Back office support equipment

Abstract

To provide an automatic processing means for supporting the introduction and improvement of back office operations utilizing knowledge inside and outside the device. [Solution] The back-office business support device 1 of the present invention comprises a business information management unit 112, a human resources information management unit 114, a terminal input acquisition unit 113 that acquires user input data entered into a user terminal TU, a text generation unit 115 that causes a large-scale language model to generate text proposing back-office business based on multiple pieces of business information, and an information provision unit 116 that identifies human resources related to the back-office business proposed by the text generation unit 115 based on multiple pieces of human resources information and transmits the business information related to the back-office business to the human resources terminal TP used by the human resources, and the text generation unit 114 is configured to be able to consider further suggestions based on the prompts generated by the large-scale language model.

Description

[Technical Field]

[0001] The present invention relates to a support device for back office operations. [Background technology]

[0002] Back office operations, such as accounting and general affairs, are essential tasks regardless of the size or other attributes of a company. However, back office operations can easily become complicated and often depend on individuals, making it difficult to adapt to changes in scale and phase. Therefore, in order to strengthen organizational capabilities and achieve sustainable corporate growth, it is necessary to introduce and improve the operation of back office operations that are tailored to the size, phase, and other requirements of the company. As a result, efforts have been made to develop technologies that support the introduction and improvement of back office operations.

[0003] Regarding technology for supporting operational optimization of back-office operations in conventional services and conventional technologies, Patent Document 1 discloses a back-office operations support device that includes a business information management unit that manages business information that associates company attributes with back-office operations performed by the company, a target attribute acquisition unit, and a business proposal unit that proposes back-office operations to the target company based on multiple business information and the target attributes managed in the business information management unit, wherein the business proposal unit has a configuration for proposing the back-office operations that are associated with the attributes in the business information that match the target attributes, and other configurations.

[0004] The technology described in Patent Document 1 can support users in selecting and appropriately operating the necessary back-office operations, regardless of the level of knowledge accumulated regarding back-office operations. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 7557812 Summary of the Invention [Problem to be solved by the invention]

[0006] The technology described in Patent Document 1 proposes back-office tasks based on task information when task information with matching attributes is under the management of the support device. However, because the attributes of companies are diverse, the support device related to this technology may not have sufficient stored task information for companies with matching attributes. To address the lack of suggestions in such cases, it is first considered to more broadly utilize the knowledge managed by the support device. Furthermore, to address this lack, it is also useful to utilize external knowledge, such as the knowledge possessed by personnel related to the back-office tasks. However, due to the above-described configuration, the technology described in Patent Document 1 leaves room for further improvement in terms of supporting the utilization of knowledge both within and outside the device.

[0007] An object of the present invention is to provide an automatic processing means that supports the introduction and improvement of back office operations by utilizing knowledge inside and outside the device. [Means for solving the problem]

[0008] As a result of intensive research into solving the above-mentioned problems, the inventors have found that the above-mentioned object can be achieved by using a large-scale language model (LLM) to utilize business information that is not limited to attribute matching, and by providing information to personnel involved in back-office work, or by other technical means. As a result, the inventors have completed the present invention.

[0009] One aspect of the present invention includes a business information management unit that manages business information related to a company's back office business, a human resources information management unit that manages human resources information that associates back office business with human resources related to the back office business, a terminal input acquisition unit that acquires user input data entered into a user terminal used by a user, a text generation unit that causes a large-scale language model to generate text proposing back office business based on a plurality of pieces of business information, and an information provision unit that identifies human resources related to the back office business proposed by the text generation unit based on a plurality of pieces of human resources information and provides business information related to the back office business to a human resources terminal used by the human resources, and the business information management unit is configured to manage business information related to company attributes including company size and company fiscal month. the human resources information management unit includes, as the human resources information to be managed, human resources information related to professional personnel and specialized personnel; the terminal input acquisition unit is configured to acquire user input data including the attributes; the text generation unit is configured to cause a large-scale language model to generate text proposing back-office work associated with an attribute that matches an attribute included in the user input data, and to generate text proposing back-office work associated with an attribute that is similar in meaning to an attribute included in the user input data; the text generation unit is configured to cause the large-scale language model to generate a prompt that causes the large-scale language model to further consider the proposal; and the text is configured to cause the large-scale language model to generate the text based on a plurality of the business information and the prompt.

[0010] In this assistance device, the text generation unit causes the large-scale language model to generate text not only for back-office tasks with matching attributes but also for back-office tasks with similar attributes, based on the task information, which is knowledge stored in the device. Furthermore, the assistance device allows the large-scale language model to perform further consideration based on the prompts it generates. This allows the assistance device to realize assistance that makes even greater use of the knowledge stored in the device.

[0011] By considering attribute similarity in addition to attribute matching, the scope of back-office tasks suggested by the generated text is expanded. While this expansion, which is achieved by utilizing more internal knowledge, gives users and other suggestion recipients a wider range of options, it may also make it more difficult for them to make an independent judgment about the suggestions.

[0012] In the support device of this aspect, the information providing unit transmits business information to the human resources terminal based on the human resources information managed by the human resources information management unit, regarding knowledge of professional personnel, specialized personnel, and other human resources that is knowledge outside the device. This allows the support device of this aspect to provide the knowledge possessed by the human resources to the user and support the proposal recipient in utilizing that knowledge in back-office work. Therefore, even if the scope of the proposal expands, the support device of this aspect can support the proposal recipient in utilizing knowledge outside the device to introduce and improve back-office work based on the proposal.

[0013] Therefore, this aspect can provide an automatic processing means that supports the introduction and improvement of back office operations by utilizing knowledge inside and outside the device.

[0014] In addition, the present invention can take various forms, such as providing support based on the results of query classification based on a database, proposing back-office work procedures based on a database, prompting the provision of missing information, having humans and large-scale language models utilize frameworks and templates for optimal data configurations in work and other knowledge, automatically collecting information related to the above knowledge, etc. These various forms, each with their own unique configurations, contribute to providing automatic processing means for supporting the introduction and improvement of back-office work by utilizing knowledge inside and outside the device. [Effects of the Invention]

[0015] As described above, the present invention can provide an automatic processing means that supports the introduction and improvement of back office operations by utilizing knowledge inside and outside the device. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a block diagram showing an example of the hardware and software configuration of a system S according to this embodiment. [Figure 2] FIG. 2 is an example of the attribute information database 131. [Figure 3] FIG. 3 is an example of the human resources information database 132. [Figure 4] FIG. 4 is an example of the advice information database 133. [Figure 5] FIG. 5 is an example of the inquiry classification database 134. [Figure 6] FIG. 6 is an example of the business procedure database 135. [Figure 7] FIG. 7 is an example of the business knowledge database 136. [Figure 8] FIG. 8 is a main flowchart showing an example of a preferable flow of the assistance process executed by the assistance device 1 of this embodiment. [Figure 9] FIG. 9 is a continuation of the previous figure. [Figure 10] FIG. 10 is a continuation of the previous figure. [Figure 11] FIG. 11 is a continuation of the previous figure. [Figure 12] FIG. 12 is a continuation of the previous figure. [Figure 13] FIG. 13 is a continuation of the previous figure. [Figure 14] FIG. 14 is a continuation of the previous figure. [Figure 15] FIG. 15 is a continuation of the previous figure. [Figure 16] FIG. 16 is a continuation of the previous figure. DETAILED DESCRIPTION OF THE INVENTION

[0017] First, although the following disclosure, diagrams, and / or claims may be described as being presented alone or in combination with one or more other aspects, the subject matter of the immediate disclosure is not intended to be so limited. That is, the immediate disclosure, diagrams, and claims are intended to encompass the various aspects described herein, each alone or in one or more combinations with each other. For example, even if the immediate disclosure describes and illustrates a first, second, and third embodiment in such a way that the first embodiment is described and illustrated specifically in conjunction with the second embodiment, or the second embodiment is described and illustrated only in conjunction with the third embodiment, the immediate disclosure and illustrations are not so limited and may include only the first embodiment, only the second embodiment, only the third embodiment, or one or more combinations of the first, second, and / or third embodiments, such as the first and second embodiments, the first and third embodiments, the second and third embodiments, or the first, second, and third embodiments.

[0018] The use of the phrase "or" in this document shall mean a "non-exclusive" arrangement unless expressly specified otherwise. For example, when we say "item x is A or B," we mean either: (1) item x is either A or B, but not both; or (2) item x is both A and B. In other words, the word "or" is not used to define an "exclusive" arrangement.

[0019] Additionally, the phrases "comprising at least one of" and "comprising at least one of the following," when used in conjunction with a system or element, mean that the system or element includes one or more of the elements listed after the phrase. For example, if there are three types of elements, element 1 through element 3, the phrases "comprising at least one of" and "comprising at least one of the following" are to be interpreted as any of the following structural arrangements: a device including the first element, a device including the second element, a device including the third element, a device including the first and second elements, a device including the first and third elements, a device including the second and third elements, or a device including the first, second, and third elements.

[0020] A similar interpretation is intended when the phrase "used in at least one of the following" is used in this context. Furthermore, as used in this context, "and / or" is used as a verbal conjunction to indicate that one or more of the listed elements or conditions are included or occur. For example, a device including a first element, a second element, and / or a third element is to be interpreted as any of the following structural arrangements: a device including the first element, a device including the second element, a device including the third element, a device including the first element and the second element, a device including the first element and the third element, a device including the second element and the third element, or a device including the first element, the second element, and the third element.

[0021] In addition, the use of the phrase "and / or" in the text means a "non-exclusive" agreement, as stipulated in the Japanese Industrial Standards (JIS) "Format and preparation method of standard sheets JIS Z 8301."

[0022] Hereinafter, an example of an embodiment of the present invention will be described in detail with reference to the drawings.

[0023] <System S> Fig. 1 is a block diagram showing an example of the hardware configuration and software configuration of a system S of this embodiment. The system S of this embodiment is a back office work support system. The system S is configured to include at least a back office work support device 1.

[0024] In the system S, the support device 1 is preferably configured to communicate via a network N with a user terminal TU used by a user (customer) related to the target company and a human resources terminal TP used by human resources related to back office operations.

[0025] [Support Device 1] The support device 1 includes a control unit 11, a storage unit 13, and a communication unit 14. The type of the support device 1 is not particularly limited, and may be, for example, a server device, a cloud server, or the like.

[0026] [Control Unit 11] The control unit 11 includes a central processing unit (CPU), random access memory (RAM), read only memory (ROM), and other hardware components.

[0027] The control unit 11 cooperates with at least one of the storage unit 13 and the communication unit 14 as necessary. The control unit 11 then realizes various software components of the program of this embodiment that is executed by the assistance device 1.

[0028] The software components include, for example, a public information acquisition unit 111, a business information management unit 112, a terminal input acquisition unit 113, a human resources information management unit 114, a text generation unit 115, an information provision unit 116, an information linkage unit 117, a data inspection unit 118, a data generation unit 119, and a task management unit (not shown). Details of each software component will be described later using the main flowchart.

[0029] [Memory Unit 13] The memory unit 13 is a device for storing data and / or files, and has a storage unit that non-temporarily stores data using a memory member (e.g., a hard disk, a semiconductor memory, a recording medium, a memory card). The memory unit 13 stores data of the program executed by the control unit 11. It is also preferable that the memory unit 13 further stores various data used in the assistance processing.

[0030] To enable common data to be used between users, human resources, and other humans and large-scale language models, it is preferable that at least a portion of the data be structured as a database containing text that can be made readable by users and human resources, an index for searching the text in natural language (e.g., a keyword search index), and an index for referencing the text by large-scale language models (e.g., a semantic vector).

[0031] The databases are configured to include, for example, an attribute information database 131, a human resources information database 132, an advice information database 133, an inquiry classification database 134, a business procedure database 135, a business knowledge database 136, and a missing information database (not shown). These databases may be combined into one, or may be divided into multiple databases.

[0032] In the following, an example will be described in which various data are stored in the above-mentioned multiple database groups, but a person skilled in the art will be able to easily conceive of other configurations based on the above description. Note that the storage unit 13 may also store data related to a large-scale language model (e.g., trained parameters, vocabulary data, auxiliary data for inference).

[0033] (Attribute Information Database 131) The attribute information database 131 stores attribute information that associates the attributes of a company with data indicating the back-office operations performed by the company, and is managed by the business information management unit 112. This allows the support device 1 to refer to the attribute information database 131 based on the attributes of the target company and identify the back-office operations that should be proposed to the target company. Because the attribute information is managed by the business information management unit 112, it is positioned as a type of business information. In the following, the attribute information will be described as being stored in association with unique identification information. The identification information is, for example, a company ID that is unique to each company.

[0034] The company attributes include, for example, the company size (e.g., number of employees), the company's fiscal year end, and the industry of the target company. The back office operations include, for example, bank transfers, bookkeeping, expense reimbursement, and applications for public support (e.g., subsidies and grants).

[0035] The data indicating the back office operations preferably includes the timing, demand, or urgency of when the operations should be performed. By including the timing, the support device 1 can propose back office operations that include the timing. By including the demand or urgency, the support device 1 can allocate back office operations accordingly.

[0036] 2 is an example of the attribute information database 131. In this example, attribute information is stored that associates a small business identified by a company ID "C0001," with a size of "20 employees," a fiscal year end of "May," and an industry of "information and communications," with data related to back-office operations performed by the company ("inventory work (starting around May △ day)," "depreciable asset processing work (starting around May △ day)," "accrued account processing work (starting around May △ △ day)," "small business △ △ △ subsidy application preparation (April - May 27)," "IT △ △ subsidy application preparation (January onward)").

[0037] In this example, attribute information is stored that associates a small or medium-sized enterprise identified by company ID "C0002," with a size of "50 employees," a fiscal year end of "April," and an industry of "information and communications," with data related to the back-office work carried out by the enterprise ("Work to prepare settlement sheets (starting around △△ day in April)," "Work to prepare account item breakdowns (starting around △△ day in April)," and "Preparation for application for small and medium-sized business △△ support subsidy (June - July 16)").

[0038] Additionally, in this example, attribute information is stored that associates the company, which is identified by company ID "C0003," is a small or medium-sized enterprise with a size of "100 employees," has a fiscal year end of "April," and is in the "food manufacturing" industry, with data related to the back-office work carried out by the company ("Work to prepare settlement sheets (starting around △△ April)," "Work to prepare account item breakdowns (starting around △△ April)," "Preparation to apply for small and medium-sized business △△ support subsidy (June - July 16)," and "Preparation to apply for IT △△ subsidy (January onwards)").

[0039] By storing this attribute information in the attribute information database 131, the support device 1 can propose back office operations to the target company by referring to attribute information that matches or is similar in terms of size, fiscal month, or industry.

[0040] (Human Resources Information Database 132) Human resources information in which data indicating back office operations are associated with data indicating personnel who perform the back office operations is stored in the human resources information database 132, and is managed by the human resources information management unit 114. In the following, the human resources information is described as being stored in association with unique identification information. The identification information is, for example, a human resources ID unique to each personnel.

[0041] The data indicating the personnel preferably includes data indicating the contribution of the personnel related to the personnel information to the back-office operations. This allows the support device 1 to narrow down the personnel who have a track record in a specific back-office operation and narrow down the back-office operations that are appropriate to be assigned to the specific personnel.

[0042] The contribution level referred to here is not particularly limited, and may be, for example, various indicators indicating the contribution level of back office work. Examples of such indicators include indicators indicating the experience of the person in charge of the back office work (number of times assigned, total work time, etc.), indicators indicating the time efficiency of the back office work (work time per case, amount of work per certain period, etc.), and indicators indicating the accuracy of the back office work (error rate, number of returns, return frequency, number of complaints, complaint rate, etc.). Furthermore, it is preferable that the personnel information database 132 includes, as indicators, past work performance and third-party evaluations of the performance (e.g., feedback from service users and feedback from the device operator) so that the support device 1 can perform the process related to personnel selection based on the objectively determined contribution level.

[0043] The human resources in the human resources information preferably include internal human resources and / or external human resources of the target company. By including data related to internal human resources, the support device 1 can make proposals that appropriately combine internal human resources and back office work. By including data related to external human resources, the support device 1 can make proposals related to the outsourcing of back office work. By including data related to internal human resources and external human resources, the support device 1 can refer to the data and support the decision on whether to proceed with in-house production or outsourcing.

[0044] The internal and external human resources managed in the human resources information database 132 are also referred to as a talent pool. The talent pool includes various types of human resources (e.g., back office human resources in charge of back office operations at each target company, human resources providing support related to back office operations, specialized human resources with specialized skills, and human resources related to professional offices). In order to be able to distinguish between internal and external human resources for each target company, the human resources information preferably includes information related to the affiliation of the human resources.

[0045] To enable estimation of future contributions according to the work situation, the human resource information preferably includes the work situation of the human resource (for example, the work the human resource is currently in charge of, the current workload).

[0046] 3 is an example of the human resources information database 132. This example stores human resources information that associates in-house human resources "Tanaka △△" of company "Company A" identified by human resources ID "P0001" with back office tasks and their contributions "bank transfer tasks (number of times handled: △△△ times, total work time △△△ hours, work time per item: △△, complaint occurrence rate: △%)" and "bookkeeping tasks (number of times handled: △△△ times, total work time △△△ hours, work time per item: △△, complaint occurrence rate: △%)."

[0047] In addition, this example stores personnel information associated with in-house personnel "Abe △△" of company "Company A" identified by personnel ID "P0002," and back office operations and their contributions to "transfer operations (number of times handled: △△ times, total work time △△ hours, work time per item: △△, complaint occurrence rate: △%)" and "bookkeeping operations (number of times handled: △△△ times, total work time △△△ hours, work time per item: △△, complaint occurrence rate: △%)."

[0048] In addition, this example stores personnel information relating to the personnel "Suzuki △△" of the company "Accounting Firm B" identified by the personnel ID "P0003," and the back office work and its contribution "Subsidy application (number of times handled: 3, total work time △△△ hours, number of acceptances: 2)."

[0049] By storing this personnel information in the personnel information database 132, the support device 1 can, for example, refer to the personnel information database 132 and suggest that the transfer work of Company A be preferentially assigned to Mr. Tanaka, who has a lot of experience, the bookkeeping work of Company A, which has a low incidence of complaints, and the subsidy application work of Company A be assigned to Mr. Suzuki of Accounting Firm B, who is an external personnel with a proven track record.

[0050] (Advice Information Database 133) The advice information database 133 stores advice information that associates financial situations with management advice for companies with those financial situations, and is managed by the business information management unit 112. Because the advice information is managed by the business information management unit 112, it is positioned as a type of business information. In the following, the advice information will be described as being stored in association with unique identification information. The identification information is, for example, an advice ID that is unique for each piece of advice.

[0051] The financial situation is, for example, data on various financial indicators that show the financial situation (for example, gross profit margin on sales, operating profit margin on sales, return on assets (ROA), return on equity (ROE), total asset turnover, accounts receivable turnover period, break-even point, current ratio, quick ratio, equity ratio, liquidity ratio, fixed ratio, fixed long-term maturity ratio, operating cash flow, investing cash flow, financial cash flow, debt ratio, value added, value added rate, labor distribution rate, labor productivity, value added rate on sales, tangible fixed asset turnover rate, sales per employee, equipment productivity, sales growth rate (revenue growth rate), operating profit growth rate, total capital growth rate, research and development expense ratio on sales, employee increase rate).

[0052] Examples of management advice include advice recommending back-office operations that should be introduced to improve the financial situation, advice proposing back-office operations that should be optimized to improve the financial situation and the procedures for optimizing them, and advice recommending subsidies, etc. that should be applied for to improve the financial situation.

[0053] 4 is an example of the advice information database 133. In this example, as advice information related to advice ID "A0001," information is stored that associates a financial situation in which the return on equity (ROE) is "△% or less" with management advice to a company in that financial situation, "Optimize equity capital through the sale of unnecessary assets and capital restructuring."

[0054] In addition, in this example, advice information related to advice ID "A0002" is stored, which associates a financial situation in which the accounts receivable turnover period is "△ or more" with management advice to a company in that financial situation, "Introduce an accounts receivable management system."

[0055] By storing this advice information in the advice information database 133, the support device 1 can refer to the advice information database 133 and advise Company C, whose return on equity (ROE) is "△% or less," to optimize its equity capital. Also, the support device 1 can refer to the advice information database 133 and advise Company D, whose accounts receivable turnover period is △ or more, to introduce an accounts receivable management system.

[0056] (Inquiry classification database 134) The inquiry classification database 134 stores inquiry classification information including correspondence between inquiries related to back office operations and the classification of the inquiries, and is managed by the business information management unit 112. Because the inquiry classification information is managed by the business information management unit 112, it is positioned as a type of business information. In the following, the explanation will be given assuming that the classification information is stored in association with unique identification information. The identification information is, for example, a classification ID unique to each correspondence.

[0057] Examples of inquiry classification include classification by the nature of the inquiry (e.g., classification into requests, questions, issues, and other characteristics), classification by the type of response required (e.g., classification into requests for business proposals, requests for procedure proposals, requests for knowledge provision, and other types), classification by the type of back-office work related to the inquiry (e.g., classification by accounting, payroll calculation, and other types), and classification based on a composite perspective (e.g., classification that combines nature classification and response classification).

[0058] A request is, for example, an inquiry seeking specific back-office operations that will solve a problem the target company has and a proposal from a contractor to do the work. A question is, for example, an inquiry seeking knowledge to resolve an unclear point related to a back-office operation being performed or being considered for introduction by the target company. A problem is, for example, an inquiry seeking a proposal for a procedure to solve a problem that exists in the back-office operation being performed by the target company.

[0059] A request for business proposal is an inquiry that can be interpreted as a request to propose appropriate back-office work in accordance with the circumstances. A request for procedure proposal is an inquiry that can be interpreted as a request to propose appropriate procedures for back-office work in accordance with the circumstances. A request for knowledge provision is an inquiry that can be interpreted as a request to provide knowledge that will help resolve questions about back-office work.

[0060] The correspondence relationships stored in the query classification database 134 are not particularly limited, and may be, for example, a correspondence relationship between a query text and a classification of the text, a correspondence relationship between the characteristics of a query written in natural language and a classification of a query having the characteristics, or a correspondence relationship between a vector in a semantic space indicating the characteristics of the query text and a classification of the query.

[0061] 5 is an example of the inquiry classification database 134. This example stores a correspondence between an inquiry "Text: I want to set up a new accounting department" identified by the classification ID "Q0001" and the classification "Request > Establishment of an accounting department." This example also stores a correspondence between an inquiry "Characteristics: Text requesting the establishment of a new accounting department" identified by the classification ID "Q0002" and the classification "Request > Establishment of an accounting department." This example also stores a correspondence between an inquiry "Semantic vector: (salary, calculation, question)" identified by the classification ID "Q0003" and the classification "Question > Payroll calculation."

[0062] In addition, this example stores the correspondence between the query "Text: Please tell me how to revise social insurance premiums in payroll calculations" identified by the classification ID "Q0004" and the classification "Knowledge request > Social insurance premium calculations."

[0063] Furthermore, in this example, the correspondence between the inquiry "Characteristics: Question text regarding calculation and revision methods for social insurance premiums" identified by the classification ID "Q0005" and the classification "Knowledge request > Social insurance premium calculation" is stored.

[0064] In this way, the query classification database 134 stores correspondence relationships between query texts and classifications, correspondence relationships between query features written in natural language and classifications, and correspondence relationships between semantic vectors indicating query features and classifications, each of which is associated with a unique classification ID. This allows the support device 1 to flexibly apply a variety of classification decisions.

[0065] (Business Procedure Database 135) Business procedure information, including the correspondence between information specifying the type of back-office business or limiting conditions and the procedures of the back-office business, is stored in the business procedure database 135, and is managed by the business information management unit 112. Because it is managed by the business information management unit 112, business procedure information is considered to be a type of business information. In the following, it is assumed that business procedure information is stored in association with unique identification information. The identification information is, for example, a procedure ID unique to each correspondence. Business procedure information is preferably stored in a format that conforms to a standard format that indicates a hierarchical overall structure in which there is a business, a process dependent on that business, and an operation dependent on that process. Using a format that prevents registration of subordinate lower levels unless a higher level is registered, the information specifies a processing procedure in which information is registered in the order of business, process, and operation.

[0066] The information specifying the type of back office work or the limiting conditions is, for example, data indicating the classification of back office work explained in the inquiry classification database 134. Depending on its granularity, the classification corresponds to either the information specifying the back office work or the limiting conditions.

[0067] The procedure includes, for example, the order of each included task, the method for calculating deadlines, and the final destination as a goal. Examples of each included task include the task of requesting each department to submit necessary documents, the task of checking whether all necessary documents are available, the task of creating documents to be submitted by referring to the documents, the task of verifying whether the documents to be submitted are complete, the task of seeking approval for the documents to be submitted, and the task of submitting the documents to be submitted.

[0068] The method of calculating the due date may be, for example, a method of calculating the due date using the fiscal year closing month, or a method of using the date determined for each fiscal year for the back office work (e.g., a method of referencing the due date for final tax returns).The method of calculating the due date may also be a method of calculating multiple due dates (e.g., a method of calculating the due date for task A, the due date for task B, and the final due date separately).

[0069] Examples of landing points include requirements that submitted documents must meet (e.g., the requirement that the balance sheet balances), requirements that personnel allocation must meet (e.g., the requirement that the number of qualified personnel is within an appropriate range), and requirements that resource allocation must meet (e.g., a requirement that includes the appropriate range of personnel numbers according to the amount of work estimated from the content).

[0070] 6 is an example of the business procedure database 135. In this example, as a correspondence relationship identified by the procedure ID "P0001", the following correspondence relationships are stored: a back-office task "Financial report" identified by the classification ID "Q0006", a work sequence "request for submission of necessary documents → collection and confirmation of documents → preparation of financial statements → verification of discrepancies → approval by the accounting manager → submission to the tax office", a deadline calculation method "submission request: △ months before the closing month, ..., final submission: by the deadline △ month △ day specified by the tax office", and a destination "final statements must satisfy the requirements ... stipulated in the relevant laws and regulations".

[0071] In this way, the business procedure database 135 stores the correspondence between back office tasks and their procedures, along with a unique procedure ID for each correspondence. This allows the assistance device 1 to provide text that suggests procedures by referring to the database.

[0072] (Business Knowledge Database 136) The business knowledge database 136 stores knowledge information including knowledge related to back office operations and is managed by the business information management unit 112. Because it is managed by the business information management unit 112, the knowledge information is considered to be a type of business information. The "knowledge" referred to here is not particularly limited, and corresponds to, for example, provisions of relevant laws and regulations, guidelines, business practices, precautions, past cases, and other general knowledge related to back office operations. In the following, the business knowledge information is described as being stored in association with unique identification information. The identification information is, for example, a knowledge ID unique to each piece of knowledge.

[0073] Examples of knowledge include the names and article numbers of laws and regulations related to back-office operations (e.g., Article △ of the Corporation Tax Act), points to note when interpreting laws and regulations (e.g., exceptions under certain conditions), practical practices (e.g., the recommended timing for prior consultation with the tax office regarding financial statements), and past trouble cases (e.g., cases where required documents were resubmitted due to incomplete information and how to avoid such cases).The format of the knowledge is not particularly limited, and may be various formats exemplified by documents, checklists, or short sentences listing key points.

[0074] The knowledge preferably includes knowledge obtained from information published on the Internet (e.g., a database of laws and regulations, documents published by government agencies, guidelines from industry associations). From the perspective of reducing the effort required by users and / or administrators to obtain knowledge and enabling access to a wide range of knowledge on the Internet with less effort, the information published on the Internet preferably includes information automatically obtained from an MCP server that provides access to information via the Model Context Protocol (MCP).

[0075] Furthermore, the knowledge preferably includes knowledge acquired from the human resource input data entered into the human resource terminal TP and based on the knowledge content included in the data. This allows the support device 1 to effectively utilize the knowledge possessed by various human resources (e.g., human resources related to professional occupations, specialized human resources) who use the human resource terminal TP. In addition, the knowledge may include knowledge acquired from data uploaded as a file and based on the knowledge content included in the data. Of the knowledge, knowledge belonging to a manual is preferably stored in a format that allows procedures related to the work to be added one by one, and has an editing function specialized for manuals, such as adding a red frame to screenshots.

[0076] FIG. 7 is an example of the business knowledge database 136. In this example, the relevant law "Corporate Tax Law Article △..." and its information source "http:△△△" are stored as knowledge information identified by knowledge ID "K0001." Also, in this example, the note "Confirming consistency of income and expenses in submitted documents..." and its information source "file:△△△.pdf" are stored as knowledge information identified by knowledge ID "K0002." Additionally, in this example, the past case "Case in which late payment tax was incurred due to delayed submission of financial statement..." and the information source "Human Resources ID: P0003, △△△△ year △△ month △△ day △△:△△:△△ input data" are stored as knowledge information identified by knowledge ID "K0003."

[0077] In this way, knowledge information related to back-office operations derived from various information sources is stored together with a unique knowledge ID for each piece of knowledge information in the business knowledge database 136. This allows the assistance device 1 to refer to the database in a large-scale language model and generate text that indicates an answer based on the knowledge.

[0078] (Required Information Database) The required information database stores necessary information including the correspondence between inquiries related to back office operations and information required to identify procedures related to the back office operations, and is managed by the business information management unit 112. Because the necessary information is managed by the business information management unit 112, it is positioned as a type of business information. Examples of necessary information stored in the necessary information database include frameworks and templates for optimal data structures in business.

[0079] The above-mentioned correspondence relationship is not particularly limited, and may be, for example, a correspondence relationship between part or all of the query text and the required information, a correspondence relationship between the features of the query written in natural language and the required information, or a correspondence relationship between a vector in a semantic space indicating the features of the query text and the required information. Furthermore, the data indicating the "required information" is not particularly limited, and may be, for example, a label name of the information (e.g., "delivery date" or "budget size") or text explaining what the information is.

[0080] By storing the above-mentioned necessary information in the necessary information database, the assistance device 1 can refer to the database to identify the missing information and generate text requesting the provision of the missing information.

[0081] [Communication Unit 14] The specific configuration of the communication unit 14 is not particularly limited as long as it is capable of connecting the support device 1 to the network N and performing communication. The communication unit 14 may be configured using, for example, a network card compatible with the Ethernet standard, a communication device compatible with wireless LAN, or other communication devices.

[0082] [Server 2] The system S is preferably configured to include a server 2 that executes natural language processing using a large-scale language model as a device separate from the assistance device 1 (Fig. 1). This allows the assistance device 1 to separate part of the processing and management related to the large-scale language model from the assistance device 1 itself. The server 2 may be in the form of a cloud server. The server 2 may be managed by a management entity different from that of the assistance device 1.

[0083] [Large-Scale Language Model] The large-scale language model (LLM) is not particularly limited, and examples thereof include a generation support service having an LLM as a backend, or an inference engine related to the LLM executed in the support device 1.

[0084] Examples of such generation assistance services include ChatGPT (registered trademark), Claude, Gemini, and Llama, which have the ability to generate natural language output based on given prompts and available data, and can be used to generate text according to the present invention.

[0085] The large-scale language model may be pre-trained or fine-tuned using training data based on various data stored in the various databases described above, and may be combined with prepared prompts.

[0086] [Network N] The type of network N is not particularly limited as long as it is a network through which the support device 1 and other devices can communicate. The access network related to the network N is, for example, a wired LAN, a wireless LAN, or a mobile line. The logical network related to the network N is, for example, the Internet, an intranet, or a cloud service network.

[0087] [User Terminal TU] The user terminal TU is used by users involved in back office operations. The users are, for example, internal personnel of the target company (e.g., employees in the general affairs department, employees in the accounting department), and personnel in the target company who consider how to proceed with back office operations (e.g., managers and executives of the target company, personnel in the information systems department, personnel in the corporate planning department). The hardware configuration of the user terminal TU is not particularly limited, and may be, for example, a mobile terminal (e.g., smartphone, tablet terminal, laptop computer) or a fixed terminal (e.g., desktop PC).

[0088] [Human Resources Terminal TP] The human resources terminal TP is used by human resources involved in back office operations. Human resources are, for example, external human resources of the target company. External human resources include, for example, human resources who provide support for back office operations, specialized human resources with specialized skills, and human resources involved in professional offices (e.g., tax accountant offices, social insurance labor consultant offices). The hardware configuration of the human resources terminal TP may be the same as that of the user terminal TU.

[0089] The user terminal TU, the human resource terminal TP, and other terminals may be collectively referred to as terminal T.

[0090] <Main Flowchart of Support Processing> Figure 8 is a main flowchart showing an example of a preferred flow of the support processing executed by the support device 1 of this embodiment. Figures 9, 10, 11, 12, 13, 14, 15, and 16 are each figures continuing from the previous figure. The following is an example of a preferred flow of the support processing executed by the support device 1 of this embodiment using Figures 8 to 16.

[0091] The support process preferably includes a series of steps for acquiring public information related to knowledge about back-office operations from the Internet. Steps S1 to S3 are an example of such steps.

[0092] [Step S1: Determine whether to acquire public information] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to enable the public information acquisition unit 111. Then, the control unit 11 executes a process of determining whether to acquire public information on the Internet related to the above-mentioned knowledge by the public information acquisition unit 111 (public information acquisition determination step). If the control unit 11 determines that the information should be acquired, it proceeds to step S2, and if not, it proceeds to step S4.

[0093] The determination in this step is not particularly limited, and examples include a procedure of determining to acquire specific public information designated by a user, human resource, and / or administrator, or public information that similarly meets designated requirements, when instructed to do so, or a procedure of determining to acquire when a pre-designated acquisition timing is met. Furthermore, the "public information on the Internet related to knowledge" is not particularly limited, and may be the same as the public information described in the section on business knowledge database 136.

[0094] [Step S2: Acquire Public Information] The control unit 11 executes a process of acquiring, from the Internet, the public information determined to be acquired in the previous step by the public information acquisition unit 111 (public information acquisition execution step). The control unit 11 proceeds to step S3.

[0095] The acquisition in this step is realized, for example, by a procedure of identifying an address on the Internet corresponding to the public information based on the above-mentioned specification, accessing the address, and downloading the public information, or by other conventional procedures. From the viewpoint of reducing the effort required of users, personnel, and / or administrators to acquire knowledge and enabling access to a wide range of knowledge on the Internet with less effort, it is preferable that the acquisition in this step includes a procedure of automatically acquiring public information from an MCP server.

[0096] [Step S3: Start management of corresponding business information] The control unit 11 cooperates with the storage unit 13 to enable the business information management unit 112. Then, the control unit 11 executes a process to start management of business information corresponding to the public information acquired in the previous step by the business information management unit 112 (public information management start step). The control unit 11 proceeds to step S4.

[0097] The start of management in this step is realized by, for example, a procedure of storing the public information itself as knowledge in the business knowledge database 136, a procedure of storing the main part of the public information extracted by the large-scale language model as knowledge in the business knowledge database 136, or other procedures. From the viewpoint of providing users and / or human resources with a means of verifying the public information, it is preferable that the start of management in this step includes a procedure of storing the knowledge in the business knowledge database 136 in association with its source.

[0098] The support process includes a series of procedures for managing personnel information that associates back-office operations with personnel involved in the operations. Steps S4 to S5 are an example of the procedures for updating personnel information.

[0099] [Step S4: Determine whether to update personnel information] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to enable the personnel information management unit 114. Then, the control unit 11 executes a process to determine whether to update the personnel information using the personnel information management unit 114 (personnel information update determination step). If the control unit 11 determines that the personnel information should be updated, it proceeds to step S5, and if not, it proceeds to step S6.

[0100] The procedure for determining whether or not an update is required in this step is not particularly limited, and includes, for example, a procedure for determining that the personnel information should be updated when data and instructions related to updating the personnel information are received.

[0101] [Step S5: Updating personnel information] The control unit 11 executes processing to update personnel information by the personnel information management unit 114 (personnel information update step). The control unit 11 moves the processing to step S6.

[0102] The update is realized, for example, by a procedure for storing the personnel information in the personnel information database 132, a procedure for updating the personnel information stored in the personnel information database 132, or other procedures. Those skilled in the art who have read this explanation will understand that the start of personnel information management (storing in the database), the end of personnel information management (deletion from the database), and other management procedures can be similarly realized.

[0103] The support process preferably includes a series of procedures for managing advice information that associates financial situations with management advice for the company based on the financial situations. Steps S6 to S7 are an example of the procedures for updating the advice information.

[0104] [Step S6: Determine whether to update advice information] The control unit 11 executes a process of determining whether to update advice information using the business information management unit 112 (advice information update determination step). If the control unit 11 determines that the advice information should be updated, it proceeds to step S7, and if not, it proceeds to step S8.

[0105] The procedure for determining whether or not the advice information is to be updated is not particularly limited, and includes, for example, a procedure for determining that the advice information is to be updated when data and instructions related to updating the advice information are received.

[0106] [Step S7: Updating Advice Information] The control unit 11 executes a process of updating the advice information in the advice information database 133 by the business information management unit 112 (advice information update step). The control unit 11 moves the process to step S8.

[0107] The update is realized, for example, by a procedure for storing the advice information in the advice information database 133, a procedure for updating the advice information stored in the advice information database 133, or other procedures. Those skilled in the art who have read this description will understand that the start of management of advice information (storing in the database), the end of management of advice information (deletion from the database), and other management can also be realized in a similar manner.

[0108] [Step S8: Determine whether to acquire user input data] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to enable the terminal input acquisition unit 113. Then, the control unit 11 executes a process to determine whether to acquire user input data from the user terminal TU using the terminal input acquisition unit 113 (user input acquisition determination step). If the control unit 11 determines that the data should be acquired, it proceeds to step S9, and if not, it proceeds to step S30.

[0109] The procedure for determining whether or not to acquire data in this step is not particularly limited. For example, this procedure includes a procedure for determining, when input data is received from the user terminal TU, that the data is to be acquired.

[0110] [Step S9: Obtain User Input Data] The control unit 11 executes a process of obtaining user input data from the user terminal TU by the terminal input obtaining unit 113 (user input obtaining execution step). The control unit 11 moves the process to step S10.

[0111] In this step, the terminal input acquisition unit 113 acquires user input data including company attributes. The "attributes" here include, for example, the company size, the company's fiscal year end, the type of business, and the status of back office operations.

[0112] [Step S10: Determine whether the data contains an inquiry related to back office operations] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to enable the text generation unit 115. Then, the control unit 11 executes a process to determine whether the user input data acquired in the previous step contains an inquiry related to back office operations using the text generation unit 115 (inquiry determination step). If the control unit 11 determines that the data contains an inquiry, it proceeds to step S11, and if not, it proceeds to step S30.

[0113] The determination in this step is realized by a process including a procedure of inputting the user input data into a regular expression engine, a large-scale language model, or other classifier, and causing the classifier to output a determination result. For example, the assistance device 1 can input a prompt asking whether the following text data is an inquiry related to back-office work and the user input data into the large-scale language model, thereby causing the large-scale language model to function as a classifier that outputs a determination result.

[0114] The support process preferably includes a series of steps for determining the classification of the inquiry and providing appropriate support based on the classification. Steps S11 to S20 are an example of a series of steps for determining whether the inquiry is requesting a business proposal, and for providing a text proposing a business or other support if the inquiry corresponds to the inquiry.

[0115] [Step S11: Determine whether the query is a request for a business proposal] The control unit 11 executes a process to determine whether the user input data determined in the previous step to include a query is a query requesting a back-office business proposal (business proposal determination step) using the text generation unit 115. If the control unit 11 determines that the data is a query, it proceeds to step S12, and if not, it proceeds to step S21.

[0116] The determination in this step is realized, for example, by a process including a procedure of inputting user input data into a large-scale language model and outputting a query classification. Preferably, this procedure involves outputting a query classification from the large-scale language model based on the user input data and classification information, which is business information stored in the query classification database 134. This allows the assistance device 1 to perform appropriate classification based on the business information.

[0117] The classification related to this step may include not only the classification related to whether the inquiry is a request for a business proposal or not, but also the various classifications described in the section on the inquiry classification database 134. This allows the support device 1 to perform not only the classification related to this step but also the various classifications performed in the subsequent processing in the same process.

[0118] The classification in this step may be performed by a person (e.g., a user or a human resource), thereby allowing the assistance device 1 to process based on a more appropriate classification.

[0119] The support process preferably includes a series of processes for determining whether or not information is missing to identify the proposed business, and, if so, identifying the missing information and requesting its provision. Steps S12 to S15 are an example of such processes. This allows the text generator 115 to autonomously determine which information to use for the decision and, if necessary, request the provision of the missing information.

[0120] [Step S12: Determine whether necessary information is missing] The control unit 11 executes a process of determining whether information necessary to identify the business to be proposed is missing (business proposal missing information determination step) using the text generation unit 115. If the control unit 11 determines that the information is missing, it proceeds to step S13, and if not, it proceeds to step S16.

[0121] The determination in this step is realized, for example, by a procedure of determining whether or not information necessary for specifying a task is lacking based on a plurality of pieces of task information and user input data. This procedure is realized, for example, by inputting a plurality of pieces of necessary information related to a necessary information database and the above-mentioned user input data into a large-scale language model, and outputting whether or not there is a lack of information.

[0122] [Step S13: Identifying Missing Information] The control unit 11 executes a process of identifying missing information, which is information necessary to identify a proposed business that is not included in the user input data and is therefore missing, by the text generation unit 115 (business proposal missing information identification step). The control unit 11 then proceeds to step S14.

[0123] The identification in this step is realized, for example, by a procedure of identifying the missing information based on a plurality of pieces of business information and user input data. This procedure is realized, for example, by inputting the plurality of pieces of business information including the necessary information related to the necessary information database and the above-mentioned user input data into a large-scale language model, and outputting text indicating that the missing information is identified.

[0124] [Step S14: Generate text requesting provision of missing information] The control unit 11 executes a process of generating text requesting provision of the missing information identified in the previous step by the text generation unit 115 (business proposal missing information generation step). The control unit 11 proceeds to step S15.

[0125] The generation in this step is realized, for example, by a procedure in which a prompt including the missing information identified in the previous step is input to a large-scale language model, and the above-mentioned text is output.

[0126] [Step S15: Transmitting a text requesting the provision of missing information] The control unit 11, in cooperation with the storage unit 13 and the communication unit 14, executes a process of transmitting a text requesting the provision of missing information to the user terminal TU (a step of requesting the provision of missing information in a business proposal). The control unit 11 then proceeds to step S21.

[0127] [Step S16: Generate text proposing a business] The control unit 11 executes a process of generating, by the text generation unit 115, a text proposing a business for which a proposal is requested in the inquiry related to the business proposal determination step, based on a plurality of pieces of business information (business proposal text generation step). The control unit 11 proceeds to step S17.

[0128] The generation in this step is realized, for example, by inputting a prompt including text based on the query related to the business proposal determination step into a large-scale language model and outputting the above-mentioned text. The text based on the query may be, for example, the text of the query itself or text indicating the classification of the query (e.g., the classification output by the large-scale language model (described above)).

[0129] In this case, if there is a back-office task associated with an attribute that matches an attribute included in the input user input data in addition to the plurality of task information, the large-scale language model generates text that suggests the task.Furthermore, if there is a back-office task associated with an attribute that is similar in meaning to an attribute included in the input user input data in addition to the plurality of task information, the large-scale language model generates text that suggests the task.

[0130] The plurality of pieces of business information may be identified based on a required information database and acquired from various databases. Alternatively, the plurality of pieces of business information may be identified by the large-scale language model through autonomous judgment based on the required information database or prior learning and acquired from various databases. In addition, in this step, the text generation unit 115 may be configured to trigger further processing of the large-scale language model. This allows the assistance device 1 to perform continuous review processing through autonomous judgment. This configuration is realized, for example, by a configuration capable of generating a prompt for reviewing a proposal and, when the prompt is generated, causing the large-scale language model to generate text proposing the business based on the plurality of pieces of business information and the prompt.

[0131] [Step S17: Providing Text to User Terminal] The control unit 11, in cooperation with the storage unit 13 and the communication unit 14, executes a process of transmitting the text generated in the previous step to the user terminal TU involved in the user input acquisition execution step (business proposal text transmission step). The control unit 11 then proceeds to step S18.

[0132] The support process preferably includes a series of processes for identifying personnel related to the back-office work proposed by the text generator 115 based on multiple pieces of personnel information, and providing the business information related to the back-office work to the personnel terminal TP used by the personnel. Steps S18 to S20 are an example of this process.

[0133] [Step S18: Determine whether to provide information] The control unit 11 makes the information providing unit 116 available in cooperation with the storage unit 13 and the communication unit 14. Then, the control unit 11 executes a process to determine whether to provide information to the human resource terminal TP using the information providing unit 116 (information provision determination step). If the control unit 11 determines that information should be provided, it proceeds to step S19, and if not, it proceeds to step S21.

[0134] The determination in this step is realized, for example, by a procedure in which it is determined that information should be provided when data indicating a desire to provide information to the human resource is received from the user terminal TU.

[0135] [Step S19: Identify Human Resources] The control unit 11 executes a process in which the information providing unit 116 identifies human resources for the back office operations proposed by the text generating unit 115 based on multiple pieces of human resource information (human resource identification step). The control unit 11 shifts the process to step S20.

[0136] The identification in this step is realized, for example, by a procedure of identifying the above-mentioned personnel by referring to a plurality of pieces of personnel information stored in the personnel information database 132 using the back office work as a key.

[0137] [Step S20: Providing Business Information to Human Resource Terminal] The control unit 11 executes a process in which the information providing unit 116 provides the business information related to the back-office business proposed by the text generating unit 115 to the human resource terminal TP corresponding to the human resource identified in the previous step (information provision execution step). The control unit 11 proceeds to step S21.

[0138] When the support process includes a series of processes for providing business information related to back-office work to the human resources terminal TP, it is preferable that the support process also includes a process for providing advice related to back-office work provided by the human resources who received the business information via the human resources terminal TP to the user terminal TU. This allows the support device 1 to support the user in utilizing the knowledge possessed by the human resources in back-office work.

[0139] As described above, the support process preferably includes a series of steps for determining the classification of an inquiry and providing appropriate support based on the classification. Steps S21 to S30 are an example of a series of steps for determining whether the inquiry is requesting a procedure related to back-office work, and, if the inquiry corresponds to the inquiry, providing a text proposing the procedure or other support.

[0140] [Step S21: Determine whether the query is a query requesting a procedure proposal] The control unit 11 executes a process in which the text generation unit 115 determines whether the user input data determined to include a query in the query determination step corresponds to a query requesting a procedure proposal related to back-office operations (procedure proposal determination step). If the control unit 11 determines that the data corresponds, it proceeds to step S22, and if not, it proceeds to step S31.

[0141] The determination in this step is realized, for example, by a process including a procedure of inputting user input data into a large-scale language model and outputting a query classification. The procedure preferably includes a step of outputting a query classification from the large-scale language model based on the user input data and classification information, which is business information stored in the query classification database 134. This allows the assistance device 1 to perform appropriate classification based on the business information.

[0142] The classification related to this step preferably includes not only a classification as to whether the inquiry is a request for procedure suggestions or not, but also the various classifications described in the section on the inquiry classification database 134. This allows the assistance device 1 to perform not only the classification related to this step but also the various classifications performed in subsequent processing in the same process.

[0143] The classification in this step may be performed by a person (e.g., a user or a human resource), thereby allowing the assistance device 1 to process based on a more appropriate classification.

[0144] If the business information management unit 112 is configured to further manage business information including correspondence between inquiries and information necessary for specifying procedures, the support process preferably includes a series of processes in which the large-scale language model determines whether or not information necessary for specification is lacking based on multiple pieces of business information and user-input data, and if so, identifies the missing information and generates text requesting the provision of the missing information. Steps S22 to S25 are an example of such processes.

[0145] [Step S22: Determine whether necessary information is missing] The control unit 11 executes a process of determining whether information necessary to identify the procedure to be proposed is missing (step of determining missing information for procedure proposal) using the text generation unit 115. If the control unit 11 determines that the information is missing, it proceeds to step S23, and if not, it proceeds to step S26.

[0146] The determination in this step is realized, for example, by a procedure of determining whether or not information necessary for identifying the procedure is insufficient based on a plurality of pieces of business information and user input data. This procedure is realized, for example, by inputting a plurality of pieces of necessary information related to the necessary information database and the above-mentioned user input data into a large-scale language model and outputting whether or not there is a shortage.

[0147] [Step S23: Identifying Missing Information] The control unit 11 executes a process of identifying missing information, which is information necessary for identifying a procedure to be proposed that is not included in the user input data and is therefore missing, by the text generation unit 115 (procedure proposal missing information identifying step). The control unit 11 then proceeds to step S24.

[0148] The identification in this step is realized, for example, by a procedure of identifying the missing information based on a plurality of pieces of business information and user input data. This procedure is realized, for example, by inputting the plurality of pieces of business information including the necessary information related to the necessary information database and the above-mentioned user input data into a large-scale language model and outputting text indicating the missing information.

[0149] [Step S24: Generate text requesting provision of missing information] The control unit 11 executes a process of generating text requesting provision of the missing information identified in the previous step by the text generation unit 115 (procedure proposal missing information generation step). The control unit 11 proceeds to step S25.

[0150] The generation in this step is realized, for example, by a procedure in which a prompt including the missing information identified in the previous step is input to a large-scale language model, and the above-mentioned text is output.

[0151] [Step S25: Transmitting a text requesting the provision of missing information] The control unit 11, in cooperation with the storage unit 13 and the communication unit 14, executes a process of transmitting a text requesting the provision of missing information to the user terminal TU (step of requesting the provision of missing information in a procedure proposal). The control unit 11 then proceeds to step S31.

[0152] [Step S26: Generate text showing procedures] The control unit 11 executes a process in which the text generation unit 115 causes the large-scale language model to classify back-office tasks based on the multiple pieces of task information and user input data, and generates text including procedures for the back-office tasks that belong to the classification based on the multiple pieces of task information and user input data (procedure proposal text generation step). The control unit 11 then proceeds to step S27.

[0153] The generation in this step is realized, for example, by inputting text based on a query related to the procedure proposal determination step, text commanding classification, and a prompt to the large-scale language model that instructs the large-scale language model to refer to the business procedure database 135 based on the classification and generate text indicating a procedure based on the information stored in the database, and then outputting the above-mentioned text.

[0154] The query-based text may be, for example, the text of the query itself, text indicating a classification of the query (e.g., a classification output by a large-scale language model (described above)), etc. If the prompt includes text indicating a classification, the prompt need not include text directing the classification.

[0155] [Step S27: Providing a text indicating the procedure to the user terminal] The control unit 11, in cooperation with the storage unit 13 and the communication unit 14, executes a process of transmitting the text generated in the previous step to the user terminal TU related to the user input acquisition execution step (procedure proposal text transmission step). The control unit 11 proceeds to step S28.

[0156] The support process preferably includes a series of processes for adding tasks related to the above-mentioned procedures to a task list, so that the support device 1 can add back-office tasks for which appropriate procedures have been identified and generated to the task list and support progress management based on the task list.

[0157] The data related to the task preferably includes information for identifying the target (e.g., user, human resource, program, computer) that needs to be notified about the task. This allows the assistance device 1 to notify the corresponding target of the work required for the task at an appropriate time, and to support the appropriate progress of the task.

[0158] [Step S28: Determine Whether to Add a Task] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to make the information providing unit 116 available. Then, the control unit 11 executes a process of determining whether to add the above-mentioned task using the information providing unit 116 (task addition determination step). If the control unit 11 determines that the task should be added, it proceeds to step S29, and if not, it proceeds to step S31.

[0159] The determination in this step is realized, for example, by a procedure in which it is determined that a task is to be added when data indicating a desire for task management is received from the user terminal TU.

[0160] [Step S29: Identify Human Resources] The control unit 11 executes a process of identifying human resources related to the above-mentioned procedure based on multiple pieces of human resource information using the information providing unit 116 (task human resource identification step). The control unit 11 proceeds to step S30.

[0161] The identification in this step is realized, for example, by a procedure for identifying the above-mentioned personnel by referring to a plurality of pieces of personnel information stored in the personnel information database 132 using the back office work related to the task as a key.

[0162] [Step S30: Adding a Task to the Task List] The control unit 11 executes a process of adding the above-mentioned task to the task list (task adding step) using the information providing unit 116. The control unit 11 moves the process to step S31.

[0163] [Step S31: Determine whether to issue a notification regarding a task] The control unit 11 executes a process of determining whether to issue a notification regarding a task added to the above-mentioned task list using the information providing unit 116 (task notification determination step). If the control unit 11 determines that a notification should be issued, it proceeds to step S32, and if not, it proceeds to step S33.

[0164] The determination in this step is realized, for example, by a procedure of determining that notification is to be performed when the requirements for task notification (for example, time, completion of prerequisite work) are met.

[0165] [Step S32: Notify Task] The control unit 11 executes a process of notifying the task determined to be notified in the previous step by the information providing unit 116 (task notification execution step). The control unit 11 moves the process to step S33.

[0166] The notification related to the step preferably includes information related to the task. The information includes, for example, part of the data (e.g., task information, task procedures) related to the task managed by the task information management unit 112. This allows the target of the notification to use information necessary and / or useful for performing the work related to the task to perform the work appropriately.

[0167] The support process preferably includes a series of processes for creating a database of various information related to back-office operations and utilizing the database. Steps S33 to S39 are an example of such processes.

[0168] The "various information" referred to here preferably includes not only information related to companies involved in back-office operations, but also various knowledge related to general back-office operations. Examples of "various knowledge" include rules (e.g., accounting standards), laws (e.g., various tax laws), and practical knowledge (e.g., bookkeeping) related to back-office operations. It is preferable that this knowledge be compiled into a database so that it can be used by both people (e.g., users, human resources) and AI (e.g., large-scale language models).

[0169] [Step S33: Determine whether to acquire talent input data] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to enable the terminal input acquisition unit 113. Then, the control unit 11 executes a process to determine whether to acquire talent input data from the talent terminal TP using the terminal input acquisition unit 113 (talent input acquisition determination step). If the control unit 11 determines that the talent input data should be acquired, it proceeds to step S34, and if not, it proceeds to step S40.

[0170] The procedure for determining whether or not to acquire data in this step is not particularly limited. For example, this procedure includes a procedure for determining that input data is to be acquired when the data is received from the human resource terminal TP.

[0171] [Step S34: Acquire personnel input data] The control unit 11 executes a process of acquiring personnel input data from the personnel terminal TP by the terminal input acquisition unit 113 (personnel input acquisition execution step). The control unit 11 moves the process to step S35.

[0172] [Step S35: Determine whether the data contains an inquiry related to back office operations] The control unit 11 executes a process to determine whether the personnel input data acquired in the previous step contains an inquiry related to back office operations (personnel inquiry determination step). If the control unit 11 determines that the data contains an inquiry, the process proceeds to step S36; if not, the process proceeds to step S38.

[0173] The determination in this step is realized by a process including a procedure of inputting the personnel input data into a regular expression engine, a large-scale language model, or other classifier, and causing the classifier to output a determination result. For example, the support device 1 can input personnel input data accompanied by a prompt asking whether the following text data is an inquiry related to back-office work into the large-scale language model, thereby causing the large-scale language model to function as a classifier that outputs a determination result.

[0174] [Step S36: Generate text including answer] The control unit 11 executes a process of causing the large-scale language model to generate text including an answer to the query related to the previous step, using the text generation unit 115 (answer text generation step). The control unit 11 proceeds to step S37.

[0175] The generation in this step is realized, for example, by configuring a large-scale language model to generate text based on the above-mentioned "knowledge" by referencing business information related to the above-mentioned "knowledge" (e.g., business information stored in the business knowledge database 136), inputting a prompt including text based on the query in the previous step into the large-scale language model, and outputting the above-mentioned text. The text based on the query may be, for example, the text of the query itself or text indicating the classification of the query (e.g., the classification output by the large-scale language model (described above)). This step may also include a procedure of sending the text generated by the large-scale language model to a human resource terminal TP used by the human resource related to the query before sending it to the person who made the query. This allows internal or external human resources to confirm the answer and provide task instructions based on the content. This also allows the human resource to correct any errors in the answer. Therefore, this procedure allows the human resource to compensate for incompleteness in the answer of the large-scale language model.

[0176] [Step S37: Provide Answer to Talent Terminal] The control unit 11, in cooperation with the storage unit 13 and the communication unit 14, executes a process of transmitting the text of the answer generated in the previous step to the talent terminal TP related to the talent input acquisition execution step (answer text transmission step). The control unit 11 proceeds to step S38.

[0177] [Step S38: Determine whether knowledge related to back office operations is included] The control unit 11 executes a process to determine whether the human resource input data acquired in the previous step includes knowledge related to back office operations (human resource knowledge determination step) by the information providing unit 116. If the control unit 11 determines that the knowledge is included, the process proceeds to step S39, and if not, the process proceeds to step S40.

[0178] The determination in this step is realized by a process including a procedure of inputting the human resource input data into a regular expression engine, a large-scale language model, or other classifier, and causing the classifier to output a determination result. For example, the assistance device 1 can input the human resource input data accompanied by a prompt asking whether the following text data contains knowledge related to back-office operations into the large-scale language model, thereby causing the large-scale language model to function as a classifier that outputs a determination result.

[0179] [Step S39: Start management of business information corresponding to knowledge] The control unit 11 executes a process of starting management of business information corresponding to the knowledge determined in the previous step by the business information management unit 112 (human resource knowledge management start step). The control unit 11 proceeds to step S40.

[0180] The start of management in this step is realized by, for example, a procedure of storing the human resource input data itself as knowledge in the business knowledge database 136, a procedure of storing the main part of the human resource input data extracted by the large-scale language model as knowledge in the business knowledge database 136, or other procedures. From the viewpoint of providing users and / or human resources with a means of verifying the knowledge, it is preferable that the start of management in this step includes a procedure of storing the knowledge in the business knowledge database 136 in association with its source.

[0181] In order to support the proper operation of document inspection regardless of the level of knowledge accumulated regarding the inspection of data such as documents, invoices, monthly bookkeeping data, etc., the support process preferably includes a series of processes for inspecting data related to back-office operations. Steps S40 to S42 are an example of such processes.

[0182] [Step S40: Determine whether to inspect data] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to enable the data inspection unit 118. Then, the control unit 11 executes a process to determine whether to inspect data related to back-office operations using the data inspection unit 118 (data inspection and determination step). If it is determined that inspection should be performed, the control unit 11 shifts the process to step S41. If it is not determined that inspection should be performed, the control unit 11 shifts the process to step S43.

[0183] The procedure for determining whether or not to inspect data in the data inspection and determination step is not particularly limited, and includes, for example, a procedure for determining that inspection is to be performed when an instruction to inspect data related to back-office operations is received.

[0184] [Step S41: Inspect Data] The control unit 11 executes processing to inspect the content of the data related to the back-office business that was determined to be inspected in the data inspection and determination step, using the data inspection unit 118 (data inspection step). The control unit 11 moves the processing to step S42. In the data inspection step, the data inspection unit 118 executes processing to output the inspection results of the data related to the above inspection, using a machine learning model that has been pre-trained using training data that includes the content of the data as an explanatory variable and the inspection results of the data as a target variable.

[0185] The machine learning model for the data inspection step is not particularly limited. For example, the machine learning model is a machine learning model for robotic process automation (RPA). The machine learning model for RPA is pre-trained using training data that includes, as explanatory variables, a document with multiple entry fields and the contents of the entry fields, and inspection results based on the presence or absence of each entry, the character type of the entry, the length of the entry, the format of the entry, etc., as objective variables.

[0186] In order to enable testing of similar data in addition to data included in the training data of the pre-training, the machine learning model involved in the data testing step is preferably a large-scale language model. In this case, in order to output appropriate testing results for the content of a wide variety of data written in natural language, the large-scale language model is preferably pre-trained using training data that includes the content of the data as an explanatory variable and the test results of the data as a target variable. This allows the large-scale language model to output appropriate testing results for other content that can be inferred from the content of the data and the corresponding test results, thanks to the emergent capabilities and pre-training inherent in the large-scale language model itself.

[0187] In order to support the inspection of a wide range of data, not limited to documents, it is preferable that the inspection of data relating to the data inspection step includes the inspection of data relating to manual data entry work, such as data recorded monthly from invoices.

[0188] [Step S42: Providing Inspection Results] The control unit 11 executes a process of providing the inspection results related to the data inspection step to the user terminal TU and other terminals by the data inspection unit 118 (data inspection result providing step). The control unit 11 proceeds to step S22.

[0189] In order to support appropriate data creation regardless of the level of knowledge accumulated in data creation, the support process preferably includes a series of processes for generating data related to back-office operations. Steps S43 to S45 are an example of such processes.

[0190] [Step S43: Determine whether to generate data] The control unit 11 cooperates with the storage unit 13 and the communication unit 14 to enable the data generation unit 119. Then, the control unit 11 executes a process to determine whether to generate data related to back-office operations using the data generation unit 119 (data generation determination step). If it is determined that data should be generated, the control unit 11 moves the process to step S44. If it is not determined that data should be generated, the control unit 11 returns the process to step S1 and repeats the processes from step S1 to step S45.

[0191] The procedure for determining whether or not to generate data in the data generation determination step is not particularly limited. For example, the procedure includes a procedure for determining to generate data when an instruction to generate data related to back-office operations is received.

[0192] [Step S44: Generate Data] The control unit 11 executes a process of generating data related to back-office operations, which is determined to be generated in the data generation determination step, by the data generation unit 119 (data generation step). The control unit 11 proceeds to step S45.

[0193] In the data generation step, the data generation unit 119 executes a process of generating the data by processing using a large-scale language model that has been pre-trained with training data that includes back-office operations (e.g., types and related information) as explanatory variables and data corresponding to the back-office operations as objective variables. As a result, the large-scale language model can output appropriate data not only for the data corresponding to the back-office operations included in the training data, but also for other data content for which appropriate data can be inferred from the back-office operations and the corresponding data, due to the emergent capabilities of the large-scale language model itself.

[0194] [Step S45: Providing the Generated Data] The control unit 11 executes a process of providing the data generated in the data generation step to the user terminal TU and other terminals by the data generation unit 119 (generated data provision step). The control unit 11 returns the process to step S1 and repeats the processes from step S1 to step S45.

[0195] In order to provide support for the provision of a wide range of data, not limited to documents, it is preferable that the provision of data related to the generated data provision step be carried out in a manner that automates data entry work that is currently done manually, such as automatic entry of data recorded monthly from invoices.

[0196] [Information sharing step] In order to support information sharing related to communication, workflow, and data exchange between related parties within the user's company, it is preferable that the support process includes an information sharing step in which data provided by internal personnel of the target company is sent to other internal personnel related to the target company or external personnel related to the target company.

[0197] The information linking step is realized, for example, by a procedure in which the control unit 11 cooperates with the storage unit 13 and the communication unit 14 to execute the information linking unit 117, and the information linking unit 117 executes a process of transmitting data provided by an internal human resource of the target company to other internal human resources related to the target company or external human resources related to the target company. In order to realize information linking between internal human resources and external human resources, the information linking step preferably includes a procedure of transmitting the above-mentioned data to a destination selected from candidate destinations including one or more internal human resources and one or more external human resources.

[0198] The data related to the information linking unit 117 provided by the internal personnel of the target company includes data related to communication related to back office operations, data related to workflow related to back office operations, data related to information linking related to data exchange, etc. The internal personnel related to the information linking unit 117 are users who use the support device 1. The external personnel related to the information linking unit 117 include, for example, the operator of the support device 1 or a business outsourcing partner related to back office operations.

[0199] [Regarding Data Acquisition via API] The support process is preferably configured to acquire data by linking with an external service (e.g., an accounting system or other support tool) using an Application Programming Interface (API). This configuration is realized, for example, by the control unit 11 working in cooperation with the storage unit 13 and communication unit 14 to execute an API unit (not shown), which acquires data from an external device via the API associated with the external device. The API unit acquires the necessary data from the external service using a known procedure for acquiring data via an API.

[0200] [Business status analysis step] The support process preferably includes a business status analysis step of generating analysis results of business information managed in the business information management unit 112 in order to visualize, analyze, or report the business status, etc., managed in the business information management unit 112.

[0201] The work situation analysis step is realized, for example, by the control unit 11 working in cooperation with the storage unit 13 and the communication unit 14 to execute a work situation analysis unit (not shown), which generates analysis results of the work information managed in the work information management unit 112. The analysis results of the work information generated by the work situation analysis unit include, for example, a report that visualizes the work situation using charts or the like, a report that analyzes the work situation using statistical processing or processing using a language model, or a report that summarizes the work situation using a large-scale language model.

[0202] [Indexing Step] The support process preferably includes an indexing step of indexing business knowledge. The indexing step is realized by, for example, a procedure for creating a database of business knowledge in a manner that can be referenced by headings, categories, companies, etc.

[0203] [Categorization Step] In order to facilitate the use of knowledge related to similar cases, the support process preferably includes a categorization step of categorizing the business knowledge. In this case, in order to facilitate the use of knowledge related to similar cases or related cases, the categorization step preferably includes a procedure of structuring the business knowledge into a tree structure or the like.

[0204] [BPO Business Registration Step] In order to allow knowledge including outsourced business to be accumulated and referenced, the support process preferably includes a procedure for updating the business knowledge database 136 so that the business knowledge includes knowledge related to the scope of business that has been outsourced (BPO).

[0205] [Contribution Visualization Step] In order to clarify the tasks performed by personnel and visualize their contributions, the support process preferably includes a contribution visualization step in which the task knowledge includes a correspondence between personnel and the back-office tasks performed by the personnel. The contribution visualization step is realized, for example, by a procedure for obtaining input from a person in a position to evaluate the contributions, a procedure for automatically extracting descriptions indicating the contributions from documents or data similar to daily reports, etc. This procedure is realized by a procedure using RPA, a procedure using a large-scale language model, etc.

[0206] [Manualization Step] In order to eliminate the dependency of back office operations on individuals, the support process preferably includes a manualization step in which knowledge related to the back office operations is made into content or a manual. The manualization step is realized, for example, by accumulating knowledge for ambiguity that cannot be addressed by existing manuals, stylizing the knowledge, etc. into a fixed form, manualizing the stylized knowledge, and indexing the manual. The manual and knowledge may be in the form of documents, but more preferably include video formatting. By executing the support process including the manual step, the support device 1 can display the knowledge that has been made into content or a manual and support the back office operations.

[0207] [Chat Step] In order to get an overview of the business by utilizing the views of external experts, it is preferable that the support process includes a chat step that provides chat between internal personnel in charge of back office operations and external experts.

[0208] [Task Management Step] In order to resolve the difficulty in task management due to lack of knowledge, the support process preferably includes a task management step for managing ongoing tasks related to back-office operations and notifying progress, deadlines, and the like.

[0209] [Member Management Step] In order to prevent database contamination by information provided from unreliable sources, such as non-members, and to provide high-quality services and generate revenue, the support process preferably includes a member management step for managing members under corporate contracts, individual members, free members, etc. The member management system is realized by a conventionally known account management system, etc.

[0210] [Funding collaboration step] When generating text containing advice for the target company to obtain credit, advice for proceeding with fundraising, and advice for improving cash flow, it is preferable that the support process provides the financial transactions for the advice, such as making advance payments on behalf of the company for the fees for client operations, as well as the associated receivables management and collaboration and intermediation functions with businesses involved in fundraising.

[0211] Specifically, the support process preferably includes a fundraising coordination step of, for example, managing data related to payments made by the target company, sending instructions related to the payment agency to the agent, and sending a report from the agent to the target company. The fundraising coordination step is realized, for example, by the control unit 11 operating a fundraising coordination unit (not shown) in cooperation with the storage unit 13 and the communication unit 14, and by the fundraising coordination unit managing data related to payments made by the target company, sending instructions related to the payment agency to the agent, and sending a report from the agent to the target company.

[0212] Management of data regarding payments related to the target company is realized, for example, by a series of processes in which, when the above-mentioned advice is accepted, instructions regarding the agent's payment of the fee for the client's work related to the advice are stored in memory unit 13, and data regarding the operator's report regarding the agent's payment is stored in memory unit 13.

[0213] The transmission of instructions for making a payment on behalf of the agent is realized, for example, by a process of sending data to the agent instructing the agent to pay the fee for the client work related to the advice on behalf of the agent if the advice is accepted. The transmission of a report by the agent is realized, for example, by a process of receiving data from the agent reporting that the payment related to the action on behalf of the agent has been made and sending report data based on the data to the target company.

[0214] By including the above-described series of processes in the support process, the support device 1 can provide functions for money transactions, such as making advance payments to companies on behalf of clients for their operations, as well as the associated credit management and collaboration and intermediation with businesses related to fundraising. As a result, the support device 1 can not only provide advice on the user's fundraising and cash flow, but also provide support through concrete means such as acting as a payment agent and collaborating and intermediating with businesses related to fundraising.

[0215] In other words, the support device 1, which executes the support process including the series of processes described above, has the function of collecting and analyzing financial information of the target company with which interactions occur while the support device 1 is carrying out its business, and by determining that the target company needs to raise funds or has the potential to grow if it has the funds, it can help investors make fundraising decisions or contact them regarding negotiations through this device, and can support the target company by acting as a hub that acts as an intermediary between the investor and the target company.

[0216] Therefore, the support device 1 that executes the support process including the above-mentioned series of processes can provide further support to users through specific means regarding fund raising and cash flow, so that users can select and properly operate the necessary back office operations, regardless of the level of knowledge they have accumulated regarding back office operations.

[0217] [Effects of the Support Process] The support device 1 that executes the above support process causes the text generation unit 115 to generate text in the large-scale language model not only for proposals of back-office tasks with matching attributes but also for proposals of back-office tasks with similar attributes (steps S11 to S17) based on the task information, which is knowledge stored in the support device 1. This enables the support device 1 to realize support that makes even greater use of the knowledge stored in the device.

[0218] Furthermore, in the support device 1, the information provider 116 transmits business information to the human resource terminal TP based on the human resource information managed by the human resource information management unit 114 regarding the knowledge of professional personnel, specialized personnel, and other human resources, which is knowledge outside the device, and supports the human resources in utilizing their knowledge in the back office work of the user (steps S18 to S20). This enables the support device 1 to realize support that makes even greater use of knowledge outside the device.

[0219] Therefore, the support device 1 that executes the above support processing can provide an automatic processing means that supports the introduction and improvement of back office operations by utilizing knowledge inside and outside the device.

[0220] In one example aspect of the present invention, the business information management unit 112 may be configured to further manage business information including correspondence between inquiries related to back office business and the classification of the inquiries, as described in the section on inquiry classification database 134.

[0221] In this configuration, when the user input data includes an inquiry related to a back-office business (step S11: Yes), the text generation unit 115 can be configured to cause the large-scale language model to classify the inquiry based on multiple pieces of business information and the user input data, and generate text that suggests a back-office business to the user based on the classification (steps S11 to S17). Furthermore, for this classification, the support device 1 can be configured to use classification by people (e.g., users, personnel).

[0222] As a result, the support device 1 can provide a function in which humans and AI (large-scale language models) can interpret the types of "requests" and "questions / issues" received from users based on the database managed by the business information management unit 112. Therefore, the support device 1 can further support the introduction and improvement of back-office operations that utilize knowledge inside and outside the device.

[0223] In one example aspect of the present invention, the business information management unit 112 may be configured to further manage business information including correspondence between business classifications related to back office business and the procedures of back office business belonging to those business classifications, as described in the section on business procedure database 135.

[0224] In this configuration, the text generation unit 115 can be configured to cause the large-scale language model to classify back office tasks based on multiple pieces of task information and user input data, and to generate text including procedures for back office tasks that belong to the task classification related to the classification based on the multiple pieces of task information and user input data.

[0225] This allows the support device 1 to provide a function that interprets the type of "request" or "question / problem" received from the user and determines how the request should be handled (e.g., order, deadline, and resolution point). Therefore, the support device 1 can further support the introduction and improvement of back-office operations that utilize knowledge inside and outside the device.

[0226] In one example aspect of the present invention, the business information management unit 112 may be configured to further manage business information including correspondence between inquiries and information required to identify the procedure, as described in the section on required information database.

[0227] In this configuration, the text generation unit 115 can be configured to cause the large-scale language model to determine whether or not information required for identification is missing based on multiple pieces of business information and user input data, and if so, to identify the missing information and generate text requesting the provision of the missing information.

[0228] This allows the support device 1 to provide a function for extracting information through dialogue with the customer (user) when the information required for the above-mentioned classification or other judgments is insufficient. Therefore, the support device 1 can further support the introduction and improvement of back-office operations that utilize knowledge inside and outside the device.

[0229] In one embodiment of the present invention, the business information management unit 112 may be configured to further manage business information including knowledge related to back office business, as described in the business knowledge database 136 .

[0230] In this configuration, the terminal input acquisition unit 113 is configured to acquire human resource input data entered into the human resource terminal TP, and the text generation unit 115 is configured to cause the large-scale language model to generate text that conforms to the knowledge by referring to business information related to the knowledge, and can be configured to cause the large-scale language model to generate text that indicates an answer that conforms to the knowledge when the human resource input data includes an inquiry related to back-office work.

[0231] As a result, the support device 1 can store not only information related to the target company but also information ("knowledge") such as rules (accounting standards), laws (tax laws), and practical knowledge (bookkeeping) related to general back offices in a database that can be interpreted by humans and AI (large-scale language models), and can perform processing using this information. For example, the support device 1 can provide a function for proposing the above-mentioned back office operations or procedures based on information in the database that can be interpreted by AI in addition to conventional information.

[0232] In one example of the aspect of the present invention, the support device 1 further includes a public information acquisition unit 111 that acquires public information on the Internet related to the above-mentioned "knowledge" and causes the business information management unit 112 to start managing business information including knowledge based on the public information, and the business information management unit 112 can be configured to manage business information based on the knowledge content included in the human resources input data.

[0233] This allows the support device 1 to provide a function of collecting information required for the database creation from public information and human resources, and creating a database with the data being related to each other.

[0234] As described above, each of the various exemplary embodiments can provide an automatic processing means that supports the introduction and improvement of back office operations by utilizing knowledge inside and outside the device, with its own unique configuration.

[0235] <Example of Use of the Support Device 1 of the Present Embodiment> The following is an example of use of the support device 1 of the present embodiment.

[0236] [Acquisition of Public Information] The support device 1 acquires public information at the timing instructed by the administrator or at predetermined regular timing, stores it in the business knowledge database 136, and starts management by the business information management unit 112.

[0237] [Confirming Necessary Back-Office Operations] The user sends a command to the support device 1 via the terminal T to propose back-office operations with the company's attributes attached. The support device 1 refers to the attribute information database 131, searches for back-office operations that match or are similar to the attributes, and proposes them. The user selects the necessary back-office operations based on the proposed back-office operations.

[0238] Furthermore, when proposing the back office business, the support device 1 refers to the business knowledge database 136 via a large-scale language model and makes a proposal based on the knowledge stored in the database. If the information required to identify the back office business is insufficient, the support device 1 determines this by referring to the necessary information database and transmits a text message to the user terminal TU requesting the provision of the missing information. The user provides the missing information and requests a new proposal for the back office business.

[0239] [Proposal of back office work procedures] A user sends a command to inquire about back office work procedures, along with company attributes, to the support device 1 via the user terminal TU. The support device 1 classifies the inquiry by referring to the inquiry classification database 134, and searches for and proposes back office work procedures by referring to the work procedure database 135 using this classification. The user performs the necessary back office work by referring to the proposed procedures.

[0240] Furthermore, when proposing the procedure, the assistance device 1 refers to the business knowledge database 136 via a large-scale language model and makes a proposal based on the knowledge stored in the database. If the information required to identify the procedure is insufficient, the assistance device 1 determines this by referring to the necessary information database and transmits a text message to the user terminal TU requesting the provision of the missing information. The user provides the missing information and requests a new procedure proposal.

[0241] [Document inspection / draft] A user instructs the assistance device 1 to inspect a document related to back office operations via the user terminal TU. The assistance device 1 uses a machine learning model to provide the inspection results of the document. The user uses the inspection results to improve the document. The user also instructs the assistance device 1 to generate a document related to back office operations via the user terminal TU. The assistance device 1 generates and provides the document using a large-scale language model. The user uses the document as a draft to efficiently create the document.

[0242] <Note> It should be noted that within the scope of the concept of the present invention, a person skilled in the art may conceive of various modifications and alterations. Therefore, it is understood that such modifications and alterations fall within the scope of the present invention. For example, even if a person skilled in the art appropriately adds or deletes components or modifies the design of the above-described embodiment, or adds or omits steps or modifies conditions, such modifications are also included within the scope of the present invention as long as they maintain the gist of the present invention. [Explanation of symbols]

[0243] S System 1 Support equipment 11 Control section 111 Public Information Acquisition Department 112 Business Information Management Department 113 Terminal input acquisition unit 114 Human Resources Information Management Department 115 Text Generation Unit 116 Information Provision Department 117 Information and Communications Department 118 Data Inspection Department 119 Data Generation Unit 13 Storage section 131 Attribute Information Database 132 Human Resources Information Database 133 Advice Information Database 134 Inquiry Classification Database 135 Business Procedure Database 136 Business Knowledge Database 14 Communications Department 2 Server N Network TU user terminal TP Human Resources Terminal

Claims

1. a business information management department that manages business information related to the company's back office operations; a human resources information management unit that manages human resources information that associates back office operations with human resources related to the back office operations; a terminal input acquisition unit that acquires user input data input to a user terminal used by a user; a text generation unit that causes a large-scale language model to generate text proposing back-office tasks based on a plurality of pieces of task information; an information providing unit that identifies human resources related to the back office work proposed by the text generating unit based on the plurality of pieces of human resources information, and provides work information related to the back office work to a human resources terminal used by the human resources; Equipped with the business information management unit is configured to manage business information relating to company attributes including company size and business information relating to company attributes including company fiscal month; The human resources information management unit manages human resources information related to professional personnel and specialized personnel, the terminal input acquisition unit is configured to acquire user input data including the attributes; The text generation unit The method is configured to cause a large-scale language model to generate text suggesting back-office operations associated with attributes that match attributes included in the user input data, and generate text suggesting back-office operations associated with attributes that are similar in meaning to attributes included in the user input data, configured to cause the large-scale language model to generate prompts that cause the large-scale language model to consider further suggestions; and and causing a large-scale language model to generate the text based on a plurality of the task information and the prompt. A device to support back office operations.

2. the business information management unit is configured to further manage business information including a correspondence relationship between an inquiry related to a back office business and a classification of the inquiry; the text generation unit is configured to, when the user input data includes an inquiry related to a back office business, cause the large-scale language model to classify the inquiry based on the plurality of pieces of business information and the user input data, and generate text suggesting a back office business to the user based on the classification. The support device according to claim 1 .

3. the business information management unit is configured to further manage business information including a correspondence relationship between a business classification related to a back office business and a procedure of the back office business belonging to the business classification; the text generation unit is configured to cause the large-scale language model to classify back office tasks based on the plurality of task information and the user input data, and to generate text including procedures for back office tasks belonging to a task classification related to the classification based on the plurality of task information and the user input data. The support device according to claim 2 .

4. the business information management unit is configured to further manage business information including a correspondence relationship between the inquiry and information necessary for identifying the procedure; the text generation unit is configured to cause the large-scale language model to determine whether or not information necessary for the identification is lacking based on the plurality of pieces of business information and the user input data, and if so, to identify the missing information and generate text requesting the provision of the missing information. The support device according to claim 3 .

5. the business information management unit is configured to further manage business information including knowledge related to back office business; the terminal input acquisition unit is configured to acquire personnel input data input to a personnel terminal; the text generation unit is configured to cause a large-scale language model to generate text in accordance with the knowledge by referring to business information related to the knowledge, and is configured to cause the large-scale language model to generate text indicating an answer in accordance with the knowledge when the human resources input data includes an inquiry related to a back-office business. The support device according to claim 1 .

6. a public information acquisition unit that acquires public information on the Internet related to the knowledge and causes the business information management unit to start managing business information including the knowledge based on the public information; the business information management unit manages business information based on knowledge content included in the human resource input data; The support device according to claim 5.

Images