Apparatus, method, and program for verifying the contents of an application

Abstract

The application form is reviewed before submission to reduce the likelihood of applications being rejected after submission. [Solution] The device 100 receives the selection of the application type from the applicant terminal 110 (S201). Next, the device 100 displays a form for the said type (S202) and receives input into multiple input fields of the form (S203). Then, the device 100 uses a generating AI model to verify the validity of at least one of the multiple input fields (S204 to S207). Upon receiving the verification result, the device 100 transmits the verification result to the applicant terminal 110 (S208). If the validity is denied, the applicant terminal 110 displays at least one of the reasons for the denial of validity and a request to correct the input in the said input field on its display unit (S209), and the device 100 receives a new input in the said input field (S210).

Description

[Technical Field]

[0001] The present invention relates to an apparatus, a method, and a program for confirming the contents of an application. [Background Art]

[0002] If there is an error in an expense application, work such as sending the application back to the applicant and rechecking the corrected application arises. [Summary of the Invention] [Problem to be Solved by the Invention]

[0003] If errors in the initial application can be reduced, subsequent work will proceed smoothly; however, even when an electronic application form is used, it has been difficult to perform detailed preliminary confirmation beyond checking whether required input fields have been filled in. The same problem is not limited to so-called expenses, and also occurs in applications for monetary payment, and furthermore, the same applies not only to payments but also to various types of applications such as purchase applications and approval petitions.

[0004] The present invention has been made in view of such circumstances, and an object thereof is to confirm input to an application form before application, thereby suppressing the occurrence of returned applications after submission.

[0005] In the present specification, the term "application" refers to an act of requesting approval, and as long as it is before confirmation by an approver, even if input for the application has been transmitted to the destination, it does not constitute an application. [Means for Solving the Problem]

[0006] To achieve this objective, a first aspect of the present invention is a method for verifying the contents of an application, comprising: receiving input from an applicant terminal to each of a plurality of input fields having a form for a predetermined type of application; creating instructions for at least one of the plurality of input fields to verify the validity of the input to that input field under criteria defined for that input field; making a request to a generating AI model including the instructions; receiving the result of the validity verification; and, if the validity is denied, sending a notification to the applicant terminal including at least one of the reasons for the denial of validity and a request to correct the input.

[0007] Furthermore, a second aspect of the present invention is the method of the first aspect, wherein the criterion is whether or not the input to the input field includes all of the multiple elements set in the instruction.

[0008] Furthermore, a third aspect of the present invention is the method of the first aspect, wherein the criterion is whether or not the input to the input field includes at least one of the one or more elements set in the instruction.

[0009] Furthermore, a fourth aspect of the present invention is the method of the second or third aspect, wherein the input to the input field is text.

[0010] Furthermore, a fifth aspect of the present invention is the method of the second or third aspect, wherein the input to the input field is an image.

[0011] Furthermore, a sixth aspect of the present invention is the method of the first aspect, wherein the criterion is whether or not the image entered into the input field belongs to the category set in the instruction.

[0012] Furthermore, a seventh aspect of the present invention is a method according to any of the first to sixth aspects, further comprising the step of receiving a new input for the input field from the applicant terminal if the above validity is denied.

[0013] Furthermore, an eighth aspect of the present invention is a method according to any of the first to seventh aspects, wherein the instruction includes one or more output examples of the confirmation result.

[0014] Furthermore, a ninth aspect of the present invention is a method according to any of the first to eighth aspects, wherein the type is transportation expenses, meeting expenses, or entertainment expenses.

[0015] Furthermore, a tenth aspect of the present invention is a method according to any of the first to ninth aspects, wherein the criterion is a criterion that takes into account the correlation between the input to the input field and the input to other input fields.

[0016] Furthermore, an eleventh aspect of the present invention is a method according to any of the first to tenth aspects, further comprising the step of receiving the selection of the predetermined type from the applicant terminal.

[0017] A twelfth aspect of the present invention is a program for causing a computer to perform a method for verifying the contents of an application, the method comprising: receiving input from an applicant terminal to each of a plurality of input fields having a form for a predetermined type of application; creating instructions for at least one of the plurality of input fields to verify the validity of the input to that input field under criteria defined for that input field; making a request to a generating AI model including the instructions; receiving the result of the validity verification; and, if the validity is denied, sending a notification to the applicant terminal including at least one of the reasons for the denial of validity and a request to correct the input.

[0018] Furthermore, a thirteenth aspect of the present invention is a device for verifying the contents of an application, which receives input from an applicant terminal to each of a plurality of input fields having a form for a predetermined type of application, creates an instruction for at least one of the plurality of input fields to verify the validity of the input to that input field under a standard defined for that input field, makes a request to a generating AI model including the instruction, receives the result of the validity verification, and, if the validity is denied, sends a notification to the applicant terminal including at least one of the reason for the denial of validity and a request to correct the input. [Effects of the Invention]

[0019] According to one aspect of the present invention, by using a generative AI model to verify the validity of input fields in an application form under defined criteria, it is possible to suppress the rejection of applications by approvers. [Brief explanation of the drawing]

[0020] [Figure 1] This is a diagram showing a system according to one embodiment of the present invention. [Figure 2] This diagram shows the flow of a method for verifying the contents of an application relating to one embodiment of the present invention. [Figure 3] This figure shows an example of instructions for validation according to one embodiment of the present invention. [Modes for carrying out the invention]

[0021] Embodiments of the present invention will be described in detail below with reference to the drawings.

[0022] Figure 1 shows an apparatus according to an embodiment of the present invention. The apparatus 100 supports confirmation of application content by communicating via an IP network such as the Internet with an applicant terminal 110 used by an applicant making an application and a platform 120 that provides a generative AI model. In Figure 1, the generative AI model is provided by the platform 120, but an application for providing the generative AI model may be executed on the apparatus 100.

[0023] The apparatus 100 includes a communication unit 101 such as a communication interface, a processing unit 102 such as a processor and a CPU, and a storage unit 103 including a storage device or storage medium such as a memory and a hard disk. The apparatus 100 can be configured by executing a program for performing each process or each operation in the processing unit 102. The apparatus 100 may include one or more devices, a computer or a server. In addition, the program may include one or more programs, and may be recorded on a computer-readable storage medium as a non-transitory program product. The program may be stored in a storage device or storage medium such as a database 104 accessible via an IP network from the storage unit 103 or the apparatus 100, and instructions included in the program can be executed by at least one processor of the processing unit 102. Data described below as being stored in the storage unit 103 may be stored in a storage device or storage medium such as the database 104, and vice versa.

[0024] First, if necessary, the apparatus 100 receives a selection of an application type from the applicant terminal 110 (S201). Although not shown in Figure 2, for example, the apparatus 100 may cause the applicant terminal 110 to display a screen for type selection, and allow the applicant to select a type from a plurality of options. Examples of the type include expense settlement application, payment application, purchasing application and approval application.

[0025] Next, the apparatus 100 causes the applicant terminal 110 to display a form for said type (S202), and receives inputs for each of the plurality of input fields of said form (S203). Inputs for the plurality of input fields may be transmitted to the apparatus 100 via a single transmission from the applicant terminal 110, or may be transmitted sequentially for one or a plurality of partial input fields. Here, the expression "each of the plurality of input fields" does not necessarily mean all of N input fields when the form has N (N is an integer of 3 or more) input fields, and also means all of two or more and less than N input fields included in said form.

[0026] Next, the apparatus 100 generates an instruction (prompt) for confirming the validity of an input in at least any one of the plurality of input fields (S204).

[0027] FIG. 3 shows an example of an instruction for validity confirmation according to an embodiment of the present invention. As an example of output omitted in FIG. 3, for example, the following example of entertainment expenses can be cited. { "extracted_results": { "entity": "Name", "values": ["Keisuke Yamada, "Makoto Takahashi", "Hiroshi Nakamura"], }, { "entity": "Company Name", "values": ["ABC Co., Ltd."], } , "check_result": true, "explanation": ""Keisuke Yamada", "Makoto Takahashi" and "Hiroshi Nakamura" as "Name", and "ABC Co., Ltd." as "Company Name" were correctly included in the application content." }

[0028] The instructions specify that the application content to be reviewed, i.e., verified, is set in the variable "review_target", the information to be extracted from the application content is set in the variable "extracted_entities", and the conditions for the review are set in the variable "review_condition", and an example output is also provided.

[0029] The variable "review_target" contains at least a portion of the text, images, and other inputs made by the applicant to the input field to be reviewed. The variable may also contain the results of processing the applicant's input by the device 100. Such processing may include deleting or replacing some characters.

[0030] The variable "extracted_entities" can be set to contain one or more elements (entities) to be extracted from the value set in the variable "review_target," and one or more different values ​​can be set depending on the input field to be reviewed. The above output example is an example where "Recipient Company Name" and "Participant Name" are set in list format as follows. Information to extract: ["Name of the other company", "Name of the participant"]

[0031] The variable "review_condition" allows you to set conditions for validating the input. In the example in Figure 3, the types of "review conditions" that can be set in the instructions and the definitions of each "review condition" are listed. By making the "review conditions" variable along with the "information to extract," it becomes possible to define the "information to extract" and "review conditions" in relation to each input field and use the same instruction template for multiple input fields. At least one of the "information to extract" and the "review conditions" may be explicitly stated in the instructions for each input field to be checked. More generally, it is sufficient for the criteria for validation to be stated in the instructions. Furthermore, these criteria may take into account the correlation between the input to the input field to be checked and the input to other input fields. For example, in the case of transportation expenses, the criterion could be that the departure and arrival locations are not the same.

[0032] If necessary, the instructions may include specific examples in addition to or instead of the output examples, as shown in Figure 3. Although omitted in Figure 3, specific examples include the following. By including output examples when the validity of the input is denied in this way, the response of the generative AI model based on the instructions will be closer to what is expected. Input example: Application details: "ABC Co., Ltd., 3 people" Information to extract: ["Name of the other company", "Name of the participant"] Review criteria: "Include all" Expected output example: { "extracted_results": [ { "entity": "Name of the other company" "values": ["ABC Corporation"], }, { "entity": "Participant's name", "values": [], } ], "check_result": false, "explanation": "The application included "ABC Co., Ltd." as the "Name of the counterparty company," but the application did not include any value corresponding to the "Name of the participant." }

[0033] In the example in Figure 3, the system checks whether the entered application content contains multiple values ​​that correspond to all of the multiple elements set in the application. Alternatively, it may be used to check whether the entered application content contains values ​​that correspond to at least one of the one or more elements set in the application. An example of this is as follows: The system checks whether the input in the reason input field of the travel expense application form contains the specified element "specific reason for travel," and explains that it is not valid input because it does not contain the element. Input example: Application details: "I just got on the train on a whim." Information to extract: ["Specific reason for boarding"] Review conditions: "Includes" Expected output example: { "extracted_results": [ { "entity": "Specific reason for boarding", "values": [], } ], "check_result": false, "explanation": "The application did not include a specific reason for boarding. 'I just boarded for no particular reason' is insufficient. Please describe your reason specifically." }

[0034] As a further example regarding expense claims for transportation, an image may be submitted, and whether or not the employee number is written on that image can be used as a criterion for validity verification. Some companies may require employees to write their employee number on taxi receipts when submitting expense claims in order to prevent duplicate claims. Another example of an image being submitted is whether or not the submitted image belongs to a set category, which can be used as a criterion for validity verification. For example, by setting taxi receipts as the category, it is possible to verify whether or not the necessary supporting documents are attached.

[0035] The device 100 then makes a request to the generating AI model, which includes the instructions (S205). The request to the generating AI model can be made by executing code containing instructions written in natural language and calling the OpenAI API, and the generating AI model generates a response to the prompt. The OpenAI API is an example, and other APIs may be used. More specifically, the code for making the request is stored in the memory of the device on which the code is executed, and the device can retrieve this code and make the request by executing code containing instructions obtained by setting the required values ​​to the variables contained in the code.

[0036] Upon receiving the request, platform 120 verifies the validity of the input (S206) and transmits the verification result to device 100 (S207). Device 100 then transmits the verification result to applicant terminal 110 (S208).

[0037] In this specification, "AI model" refers to a machine learning model that has been trained to predict an output for a given input, and "generative AI model" refers to a large-scale language model (LLM) that has been trained using text data to generate an output not included in the input for a given input. As a generative AI model, an LLM applying a transformer architecture is particularly preferred, but it is expected that the name of the architecture may change as technology advances. Therefore, in this specification, "transformer architecture" includes architectures that use one or more features of a transformer architecture or improvements thereof. In this specification, whether or not "generative AI models" are the same is determined by whether or not the type of generative AI model specified by the user is the same. In the case of the Open AI API, for GPT-4, if the value of the variable "model" is the same, it is expressed as the same generative AI model. Generative AI models used in different processes can be the same, but if they are not the same, they may be provided on the same platform 120 or device 100, or on different devices. Furthermore, it goes without saying that each request for a generative AI model may include multiple requests, and may also include one or more processes performed on the device making the request other than the requests for the generative AI model.

[0038] Upon receiving the confirmation result, if the validity is denied, the device 100 sends a notification to the applicant terminal 110 including at least one of the reasons for the denial of validity and a request to correct the input in the input field (S208). Upon receiving the notification, the applicant terminal 110 displays the reason or the correction request on its display unit (S209) and can receive a new input in the input field from the applicant terminal 110 (S210). Here, the confirmation result received by the device 100 and the confirmation result transmitted by the device 100 do not necessarily have to be identical in at least one of their contents and format; it is sufficient that the latter is based on the former.

[0039] After the validity of all one or more input fields subject to validation has been confirmed, the application content is sent to the device 100 in response to the applicant pressing the "Submit" button, etc., and the device 100 notifies the approver's terminal (not shown) of the application content request, thereby suppressing rejection by the approver. Alternatively, the device 100 may notify the approver's terminal of the approval request when the validity of all one or more input fields subject to validation has been confirmed, without requiring the applicant to press the "Submit" button, etc.

[0040] Furthermore, in the embodiments described above, unless the word "only" is used, such as "based only," "depending only," "in the case of only," or "referencing only," it is assumed in this specification that additional information may also be considered. Also, as an example, the statement "if a, then b" does not necessarily mean "always b in the case of a" or "b immediately after a," unless explicitly stated otherwise. In addition, the statement "each a constituting A" does not necessarily mean that A is composed of multiple components, but includes the possibility that the component is singular.

[0041] Furthermore, it should be noted that the embodiments of the present invention described above are included in the disclosure herein, in any way that they are not inconsistent with each other.

[0042] Furthermore, for the sake of clarity, even if there are aspects of operation in some method, program, terminal, device, server, or system (hereinafter referred to as "method, etc.") that differ from the operation described herein, each aspect of the present invention is intended to cover the same operation as any of the operations described herein, and the existence of operation different from the operation described herein does not mean that such method, etc. is outside the scope of each aspect of the present invention. [Explanation of Symbols]

[0043] 100 devices 101 Communications Department 102 Processing Unit 103 Storage section 104 Databases 110 Applicant terminal 120 platforms

Claims

1. A method for verifying the contents of an application, The computer receives input from the applicant's terminal for each of the multiple input fields in a form for a specified type of application, The steps include: the computer creating instructions for at least one of the plurality of input fields to verify the validity of the input to that input field under the criteria set for that input field; The computer makes a request including the instructions for the generated AI model, The computer receives the validation result, The computer, if the validity is rejected, sends a notification to the applicant terminal including at least one of the reasons for the rejection and a request to correct the input; The computer, if the validity is confirmed, notifies the approver terminal used by the approver of a request for approval of the application contents for the form. Includes, The aforementioned criteria include one or more elements to be extracted from the input.

2. The method according to claim 1, The aforementioned criteria further include, as a condition for the confirmation, whether the input to the input field includes all of the one or more elements set in the instruction.

3. The method according to claim 1, The aforementioned criteria further include, as a condition for the confirmation, whether the input to the input field includes at least one of the one or more elements set in the instruction.

4. The method according to claim 2 or 3, The input for this input field is text.

5. The method according to claim 2 or 3, The input for this input field is an image.

6. The method according to claim 1, The aforementioned criterion is whether or not the image entered into the input field belongs to the category set in the instruction.

7. A method according to any one of claims 1 to 3, If the aforementioned validity is denied, the method further includes the step of receiving new input for the input field from the applicant's terminal.

8. A method according to any one of claims 1 to 3, The instructions include one or more output examples of the verification results.

9. A method according to any one of claims 1 to 3, The aforementioned categories are transportation expenses, meeting expenses, or entertainment expenses.

10. A method according to any one of claims 1 to 3, The process further includes the step of receiving the selection of the predetermined type from the applicant's terminal.

11. A program for causing a computer to perform a method for verifying the contents of an application, wherein the method is The steps include receiving input from the applicant's terminal for each of the multiple input fields in a form for a specified type of application, A step of creating instructions for at least one of the aforementioned multiple input fields to verify the validity of the input to that input field under the criteria set for that input field, A step of making a request including the instructions for the generated AI model, The steps include receiving the results of the validation check, If the validity is denied, the step of sending a notification to the applicant's terminal that includes at least one of the reasons for the denial of validity and a request to correct the input, If the validity is confirmed, the approver's terminal is notified of the request for approval of the application contents for the form. Includes, The aforementioned criteria include one or more elements to be extracted from the input.

12. A device for verifying the contents of an application, The system receives input from the applicant's terminal into each of the multiple input fields in the form for the specified type of application. For at least one of the aforementioned plurality of input fields, create an instruction to verify the validity of the input to that input field under the criteria set for that input field, and make a request to the generating AI model including the said instruction. Upon receiving the results of the validation check, If the above validity is denied, a notification is sent to the applicant's terminal including at least one of the reasons for the denial of validity and a request to correct the input. If the validity is confirmed, the approver's terminal used by the approver is configured to notify the approver of the application content for the form, The aforementioned criteria include one or more elements to be extracted from the input.

Images