Business support device and program

Abstract

A business support device and program are provided that can reduce the time and effort required of participants in confirming syndicated loan contracts. [Solution] The business support device includes an acquisition unit that acquires syndicated loan contract data regarding syndicated loans created by arrangers who form syndicates for fundraising companies; an identification unit that identifies the arranger based on the syndicated loan contract data; a generation unit that generates confirmation result information for the syndicated loan contract data using one learning model corresponding to the financial institution that is the arranger identified by the identification unit from among multiple learning models generated for each financial institution that can become an arranger and machine-learned using training data including syndicated loan contract data created by the financial institutions and confirmation results of the syndicated loan contract data; and an output unit that outputs the confirmation result information.

Description

[Technical Field]

[0001] The embodiments disclosed in the present specification and drawings relate to a business support device and a program. [Background technology]

[0002] Traditionally, syndicated loans have been conducted in which multiple financial institutions form a syndicate (a group of cooperating lenders) and provide credit under the same conditions based on the same loan agreement. First, in a syndicated loan, the company raising funds (hereinafter referred to as the fund-raising company) receives the syndicate formation conditions from the arranger. Then, once the fund-raising company agrees to the syndicate formation conditions, it entrusts the arranger with the formation of the syndicate.

[0003] Next, the arranger, as the lead manager of the syndicate, considers loan conditions, invites financial institutions, prepares contracts, etc., and coordinates between the company raising funds and the syndicate. The participants (lenders without a special role), which are financial institutions that join the syndicate, then review the contents of the syndicated loan agreement, which is the loan agreement prepared and adjusted by the arranger, and consider whether to sign it.

[0004] In such syndicated loan agreements, the matters that participants must confirm are generally agreed upon to a certain extent. However, the structure and wording of syndicated loan agreements may differ depending on the arranger. Furthermore, due to the wide range of contract contents, syndicated loan agreements may contain an enormous amount of information. For this reason, even if the matters that must be confirmed are generally agreed upon to a certain extent, it is time-consuming for participants to search for the matters that must be confirmed in the syndicated loan agreement and confirm the details of those matters. Therefore, it is desirable to reduce the time and effort required for participants to confirm syndicated loan agreements. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2006-285589 Summary of the Invention [Problem to be solved by the invention]

[0006] An object of the present invention is to provide a business support device and program that can reduce the effort required of participants in confirming syndicated loan contracts. [Means for solving the problem]

[0007] The business support device according to the present invention includes: an acquisition unit that acquires syndicated loan contract data, which is contract data relating to a syndicated loan prepared by an arranger who organizes a syndicate for a company raising funds; an identification unit that identifies the arranger based on the syndicated loan agreement data; a generation unit that generates confirmation result information indicating the confirmation result of the syndicated loan contract data using one learning model corresponding to the financial institution that is the arranger identified by the identification unit from among the plurality of learning models for confirming the syndicated loan contract data, the learning models being generated for each financial institution that can be the arranger, and machine-learned using training data including the syndicated loan contract data created by the financial institution and the confirmation result of the syndicated loan contract data; an output unit that outputs the confirmation result information; Equipped with. [Effects of the Invention]

[0008] According to the present invention, it is possible to provide a business support device and a program that can reduce the effort required of participants in checking syndicated loan contracts. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram illustrating an example of a configuration of a business assistance device according to an embodiment. [Figure 2] FIG. 10 is a diagram showing an example of syndicated loan contract data. [Figure 3] FIG. 3 is a diagram showing a part of the contract contents included in the syndicated loan contract data shown in FIG. 2. [Figure 4] FIG. 10 is a diagram illustrating an example of checklist data. [Figure 5] FIG. 5 is a diagram showing some of the check items included in the checklist data shown in FIG. [Figure 6] FIG. 10 is a diagram illustrating an example of training data for a learning model. [Figure 7] FIG. 10 is a diagram illustrating an example of a syndicated loan agreement confirmation process according to an embodiment. [Figure 8] FIG. 10 is a diagram illustrating an example of a home screen. [Figure 9] FIG. 10 is a diagram illustrating an example of a processing execution screen. [Figure 10] FIG. 10 is a diagram illustrating an example of a processing execution screen on which a target model is selected. [Figure 11] FIG. 10 is a diagram illustrating an example of a file selection screen. [Figure 12] FIG. 10 is a diagram showing an example of a processing execution screen on which syndicated loan contract data has been uploaded. [Figure 13] FIG. 10 is a diagram illustrating an example of a destination folder designation screen. [Figure 14] FIG. 10 is a diagram showing an example of a screen indicating that a file is being analyzed. [Figure 15] FIG. 10 is a diagram illustrating an example of a processing completion message screen. [Figure 16] FIG. 10 is a diagram illustrating an example of a destination folder output screen. [Figure 17] FIG. 4 is a diagram showing an example of checklist data generated based on the syndicated loan contract data shown in FIGS. 2 and 3. [Figure 18]FIG. 18 is a diagram showing some of the check items included in the generated checklist data shown in FIG. 17. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0011] A business support device 1 according to this embodiment will be described with reference to FIG. 1. FIG. 1 is a block diagram showing an example of the configuration of the business support device 1 according to one embodiment. The business support device 1 is a device for supporting business related to syndicated loan contract data. The syndicated loan contract data is contract data related to a syndicated loan prepared by an arranger who forms a syndicate for a company raising funds. The syndicate is formed by multiple financial institutions. The financial institutions are, for example, banks. Note that the financial institutions are not limited to banks, but may also include insurance companies, securities companies, trust companies, etc. In other words, the syndicate is not limited to banks, but may also include insurance companies, securities companies, trust companies, etc. Forms of syndicated loans include, for example, commitment lines, term loans, and committed term loans.

[0012] The business support device 1 is, for example, an on-premise type. However, the business support device 1 may also be a cloud type. When the business support device 1 is a cloud type, the business support device 1 may provide the functions and processes described below, for example, in the form of SaaS (Software as a Service) or cloud computing. Furthermore, the business support device 1 is used, for example, by a syndicated loan officer at a financial institution. As shown in FIG. 1, the business support device 1 includes a control unit 10, a communication unit 20, an operation input unit 30, a display unit 40, and a memory unit 50.

[0013] The control unit 10 includes an acquisition unit 11, an identification unit 12, a generation unit 13, an output unit 14, a determination unit 15, and a notification unit 16. Each unit will be described in detail later. In this embodiment, each unit of the control unit 10 is realized by the processor of the business assistance device 1 executing a predetermined program. Note that at least one of the units of the control unit 10 may be configured by hardware. Furthermore, the function of the control unit 10 may be realized by cooperation between multiple business assistance devices that can communicate with each other.

[0014] The communication unit 20 is an interface for transmitting and receiving data and information to and from other business support devices (such as a web server of a financial institution) via wireless or wired communication. The communication method and standard used by the communication unit 20 are not particularly limited.

[0015] The operation input unit 30 is an interface for the user of the business support device 1 to input various information, and is composed of information input means such as a keyboard, a mouse, a touch panel, etc. The user can, for example, select a target model or upload syndicated loan contract data via the operation input unit 30.

[0016] The syndicated loan contract data CD according to this embodiment will be described with reference to Figures 2 and 3. Figure 2 is a diagram showing an example of the syndicated loan contract data CD. Figure 3 is a diagram showing part of the contract content included in the syndicated loan contract data shown in Figure 2. In the example shown in Figure 2, the syndicated loan contract data CD is syndicated loan contract data relating to a commitment line, which is one form of syndicated loan. As shown in Figure 2, the syndicated loan contract data CD is made up of multiple document data. Each of the multiple document data includes information relating to the contract content, such as information on the form of the syndicated loan, information on the borrower's name, information on the arranger's name, information on the lender's name, and information on the contract terms.

[0017] In the example shown in Figure 3(a), the syndicated loan contract data CD according to this embodiment includes "commitment line agreement" as information C1 on the form of the syndicated loan, which is one piece of information on the contract content, and includes "XX Bank, Ltd." as information C2 on the arranger's name, which is one piece of information on the contract content. Also, in the example shown in Figure 3(b), the syndicated loan contract data CD according to this embodiment includes a clause on net assets in the clause on financial status, "The borrower shall close each fiscal year at the end of the fiscal year and...", as information C3 on one contract clause out of multiple contract clauses, which is one piece of information on the contract content.

[0018] The display unit 40 has a video display means such as a liquid crystal display or an organic EL display, and displays a screen for uploading the syndicated loan contract data CD, checklist data, and the like.

[0019] The checklist data CL according to this embodiment will be described with reference to Figures 4 and 5. Figure 4 is a diagram showing an example of the checklist data CL. Figure 5 is a diagram showing some of the check items included in the checklist data CL shown in Figure 4. As shown in Figure 4, the checklist data CL stores, in association with each other, an item number, a check item, a check content, a check judgment result, an alert, a reference document citation, a reference text category, a text referred to at the time of judgment, a sales office judgment, and a sales office opinion.

[0020] The check items are items to be confirmed in the syndicated loan contract data CD. The check items are indicated, for example, by item names. In the example shown in FIG. 4, the check items include items related to clauses regarding financial status, items related to collateral, and items related to assets. These check items correspond to the confirmation items in this embodiment.

[0021] 4 include items related to financial condition clauses, items related to security clauses, and items related to assets, but the check items are not limited to items related to financial condition clauses, items related to security clauses, and items related to assets. In other words, the content of the check items is arbitrary, and the check items may include at least one of items related to financial condition clauses, items related to security clauses, and items related to assets, or may include items other than items related to financial condition clauses, items related to security clauses, and items related to assets.

[0022] The check number is a number associated with the check content in the check item. As shown in Figure 4, taking the check number for the check item "Clause on financial status" as an example, the numbers 1 to 3 are assigned to each check content included in the check item. In other words, the check item "Clause on financial status" is associated with three check contents.

[0023] The check content is the specific content of the check item. One or more check contents are associated with a check item. In the example shown in Figures 4 and 5, taking the check content related to the check item of the clause regarding financial status as an example, the check content is "A clause regarding net assets is specified" and "A clause regarding profits is specified."

[0024] Note that the check contents for the check items are not limited to the examples shown in Figures 4 and 5, and the check contents for the check items may include contents other than the examples shown in Figures 4 and 5. Furthermore, the check contents in the checklist data CL may include, in addition to the contents checked in the syndicated loan contract data CD, reference contents that are contents that were referred to when checking the check contents in the syndicated loan contract data CD. In other words, it may be determined whether the reference contents are included in the syndicated loan contract data CD.

[0025] The check judgment result is information indicating whether the syndicated loan contract data CD includes the check content. For example, if the syndicated loan contract data CD includes the check content, the check judgment result includes the symbol "○", and if the syndicated loan contract data CD does not include the check content, the check judgment result includes the symbol "×". Note that the check judgment result corresponding to the reference content in the checklist data CL may include the reference information used to derive the check judgment result. This check judgment result corresponds to the judgment result of the confirmation item in this embodiment.

[0026] The alert holds alert information related to the alert when the syndicated loan contract data does not contain any content corresponding to a check item, i.e., when the syndicated loan contract data CD does not contain any check content. The alert information holds wording information such as, for example, "The above conditions may not be met" or "No sentence corresponding to the item was found. It may not be specified." Note that the alert information is not limited to wording information. In other words, the content of the alert information is arbitrary. When wording information is held in this alert, the entire row of the check item containing the wording information in the alert is colored. This allows the user to easily confirm the check item containing the wording information in the alert. Note that the alert is not limited to coloring the entire row of the check item containing the wording information in the alert; only the row of the check content containing the wording information in the alert may be colored.

[0027] The reference location is location information that indicates the location of a check item in the syndicated loan contract data CD when the syndicated loan contract data CD contains content that corresponds to the check item, i.e., when the syndicated loan contract data CD includes check content. The reference location is indicated, for example, by the article number in which the check content related to the check item is stated. Note that the reference location may be indicated, for example, by the page number and paragraph number instead of or in addition to the article number.

[0028] The reference text category is information about the category in which the text describing the check contents in the syndicated loan contract data CD is written. For example, the reference text category holds the category name in the syndicated loan contract data CD.

[0029] The sentence referenced at the time of determination holds information about at least a part of the sentence written in the reference sentence description portion.

[0030] The sales department judgment and sales department opinion hold information related to the judgment result of the user. The sales department judgment and sales department opinion are input by the user via the operation input unit 30.

[0031] Note that the checklist data CL shown in FIG. 4 is configured to store, in association with each other, item numbers, check items, check details, check judgment results, alerts, reference document locations, reference text categories, texts referenced during the judgment, sales office judgments, and sales office opinions. However, the contents of the checklist data CL are not limited to item numbers, check items, check details, check judgment results, alerts, reference document locations, reference text categories, texts referenced during the judgment, sales office judgments, and sales office opinions. In other words, the contents of the checklist data CL are arbitrary. For example, the checklist data CL may store, in association with each other, only check items and check judgment results. In other words, the checklist data CL is required to include at least check items and judgment results for the check items in the syndicated loan contract data CD. Since the checklist data CL includes at least check items and the judgment results of the check items in the syndicated loan contract data CD, the participant can at least confirm whether or not the items that need to be confirmed are described in the syndicated loan contract data.

[0032] Furthermore, the checklist data CL may, for example, hold only the check items, check contents, and check judgment results in association with each other, or may hold only the check items, check contents, check judgment results, alerts, reference literature locations, reference text categories, and texts referenced at the time of judgment in association with each other, or may include content other than the item number, check items, check contents, check judgment results, alerts, reference literature locations, reference text categories, texts referenced at the time of judgment, sales office judgment, and sales office opinion.

[0033] The storage unit 50 is configured with a semiconductor memory and / or a hard disk drive. The storage unit 50 stores, for example, a program for processing executed by the control unit 10 and data required for the processing. As shown in FIG. 1 , the storage unit 50 also stores multiple learning models. The multiple learning models are generated for each potential arranger financial institution and trained by machine learning using training data including syndicated loan contract data CD prepared by the financial institution and check results for the syndicated loan contract data CD. The learning method for machine learning may be any of supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, and deep learning, or a combination of these learning methods. As shown in FIG. 1 , the storage unit 50 according to this embodiment stores multiple learning models, including a first learning model 51, a second learning model 52, and a third learning model 53. The check results correspond to the confirmation results in this embodiment.

[0034] The first learning model 51 is a learning model corresponding to the first financial institution. Specifically, the first learning model 51 is a machine learning model for verifying the syndicated loan contract data CD created by the first financial institution, which is machine-learned using training data including the syndicated loan contract data CD created by the first financial institution and the check results of the syndicated loan contract data CD.

[0035] The second learning model 52 is a learning model corresponding to the second financial institution. Specifically, the second learning model 52 is a machine learning model for verifying the syndicated loan contract data CD created by the second financial institution, which is machine-learned using training data including the syndicated loan contract data CD created by the second financial institution and the check results of the syndicated loan contract data CD.

[0036] The third learning model 53 is a learning model corresponding to a third financial institution. Specifically, the third learning model 53 is a machine learning model for checking the syndicated loan contract data CD created by the third financial institution, which is machine-learned using training data including the syndicated loan contract data CD created by the third financial institution and the check results of the syndicated loan contract data CD.

[0037] The checklist data CL generated in each of the multiple learning models includes the same check items. Specifically, in this embodiment, the checklist data CL generated in each of the first learning model 51, the second learning model 52, and the third learning model includes, as the same check items, at least an item related to a clause regarding financial status, an item related to a clause regarding collateral, and an item related to a clause regarding assets, because the items that participants must confirm in the syndicated loan contract data CD are generalized to a certain extent.

[0038] In this embodiment, the check items included in the checklist data CL generated in each of the first learning model 51, the second learning model 52, and the third learning model are all the same, but this is not limited to this. For example, some of the check items included in the checklist data CL generated in each of the first learning model 51, the second learning model 52, and the third learning model may be the same.

[0039] Furthermore, because the checklist data CL generated in each of the multiple learning models includes the same check items, the check items confirmed in the check results of the syndicated loan contract data CD included in the training data used when generating each of the multiple learning models include the same check items. In other words, training data including the check results of the syndicated loan contract data CD confirmed by the same check items is used when generating the multiple learning models.

[0040] 1, three learning models are stored in the storage unit 50 as the multiple learning models, but the number of learning models stored in the storage unit 50 is not limited to three. That is, the number of learning models stored in the storage unit 50 is arbitrary, and may be two, four or more.

[0041] FIG. 6 is a diagram illustrating an example of training data for a learning model. The training data TD illustrated in FIG. 6 is training data used in machine learning to generate a machine learning model for verifying syndicated loan contract data CD created by a first financial institution, "XX Bank Co., Ltd.", i.e., a first learning model 51. In the example illustrated in FIG. 6, the training data TD includes n datasets, from a first training dataset TD1 to an nth training dataset. As illustrated in FIG. 6, taking the first training dataset TD1 as an example, the first training dataset TD1 is a dataset including training syndicated loan contract data TCD1 created by the first financial institution, "XX Bank Co., Ltd.", and checklist data TCL1 including the check results for the training syndicated loan contract data TCD1 created by the first financial institution, "XX Bank Co., Ltd." In other words, each of the n datasets is a dataset including training syndicated loan contract data and checklist data including the check results for the training syndicated loan contract data. Then, machine learning is performed using these n data sets to generate a first learning model 51. Similarly, machine learning is performed using training data including training syndicated loan agreement data created by a second financial institution and n data sets including the check results of the training syndicated loan agreement data created by the second financial institution to generate a second learning model 52, and machine learning is performed using training data including training syndicated loan agreement data created by a third financial institution and n data sets including the check results of the training syndicated loan agreement data created by the third financial institution to generate a third learning model 53.

[0042] Next, each part of the control unit 10 will be described in detail.

[0043] The acquisition unit 11 acquires syndicated loan contract data CD. The identification unit 12 identifies an arranger based on the syndicated loan contract data CD.

[0044] The generation unit 13 generates checklist data CL indicating the check results of the syndicated loan contract data CD using one of the multiple learning models that corresponds to the financial institution that is the arranger identified by the identification unit 12. The output unit 14 outputs the checklist data CL. This checklist data CL is an example of confirmation result information in this embodiment.

[0045] The determination unit 15 determines whether or not there is a learning model among the multiple learning models that corresponds to the financial institution that is the arranger identified by the identification unit 12. If there is no learning model that corresponds to the financial institution, the notification unit 16 notifies the user that there is no learning model.

[0046] 7 is a flowchart illustrating syndicated loan contract confirmation processing according to one embodiment. This syndicated loan contract confirmation processing includes outputting a home screen, outputting a processing execution screen, acquiring syndicated loan contract data CD, identifying an arranger, generating checklist data CL, and outputting the checklist data CL. This syndicated loan contract confirmation processing is executed when a selection for business efficiency support is accepted on the home screen.

[0047] As shown in Fig. 7, first, the output unit 14 in the control unit 10 outputs a home screen (step S11). Specifically, the output unit 14 outputs the home screen to the display unit 40. Fig. 8 is a diagram showing an example of the home screen. As shown in Fig. 8, a business efficiency improvement support button B1 is output in a selectable manner on the home screen SC1.

[0048] Next, as shown in Fig. 7, the determination unit 15 determines whether or not a selection of business efficiency improvement support has been accepted (step S13). Specifically, the determination unit 15 determines whether or not a selection of business efficiency improvement support has been accepted by determining whether or not a selection of business efficiency improvement support button B1 output on the home screen SC1 shown in Fig. 8 has been accepted from the user via the operation input unit 30. Then, if the selection of business efficiency improvement support has not been accepted in step S13 (step S13: NO), the process of step S13 is repeated and the unit waits until the selection of business efficiency improvement support is accepted.

[0049] On the other hand, if the selection of business efficiency improvement support is accepted in step S13 (step S13: YES), the output unit 14 outputs a processing execution screen (step S15). Specifically, the output unit 14 outputs a processing execution screen for executing a process to confirm the syndicated loan contract data CD to the display unit 40.

[0050] Fig. 9 is a diagram showing an example of a processing execution screen. In the example shown in Fig. 9, the processing execution screen SC2 includes a model selection button B2 for selecting a target model, a reception window W1, a file selection button B3, information IF1 regarding uploaded files, checklist information IF2 regarding the contents of the checklist, and a processing start button B4.

[0051] The model selection button B2 on the process execution screen SC2 is a button for selecting multiple learning models related to business efficiency improvement support. When the model selection button B2 is selected, a list of selectable models related to business efficiency improvement support is output on the process execution screen SC2.

[0052] The reception window W1 on the processing execution screen SC2 accepts file uploads. Specifically, the user uploads the syndicated loan contract data CD by dragging and dropping the syndicated loan contract data CD into the reception window W1 via the operation input unit 30, or by selecting the file selection button B3 displayed in the reception window W1 and selecting the syndicated loan contract data CD to be uploaded from the storage unit 50.

[0053] The information IF1 relating to the uploaded file on the processing execution screen SC2 is, for example, the file path of the uploaded syndicated loan contract data CD. When the acquisition unit 11 acquires the syndicated loan contract data CD by accepting a file upload via the reception window W1, the output unit 14 outputs information relating to the uploaded file.

[0054] Furthermore, the checklist information IF2 on the processing execution screen SC2 stores check items and confirmation items for the check items in association with each other. That is, the output unit 14 outputs a list of check items to the display unit 40. In the example shown in FIG. 9, the processing execution screen SC2 outputs, as check items related to the checklist information IF2, clauses related to financial status, clauses related to security, and clauses related to assets, and, as confirmation items related to the checklist information IF2, specific check contents for the clauses related to financial status, clauses related to security, and clauses related to assets. As described above, since the check items of the checklist data CL generated by each of the multiple learning models according to this embodiment are the same, it is not necessary to output checklist information IF2 corresponding to each of the multiple learning models.

[0055] The process start button B4 is a button for starting a process related to business efficiency improvement support.

[0056] 7, the determination unit 15 determines whether or not selection of the model selection button B2 has been accepted (step S17). Specifically, the determination unit 15 determines whether or not selection of the model selection button B2 has been accepted by determining whether or not selection of the model selection button B2 has been accepted from the user via the operation input unit 30. Then, in step S17, if selection of the model selection button B2 has not been accepted (step S17: NO), the determination unit 15 waits and repeats the process of step S17 until selection of a model is accepted.

[0057] On the other hand, if the selection of the model selection button B2 is accepted in step S17 (step S17: YES), the output unit 14 outputs a target model list (step S19). Specifically, the output unit 14 outputs a pull-down menu including the names of models related to business efficiency improvement support as the target model list.

[0058] 7, the determination unit 15 determines whether or not a contract check has been received (step S21). Specifically, the determination unit 15 determines whether or not a contract check has been received by determining whether or not the user has selected contract check in the target model list via the operation input unit 30. If the contract check has not been received (step S21: NO), the determination unit 15 waits and repeats the process of step S21 until the contract check is received.

[0059] On the other hand, if a contract check is accepted in step S21 (step S21: YES), the output unit 14 outputs the name of the target model (step S22). Specifically, in step S21, the output unit 14 outputs the name of the selected target model by superimposing text information of "contract check" on the model selection button B2 for selecting the target model. FIG. 10 is a diagram showing an example of the process execution screen SC2 in which the target model has been selected. As shown in FIG. 10, the text information of "contract check" is superimposed on the model selection button B2.

[0060] Next, as shown in Fig. 7, the determination unit 15 determines whether or not a file upload has been accepted (step S23). Specifically, the determination unit 15 determines whether or not a file upload of the syndicated loan contract data CD has been accepted by the user dragging and dropping the syndicated loan contract data CD onto the reception window W1 via the operation input unit 30, or by the user selecting the file selection button B3 output on the reception window W1 and selecting the syndicated loan contract data CD to be uploaded from the storage unit 50. Then, if the file upload has not been accepted (step S23: NO), the determination unit 15 waits and repeats the processing of step S23 until a file upload is accepted.

[0061] FIG. 11 is a diagram showing an example of a file selection screen. In the example shown in FIG. 11, when a user selects the file selection button B3 displayed in the reception window W1, the output unit 14 outputs a file selection screen SC3 so as to be superimposed on the processing execution screen SC2. Then, the user selects, via the operation input unit 30, a file of syndicated loan contract data CD to be uploaded from among the syndicated loan contract data CD displayed on the file selection screen SC3. In the example shown in FIG. 11, the user selects, via the operation input unit 30, the syndicated loan contract data CD with the file name "aaa." Then, when file selection is complete, the user selects the OK button in the file selection screen SC3 via the operation input unit 30 to end file selection.

[0062] On the other hand, if a file upload is accepted in step S23 (step S23: YES), the acquisition unit 11 acquires the syndicated loan contract data CD (step S25). Specifically, the acquisition unit 11 acquires the syndicated loan contract data CD for which the file upload was accepted in step S23 from the storage unit 50.

[0063] 7, the output unit 14 outputs the file path (step S27). Specifically, the output unit 14 outputs the file path of the syndicated loan contract data CD1 acquired by the acquisition unit 11 in step S25 to the processing execution screen SC2.

[0064] Figure 12 is a diagram showing an example of a processing execution screen on which syndicated loan contract data CD has been uploaded. As shown in Figure 12, because the user selected syndicated loan contract data CD with the file name "aaa" from the file selection screen SC3 shown in Figure 11, the file name "aaa" of the syndicated loan contract data is output as the file path of the syndicated loan contract data CD on the processing execution screen SC2. By outputting the file name in this way, it is possible to prevent confusion over the file to be processed.

[0065] 7, the determination unit 15 determines whether or not selection of the process start button B4 has been accepted (step S29). Specifically, the determination unit 15 determines whether or not selection of the process start button B4 has been accepted by the user via the operation input unit 30. If selection of the process start button B4 has not been accepted (step S29: NO), the determination unit 15 waits and repeats the process of step S21 until selection of the process start button B4 is accepted.

[0066] On the other hand, in step S29, when the selection of the process start button B4 is accepted (step S29: YES), the output unit 14 outputs a destination folder designation screen SC4 (step S31). Specifically, the output unit 14 outputs the destination folder designation screen SC4 for designating a destination folder for the checklist data CL generated by the generation unit 13 to the display unit 40.

[0067] 13 is a diagram showing an example of a destination folder designation screen SC4. As shown in FIG. 13, the user designates a destination folder from the folder name output on the destination folder designation screen SC4 via the operation input unit 30. In the example shown in FIG. 13, the user designates the folder name "100_ZZZ folder" via the operation input unit 30, and the folder with the folder name "100_ZZZ folder" is opened. Then, when the selection of the destination folder is complete, the user selects the OK button on the destination folder designation screen SC4 via the operation input unit 30, thereby completing the designation of the destination folder.

[0068] 7, the determination unit 15 determines whether or not a destination folder has been designated (step S33). Specifically, the determination unit 15 determines whether or not a destination folder has been designated by determining whether or not the user has designated a destination folder on the destination folder designation screen SC4 via the operation input unit 30. If a destination folder designation has not been accepted (step S33: NO), the determination unit 15 waits and repeats the process of step S33 until a destination folder designation is accepted.

[0069] On the other hand, if a destination folder has been specified in step S33 (step S33: YES), the identifying unit 12 identifies the arranger (step S35). Specifically, the identifying unit 12 identifies the arranger by identifying the arranger name included in the syndicated loan contract data CD based on the syndicated loan contract data CD acquired in step S25. More specifically, if the syndicated loan contract data CD shown in Figures 2 and 3 has been acquired in step S25, the identifying unit 12 identifies the arranger name "XX Bank, Ltd." by identifying information related to the arranger name in the syndicated loan contract data CD shown in Figures 2 and 3, thereby identifying the arranger.

[0070] 7, the determination unit 15 determines whether or not a learning model exists (step S37). Specifically, the determination unit 15 determines whether or not a learning model exists corresponding to the financial institution that is the arranger, based on the arranger identified by the identification unit 12 in step S35. More specifically, the determination unit 15 determines whether or not a learning model exists corresponding to the financial institution that is the arranger, among the multiple learning models stored in the storage unit 50.

[0071] Then, if there is no learning model in step S37 (step S37: NO), the notification unit 16 notifies the user that there is no learning model (step S39). Specifically, the notification unit 16 notifies the user that there is no learning model by superimposing on the processing execution screen SC2 and outputting to the display unit 40 a message indicating that a learning model corresponding to the financial institution that is the arranger identified in step S35 is not stored in the memory unit 50. Note that the method of notifying the user that there is no learning model in step S39 is not limited to outputting to the display unit 40. In other words, the method of notifying the user that there is no learning model is arbitrary, and the message indicating that there is no learning model may be output to a speaker (not shown) or a printer (not shown).

[0072] On the other hand, if a learning model is found in step S37 (step S37: YES), the generation unit 13 generates a checklist (step S41). Specifically, the generation unit 13 generates checklist data CL for the syndicated loan contract data CD using one learning model corresponding to the identified financial institution that is the arranger. More specifically, the generation unit 13 generates checklist data CL by inputting the syndicated loan contract data CD into one learning model corresponding to the identified financial institution that is the arranger, which is stored in the memory unit 50.

[0073] Next, as shown in Fig. 7, the output unit 14 outputs a processing completion message (step S43). Specifically, the output unit 14 outputs the processing completion message so as to be superimposed on the processing execution screen SC2.

[0074] Fig. 14 is a diagram showing an example of a screen indicating that a file is being analyzed, and Fig. 15 is a diagram showing an example of a processing completion message screen. As shown in Fig. 14, when the determination unit 15 receives the designation of a destination folder in step S33, the determination unit 15 outputs a screen SC5 indicating that the file is being analyzed to the display unit 40 so as to be superimposed on the processing execution screen SC2 during the execution of the processes of steps S35 to S37 and step S41. Then, as shown in Fig. 15, when the generation of checklist data CL is completed in step S41, the output unit 14 outputs a processing completion screen SC6 to the display unit 40 so as to be superimposed on the screen SC5 indicating that the file is being analyzed. The processing completion screen SC6 shown in Fig. 15 includes a processing completion message and an OK button B5 for inputting confirmation of the processing completion message.

[0075] 7, the output unit 14 outputs the checklist (step S45). Specifically, the output unit 14 outputs the checklist data CL to the destination folder specified in step S33.

[0076] 7, the output unit 14 displays the destination folder (step S47). Specifically, the output unit 14 displays on the display unit 40 the destination folder specified in step S33.

[0077] 16 is a diagram showing an example of a destination folder output screen. As shown in Fig. 16, the output unit 14 outputs a destination folder output screen SC7 including the destination folder specified in step S33 to the display unit 40. In the example shown in Fig. 16, checklist data CL with the file name "aaa checklist" is stored in a folder named "100_ZZZ folder."

[0078] 7, the determination unit 15 determines whether or not the display of the checklist data CL has been accepted (step S49). Specifically, the determination unit 15 determines whether or not the display of the checklist data CL has been accepted by accepting an input operation from the user, via the operation input unit 30, to select the checklist data CL output in step S45 from the destination folder displayed in step S47. If the display of the checklist data CL has not been accepted (step S49: NO), the determination unit 15 waits and repeats the process of step S49 until the display of the checklist data CL is accepted.

[0079] On the other hand, if a request to display the checklist data CL is received in step S49 (step S49: YES), the output unit 14 displays the checklist data CL (step S51). Specifically, the output unit 14 displays the checklist data CL by outputting the checklist data CL received to be displayed in step S49 to the display unit 40.

[0080] The checklist data will be explained in detail using Figures 17 and 18. Figure 17 is a diagram showing an example of checklist data generated based on the syndicated loan contract data CD shown in Figures 2 and 3, and Figure 18 is a diagram showing some of the check items included in the generated checklist data shown in Figure 17. As shown in Figure 17, the generated checklist data CL includes the check judgment result, alert, reference sentence description location, reference sentence category, and the check result for each check item in the sentence referred to during the judgment.

[0081] As shown in Figure 18, taking the check item "Financial Condition Clause" as an example, the check item "Financial Condition Clause" check item "Net assets are specified" includes the check result of "○", the reference text location "Article 21 (Borrower's Commitment) - (5)", the reference text category "Net assets", and the text referenced at the time of the check result "The borrower shall end each fiscal year and...". In other words, the check result was made based on the content of "The borrower shall end each fiscal year and...", which is the content of information C3 regarding contract clause 1 shown in Figure 3, and the syndicated loan contract data CD contains content that satisfies the criterion "Net assets are specified". Note that because the check result is "○", the alert is left blank.

[0082] On the other hand, as shown in Figure 18, for example, the check item "Financial Status Clause" and the check content "Profits are stipulated" include the check result "X", the alert "No relevant text was found for the item. It may not be stipulated.", the reference text category "Profits", and the check result for the text referenced during the assessment. The reference text location and the text referenced during the assessment are blank. In other words, this indicates that the syndicated loan contract data CD in the checklist data does not contain any content that satisfies the "Profits are stipulated" condition. Also, because the alert holds the wording information for the check content "Profits are stipulated," the entire row of the check item "Financial Status Clause," which includes the wording information in the alert, is colored.

[0083] By executing the processing of step S39 or the processing of step S51, the syndicated loan contract confirmation processing is completed.

[0084] As described above, the business support device 1 according to this embodiment acquires syndicated loan contract data, identifies an arranger based on the syndicated loan contract data, generates confirmation result information indicating the confirmation result of the syndicated loan contract data using one learning model corresponding to the identified arranger financial institution out of multiple learning models generated for each potential arranger and used to confirm the syndicated loan contract data CD, and outputs the confirmation result information. This reduces the effort required for a participant to find the items to be confirmed from the syndicated loan contract and confirm the details of the items to be confirmed when confirming the syndicated loan contract data. In other words, it reduces the effort required for a participant to confirm a syndicated loan contract.

[0085] In step S45 of the syndicated loan contract confirmation process described above, the output unit 14 outputs the checklist data CL to the destination folder, but the output destination of the checklist data CL is not limited to the destination folder. In other words, the output destination of the checklist data CL is arbitrary, and the output unit 14 may output the checklist data CL only to the display unit 40, or may output it to both the destination folder and the display unit 40. When outputting the checklist data CL to the display unit 40, steps S47 and S49 in the syndicated loan contract confirmation process do not have to be executed.

[0086] The above describes the embodiments of the present invention. Based on the above description, a person skilled in the art may be able to conceive additional effects and various modifications of the present invention, but the aspects of the present invention are not limited to the above-described embodiments. Various additions, modifications, and partial deletions are possible within the scope of the conceptual idea and spirit of the present invention, which can be derived from the content defined in the claims and their equivalents.

[0087] For example, each part of the control unit 10 may be realized by a processor of the task support device 1 executing a predetermined program and using hardware resources to perform software processing, or may be realized by the implemented hardware itself. When configured by software, a program that realizes at least some of the functions of the control unit 10 may be stored on a recording medium such as a flexible disk or CD-ROM and read and executed by a computer. The recording medium is not limited to removable media such as magnetic disks and optical disks, but may also be fixed recording media such as hard disk drives and memories.

[0088] In addition, a program that realizes at least a part of the functions of the control unit 10 may be distributed via a communication line (including wireless communication) such as the Internet. Furthermore, the program may be encrypted, modulated, or compressed and distributed via a wired line or wireless line such as the Internet, or stored on a recording medium. [Explanation of symbols]

[0089] 1...business support device, 10...control unit, 11...acquisition unit, 12...identification unit, 13...generation unit, 14...output unit, 15...determination unit, 16...notification unit, 20...communication unit, 30...operation input unit, 40...display unit, 50...storage unit, 51...first learning model, 52...second learning model, 53...third learning model

Claims

1. an acquisition unit that acquires syndicated loan contract data, which is contract data regarding syndicated loans prepared by an arranger who organizes a syndicate for a company seeking financing; an identification unit that identifies the arranger based on the syndicated loan agreement data; a generation unit that generates confirmation result information indicating the confirmation result of the syndicated loan contract data using one learning model corresponding to the financial institution that is the arranger identified by the identification unit from among the plurality of learning models for confirming the syndicated loan contract data, the learning models being generated for each financial institution that can be the arranger, and machine learning has been performed using training data including the syndicated loan contract data created by the financial institution and the confirmation result of the syndicated loan contract data; an output unit that outputs the confirmation result information; A business support device comprising:

2. a determination unit that determines whether or not there is a learning model corresponding to the financial institution that is the arranger identified by the identification unit, among the plurality of learning models; a notification unit that notifies a user that there is no learning model corresponding to the financial institution when there is no learning model; The task support device according to claim 1 , further comprising:

3. The business support device according to claim 1 , wherein the generation unit generates, as the confirmation result information, checklist data including at least confirmation items and judgment results of the confirmation items in the syndicated loan contract data.

4. The business support device of claim 3, wherein the checklist data includes location information indicating the location of the confirmation item in the syndicated loan contract data when the syndicated loan contract data contains content that corresponds to the confirmation item.

5. 5. The business support device according to claim 3, wherein the checklist data includes alert information regarding an alert when the syndicated loan contract data does not contain content corresponding to a confirmation item.

6. The business support device according to claim 3 , wherein the checklist data includes reference information used to derive a determination result for the check item.

7. The task assistance device according to claim 1 , further comprising a storage unit that stores the plurality of learning models.

8. The business support device according to claim 1 , wherein the identifying unit identifies the arranger by identifying the arranger name included in the syndicated loan contract data.

9. 4. The business support device according to claim 3, wherein the confirmation items include at least one of an item relating to a clause relating to financial status, an item relating to security, and an item relating to assets.

10. The business support device according to claim 1 , wherein the financial institution is a bank.

11. The business support device according to claim 3 , wherein the output unit outputs the list of the confirmation items to a display unit.

12. The business support device according to claim 3 , wherein the checklist data generated in each of the plurality of learning models includes the same check items.

13. To the computer A process of acquiring syndicated loan contract data, which is contract data regarding a syndicated loan prepared by an arranger who organizes a syndicate for a fundraising company; A process of identifying the arranger based on the syndicated loan agreement data; a process of generating confirmation result information indicating the confirmation result of the syndicated loan contract data using one learning model corresponding to the identified financial institution that is the arranger from among the plurality of learning models for confirming the syndicated loan contract data, the learning models being generated for each financial institution that can be the arranger, and machine-learned using training data including the syndicated loan contract data created by the financial institution and the confirmation result of the syndicated loan contract data; and a process of outputting the confirmation result information.

Images