Business support device, and business support method

Abstract

To support creation of business record which can estimate future reception of orders.SOLUTION: A business support device supports creation of business record by a user. The business support device is configured by a computer having one or more arithmetic devices, one or more memory resources, and one or more storage devices. The arithmetic device comprises a business record correction unit which corrects a business record scheme created by the user, by using a language model. The arithmetic device comprises a business record evaluation unit which evaluates quality of the business record scheme created by the user and the corrected business record scheme. The arithmetic device comprises a business record creation support unit which provides business record creation support information for correcting the business record scheme for the user on the basis of the business record scheme created by the user, a correction result of the corrected business record scheme, and an evaluation result of the business record scheme by the business record evaluation unit.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a sales support device and a sales support method.

Background Art

[0002] In the sales business of proposing and selling products to customers, sales records such as daily reports recording the negotiation results with customers are expected to be utilized in various ways, such as reviewing the work for salespersons, checking the progress for managers of salespersons, and formulating strategies for business operators. However, due to factors such as the vastness and complexity of information, the sufficient utilization of sales records has not advanced.

[0003] Regarding the utilization of sales records, for example, Patent Document 1 describes "an information processing device including a memory and one or more processors, the memory storing a document file created for a case or a group of messages exchanged among a plurality of users regarding the case, the processor extracting information about a target case from the document file or the group of messages stored in the memory, and predicting a conclusion of the target case from the extracted information by an artificial intelligence that has learned to predict a conclusion of the case corresponding to the information in the document file or the group of messages for each case in advance by machine learning."

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] According to the information processing apparatus described in Patent Document 1, it is possible to predict, by machine learning, the conclusion of a case corresponding to a document file for each case or information in a message group, such as whether an agreement is possible or not. However, in order to perform such prediction, it is premised that appropriate information corresponding to the case is accurately and comprehensively recorded, and it does not support the creation of such business records. In particular, it is not possible to support the creation of business records such as daily reports based on past performance such as the trend of the presence or absence of orders.

[0006] The present invention has been made in view of the above points, and an object thereof is to enable the creation of business records that can predict future orders.

Means for Solving the Problems

[0007] This application includes a plurality of means for solving at least a part of the above-described problems. For example, they are as follows.

[0008] In order to solve the above-described problems, a business support apparatus according to an aspect of the present invention is a business support apparatus that supports the creation of business records by a user, and is configured by a computer having one or more arithmetic units, one or more memory resources, and one or more storage devices. The arithmetic unit includes a business record proofreading unit that proofreads a business record draft created by the user using a language model, a business record evaluation unit that evaluates the quality of the business record draft and the proofread business record draft created by the user, and a business record creation support unit that provides the user with business record creation support information for correcting the business record draft based on the proofreading result of the business record draft and the proofread business record draft created by the user and the evaluation result of the business record draft by the business record evaluation unit.

Effects of the Invention

[0009] According to the present invention, it becomes possible to support the creation of business records that can predict future orders.

[0010] Problems, configurations, and effects other than those described above will be clarified by the description of the following embodiments.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Best Mode for Carrying Out the Invention

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In all the drawings for explaining an embodiment, the same members are basically denoted by the same reference numerals, and the repeated description thereof will be omitted. Further, in the following embodiments, the components (including element steps, etc.) are not necessarily essential unless otherwise explicitly stated or considered to be clearly essential in principle. Also, when it is said that "comprising A", "consisting of A", "having A", or "including A", other elements are not excluded unless it is explicitly stated that only that element is involved. Similarly, in the following embodiments, when referring to the shape, positional relationship, etc. of components, etc., unless otherwise explicitly stated or considered not to be so in principle, those substantially approximating or similar to the shape, etc. are included.

[0013] <Configuration example of sales support system 10 including sales support device 100> FIG. 1 shows a configuration example of a sales support system 10 including a sales support device 100 according to an embodiment of the present invention.

[0014] The sales support system 10 assumes a salesperson who conducts sales operations of proposing products to customers and selling them as a user, and supports the user in creating sales records such as daily reports. In this embodiment, insurance is assumed as an example of a product for explanation, but the product is not limited to insurance.

[0015] The sales support system 10 includes a sales support device 100, a CRM (Customer Relationship Management) / daily report system 140, and a language model API (Application Programming Interface) 150 connected via a network 130.

[0016] The business support device 100 includes an input device 101, an output device 102, a communication device 103, a memory 104, an arithmetic device 110, and an auxiliary storage device 120, which are connected via a data bus 105.

[0017] The business support device 100 is realized by a general computer such as a personal computer or a server computer, for example. The input device 101 includes a keyboard, a mouse, a media drive, etc. provided in the computer. The output device 102 includes a display, a speaker, etc. provided in the computer. The communication device 103 includes an Ethernet (trademark) card, a Wi-Fi (trademark) adapter, etc. provided in the computer. The memory 104 includes a memory resource such as a DRAM (Dynamic Random Access Memory) provided in the computer. The arithmetic device 110 includes a processor such as a CPU (Central Processing Unit) provided in the computer. The auxiliary storage device 120 includes a storage such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive).

[0018] The input device 101 receives various inputs from the user. The output device 102 displays, for example, a UI screen 900 (FIG. 9). The communication device 103 communicates various data with a CRM / daily report system 140, a language model API 150, etc. via a network 130.

[0019] The arithmetic device 110 realizes each functional block of a sales record evaluation unit 111, a sales record proofreading unit 112, a sales record creation support unit 113, a sales record evaluation model learning unit 114, a sales record case input support unit 115, and a sales record proofreading result writing unit 116 by executing a predetermined program 1041 stored in the memory 104.

[0020] Note that the program 1041 executed by the arithmetic unit 110 may be stored in the memory 104 in advance, or may be downloaded from a predetermined server or the like via a removable medium (such as a CD-ROM or a flash memory) or a network such as the Internet, and stored in the auxiliary storage device 120 which is a non-temporary storage medium, and read from the auxiliary storage device 120 into the memory 104 when necessary. Therefore, it is desirable that the computer as the sales support device 100 be provided with an interface for reading data from a removable medium.

[0021] Also, the sales support device 100 may be realized by one physical or logical computer, or may be realized by two or more physical or logical computers. Two or more physical or logical computers may be distributed and arranged on the network.

[0022] The sales record evaluation unit 111 uses the learned sales record evaluation model stored in the sales record evaluation model storage unit 125 to evaluate the quality of the sales record (past sales record, sales record draft before proofreading, sales record draft after proofreading), and stores the evaluation result in the sales record evaluation result storage unit 122 of the auxiliary storage device 120. Also, the sales record evaluation unit 111 evaluates the quality of the sales record draft based on at least one of the information on the business performance corresponding to the past sales record similar to the sales record draft before or after proofreading, and the attribute information of the creator of the past sales record similar to the sales record draft before or after proofreading. Here, the quality of the sales record refers to the quality of the text, whether there is an excess or deficiency in the description items according to the proofreading prompt template, whether the date is accurately described, whether the reason for the business result is described, whether the next plan is described, etc.

[0023] The sales record proofreading unit 112 requests the language model to proofread the sales record by sending the proofreading prompt (including the sales record to be proofread) created by the sales record creation support unit 113 to the language model API 150 via the network 130, and obtains the proofreading result of the sales record from the language model.

[0024] The business record creation support unit 113 creates a proofreading prompt (command sentence) for requesting the proofreading of a business record from a language model (described later) using the proofreading prompt template stored in the proofreading prompt template storage unit 124. Also, the business record creation support unit 113 outputs a business record (past business record or business record draft) via the language model API 150 to calculate an embedding vector representing the characteristics of the business record. Furthermore, the business record creation support unit 113 causes the business record evaluation unit 111 to evaluate the proofread business record and presents the evaluation result to the user, thereby supporting the creation (revision) of the business record by the user. Also, the business record creation support unit 113 may create or modify a proofreading prompt template presented to the user when creating a business record draft based on a business record draft and a proofreading example with good evaluation results by the business record evaluation unit 111, and store the created or modified proofreading prompt template in the proofreading prompt template storage unit 124.

[0025] The business record evaluation model learning unit 114 learns the relationship between a business record and the business result for that business record, and creates a business record evaluation model. For example, a business record evaluation model is created by learning past business records, attribute information of the user who created the past business records, customer information, etc. as explanatory variables, and business results (closing rate, number of days required for the contract, cancellation rate, etc.) corresponding to the past business records as objective variables, and the learned business record evaluation model is stored in the business record evaluation model storage unit 125 of the auxiliary storage device 120.

[0026] The business record draft input support unit 115 supports the input of a business record draft by the user through interactive questions.

[0027] The business record proofreading result writing unit 116 associates the business record draft before and after proofreading and the business record revised (e.g., additional information added) by the user based on the proofreading result, and stores it in the business record proofreading result storage unit 123 of the auxiliary storage device 120.

[0028] The auxiliary storage device 120 includes a business record storage unit 121, a business record evaluation result storage unit 122, a business record proofreading result storage unit 123, a proofreading prompt template storage unit 124, and a business record evaluation model storage unit 125.

[0029] The business record storage unit 121 stores business records read from the CRM / daily report system 140. Further, the business record storage unit 121 may store in advance a business record template that serves as a sample of an ideal business record.

[0030] The business record evaluation result storage unit 122 stores the evaluation results of business records. The business record proofreading result storage unit 123 stores the proofreading results of business records. The proofreading prompt template storage unit 124 stores proofreading prompt templates in advance.

[0031] The business record evaluation model storage unit 125 stores the business record evaluation model learned by the business record evaluation model learning unit 114. Note that the auxiliary storage device 120 may store information and data other than those described above.

[0032] The network 130 is a two-way communication network such as a LAN (Local Area Network) or a WAN (Wide Area Network).

[0033] The CRM / daily report system 140 manages a customer information DB (Data Base) 141 and a business information DB 142. Customer information is stored in the customer information DB 141. Business records 500 (FIG. 5), case event information 600 (FIG. 6), etc. are stored in the business information DB 142. The CRM / daily report system 140 corresponds to an external system of the present invention.

[0034] The language model API 150 outputs the business record transmitted from the business record creation support section 113 to a language model (not shown), and transmits an embedding vector representing the features of the business record obtained from the language model to the business record creation support section 113. Further, the language model API 150 outputs the proofreading prompt (including the business record to be proofread) transmitted from the business record creation support section 113 to the language model, and transmits the proofreading result of the business record obtained from the language model to the business record creation support section 113.

[0035] The language model is a large language model (LLM) that has learned text data, or a generative AI (Artificial Intelligence) model such as GPT (Generative Pre-trained Transformer). The language model calculates an embedding vector of the business record in response to the input of the business record via the language model API 150 and outputs it to the language model API 150. Further, the language model outputs, in natural language, a response according to the proofreading prompt (specifically, the proofreading result of the business record) according to the input of the proofreading prompt described in natural language via the language model API 150. The language model may be arranged on a WAN such as the Internet, or may be arranged on the same LAN as the business support device 100.

[0036] <Sequence of business record pre-evaluation process by business support system 10> Figure 2 shows an example of the sequence of the business record pre-evaluation process by the business support system 10.

[0037] The business record pre-evaluation process is for learning a business record evaluation model and evaluating past business records using the learned business record evaluation model by being executed prior to the business record creation support process (Figure 3) described later. The business record pre-evaluation process is started in response to a predetermined operation from the user.

[0038] First, the user requests the CRM / daily report system 140 to call up past business records for use in learning the business record evaluation model or evaluating them as past cases using the input device 101 and output device 102 of the business support device 100 (step S201). In response to this request, the CRM / daily report system 140 reads the past business records from the business information DB 142 and transmits them to the business support device 100 (step S202). In the business support device 100, the business record creation support section 113 outputs the past business records to the language model via the language model API 150 to calculate an embedding vector representing its features, and the business record storage section 121 associates and registers (stores) the embedding vector with the past business records and notifies the output device 102 of the registration completion (step S203).

[0039] Next, the user who has confirmed the completion of the registration of the business records specifies the learning conditions when learning the business record evaluation model using the input device 101 and output device 102 (step S204). Specifically, as the learning conditions, the business records used for learning, the attribute information of the user who created the business records, customer information, etc. are specified as input variables, the business results corresponding to the business records (conversion rate, number of days required for the contract, cancellation rate, etc.) are specified as target variables, and the hyperparameters of the business record evaluation model are specified.

[0040] Next, in response to the specification of the learning conditions, the business record evaluation model learning section 114 requests the business record storage section 121 to call the specified business records (step S205), and the business record storage section 121 outputs the business records to the business record evaluation model learning section 114 (step S206).

[0041] Next, the business record evaluation model learning unit 114 learns a business record evaluation model according to the learning conditions specified by the user, and outputs the obtained learned business record evaluation model to the business record evaluation model storage unit 125 (step S207). Then, the business record evaluation model storage unit 125 registers (stores) the learned business record evaluation model, and outputs a registration completion notice to the business record evaluation model learning unit 114 (step S208). Then, the business record evaluation model learning unit 114 outputs a learning completion notice of the business record evaluation model to the output device 102 (step S209).

[0042] Next, the user who has confirmed the completion of the learning of the business record evaluation model specifies the evaluation conditions when evaluating the business record using the input device 101 and the output device 102 (step S210). Specifically, the attribute information of the user who created the past business record to be evaluated, customer information, the specification of the business record evaluation model used for evaluation, output variables (factor scores, factor analysis results, etc.) are specified as evaluation conditions.

[0043] Next, according to the specified evaluation conditions, the business record evaluation unit 111 requests the business record storage unit 121 to call the past business record to be evaluated (step S211), and the business record storage unit 121 outputs the business record to the business record evaluation unit 111 (step S212). Next, the business record evaluation unit 111 requests the business record evaluation model storage unit 125 to call the business record evaluation model used for evaluation (step S213), and the business record evaluation model storage unit 125 outputs the business record evaluation model to the business record evaluation unit 111 (step S214).

[0044] Next, the sales record evaluation unit 111 uses the sales record evaluation model to evaluate the past sales records to be evaluated, and outputs the output variables (factor scores, factor analysis results, etc.) as the evaluation results and the embedding vectors of the past sales records to the sales record evaluation result storage unit 122 (step S125). Then, the sales record evaluation result storage unit 122 registers (stores) the evaluation results and the embedding vectors for the past sales records, and outputs a registration completion notice to the sales record evaluation unit 111 (step S216). Next, the sales record evaluation unit 111 outputs a registration completion notice for the evaluation results of the past sales records to the output device 102 (step S217). The above is the description of the sequence of the sales record pre-evaluation process by the sales support system 10.

[0045] <Sequence of the sales record creation support process by the sales support system 10> Figure 3 shows an example of the sequence of the sales record creation support process by the sales support system 10.

[0046] The sales record creation support process is started in response to a predetermined operation from a user who has confirmed that the above-mentioned sales record pre-evaluation process has been completed.

[0047] First, the user uses the input device 101 and the output device 102 to input a sales record draft (a work in progress) (step S301). The sales record draft may be, for example, text data obtained by converting the conversation voice between a salesperson and a customer into characters.

[0048] Next, the sales record creation support unit 113 calculates the embedding vector of the sales record draft using the language model, queries the sales record evaluation result storage unit 122 for past sales records (evaluated ones) that can serve as reference examples (step S302), and extracts, from the sales record evaluation result storage unit 122, those similar to the sales record draft based on the embedding vector of the sales record draft as reference past sales records (step S303).

[0049] Next, the sales record creation support department 113 queries the sales record proofreading result storage department 123 for the proofreading results of past sales records that can serve as reference past proofreading examples (step S304), and extracts from the sales record evaluation result storage department 122 the proofreading results of past sales records similar to the sales record draft as reference past proofreading examples (step S305).

[0050] Next, the sales record creation support department 113 generates a proofreading prompt and outputs it to the sales record proofreading department 112 (step S306). Next, the sales record proofreading department 112 sends the proofreading prompt to the language model via the language model API 150 to request proofreading of the sales record draft. Then, the proofreading result of the sales record draft obtained from the language model is output to the sales record creation support department 113 (step S307).

[0051] Next, the sales record creation support department 113 outputs the proofreading result input from the sales record proofreading department 112 in step S307 to the sales record evaluation department 111 and requests its evaluation (step S308). Next, the sales record evaluation department 111 evaluates the proofreading result and outputs the evaluation result to the sales record creation support department 113 (step S309). Note that the evaluation conditions for evaluating the proofreading result are the same as those specified in step S210 of the sales record pre-evaluation process (Figure 2), but they may also be specified again.

[0052] Next, the sales record creation support department 113 outputs sales record creation support information including the sales record draft, the proofreading result of the sales record draft, and the evaluation result for the proofreading result of the sales record draft to the output device 102, and the output device 102 reflects the sales record creation support information on the UI screen 900 (Figure 9) to prompt the user to make corrections (step S310). For example, the sales record creation support department can also provide the user with the proofreading results for which the evaluation value output by the sales record evaluation department or the similarity with the sales record input by the user for the evaluation result of the proofreading result is equal to or greater than a predetermined value as the proofread sales records.

[0053] Next, when the user finishes modifying the business record draft based on the proofreading result on the UI screen 900 and gives an instruction to record, the business record proofreading result writing section 116 associates the business record modified by the user based on the business record creation support information with the embedding vector of the business record draft before proofreading and registers (stores) it in the business record proofreading result storage section 123 (step S311). Then, the business record proofreading result storage section 123 outputs the completion of registration to the output device 102 (step S312). The above is the explanation of the sequence of the business record creation support process by the business support system 10.

[0054] <Flowchart of the business record creation support process by the business support device 100> Figure 4 is a flowchart showing an example of the business record creation support process by the business support device 100.

[0055] The business record creation support process is started in response to a predetermined operation from the user.

[0056] First, the business record creation support section 113 receives the business record draft (draft) input by the user using the input device 101 and the output device 102 and stores it in the business record storage section 121 (step S401 (corresponding to step S301 in FIG. 3)). Note that the business record draft may be created by the business record draft input support process (FIG. 15) in the dialogue format described later.

[0057] Next, the business record creation support section 113 executes a reference example extraction process. Specifically, it extracts past business records (evaluated ones) that can be reference examples from the business record evaluation result storage section 122, and extracts the proofreading results of past business records that can be reference past proofreading examples from the business record proofreading result storage section 123 (step S402 (corresponding to steps S302 to S305 in FIG. 3)). The details of the reference example extraction process in step S402 will be described later with reference to FIG. 10.

[0058] Next, the sales record creation support unit 113 executes proofreading prompt generation processing (step S403 (corresponding to step S306 in FIG. 3)). Specifically, a proofreading prompt is generated by embedding a sales record draft or the like in a proofreading prompt template. Details of the proofreading prompt generation processing in step S403 will be described later with reference to FIG. 11.

[0059] Next, the sales record proofreading unit 112 executes sales record proofreading processing. Specifically, the generated proofreading prompt is transmitted to the language model via the language model API 150, and the proofreading result of the sales record draft obtained from the language model is output to the sales record creation support unit 113 (step S404 (corresponding to step S307 in FIG. 3)). Details of the sales record proofreading processing in step S404 will be described later with reference to FIG. 13.

[0060] Next, the sales record creation support unit 113 executes proofreading result evaluation processing. Specifically, the sales record before proofreading and a plurality of corresponding proofreading results are output to the sales record evaluation unit 111 to request respective evaluations. Then, the sales record evaluation unit 111 evaluates the sales records before and after proofreading. More specifically, the sales performance of each of the sales records before and after proofreading is predicted as an evaluation value and scored, the two are compared, the improvement amount of the evaluation value after proofreading is calculated, and based on the improvement amount of the evaluation value after proofreading, the proofreading result to be presented to the user is selected. Also, the sales record evaluation unit 111 extracts the factors contributing to the prediction and the factors contributing to the improvement of the evaluation value, and outputs them to the sales record creation support unit 113 (step S405 (corresponding to steps S308 and S309 in FIG. 3)). Details of the proofreading result evaluation processing in step S405 will be described later with reference to FIG. 14.

[0061] Next, the sales record creation support unit 113 formats the sales record creation support information including the sales record draft, the proofreading result of the selected sales record draft, and the evaluation result for the proofreading result of the sales record draft by transmitting it to the language model via the language model API 150. Then, the formatted sales record creation support information is output to the output device 102, and the output device 102 reflects the sales record creation support information on the UI screen 900 to prompt the user for correction (corresponding to step S406 (step S310 in FIG. 3)). The above is an example description of the sales record creation support process by the sales support device 100.

[0062] <Data Structure of Various Information> FIG. 5 shows an example of the data structure of the sales record 500 read from the sales information DB 142 and stored in the sales record storage unit 121.

[0063] In the sales record 500, a case ID for identifying a case, a customer ID for identifying a customer, a sales record ID for identifying a sales record, a recorder ID for identifying a salesperson who created the sales record, a record type for identifying the type of sales business, a sales record date and time indicating the date and time when the sales record was created, and record information describing the content of the sales business are recorded in association.

[0064] FIG. 6 shows an example of the data structure of the case event information 600 stored in the sales information DB 142.

[0065] In the case event information 600, a case ID, an event record date and time, a classification representing the sales performance of the case, the contract content before the change, and the contract content after the change are recorded in association.

[0066] As the classification representing the sales performance of the case, in addition to the exemplified contract continuation and contract expiration, the number of days required from the start of business negotiation to the conclusion of the contract, etc. are recorded.

[0067] FIG. 7 shows an example of the data structure of the sales record evaluation result 700 stored in the sales record evaluation result storage unit 122.

[0068] In the business record evaluation result 700, the case ID, business record ID, contract status, evaluation model used for evaluation, contract continuation probability, and predictive contribution factor information are recorded in association with each other.

[0069] Figure 8 shows an example of the data structure of the business record proofreading result 800 stored in the business record proofreading result storage unit 123.

[0070] In the business record proofreading result 800, the business record ID for identifying the business record, the proofreading ID for identifying the proofreading result, the business record content before proofreading, the proofreading content (improvement proposal content), the revised version (revision proposal content), the contract continuation probability before proofreading, the contract continuation probability after proofreading, and the predictive factor analysis result are recorded in association with each other.

[0071] <Display example of UI screen 900> Figure 9 shows a display example of the UI screen 900 for providing business record creation support information to the user, which is displayed on the output device 102. The UI screen 900 is for the user to input a business record draft, display the proofreading result for the business record draft, and perform revised input for the business record draft.

[0072] On the UI screen 900, a "Customer Information Setting" button 901, a "Business Record Template Setting" button 902, a "Proofreading Execution" button 903, a "Proofreading Example Storage" button 904, a business record draft input field 905, an improvement proposal display field 906, a revision proposal display field 907, and a proofreading supplementary information display field 910 are provided.

[0073] The "Customer Information Setting" button 901 is a button for displaying a sub-screen (not shown) for customer information setting on the UI screen 900. The user can input information about the customer on the sub-screen.

[0074] The "Sales Record Template Setting" button 902 is a button for displaying a sub-screen (not shown) for setting the sales record template on the UI screen 900. The user can select a pre-prepared sales record template on the sub-screen, and the sales record template selected by the user is displayed in the sales record entry field 905. The user can input a sales record based on the sales record template displayed in the sales record entry field 905.

[0075] The "Proofreading Execution" button 903 is a button for instructing the proofreading of the sales record entered in the sales record entry field 905. When the "Proofreading Execution" button 903 is operated, the proofreading of the sales record is executed. As a result of the proofreading, improvement proposals indicating the parts of the sales record to be corrected are displayed in the improvement proposal display column 906. Also, in the correction proposal display column 907, correction proposals exemplifying the specific contents of the parts of the sales record to be corrected are displayed. Note that the specific correction proposals are extracted from the customer information DB 141 and the sales information DB 142 managed by the CRM / daily report system 140. Also, in the proofreading supplementary information display column 910, proofreading supplementary information representing the basis of the correction proposals displayed in the correction proposal display column 907 is displayed.

[0076] Note that only the one with the highest evaluation of the proofreading results displayed on the UI screen 900 may be displayed, or a plurality of them may be displayed in order from the highest evaluation. The user can correct the sales record in the sales record entry field 905 while referring to the improvement proposals and correction proposals.

[0077] The "Proofreading Example Memory" button 904 is a button for associating the sales record displayed in the sales record entry field 905, the improvement proposals displayed in the improvement proposal display column 906, and the correction proposals displayed in the correction proposal display column 907 and storing them as proofreading examples.

[0078] <Details of reference example extraction process> Figure 10 is a flowchart showing an example of the reference example extraction process in step S402 of Figure 4.

[0079] First, the sales record creation support unit 113 reads a sales record draft from the sales record storage unit 121 (step S1001). Next, the sales record creation support unit 113 sets filtering conditions for extracting reference cases based on the input from the user (step S1002). Specifically, the business performance of the corresponding project (such as the presence or absence of an order, the order probability, etc.) and the attributes of the salesperson (position, length of service, etc.) are set as filtering conditions.

[0080] Next, the sales record creation support unit 113 sends the sales record draft to the language model via the language model API 150 to calculate the embedding vector of the sales record draft (step S1003).

[0081] Next, the sales record creation support unit 113 refers to the sales record evaluation result storage unit 122 and calculates the similarity between the past sales records that meet the filtering conditions set in step S1002 and the sales record draft based on their respective embedding vectors (step S1004). Next, the sales record creation support unit 113 extracts the past sales records with a similarity calculated in step S1004 that is equal to or greater than a predetermined threshold as reference cases (step S1005).

[0082] Next, the sales record creation support unit 113 refers to the sales record proofreading result storage unit 123 and calculates the similarity between the sales record proofreading results that meet the filtering conditions set in step S1002 and the sales record draft based on their respective embedding vectors (step S1006). Next, the sales record creation support unit 113 extracts the sales record proofreading results with a similarity calculated in step S1006 that is equal to or greater than a predetermined threshold as reference past proofreading cases (step S1007). The above is a detailed description of the reference case extraction process.

[0083] <Details of the proofreading prompt generation process> FIG. 11 is a flowchart showing an example of the proofreading prompt generation process in step S403 of FIG. 4.

[0084] First, the sales record creation support unit 113 reads out the proofreading prompt template 1200 (FIG. 12) from the proofreading prompt template storage unit 124 (step S1101).

[0085] FIG. 12 shows an example of the proofreading prompt template 1200. The proofreading prompt template 1200 is provided with a sales record embedding field 1201, a sales record template embedding field 1202, a reference example embedding field 1203, and a reference past proofreading example embedding field 1204.

[0086] Return to FIG. 11. Next, the sales record creation support unit 113 embeds the sales record draft into the sales record embedding field 1201 of the proofreading prompt template 1200 (step S1102).

[0087] Next, the sales record creation support unit 113 refers to the sales record storage unit 121 and determines whether there is a sales record template (step S1103). If it is determined that there is a sales record template (YES in step S1103), the sales record template is embedded into the sales record template embedding field 1202 of the proofreading prompt template 1200 (step S1104). On the contrary, if it is determined that there is no sales record template (NO in step S1103), the sales record creation support unit 113 skips step S1104.

[0088] Next, the sales record creation support unit 113 determines whether the reference example (past sales record) could be extracted in step S1005 of the reference example extraction process described above (step S1105). If it is determined that the reference example could be extracted (YES in step S1105), the reference example is embedded into the reference example embedding field 1203 of the proofreading prompt template 1200 (step S1106). On the contrary, if it is determined that the reference example could not be extracted (NO in step S1105), the sales record creation support unit 113 skips step S1106.

[0089] Next, the sales record creation support unit 113 determines whether or not it was able to extract a reference past proofreading example (proofreading result of past sales records) in step S1007 of the reference example extraction process described above (step S1107). Then, when it is determined that a reference past proofreading example has been extracted (YES in step S1107), the reference past proofreading example is embedded in the reference past proofreading example embedding field 1204 of the proofreading prompt template 1200 (step S1108). On the contrary, when it is determined that a reference past proofreading example could not be extracted (NO in step S1107), the sales record creation support unit 113 skips step S1108.

[0090] Next, the sales record creation support unit 113 temporarily records the proofreading prompt template 1200 embedded with the sales record draft etc. in the auxiliary storage device 120 as a proofreading prompt (step S1109). The above is the detailed description of the proofreading prompt generation process.

[0091] <Details of the sales record proofreading process> FIG. 13 is a flowchart showing an example of the sales record proofreading process in step S404 of FIG. 4.

[0092] First, the sales record proofreading unit 112 reads out the proofreading prompt generated by the proofreading prompt generation process (FIG. 11) from the auxiliary storage device 120 (step S1301). Next, the sales record proofreading unit 112 sequentially sets within a range where the proofreading parameters for controlling the reliability of the language model can be changed, and determines whether or not proofreading results corresponding to all the proofreading parameters have been obtained (step S1302).

[0093] Here, when it is determined that the redaction results corresponding to all redaction parameters have not been obtained respectively, that is, there are remaining set values of the redaction parameters that have not been set (NO in step S1302), next, the business record redaction unit 112 sets the redaction parameters to the unset set values (step S1303), and transmits them to the language model API 150 via the network 130 together with the redaction prompt (step S1304). Next, the business record redaction unit 112 obtains the redaction result from the language model via the language model API 150 (step S1305).

[0094] After this, the business record redaction unit 112 returns the process to step S1302 and repeats steps S1302 to S1305. As a result, a plurality of redaction results with different redaction parameters are obtained from the language model.

[0095] After that, in step S1302, when it is determined that the redaction results corresponding to all redaction parameters have been obtained respectively, that is, there are no remaining set values of the redaction parameters that have not been set (YES in step S1302), the business record redaction unit 112 outputs the obtained plurality of redaction results to the business record creation support unit 113. Then, the business record creation support unit 113 temporarily records the obtained plurality of redaction results for one business record draft in the auxiliary storage device 120 (step S1306). The above is the detailed description of the business record redaction process.

[0096] <Details of the redaction result evaluation process> FIG. 14 is a flowchart showing an example of the redaction result evaluation process in step S405 of FIG. 4.

[0097] First, the business record creation support unit 113 reads out the plurality of redaction results obtained for one business record draft by the business record redaction process (FIG. 13) from the auxiliary storage device 120 (step S1401). Next, the business record creation support unit 113 determines whether evaluation results for each of the plurality of redaction results have been obtained (step S1402).

[0098] Here, when it is determined that evaluation results for each of the plurality of proofreading results have not been obtained, that is, there are still proofreading results for which evaluation results have not been obtained (NO in step S1402), next, the business record creation support unit 113 outputs the business record before proofreading and one of the plurality of proofreading results to the business record evaluation unit 111 and requests the evaluation. The business record evaluation unit 111 uses the business record evaluation model to predict and score the evaluation values (business performance) of the business records before and after proofreading, compares the two, and calculates the improvement amount of the evaluation value after proofreading. Also, factors that contributed to the prediction of the evaluation value and factors that contributed to the improvement of the evaluation value are extracted (step S1403).

[0099] After this, the process returns to step S1402, and steps S1402 and S1403 are repeated. Note that in step S1403 after the second time, the output of the business record before proofreading from the business record creation support unit 113 to the business record evaluation unit 111 can be omitted.

[0100] After that, in step S1402, when it is determined that evaluation results for each of the plurality of proofreading results have been obtained, that is, there are no remaining proofreading results for which evaluation results have not been obtained (YES in step S1402), next, the business record evaluation unit 111 selects proofreading results for which the improvement amount of the evaluation value after proofreading is equal to or greater than a predetermined threshold value (step S1404). Next, the business record evaluation unit 111 outputs the selected proofreading results, the factors that contributed to the prediction of the evaluation value, and the factors that contributed to the improvement of the evaluation value to the business record creation support unit 113 (step S1405). The above is a detailed description of the proofreading result evaluation process.

[0101] <Business record draft input support process in dialogue form> FIG. 15 is a flowchart showing an example of the business record draft input support process in dialogue form.

[0102] The business record draft input support process is arbitrarily executed prior to step S401 of the business record creation support process (FIG. 4).

[0103] First, the sales record input support unit 115 connects to the CRM / daily report system 140 via the network 130 and acquires customer information and sales information associated with the cases for which sales operations have been performed (step S1501).

[0104] Next, the sales record input support unit 115 reads a sales record template from the sales record storage unit 121 (step S1502). Next, the sales record input support unit 115 transmits the sales record template, customer information, sales information, etc. to the language model API 150 via the network 130 and requests the language model to complete the sales record template (step S1503).

[0105] Next, the sales record input support unit 115 checks the sales record template completed by the language model and determines whether there are any unentered items that the language model could not complete (step S1504). Here, if it is determined that there are unentered items (YES in step S1504), the sales record input support unit 115 uses the screen displayed on the output device 102 to prompt the user to input information regarding the unentered items in a question format and accepts the information input from the user by the input device 101 (step S1505). Next, the sales record input support unit 115 records the information input by the user in the sales record template completed by the language model (step S1506).

[0106] After that, the sales record input support unit 115 returns the process to step S1504 and repeats steps S1504 to S1506. Then, in step S1504, if it is determined that there are no unentered items that the language model could not complete (NO in step S1504), the sales record input support unit 115 outputs the sales record template completed by the language model and with the information from the user recorded as a sales record (step S1507). The above is the explanation of the sales record input support process in the dialogue format.

[0107] According to the sales support system 10 described above, it is possible to support the creation of a sales record that can predict future orders based on past sales records with results.

[0108] The present invention is not limited to the above-described embodiments, and various modifications are possible. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Also, it is possible to replace or add a part of the configuration of one embodiment with the configuration of another embodiment.

[0109] In addition, each of the above-described configurations, functions, processing units, processing means, etc. may be realized in hardware by designing part or all of them, for example, by an integrated circuit. Also, each of the above-described configurations, functions, etc. may be realized in software by a processor interpreting and executing a program for realizing each function. Information such as a program, table, file, etc. for realizing each function can be placed in a memory, a recording device such as a hard disk, SSD, or a recording medium such as an IC card, SD card, DVD. Also, control lines and information lines show those considered necessary for explanation, and not necessarily all control lines and information lines are shown on the product. In practice, it may be considered that almost all configurations are interconnected.

Explanation of Reference Numerals

[0110] 10... Business support system, 100... Business support device, 101... Input device, 102... Output device, 103... Communication device, 104... Memory, 105... Data bus, 110... Arithmetic unit, 111... Business record evaluation unit, 112... Business record correction unit, 113... Business record creation support unit, 114... Business record evaluation model learning unit, 115... Business record case input support unit, 116... Business record correction result writing unit, 120... Auxiliary storage device, 121... Business record storage unit, 122... Business record evaluation result storage unit, 123... Business record correction result storage unit, 124... Correction prompt template storage unit, 125... Business record evaluation model storage unit, 130... Network, 140... CRM / daily report system, 141... Customer information DB, 142... Business information DB, 150... Language model API, 500... Business record, 600... Project event information, 700... Business record evaluation result, 800... Business record correction result, 900... UI screen, 1041... Program, 1200... Correction prompt template

Claims

1. A sales support device that supports a user in creating a sales record, The system is configured by a computer having one or more arithmetic units, one or more memory resources, and one or more storage devices, the computing device includes a sales record correction unit that corrects the sales record draft created by the user using a language model; a sales record evaluation unit configured to evaluate the quality of the sales record draft created by the user and the corrected sales record draft; A sales support device characterized in that the calculation device comprises a sales record creation support unit that provides the user with sales record creation support information for correcting the sales record draft based on the sales record draft created by the user, the correction results of the corrected sales record draft, and the evaluation results of the sales record draft by the sales record evaluation unit.

2. The sales support device according to claim 1, The sales support device is characterized in that the sales record evaluation unit evaluates the quality of the sales record draft based on at least one of information regarding sales performance corresponding to past sales records similar to the sales record draft and attribute information of the creators of the past sales records similar to the sales record draft.

3. The sales support device according to claim 1, The sales support device is characterized in that the sales record evaluation unit uses a sales record evaluation model to evaluate the quality of the sales record draft.

4. The sales support device according to claim 3, A sales support device further comprising a sales record evaluation model learning unit that pre-learns the sales record evaluation model using created sales records, attribute information of the creator of the created sales records, and customer information corresponding to the created sales records as input variables, and sales results corresponding to the created sales records as objective variables.

5. The sales support device according to claim 1, The sales record creation support unit generates a correction prompt, The sales support device is characterized in that the sales record correction unit inputs the correction prompt into the language model and obtains, from the language model, a correction result for the sales record draft created by the user.

6. 6. The sales support device according to claim 5, The sales support device is characterized in that the sales record creation support unit generates the correction prompt by embedding what is present in the sales record draft, a sales record template, past sales records similar to the sales record draft, and the correction results for the past sales records similar to the sales record draft into a correction prompt template.

7. 7. The sales support device according to claim 6, The sales record creation support unit creates the correction prompt template to be presented to the user when creating a sales record draft, based on a sales record draft that has received good evaluation results from the sales record evaluation unit and correction examples that have received good evaluation results from the sales record evaluation unit.

8. 6. The sales support device according to claim 5, the sales record correction unit obtains a plurality of correction results corresponding to one of the sales record drafts by changing correction parameters to be output to the language model together with the correction prompt; The sales support device is characterized in that the sales record creation support unit provides the user with at least one of a plurality of correction results corresponding to one of the sales record drafts as the sales record creation support information.

9. 9. The sales support device according to claim 8, A sales support device characterized in that the sales record creation support information proposed to the user by the sales record creation support unit includes at least one of an improvement suggestion pointing out areas that need to be corrected in the draft sales record, a correction suggestion that provides examples of specific areas that need to be corrected in the draft sales record, and supplementary correction information that indicates the basis for the correction suggestion.

10. The sales support device according to claim 1, The sales support device further comprises a sales record draft input support unit that supports the user in creating the sales record draft by asking interactive questions.

11. The sales support device according to claim 1, A sales support device that reads and utilizes at least one of customer information and sales information from an external system that manages at least one of the customer information and sales information.

12. A sales support method for a sales support device that supports a user in creating a sales record, comprising: the sales support device is configured by a computer having one or more arithmetic units, one or more memory resources, and one or more storage devices; The sales support method includes: a sales record correction step in which the arithmetic device corrects the sales record draft created by the user using a language model; a sales record evaluation step in which the computing device evaluates the quality of the sales record draft created by the user and the corrected sales record draft; a sales record creation support step in which the computing device provides the user with sales record creation support information for revising the sales record draft based on the sales record draft created by the user, the correction results of the corrected sales record draft, and the evaluation results of the sales record draft in the sales record evaluation step.

Images