Generation device, generation method, and generation program

Abstract

Properly implement improvements to voice communications. [Solution] A generation device 100 extracts issues related to the usage status of voice communications by users who use voice communications via a telephone network for business activities. Using the extracted issues and acquired information related to the user's business activities, the generation device 100 inputs prompts including instructions for generating a solution corresponding to the issue and a talk script in which the solution is expressed based on a predetermined story into a large-scale language model, thereby generating the solution and the talk script.

Description

[Technical Field]

[0001] The present invention relates to a generating device, a generating method, and a generating program. [Background technology]

[0002] Voice communication via telephone networks is sometimes used in business activities. For example, when a call is made to a specific telephone number, the call is connected to an operator, who responds appropriately according to the requirements. However, with the above-mentioned voice communication via telephone networks, there are cases where an issue occurs, such as the call not being connected to an operator, which can lead to a decrease in customer satisfaction.

[0003] Therefore, a conventional technique is known that outputs analysis results of call status data for each unit time based on log data that associates the start time and end time of calls between telephones with the calling and receiving telephone numbers (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-027377 Summary of the Invention [Problem to be solved by the invention]

[0005] However, conventional technologies have problems in appropriately improving voice communications. For example, conventional technologies output analysis results of call status data per unit time, such as call volume. However, conventional technologies only output the analysis results, and it is difficult to generate solutions based on the analysis results or talk scripts for appropriately proposing such solutions to users who use voice communications for business activities. [Means for solving the problem]

[0006] Therefore, in order to solve the above-mentioned problems and achieve the objective, the generation device of the present invention is characterized by having an extraction unit that extracts issues regarding the usage status of voice communication by users who use voice communication via a telephone network for business activities, and a generation unit that uses the extracted issues and the acquired information regarding the user's business activities to input a prompt including an instruction to generate a solution corresponding to the issue and a talk script in which the solution is expressed based on a predetermined story into a large-scale language model, thereby generating the solution and the talk script. [Effects of the Invention]

[0007] The present invention provides an effect of appropriately improving voice communication. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram illustrating an overall view of the processing of a generating device according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating a configuration of a generating device according to an embodiment. [Figure 3] FIG. 3 is a table diagram illustrating an example of the usage status information according to the embodiment. [Figure 4] FIG. 4 is a diagram illustrating an example of call volume data according to the embodiment. [Figure 5] FIG. 5 is a diagram illustrating an example of a call volume analysis result according to the embodiment. [Figure 6] FIG. 6 is a table diagram illustrating an example of business information according to the embodiment. [Figure 7] FIG. 7 is a diagram illustrating an example of a potential solution according to the embodiment. [Figure 8] FIG. 8 is a table diagram illustrating an example of talk script information according to the embodiment. [Figure 9] FIG. 9 is a diagram illustrating an example of a prompt according to the embodiment. [Figure 10] FIG. 10 is a diagram illustrating an example of a solution and an output of a talk script according to the embodiment. [Figure 11]FIG. 11 is a flowchart illustrating processing by the generating device according to the embodiment. [Figure 12] FIG. 12 is a diagram illustrating an example of a computer that realizes the generating device according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention (hereinafter referred to as "embodiments") will be described with reference to the drawings. Note that the embodiments are not limited to the following description.

[0010] <Overview> (background) When a call is made to a specified telephone number, the call is connected to an operator who responds appropriately according to the request. In this way, voice communication via the telephone network may be used for business activities.

[0011] In order to address the decline in customer satisfaction in the above-mentioned voice communications via telephone networks, a reference technology is known that outputs the analysis results of call status data for each unit time based on log data that associates the start and end times of calls between telephones with the calling and receiving telephone numbers.

[0012] However, the reference technology only outputs the analysis results, and it is difficult to generate the solution or a talk script to appropriately propose the solution, and there are issues in appropriately realizing improvements in voice communications.

[0013] (Processing by generation device 100) Therefore, the generating device 100 according to this embodiment generates a solution to the problem extracted based on the usage status of voice communication and a talk script for explaining the solution to the user.

[0014] Here, an overview of the processing by the generating device 100 will be described. Fig. 1 is a diagram illustrating an overview of the processing by the generating device 100 according to an embodiment. The generating device 100 shown in Fig. 1 is an example of a computer that provides a technology for realizing the information processing described below.

[0015] First, the generating device 100 extracts issues regarding the usage status of voice communications of users who use voice communications via a telephone network for business activities, using data regarding the traffic volume related to the use of voice communications (hereinafter, sometimes referred to as "traffic volume data") contained in information regarding the usage status of the voice communications (hereinafter, sometimes referred to as "usage status information") ((1-1) in FIG. 1). The usage status information includes, for example, information such as the company name (identification information), telephone number, and traffic volume data related to the user ((1-2) in FIG. 1).

[0016] The generating device 100 uses the problem ((2-1) in Figure 1) about the usage of voice communication extracted from the call volume data and the information ((2-2) in Figure 1) about the user's business activities that has been acquired to input a prompt ((2-3) in Figure 1) that includes an instruction to generate a solution corresponding to the problem and a talk script in which the solution is expressed based on a predetermined story into a large-scale language model, thereby generating the solution and the talk script ((2-4) in Figure 1).

[0017] The information regarding the user's business activities includes information such as the business content and organizational information of the user's business activities ((2-5) in Figure 1), and may be referred to as "business activity information" hereinafter.

[0018] Then, the generating device 100 outputs the generated solution and talk script to a person in charge (hereinafter, sometimes simply referred to as "person in charge") who will propose the solution to the target user ((3) in Figure 1).

[0019] In this way, the generating device 100 according to the present embodiment generates solutions according to issues in the usage status of voice communication and generates talk scripts to be used when proposing the solutions to users, thereby enabling efficient and effective proposal of solutions to users, thereby achieving the effect of appropriately improving voice communication.

[0020] <Description of Generation Device 100> Next, the configuration of the generating device 100 according to this embodiment will be described. Fig. 2 is a diagram showing the configuration of the generating device 100 according to this embodiment. As shown in Fig. 2, the generating device 100 has a communication unit 110, a storage unit 120, and a control unit 130.

[0021] 2, the generating device 100 may include an input unit such as a keyboard or a mouse for receiving input such as operations by an administrator, etc. The generating device 100 may also include a display or the like for the generating device 100 to display information related to the generation process of the solution and the talk script to an administrator, etc.

[0022] (Communication unit 110) The communication unit 110 performs data communication related to input of usage status information collected from outside, business activity information acquired from outside, etc. The communication unit 110 also performs data communication related to output of extracted issues regarding the usage status of voice communication, information related to solutions corresponding to the issues, and information related to talk scripts. The communication unit 110 is realized by a NIC (Network Interface Card) or the like, and controls communication via telecommunication lines such as a LAN (Local Area Network) or the Internet. The communication unit 110 is connected to a network via wired or wireless connection as necessary, and can transmit and receive information bidirectionally with a terminal device or the like.

[0023] (Storage unit 120) The storage unit 120 stores data and programs used for various processes by the control unit 130, and various data acquired by the operation of the control unit 130. The storage unit 120 is realized by a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. As shown in FIG. 2 , the storage unit 120 has a usage status information DB 121, a business activity information DB 122, a solution candidate DB 123, a talk script information DB 124, and a generative model DB 125.

[0024] (Usage status information DB121) The usage status information DB 121 is a database that stores information (usage status information) related to the usage status of voice communication in the business activities of users. Specifically, the usage status information DB 121 stores information such as information for identifying users (identification information), telephone numbers, call volume data, and issues related to the extracted usage status of voice communication.

[0025] An example of usage status information stored in the usage status information DB 121 will now be described with reference to Fig. 3. Fig. 3 is a table diagram showing an example of usage status information according to an embodiment. The usage status information DB 121 associates information relating to each of the following items in a table format or the like: "No." This information identifies individual data included in the user information; "Identification information." "Telephone number." "Traffic volume data." "Traffic volume analysis results." "Tasks."

[0026] For example, as shown in Fig. 3, the usage status information DB 121 stores identification information "A", telephone number "B", traffic volume data "C", traffic volume analysis result "D", and task "E", which are identified by No. "1". Note that the letters "A to E" written in each item in the table diagram shown in Fig. 3 are legends for the information included in each item.

[0027] "Identification information" is information that identifies a user who uses voice calls for business activities, and includes, for example, information expressed by a combination of predetermined character strings, numbers, symbols, etc. Note that information that can identify an individual user may be deleted or replaced with information that can be used by known technology.

[0028] "Telephone number" is information that identifies a telephone that mediates voice calls used in business activities, and includes, for example, information expressed by a combination of predetermined numbers. "Traffic volume data" is data related to call volume obtained based on the usage status of voice communications, and includes, for example, information such as total call volume, incomplete call volume (call volume outside of business hours, call volume outside of designated areas, call volume due to mid-call abandonment, call volume encountered while busy, etc.), and completed call volume.

[0029] Here, the call volume data acquired based on the usage status of voice communication will be described with reference to Fig. 4. Fig. 4 is a diagram showing an example of call volume data according to the embodiment. Fig. 4 shows the end user 10 who makes a call and the destinations set for each call. <1> , connection destination <2> , connection destination <3> An example is shown in which ... and are connected by voice communication.

[0030] The "total call volume" shown in (1-1) or (1-2) of Figure 4 is information indicating the total volume of voice calls used per unit time for business activities (total number of calls made to the target destination). The total call volume is the sum of the "completed call volume (2-2)" and the "incomplete call volume ((3-1) of Figure 4)."

[0031] The "Completed call volume" shown in (2-1) or (2-2) of Figure 4 is information indicating the volume of voice calls used for business activities that have been successfully connected to the target destination identified by the telephone number above (the number of successful connections to the target destination).

[0032] The "uncompleted call volume" shown in (3-1) of Figure 4 is the number of cases where a connection could not be made to the target destination. The uncompleted call volume is the total value of "out-of-hours call volume," "call volume outside designated area," "call volume leaving mid-call," "call volume encountering busy calls," and "other."

[0033] The "out-of-hours call volume" shown in (4-1) or (4-2) of Figure 4 is the number of calls made outside of voice communication reception hours. The "out-of-area call volume" shown in (5-1) or (5-2) of Figure 4 is the number of calls from areas other than the designated area. The "mid-call drop-off volume" shown in (6-1) or (6-2) of Figure 4 is the number of calls that were dropped mid-call. The "busy call volume" shown in (7-1) or (7-2) of Figure 4 is the number of calls where there were no available lines at the subsequent stage. The "other" shown in (8-1) of Figure 4 is the number of calls that could not be connected for reasons other than those mentioned above.

[0034] In addition, the call volume data also includes the average call duration. "Average call duration" is information that indicates the average call duration per voice call for voice calls used in business activities.

[0035] Continuing the explanation by returning to Fig. 3, the "traffic volume analysis result" is the result of a traffic volume analysis executed by extraction unit 133 (described later), and includes, for example, information on the results of tallying up call volumes based on predetermined classifications such as total call volume, completed call volume, incomplete call volume, etc., and results such as call duration, etc., displayed in the form of a list or graph.

[0036] An example of the "traffic volume analysis result" stored in the usage information DB 121 will now be described with reference to Fig. 5. Fig. 5 is a diagram showing an example of the traffic volume analysis result according to the embodiment. As shown in Fig. 5, the usage information DB 121 stores analysis results such as a bar graph (Fig. 5(1)), a pie chart (Fig. 5(2)), and a list (Fig. 5(3)) relating to the aggregated traffic volume. Note that the display contents and display formats of the bar graph, pie chart, and list are merely examples, and the traffic volume analysis results stored in the usage information DB 121 are not particularly limited.

[0037] Furthermore, the "traffic volume analysis result" includes traffic volume problem data that satisfies preset conditions among the traffic volume data as a result of the traffic volume analysis. Specifically, the usage status information DB121 stores, as traffic volume problem data, traffic volume data that satisfies conditions such as "many abandoned calls + high abandoned call rate," "many busy calls + high busy call rate," "many after-hours calls + high after-hours rate," "long average call duration," and "high call volume (even though the data appears to be OK)."

[0038] Continuing the explanation by returning to Fig. 3, "Problem" refers to a problem regarding the usage status of voice communication extracted based on traffic volume data by an extraction unit 133 described later, and stores a problem corresponding to traffic volume problem data.

[0039] For example, the usage status information DB121 stores the issue "The guidance is complicated and difficult to understand, so many people drop out" corresponding to the traffic volume issue data "Many abandoned calls + high abandoned call rate." The usage status information DB121 also stores the issue "Insufficient number of lines, insufficient operator availability" corresponding to the traffic volume issue data "Many busy calls + high busy call rate." The usage status information DB121 also stores the issue "High demand for after-hours calls" corresponding to the traffic volume issue data "Many after-hours calls + high after-hours rate." The usage status information DB121 also stores the issue "Operators are overworked, customer harassment is occurring" corresponding to the traffic volume issue data "Long average call duration."

[0040] (Business activity information DB122) Continuing the explanation, returning to Fig. 2, the business activity information DB 122 is a database that stores information about the user's business activities (business activity information), including information about the organization related to the user (organization information) and information about the user's business content (business content).

[0041] For example, the business activity information DB 122 includes information such as an overview of the user's organization, business policy, organizational philosophy, and focus points in business activities (for example, cost reduction, customer satisfaction, sales development, etc.) The business activity information DB 122 also stores information on the business content performed in the organization of the connection destination of the voice communication and information on the type of connection method, etc. used for the voice communication at the connection destination, as the business content.

[0042] An example of task information among the business activity information stored in the business activity information DB 122 will now be described with reference to Fig. 6. Fig. 6 is a table diagram showing an example of task information according to an embodiment. As shown in Fig. 6, the business activity information DB 122 associates information relating to each item of "No.", which is information identifying individual data included in the task information, "task" indicating the content of the task performed in the organization set as the connection destination for voice communication, and "task analysis result" indicating the result of a predetermined analysis performed on the task, and stores the information in a table format or the like.

[0043] For example, as shown in FIG. 6, the business activity information DB122 stores in association the task No. "1" "Handling tasks related to login methods and operations for page XX," the task No. "1-1" subordinate to No. "1" "Regarding authentication methods," and the task No. "1-2" subordinate to No. "1" "Other."

[0044] Furthermore, the business activity information DB122 stores, as the result of the business analysis for each business, "routine" or "non-routine" for "routine / non-routine," which indicates whether the business is routine or non-routine. The business activity information DB122 also stores "required" or "not required" for "identity verification," which indicates whether identity verification is required. The business activity information DB122 also stores "yes" or "no" for "personal information handling," which indicates whether personal information is handled.

[0045] In other words, the business activity information DB 122 stores, as business information, information summarizing the type of business being performed at each connected party via voice call.

[0046] (Solution candidate DB123) The solution candidate DB 123 is a database that stores solution candidates used in the problem extraction process by the extraction unit 133, which will be described later. Specifically, the solution candidate DB 123 stores information such as hypothetical problems (hypothetical problems) regarding the usage status of voice communication that has been set in advance, and potential solutions to the problems (solution candidate) that correspond to the hypothetical problems.

[0047] Here, an example of the solution candidates stored in the solution candidate DB 123 will be described with reference to Fig. 7. Fig. 7 is a diagram showing an example of the solution candidates according to the embodiment. As shown in Fig. 7, the solution candidate DB 123 stores the solution candidates in a table format or the like, in association with "No.", which is information identifying individual data included in the solution candidate, information relating to each of the items of "hypothetical problem" and "proposed solution."

[0048] For example, as shown in FIG. 7, the solution candidate DB 123 includes a hypothetical problem "Complicated guidance / Many dropouts" identified by No. "1" and a proposed solution " <1> Connection guidance correction, <2> In the proposed solution above, if the call volume analysis results show that "there are many abandoned calls + the abandoned call rate is high" based on usage information, the hypothetical issue is assumed to be "the guidance is complicated / difficult to understand, so many people drop out," and then the solution is stored as " <1> Connection guidance correction, <2> This means that "requirements allocation for AI automatic response" is proposed as a solution. Note that explanations of the solution candidates identified by numbers 2 to 5 will be omitted.

[0049] (Talk script information DB124) The talk script information DB 124 is a database that stores information (talk script information) related to talk scripts generated by the generation unit 134 described below. Specifically, the talk script information DB 124 stores information such as the name (identification information) of the target user who uses the talk script, the generated talk script (talk script), the results of discussions regarding improvement of the generated talk script and improvement information (improvement information), and the talk script improved based on the discussion information (improved talk script). Note that the "identification information" may be the same information as the "identification information" stored in the usage status information DB 121 described with reference to FIG. 3.

[0050] An example of talk script information stored in the talk script information DB 124 will now be described with reference to Fig. 8. Fig. 8 is a table diagram showing an example of talk script information according to an embodiment. The talk script information DB 124 stores the talk script information in a table format or the like, in association with information relating to each of the items of "No.", which is information identifying individual data included in the talk script information, "identification information," "talk script," "improvement information," and "improved talk script."

[0051] For example, as shown in Fig. 8, the talk script information DB 124 stores identification information "A", talk script "F", improvement information "G", and improvement talk script "H", which are identified by No. "1". Note that the letters "A, and F to H" written in each item in the table diagram shown in Fig. 8 are legends for the information included in each item.

[0052] (Generated model DB125) The generative model DB 125 is a database that stores predetermined models used to generate solutions and talk scripts by the later-described generator 134. For example, the generative model DB 125 can store large-scale language models as generative models.

[0053] Specifically, the generating device 100 according to this embodiment can use at least one of "ChatGPT (registered trademark)", a large-scale language model with general-purpose knowledge, and "tsuzumi (registered trademark)", a predetermined large-scale language model for which adapter tuning is performed, as a large-scale language model (see, for example, References 1 and 2).

[0054] (Reference 1):ChatGPT(OpenAI),<URL:https: / / openai.com / chatgpt> ,<Searched on September 2, 2020> (Reference 2): NTT's large-scale language model "tsuzumi",<URL:https: / / www.rd.ntt / research / LLM_tsuzumi.html> ,<Searched on September 2, 2020>

[0055] (control unit 130) Here, the explanation will be continued by returning to Fig. 2. The control unit 130 has an internal memory for temporarily storing programs that define various processing procedures and the like of the generation device 100 and processing data, and is realized by electronic circuits such as a CPU (Central Processing Unit) and an MPU (Micro Processing Unit), and integrated circuits such as an ASIC (Application Specific Integrated Circuit) and an FPGA (Field Programmable Gate Array). As shown in Fig. 2, the control unit 130 has a collection unit 131, a reception unit 132, an extraction unit 133, a generation unit 134, and an output unit 135.

[0056] (Collection Department 131) The collection unit 131 collects information used by the generation device 100 to execute a predetermined process. The collection unit 131 then stores the collected information in the storage unit 120.

[0057] Specifically, the collection unit 131 collects target traffic volume data from an information processing device or the like having a function of storing traffic volume data related to the usage status of the user's voice communication. Then, the collection unit 131 stores the collected traffic volume data in the usage status information DB 121.

[0058] Additionally, the collection unit 131 collects information related to discussions on improvements made to the generated solutions and talk scripts (hereinafter, this may be referred to as "discussion information") and stores this information related to improvements to the solutions and talk scripts (hereinafter, this may be referred to as "improvement information") in the talk script information DB 124. For example, the collection unit 131 collects log data of telephone calls, emails, chats, video conferences, etc., in which discussions on the generated solutions and talk scripts were made, as the discussion information. Next, the collection unit 131 extracts improvement information from the collected log data. Then, the collection unit 131 stores the collected improvement information in the talk script information DB 124.

[0059] (Reception Department 132) The receiving unit 132 receives instructions relating to extraction of a problem, generation of a solution, and generation of a talk script inputted by a user via the display unit, input unit, and communication unit 110 described above.

[0060] (Extraction part 133) The extraction unit 133 extracts problems from data that satisfies a predetermined condition from among the traffic volume data. For example, the extraction unit 133 extracts problems in the following manner.

[0061] The extraction unit 133 acquires the usage status information stored in the usage status information DB 121. Next, the extraction unit 133 uses the acquired usage status information to perform call volume analysis including tallying numerical values ​​and predetermined statistical processing based on known techniques.

[0062] The extraction unit 133 extracts traffic volume problem data that satisfies conditions such as "many abandoned calls + high abandoned call rate," "many busy calls + high busy call rate," "many after-hours calls + high after-hours rate," "long average call duration," and "high call volume (even though there appears to be no problem with the data)" from the traffic volume data calculated by the traffic volume analysis. Then, the extraction unit 133 stores the extracted traffic volume problem data in the solution candidate DB 123 as "traffic volume problem data."

[0063] Next, the extraction unit 133 extracts problems using the extracted traffic volume problem data, and stores the extracted problems in the solution candidate DB 123 as "problems."

[0064] (Generation unit 134) The generation unit 134 inputs the problem extracted from the call volume data in the user's business activities by the extraction unit 133, the user's business activity information, and a prompt including an instruction to generate a solution and a talk script corresponding to the solution to the large-scale language model, thereby generating a solution and a talk script. Specifically, the generation unit 134 uses the prompt described below to cause the large-scale language model to acquire predetermined information and generate a solution and a talk script.

[0065] The prompt includes an instruction to cause the large-scale language model to acquire, as prior knowledge, a hypothetical problem associated with call volume data and a proposed solution (proposed solution method) for the hypothetical problem. The prompt also includes an instruction to cause the large-scale language model to acquire, as prior knowledge, at least one of the user's voice communication usage status information and the user's business activity information (business content, organizational information), which are acquired as the user's business activity information.

[0066] Using the above prompt, the generation unit 134 causes the large-scale language model to acquire the hypothetical problem and proposed solution methods stored in the solution candidate DB 123, and inputs the acquired hypothetical problem and proposed solution methods into the large-scale language model as prior knowledge.

[0067] Furthermore, the generation unit 134 uses the above prompt to cause the large-scale language model to acquire business activity information including business information and organizational information from a homepage related to the user's organization or a server device that displays the homepage, etc. Then, the generation unit 134 stores the business activity information acquired by the large-scale language model in the business activity information DB 122.

[0068] The generation unit 134 also uses the prompt to identify a proposed solution corresponding to the hypothetical problem based on the extracted problem, and generates a solution using the identified proposed solution. The prompt includes an instruction to generate a talk script in which the generated solution is expressed based on a predetermined story.

[0069] (Explanation of an example prompt) An example of a prompt to be input to the large-scale language model in this embodiment will now be described with reference to Fig. 9. Fig. 9 is a diagram showing an example of a prompt according to the embodiment. Fig. 9 shows an example of a prompt for causing the large-scale language model to generate a solution and a talk script corresponding to a problem based on the results of the problem and call volume analysis, business information, and organizational information.

[0070] The generation unit 134 inputs the "prompt" shown in (1) of Fig. 9 into the large-scale language model to acquire prior knowledge and generate a solution and a talk script. The prompt shown in (1) of Fig. 9 includes the following content:

[0071] As shown in (1-1) of Figure 9, the prompt includes a role definition: "You are a salesperson for an automated telephone response system." By using the prompt including the "role definition" as described above, the generation unit 134 can cause the large-scale language model to generate a solution and a talk script that are in line with the field and perspective desired by the person in charge.

[0072] As shown in (1-2) of Figure 9, the prompt includes a processing instruction "Please execute the task according to the following constraints." The generation unit 134 can use the prompt including the "processing instruction based on the processing constraints" as described above to constrain the contents of the solution and talk script to be generated, thereby allowing the large-scale language model to generate the solution and talk script desired by the person in charge.

[0073] In the example shown in FIG. 9, the first to third tasks are included in the prompt ((1-3) to (1-5) in FIG. 9).

[0074] For example, the first task includes the instruction to "understand the business content and organizational information," along with the constraints, "obtain business and organizational information about the X company and read the information about the business content in detail. In particular, focus on understanding the business execution policy and the type of business that is performed. Then, summarize the following," and "the management policy and approach that the organization has toward the departments that execute the business" and "the business content that the departments that execute the business handle" ((1-3) in Figure 9).

[0075] The first task allows the generation unit 134 to cause the large-scale language model to acquire "business content and organizational information" from an external information processing device or the like. The generation unit 134 can also use the acquired "business content and organizational information" as prior knowledge to perform pre-learning such as "understanding business policies (policies for carrying out business) and business content (what kind of business is being performed)." Then, the generation unit 134 generates "management policies and approaches" and "business content" to be included in the solution and talk script based on the first task.

[0076] The second task also includes the command "Understand the traffic volume analysis results and issues" and the constraint "Obtain the traffic volume analysis results and issues of company X and understand the issues of company X" ((1-4) in Figure 9).

[0077] The second task allows the generation unit 134 to cause the large-scale language model to acquire "traffic volume analysis results of the target company (XX company)" from the usage status information DB 121. Also, the generation unit 134 can cause the large-scale language model to acquire "issues of the target company (XX company)" from the usage status information DB 121. Also, the generation unit 134 can use the acquired "traffic volume analysis results and issues" as prior knowledge to perform prior learning such as "understanding the issues of XX company."

[0078] The third task includes the command "Generate a solution and a talk script" and the constraint "Create a solution that meets the challenges of Company X and a talk script to propose the solution to Company X. When doing so, create a solution and a talk script that takes into consideration the management policy and approach, business content, challenges, and call volume analysis results" ((1-5) in Figure 9).

[0079] For example, the third task allows the generation unit 134 to generate a "solution for the target company (XX company)" for a large-scale language model that has been pre-trained, based on the management policy and approach, business content, issues, and traffic volume analysis results of the XX company. Also, the third task allows the generation unit 134 to generate a "talk script for the target company (XX company)" for a large-scale language model that has been pre-trained, based on the management policy and approach, business content, issues, and traffic volume analysis results of the XX company, and the generated solution.

[0080] (Explanation of the solution and an example of a talk script) An example of a solution and a talk script generated by the generating device 100 will now be described with reference to Fig. 10. Fig. 10 is a diagram showing an example of an output of a solution and a talk script according to the embodiment. Fig. 10 shows an example of a solution and a talk script generated based on the example of the prompt shown in Fig. 9.

[0081] As described above, the generation unit 134 generates a solution and a talk script according to the problem of the user (XX company). For example, as shown in (1) of FIG. 10, the generation unit 134 generates a "solution <1> ((1-1) in Figure 10)," "Solution <2> One or more solutions are generated, such as (1-2) in Figure 10. A solution is information that shows methods for solving the extracted problem using text, figures, symbols, etc.

[0082] 10(2), the generation unit 134 generates a talk script including an "operational policy and approach" and a "proposal flow." The "operational policy and approach" includes summary information such as the policy related to organizational management of the target organization (XX company) and efforts to achieve the mission, as shown in FIG. 10(2-1).

[0083] "Proposal flow" is information that shows the flow of conversation when proposing a solution to a user, and includes, for example, "1. Opening," "2. Raising the problem (challenge)," "3. Proposing a solution," "4. Emphasizing benefits," and "5. Sharing implementation examples" ((2-2) in Figure 10).

[0084] The above "1. Opening" includes, for example, talk content such as an overview of what will be proposed and ice-breakers before explaining the solution to the user. "2. Problem Statement (Challenge)" includes, for example, talk content explaining what issues users are experiencing with the voice communications they use in their business activities. "3. Solution Proposal" includes, for example, talk content introducing means and solutions for solving the issues explained in "2. Problem Statement (Challenge)." "4. Emphasizing Benefits" includes, for example, talk content explaining the benefits of the problem-solving methods and solutions explained in "3. Solution Proposal." "5. Sharing Implementation Cases" includes, for example, talk content explaining actual implementation examples of the problem-solving methods and solutions explained in "3. Solution Proposal."

[0085] The solution shown in (1) of FIG. 10 and the contents of the talk script shown in (2) of FIG. 10 are merely examples, and are not limited to the above contents.

[0086] (An example of a solution and an improvement to the talk script) Furthermore, the generation unit 134 can execute a process of improving the solution and talk script generated by the process described with reference to Figures 9 and 10. Specifically, the collection unit 131 collects improvement information on the solution and talk script generated by the generation unit 134 from a person in charge of proposing the solution and talk script.

[0087] Next, the generation unit 134 uses the improvement information for the solution and talk script to input prompts, which are expressed in natural language text as instructions to improve the generated solution and talk script, into a large-scale language model, thereby generating improved solutions and talk scripts.

[0088] For example, the generation unit 134 generates a summary such as "Reduce the number of characters in the talk script" from conversation data (discussion information included in improvement information) such as "The generated talk script has too many characters, so I would like to reduce it further" by a person in charge stored in the talk script information DB 124. Then, the generation unit 134 inputs a prompt such as "Reduce the number of characters in the generated talk script and generate the talk script again" into the large-scale language model, and generates a talk script with a reduced number of characters (i.e., "improved").

[0089] (output unit 135) Now, the explanation will be continued by returning to Fig. 2. The output unit 135 outputs the solution and talk script generated by the generation unit 134 to a terminal device or the like operated by the user via the communication unit 110 or the like described above.

[0090] (Processing Procedure by Generation Device 100) From here, the procedure of the process realized by the generating device 100 according to this embodiment will be described with reference to Fig. 11. Fig. 11 is a diagram showing a flowchart of the process performed by the generating device 100 according to this embodiment.

[0091] The collection unit 131 collects usage status information including traffic volume data (S101). Next, the extraction unit 133 performs traffic volume analysis using the traffic volume data (S102). Next, the extraction unit 133 extracts issues based on the results of the traffic volume analysis (S103). Next, the generation unit 134 generates solutions to the issues and talk scripts based on the extracted issues and business activity information acquired by the large-scale language model (S104).

[0092] Here, if the generated solution and talk script are to be improved (Yes in S105), the collection unit 131 collects improvement information (S106). Next, the generation unit 134 generates an improved solution and talk script using the collected improvement information (S107). Then, the output unit 135 outputs the generated solution and talk script (S108). Thereafter, the generation device 100 ends the process.

[0093] On the other hand, if the generated solution and talk script are not improved (No in S105), the generating device 100 skips steps S106 and S107. Then, the output unit 135 outputs the generated solution and talk script (S108). After that, the generating device 100 ends the process.

[0094] (effect) Next, the effects of the generating device 100 according to this embodiment will be described. Conventionally, analysis results of call status data per unit time may be generated to address the decline in customer satisfaction when voice communications via a telephone network are used for business activities. However, it is difficult to generate a talk script or the like to appropriately propose a solution by simply outputting the analysis results, and there are problems in appropriately realizing improvements in voice communications.

[0095] Therefore, the extraction unit 133 of the generation device 100 according to this embodiment extracts issues regarding the usage status of voice communication of users who use voice communication via a telephone network for business activities. The generation unit 134 of the generation device 100 uses the extracted issues and the acquired business activity information of the users to input prompts including instructions for generating a solution corresponding to the issue and a talk script in which the solution is expressed based on a predetermined story into a large-scale language model, thereby generating the solution and the talk script.

[0096] Through the above-described processing, the generating device 100 can automatically generate solutions and talk scripts, whereas conventionally, solutions and talk scripts were manually generated based on analysis results using information on the usage status of voice communication, such as usage status information. Therefore, the generating device 100 according to this embodiment has the effect of appropriately improving voice communication using the generated solutions and talk scripts.

[0097] Furthermore, the generating device 100 according to this embodiment achieves the following effects by executing the following processes.

[0098] The extraction unit 133 extracts problems from data that satisfies predetermined conditions from traffic volume data related to the use of voice communications in the user's business activities. The generation unit 134 inputs the problems extracted from the traffic volume data related to the use of voice communications in the user's business activities, the user's business activity information, and prompts including instructions for generating solutions and talk scripts corresponding to the solutions into a large-scale language model, and generates solutions and talk scripts.

[0099] Through the above-described process, the generating device 100 can extract problems according to the user's voice communication usage status and generate solutions and talk scripts according to the extracted problems. Therefore, the generating device 100 has the effect of enabling the generating device 100 to generate solutions according to the user more efficiently than ever before.

[0100] Furthermore, the generating device 100 can generate a talk script for appropriately proposing a solution to a user. Therefore, the generating device 100 not only generates and outputs a solution, but also has the effect of facilitating a user to consider a solution to solve a problem in the usage situation of voice communication.

[0101] The generation unit 134 uses prompts expressed in natural language text, which include instructions to cause the large-scale language model to acquire, as prior knowledge, a hypothetical problem associated with call volume data related to the use of voice communication in the user's business activities and proposed solutions to the hypothetical problem, and instructions to cause the large-scale language model to acquire, as prior knowledge, at least one of information on the usage status of voice communication in the user's business activities and the user's business activity information (business content, organizational information, etc.).

[0102] By using the above-described prompt, the generating device 100 can cause the large-scale language model to acquire information for generating solutions and talk scripts from an external information processing device, etc. As a result, the generating device 100 can efficiently generate solutions and talk scripts by eliminating the need to manually acquire information for generating solutions and talk scripts.

[0103] The generation unit 134 uses information regarding the improvement of the solution and talk script to input prompts, in which instructions to improve the generated solution and talk script are expressed in natural language text, into a large-scale language model to generate an improved solution and an improved talk script.

[0104] Through the above-described processing, the generating device 100 can improve the solution and talk script output to the person in charge to make them more appropriate. By generating the improved solution and talk script, the generating device 100 can provide the user with a more appropriate solution and talk script.

[0105] The collection unit 131 collects information on discussions on improvements made to the generated solutions and talk scripts, and stores the information on improvements to the solutions and talk scripts.

[0106] Through the above-described processing, the generating device 100 automatically collects information about improvement discussions between staff members, such as chats, emails, voice calls, and video conferences, which is information used to improve solutions and talk scripts, and makes it possible to use this information as information for improving solutions and talk scripts. As a result, the generating device 100 has the effect of enabling solutions and talk scripts to be improved more efficiently.

[0107] <Modification> The following describes modified examples realized by the generation device 100 according to this embodiment.

[0108] (Data, etc.) The voice communication usage status, usage status information, business activity information, call volume analysis (results), business information, organizational information, solutions, talk scripts, names of functional parts of the generating device 100, steps, processes, names of steps or processes, etc. used in the description of the above embodiments are merely examples and can be changed as desired.

[0109] For example, it has been explained that the usage status information DB 121 associates information relating to each of the following items: "No.", which identifies individual data included in the user information; "Identification Information," "Telephone Number," "Traffic Volume Data," "Traffic Volume Analysis Results," and "Problems"; and stores the information in a table format or the like; however, the items and types of information stored are not limited to the content described above. It has been explained that the usage status information DB 121 stores information relating to the user's business activities (business activity information), including information relating to the organization related to the user (organization information) and information relating to the user's business content (business content); however, the items and types of information stored are not limited to the content described above. It has been explained that the solution candidate DB 123 associates information relating to each of the following items: "No.", which identifies individual data included in the solution candidate; "Hypothetical Problem" and "Proposed Solution," and stores the information in a table format or the like; however, the items and types of information stored are not limited to the content described above. It has been explained that the talk script information DB124 stores information relating to each of the items, "No.", which is information identifying individual data contained in the talk script information, "user name," "talk script," "improvement information," and "improved talk script," in correspondence with each other in a table format, etc., but the items and types of information stored are not limited to those described.

[0110] (Regarding the timing of pre-training large-scale language models) In the present embodiment, the generation unit 134 has been described as having the large-scale language model acquire prior knowledge when executing the process of generating solutions and talk scripts, but the timing of learning is not limited to this. For example, the generation unit 134 can acquire prior knowledge in advance, perform prior learning, and then execute the process of generating solutions and talk scripts.

[0111] (About the use of generative models) In the present embodiment, the model (large-scale language model) used by the generating device 100 is described as being stored in the generating model DB 125 of the storage unit 120, but this is not limiting. For example, the generating device 100 can access an external information processing device (such as a server) and use a predetermined model.

[0112] (Flowcharts, etc.) The steps in the flowcharts may be interchanged as long as there is no contradiction, and some steps may not be performed. In addition, conjunctions such as "next," "continue," "further," "at this time," and "on this occasion" used in the explanation of the flowcharts do not limit the order or timing of the execution of the processes in the flowcharts.

[0113] <Hardware configuration> The components of each device shown in the figure are conceptual functional units and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown, and all or part of each device can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc. Furthermore, all or any part of the processing functions performed by each device can be realized by a CPU and a program analyzed and executed by the CPU, or can be realized as hardware using wired logic.

[0114] Furthermore, among the processes described in this embodiment, all or part of the processes described as being performed automatically can also be performed manually using known methods. In addition, the information including the processing procedures, control procedures, specific names, various data, and parameters shown in the drawings can be changed as desired unless otherwise specified.

[0115] <Program> In one embodiment, the various devices constituting the generating device 100 can be implemented by installing a generating program as package software or online software on a desired computer. For example, by executing the above-described generating program on an information processing device, the various devices constituting the generating device 100 can be made to function. The information processing device referred to here includes desktop and notebook personal computers. In addition, the information processing device also includes mobile communication terminals such as smartphones and mobile phones, and even slate terminals such as PDAs (Personal Digital Assistants).

[0116] 12 is a diagram showing an example of a computer that realizes the generating device 100 according to the embodiment. The computer 1000 includes, for example, a memory 1010 and a CPU 1020. The computer 1000 also includes a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0117] The memory 1010 includes a ROM (Read Only Memory) 1011 and a RAM 1012. The ROM 1011 stores, for example, a boot program such as a BIOS (Basic Input Output System). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to, for example, a mouse 1110 and a keyboard 1120. The video adapter 1060 is connected to, for example, a display 1130.

[0118] The hard disk drive 1090 stores, for example, an OS (Operating System) 1091, an application program 1092, a program module 1093, and program data 1094. That is, a program that defines each process of the various devices that constitute the generating device 100 is implemented as a program module 1093 in which computer-executable code is written. The program module 1093 is stored, for example, in the hard disk drive 1090. For example, a program module 1093 for executing the same processes as the functional configurations of the various devices that constitute the generating device 100 is stored in the hard disk drive 1090. Note that the hard disk drive 1090 may be replaced with an SSD (Solid State Drive).

[0119] Furthermore, setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in the memory 1010 or the hard disk drive 1090. Then, the CPU 1020 reads the program module 1093 or the program data 1094 stored in the memory 1010 or the hard disk drive 1090 into the RAM 1012 as necessary, and executes the processing of the above-described embodiment.

[0120] The program module 1093 and program data 1094 are not limited to being stored in the hard disk drive 1090, but may also be stored in, for example, a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a LAN or a WAN (Wide Area Network)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

[0121] <Other> Although the present embodiment has been described above, the present embodiment is not limited by the descriptions and drawings that form part of the disclosure. In other words, other embodiments, examples, operational techniques, etc. that are made by those skilled in the art based on the present embodiment are all included in the scope of the present embodiment. [Explanation of symbols]

[0122] 100 generator 110 Communications Department 120 Storage section 121 Usage Information DB 122 Business activity information DB 123 Solution candidate DB 124 Talk Script Information DB 125 Generative Model DB 130 control section 131 Collection Department 132 Reception Department 133 Extraction part 134 Generation part 135 Output section

Claims

1. an extraction unit that extracts issues regarding the usage status of voice communications by users who use voice communications via a telephone network for business activities; a generation unit that generates a solution corresponding to the problem and a talk script in which the solution is expressed based on a predetermined story by inputting a prompt including an instruction to generate the solution and the talk script into a large-scale language model using the extracted problem and the acquired information on the user's business activities, The extraction unit acquire information about the usage of voice communications in the business activities of the user stored in a storage unit; Using the acquired information on the usage status of voice communications in the business activities of the user, a call volume analysis including predetermined statistical processing is performed, extracting the problem from data that satisfies a predetermined condition from data on call volume related to the use of voice communications in the user's business activities, which is included in information on the use of voice communications in the user's business activities; The generation unit inputting the problem extracted from data on call volume related to the use of voice communications in the user's business activities, information on the user's business activities, and the prompt including an instruction to generate the solution and the talk script corresponding to the solution into a large-scale language model to generate the solution and the talk script; a command to cause the large-scale language model to acquire, as prior knowledge, a hypothetical problem associated with data on call volume related to the use of voice communications in the user's business activities and a proposed solution to the hypothetical problem; an instruction to cause the large-scale language model to acquire, as prior knowledge, at least one of information on the usage status of voice communication in the business activities of the user and information on the business activities of the user, which information has been acquired as information on the business activities of the user; uses prompts expressed in natural language text, A generating device characterized by:

2. The generation unit inputting a prompt, expressed in natural language text, to the large-scale language model as an instruction to improve the generated solution and talk script using information regarding the improvement of the solution and talk script; Generate improved solutions and improved talk scripts, The generating device according to claim 1 .

3. The method further includes a collection unit that collects information regarding the generated solution and the discussion on improvements made to the talk script, and stores the information as information regarding improvements to the solution and the talk script. The generating device according to claim 2 .

4. A generation method to be executed by a generation device, An extraction step of extracting issues regarding the usage status of voice communications by users who use voice communications via a telephone network for business activities; a generation step of inputting a prompt including an instruction to generate a solution corresponding to the problem and a talk script in which the solution is expressed based on a predetermined story into a large-scale language model using the extracted problem and the acquired information on the user's business activities, thereby generating the solution and the talk script; The extraction step comprises: acquire information about the usage of voice communications in the business activities of the user stored in a storage unit; Using the acquired information on the usage status of voice communications in the business activities of the user, a call volume analysis including predetermined statistical processing is performed, extracting the problem from data that satisfies a predetermined condition from data on call volume related to the use of voice communications in the user's business activities, which is included in information on the use of voice communications in the user's business activities; The generating step includes: inputting the problem extracted from data on call volume related to the use of voice communications in the user's business activities, information on the user's business activities, and the prompt including an instruction to generate the solution and the talk script corresponding to the solution into a large-scale language model to generate the solution and the talk script; a command to cause the large-scale language model to acquire, as prior knowledge, a hypothetical problem associated with data on call volume related to the use of voice communications in the user's business activities and a proposed solution to the hypothetical problem; an instruction to cause the large-scale language model to acquire, as prior knowledge, at least one of information on the usage status of voice communication in the business activities of the user and information on the business activities of the user, which information has been acquired as information on the business activities of the user; uses prompts expressed in natural language text, A generating method characterized by:

5. An extraction step of extracting issues regarding the usage status of voice communications by users who use voice communications via a telephone network for business activities; a generation step of inputting a prompt including an instruction to generate a solution corresponding to the problem and a talk script in which the solution is expressed based on a predetermined story into a large-scale language model using the extracted problem and information on the user's business activities obtained, thereby generating the solution and the talk script, The extraction step comprises: acquire information about the usage of voice communications in the business activities of the user stored in a storage unit; Using the acquired information on the usage status of voice communications in the business activities of the user, a call volume analysis including predetermined statistical processing is performed, extracting the problem from data that satisfies a predetermined condition from data on call volume related to the use of voice communications in the user's business activities, which is included in information on the use of voice communications in the user's business activities; The generating step includes: inputting the problem extracted from data on call volume related to the use of voice communications in the user's business activities, information on the user's business activities, and the prompt including an instruction to generate the solution and the talk script corresponding to the solution into a large-scale language model to generate the solution and the talk script; a command to cause the large-scale language model to acquire, as prior knowledge, a hypothetical problem associated with data on call volume related to the use of voice communications in the user's business activities and a proposed solution to the hypothetical problem; an instruction to cause the large-scale language model to acquire, as prior knowledge, at least one of information on the usage status of voice communication in the business activities of the user and information on the business activities of the user, which information has been acquired as information on the business activities of the user; uses prompts expressed in natural language text, Generator.

Images