that the corresponding voucher data is genuine. Generate the machine learning model by performing machine learning using new learning data added to the learning data.
The data processing device according to claim 6. a storage unit that stores a plurality of determination data indicating characteristics of at least one of predetermined characters or images included in the regular voucher data in association with a plurality of issuers of the voucher;
an acquisition unit that acquires evidence data;
the characteristics of at least one of the predetermined characters or images included in each of the plurality of pieces of determination data stored in the storage unit; and the characteristics of at least one of the predetermined characters or images included in the acquired evidence data acquired by the acquisition unit. a specifying unit that specifies the determination data that indicates the most similar feature to the acquired voucher data by comparing the data with at least one of the features;
determination of whether or not an issuer identified based on at least one of characters or images included in the acquired voucher data matches the issuer corresponding to the determination data identified by the identification unit; Department and
an output unit that outputs the determination result of the determination unit;
A data processing device comprising: computer executes
a step of obtaining evidence data;
identifying the issuer of the acquired voucher data based on at least one of characters or images included in the acquired acquired voucher data;
The specified issuer is associated with a plurality of issuers of the evidence by referring to a storage unit that stores a plurality of determination data indicating characteristics of at least one of predetermined characters or images included in the regular evidence data. At least one feature of the predetermined character or image included in the determination data stored in the storage unit in association with at least one feature of the predetermined character or image included in the acquired voucher data. a step of determining whether the acquired voucher data is genuine data by comparing the data;
a step of outputting a determination result;
A data processing method comprising: computer executes
a step of obtaining evidence data;
with reference to a storage unit that stores a plurality of pieces of determination data indicating characteristics of at least one of predetermined characters or images included in the regular voucher data in association with a plurality of issuers of the vouchers; Comparing at least one feature of the predetermined character or image included in each of the plurality of determination data and at least one feature of the predetermined character or image included in the acquired acquired evidence data. identifying the determination data that is most similar to the acquired evidence data;
determining whether an issuer identified based on at least one of characters or images included in the acquired voucher data matches the issuer corresponding to the identified determination data;
a step of outputting a determination result;
A data processing method comprising: to the computer,
a step of obtaining evidence data;
identifying the issuer of the acquired voucher data based on at least one of characters or images included in the acquired acquired voucher data;
The specified issuer is associated with a plurality of issuers of the evidence by referring to a storage unit that stores a plurality of determination data indicating characteristics of at least one of predetermined characters or images included in the regular evidence data. At least one feature of the predetermined character or image included in the determination data stored in the storage unit in association with at least one feature of the predetermined character or image included in the acquired voucher data. a step of determining whether the acquired voucher data is genuine data by comparing the data;
a step of outputting a determination result;
A program to run. to the computer,
a step of obtaining evidence data;
with reference to a storage unit that stores a plurality of pieces of determination data indicating characteristics of at least one of predetermined characters or images included in the regular voucher data in association with a plurality of issuers of the vouchers; Comparing at least one feature of the predetermined character or image included in each of the plurality of determination data and at least one feature of the predetermined character or image included in the acquired acquired evidence data. identifying the determination data most similar to the determination data by;
determining whether an issuer identified based on at least one of characters or images included in the acquired voucher data matches the issuer corresponding to the identified determination data;
a step of outputting a determination result;
A program to run.

Images