Assistance device, assistance method, and program

Abstract

An assistance device according to one aspect of the present disclosure is for assisting a business in which a first person interacts with a second person. The assistance device comprises: a creation unit that creates an instruction document including at least a text representing an utterance of at least one of the first person and the second person, and a command for generating the text and information for assisting the business from the text, on the basis of a pre-created template; and a generation unit that generates the information by a trained machine learning model, on the basis of instruction document.

Description

Support device, support method, and program

[0001] The present disclosure relates to an assistance device, an assistance method, and a program.

[0002] Analysis systems for contact centers (also called call centers) have been known for some time (for example, Non-Patent Document 1), and have been introduced in many contact centers for the purposes of supporting the work of operators and supervisors, improving customer satisfaction, etc.

[0003] ForeSight Voice Mining, Internet <URL: https: / / www.ntt-tx.co.jp / products / foresight_vm / >

[0004] However, conventional techniques may not provide sufficient support for operators and supervisors.

[0005] The present disclosure has been made in consideration of the above points, and provides a technology that can support a user's business.

[0006] An assistance device according to one aspect of the present disclosure is an assistance device that assists a first person in a task of interacting with a second person, and includes: a creation unit that creates, based on a pre-created template, an instruction sheet that includes at least text representing the speech of at least one of the first person and the second person and an instruction to generate information to assist the task from the text; and a generation unit that generates the information using a trained machine learning model based on the instruction sheet.

[0007] A technology is provided that can support the user's business.

[0008] FIG. 1 is a diagram showing an example of the overall configuration of a contact center system according to the present embodiment; FIG. 2 is a diagram showing an example of the functional configuration of a business support system, an operator terminal, and a supervisor terminal according to the present embodiment; FIG. 3 is a sequence diagram showing an example of a support process for customer response business; FIG. 4 is a diagram showing an example of a call content screen; FIG. 5 is a diagram showing an example of a call content registration screen; FIG. 6 is a diagram showing an example of a template for a first prompt; FIG. 7 is a diagram showing an example of automatic check input and automatic display of detailed items; FIG. 8 is a sequence diagram showing an example of a support process for evaluation business; FIG. 9 is a diagram showing an example of a template for a second prompt; FIG. 10 is a diagram showing an example of evaluation information.

[0009] An embodiment of the present invention will be described in detail below with reference to the drawings. In the following embodiment, a contact center system 1 will be described that can support the operators' customer service work and the supervisor's evaluation work for contact center operators and supervisors. Here, the customer service work refers to the work in which the operators respond to inquiries from customers, etc. On the other hand, the evaluation work refers to the work in which the supervisor evaluates the operators' customer service work. Note that a supervisor is a person who supervises an operator, and in addition to evaluating the operators' customer service work, the supervisor also performs work such as monitoring the operator's calls and supporting the operator when any problems are likely to occur or at the operator's request.

[0010] However, the contact center operators and supervisors are merely examples, and the present invention can be similarly applied to cases where support is provided to personnel other than contact center operators and supervisors, such as personnel in an office or store. For example, the present invention can be similarly applied to cases where support is provided to customer service operations for sales personnel of a certain product or service, or for store front desk personnel, or to cases where support is provided to evaluation operations that evaluate such customer service operations. More generally, the present invention can be similarly applied to cases where support is provided to customer service operations when a certain person serves one or more other people, or to cases where support is provided to evaluation operations that evaluate such customer service operations.

[0011] <Overall Configuration Example> An overall configuration example of a contact center system 1 according to this embodiment will be described with reference to FIG.

[0012] 1, the contact center system 1 according to this embodiment includes a business support system 10, one or more operator terminals 20, one or more supervisor terminals 30, one or more telephones 40, a PBX (Private Branch eXchange) 50, a NW switch 60, and a customer terminal 70. The business support system 10, the operator terminals 20, the supervisor terminals 30, the telephones 40, the PBX 50, and the NW switch 60 are installed in a contact center environment E, which is a system environment of a contact center. Note that the contact center environment E is not limited to a system environment within the same building, and may be, for example, a system environment within multiple geographically separated buildings.

[0013] The business support system 10 is a server or a group of servers that provides functions for supporting the customer service work of the operators and the evaluation work of the supervisors. The operators and supervisors correspond to users of the business support system 10.

[0014] Here, the business support system 10 provides various functions in addition to the function for supporting the customer service work of the agents and the evaluation work of the supervisors. These functions include, for example, the functions shown in (1) to (9) below. Note that the functions shown in (1) to (9) below are all functions provided by existing technologies (see, for example, Non-Patent Document 1, etc.).

[0015] (1) Voice recognition function: A function that uses voice packets sent from the NW switch 60 to create a spoken text by converting each utterance made during a voice call between an operator and a customer into text with time information. (2) Speech text provision function: A function that provides the operator terminal 20 with the text of each utterance made during a voice call with a customer in real time when answering the phone with the customer. (3) Memo creation function: A function that creates a memo about a call with a customer when answering the phone with the customer or after the call has ended. (4) Memo provision function: A function that provides the operator terminal 20 with a memo about past calls with the customer when answering the phone with the customer. (5) Search function: A function that searches for response content information that includes various information about a call between an operator and a customer (for example, the text of the utterance made during the call), and provides the search results to the operator terminal 20 or the supervisor terminal 30. (6) Speech text analysis function: A function that uses the text of each utterance made during the call to identify the call reason (reason for receiving a call), whether the call is inbound or outbound, and the scene (the situation of the call). (7) Talk script provision function: A function to identify a talk script from the call reason, inbound or outbound, scene, etc. identified by the speech text analysis function, and provide it to the operator terminal 20. (8) Call monitoring function: A function to provide the supervisor terminal 30 with information for monitoring the operator's calls (for example, the speech text in the call and inappropriate words contained in the speech text). (9) Support request function: A function to allow the operator to request support from the supervisor when handling customer service, or to request the supervisor to take charge of the operator's own customer service.

[0016] The business support system 10 may provide all of the functions shown in (1) to (9) above, or may provide only some of the functions shown in (1) to (9) above. However, in the following, it is assumed that the business support system 10 provides at least the functions shown in (1) and (2) above. Furthermore, the functions shown in (1) to (9) above are examples, and it goes without saying that the business support system 10 may provide various other functions in addition to these.

[0017] The operator terminal 20 is a terminal such as a personal computer (PC) used by an operator who handles customer service. The supervisor terminal 30 is a terminal such as a PC used by a supervisor who handles evaluation work.

[0018] The telephone 40 is an IP (Internet Protocol) telephone (such as a fixed IP telephone or a mobile IP telephone) used by an operator. Generally, one operator terminal 20 and one telephone 30 are installed at each operator's desk.

[0019] The PBX 50 is a telephone exchange (IP-PBX) and is connected to a communication network 80 including a Voice over Internet Protocol (VoIP) network and a Public Switched Telephone Network (PSTN).

[0020] The NW switch 60 relays voice packets between the telephone 40 and the PBX 50 , and also captures the voice packets and transmits them to the business support system 10 .

[0021] The customer terminal 70 is a variety of terminals used by customers, such as a smartphone, a mobile phone, or a landline phone.

[0022] The overall configuration of the contact center system 1 shown in FIG. 1 is merely an example, and other configurations may be used. For example, in the example shown in FIG. 1, the business support system 10 is included in the contact center environment E (i.e., the business support system 10 is on-premise), but all or part of the functions of the business support system 10 may be implemented by a cloud service or the like. Similarly, in the example shown in FIG. 1, the PBX 50 is an on-premise telephone exchange, but it may be implemented by a cloud service or the like. Furthermore, although the telephone 40 exists separately from the operator terminal 20, for example, if the operator terminal 20 functions as an IP telephone, the operator terminal 20 and the telephone 40 may be integrated into one unit.

[0023] <Example of Functional Configuration> An example of the functional configuration of the business support system 10, the operator terminal 20, and the supervisor terminal 30 included in the contact center system 1 according to this embodiment will be described with reference to FIG.

[0024] <Business Support System 10> As shown in FIG. 2 , the business support system 10 according to this embodiment includes a speech recognition unit 101, a business support unit 102, and an information generation unit 103. These functional units are implemented, for example, by one or more programs installed in the business support system 10, which are executed by a computing device such as a central processing unit (CPU). The business support system 10 according to this embodiment also includes an utterance text storage unit 104, a template storage unit 105, and a response content information storage unit 106. These storage units are implemented, for example, by storage areas of storage devices such as hard disk drives (HDDs), solid state drives (SSDs), flash memory, etc. At least one of these storage units may be implemented by storage areas of a storage device (e.g., a storage device provided in a database server, etc.) communicatively connected to the business support system 10.

[0025] The voice recognition unit 101 performs voice recognition processing on the voice represented by the voice data included in the voice packet transmitted from the NW switch 60, and creates a spoken text with time information for each speaker.

[0026] Furthermore, the speech recognition unit 101 stores the spoken text with time information for each speaker (i.e., the customer's spoken text with time information and the operator's spoken text with time information) in the spoken text storage unit 104. For simplicity, the term "with time information" will be omitted below, and the spoken text with time information will be simply referred to as "spoken text."

[0027] The business support unit 102 supports the registration of customer call content as a function of supporting the operator's customer service work. More specifically, the business support unit 102 commands the information generation unit 103 to generate information regarding various items (e.g., customer name, age, birthday, address, inquiry type, etc.) to be input when registering the customer call content, and sets the information generated by the information generation unit 103 in response to this command (hereinafter also referred to as input information) in the corresponding item. This eliminates the need for the operator to input information in that item, thereby realizing, for example, improved efficiency in customer service work.

[0028] Furthermore, the operation support unit 102 realizes support for evaluating the contents of the call represented by the call content information as a support function for the supervisor's evaluation work. More specifically, the operation support unit 102 commands the information generation unit 103 to evaluate the contents of the call represented by the call content information, and outputs information generated by the information generation unit 103 in response to this command (hereinafter also referred to as evaluation information). This eliminates the need for the supervisor to evaluate the contents of the call himself, thereby realizing efficiency in the evaluation work, etc.

[0029] The information generation unit 103 generates input information and evaluation information in response to commands from the business support unit 102. Here, the information generation unit 103 is realized by, for example, a program or module called generative AI (Artificial Intelligence) (or generative AI). Generative AI is a trained machine learning model that receives commands called prompts as input and generates and outputs information in accordance with the prompts. There are various machine learning models that realize generative AI, but for example, machine learning models called large language models (LLMs) are often used.

[0030] The spoken text storage unit 104 stores the spoken text for each speaker (that is, the spoken text of the customer and the spoken text of the operator).

[0031] The template storage unit 105 stores templates of various prompts for the generation AI that realizes the information generation unit 103. Hereinafter, a prompt used to generate input information will be referred to as a "first prompt," and a prompt used to generate evaluation information will be referred to as a "second prompt." Templates of first prompts are stored in the template storage unit 105 for each automatic input component described below, in association with the identification information of the automatic input component. Specific examples of templates of first prompts and templates of second prompts will be described later. Note that prompts may also be called, for example, "instructions."

[0032] The response content information storage unit 106 stores response content information that represents the content of an operator's response to a customer. The response content information includes, for example, identification information for the response content information, identification information for the operator who responded to the customer (e.g., operator name, operator ID, etc.), text spoken by the operator during a call between the operator and the customer, text spoken by the customer during a call between the operator and the customer, and call content information that represents the content of the call registered by the operator. However, the response content information may also include various other information (e.g., audio data of the call between the operator and the customer, notes created by the operator regarding the call, etc.).

[0033] Although not shown in the example of FIG. 2, the business support system 10 has functional units and memory units necessary to realize the various functions (3) to (9) above.

[0034] 2, the operator terminal 20 according to this embodiment has a UI control unit 201. The UI control unit 201 is realized, for example, by a process in which one or more programs (e.g., a web browser) installed in the operator terminal 20 are executed by a computing device such as a CPU.

[0035] The UI control unit 201 displays various screens (such as a call content screen and a call content registration screen, which will be described later) and accepts various operations by the operator on these various screens.

[0036] 2, the supervisor terminal 30 according to this embodiment has a UI control unit 301. The UI control unit 301 is realized, for example, by a process in which one or more programs (e.g., a web browser, etc.) installed in the supervisor terminal 30 are executed by a computing device such as a CPU, etc.

[0037] The UI control unit 301 displays various screens and accepts various operations by the supervisor on these various screens.

[0038] <Support Process for Customer Response Work> The process for supporting a certain operator in a customer response work will be described below with reference to Fig. 3. It is assumed below that a voice call has started between the telephone set 40 of the operator and the customer terminal 50.

[0039] When the voice call is initiated, the UI control unit 201 of the operator terminal 20 displays a call content screen 1000 and a call content registration screen 2000 (described later) (step S101). Here, the call content screen 1000 is a screen on which the text of each speaker's utterance during a voice call between the operator and the customer is displayed in real time. On the other hand, the call content registration screen 2000 is a screen on which the operator can register the content of the call.

[0040] The following steps S102 to S112 are repeatedly executed during the voice call.

[0041] The voice recognition unit 101 of the business support system 10 performs voice recognition processing on the voice represented by the voice data included in the voice packet transmitted from the NW switch 60, and creates a speech text for each speaker (step S102).

[0042] The speech recognition unit 101 of the business support system 10 stores the utterance text created in step S102 in the utterance text storage unit 104 (step S103). As a result, the utterance text for each speaker is stored in the utterance text storage unit 104 as time-series data.

[0043] The task support unit 102 of the task support system 10 transmits the utterance text created in step S102 to the operator terminal 20 (step S104).

[0044] When the UI control unit 201 of the operator terminal 20 receives the utterance text transmitted in step S102, it displays the utterance parts representing the utterance text on the call content screen 1000 (step S105). As a result, the utterance text is displayed on the call content screen 1000 in real time.

[0045] FIG. 4 shows an example of a call content screen 1000 displayed by the UI control unit 201 of the operator terminal 20. The call content screen 1000 shown in FIG. 4 includes an utterance display field 1010, in which utterance components representing the utterance text for each speaker are displayed in real time. In the call content screen 1000 shown in FIG. 4, utterance components 1021-1023 representing the operator's utterance text and utterance components 1031-1033 representing the customer's utterance text are displayed in the utterance display field 1010. The call content screen 1000 shown in FIG. 4 also includes a scroll bar 1040 for scrolling the display range of the utterance display field 1010 and a search field 1050 for searching for utterance components displayed in the utterance display field 1010 using an arbitrary character string. The call content screen 1000 shown in FIG. 4 allows the operator to check the content of the call with the customer in text format in real time.

[0046] 5 shows an example of a call content registration screen 2000 displayed by the UI control unit 201 of the operator terminal 20. The call content registration screen 2000 shown in Fig. 5 includes items 2111 to 2120 related to the content of the call with the customer, and a registration button 2300 for registering the information set in these items 2111 to 2120. Item 2111 is an item for inputting the customer's name, item 2112 is an item for inputting the customer's age, item 2113 is an item for inputting a summary of the call content, item 2114 is an item for inputting the customer's birthday, item 2115 is an item for inputting a desired call-back date and time, item 2116 is an item for selecting the reason for an address change, item 2117 is an item for selecting confirmation items that have been confirmed by the operator, item 2118 is an item for selecting the type of inquiry, item 2119 is an item for selecting whether or not identity verification has been completed, and item 2120 is an item for inputting a changed address in the case of an address change, etc. 5 includes automatic input components 2211 to 2220 for automatically generating and setting input information for items 2111 to 2120. Note that items 2111 to 2120 are merely examples, and the items included in the call content registration screen may differ depending on the type of customer service work handled by the operator, etc.

[0047] By selecting an automatic input component, the operator can automatically generate and set input information for the item corresponding to that automatic input component. For example, when automatic input component 2211 is selected, input information representing the customer's name is generated as input information for item 2111 corresponding to that automatic input component 2211, and the input information is set in item 2111. Similarly, when automatic input component 2212 is selected, input information representing the customer's age is generated as input information for item 2112 corresponding to that automatic input component 2212, and the input information is set in item 2112. The same applies when other automatic input components are selected.

[0048] In the following steps S106 to S112, a case where a selection operation of a certain automatic input component is performed by the operator will be described.

[0049] The UI control unit 201 of the operator terminal 20 accepts the selection operation of the automatic input component (step S106).

[0050] The UI control unit 201 of the operator terminal 20 transmits the operation information of the selection operation accepted in step S106 to the business assistance system 10 (step S107). Here, the operation information includes identification information of the automatic input component selected by the selection operation accepted in step S106 (hereinafter also referred to as automatic input component identification information).

[0051] When the business support unit 102 of the business support system 10 receives the operation information transmitted in step S107, it acquires a template corresponding to the automatic input component identification information included in the operation information from among the templates of the first prompt stored in the template storage unit 105 (step S108). Here, the template of the first prompt includes at least a "#instruction" for instructing the generation AI to generate input information to be set in the item corresponding to the corresponding automatic input component when the corresponding automatic input component is selected, and a "#input statement" in which data to be used for the generation is set.

[0052] An example of a template for a first prompt is shown in Fig. 6. Template 3100 shown in Fig. 6 is a template for a first prompt corresponding to the identification information of automatic input component 2211 (i.e., the automatic input component corresponding to item 2111 for inputting a customer's name), and includes "#instruction," "#constraint," "#output format," and "#input statement." Template 3200 shown in Fig. 6 is a template for a first prompt corresponding to the identification information of automatic input component 2213 (i.e., the automatic input component corresponding to item 2113 for inputting a summary of the call content), and includes "#instruction," "#constraint," and "#input statement." Template 3300 shown in Fig. 6 is a template for a first prompt corresponding to the identification information of automatic input component 2214 (i.e., the automatic input component corresponding to item 2114 for inputting a customer's birthday), and includes "#instruction," "#constraint," "#output format," and "#input statement." The template 3400 shown in FIG. 6 is a template for the first prompt corresponding to the identification information of the automatic input component 2218 (i.e., the automatic input component corresponding to the item 2118 for inputting the inquiry type), and includes "#instruction", "#constraints", "#selection items", and "#input statement".

[0053] The business support unit 102 of the business support system 10 creates a first prompt from the template acquired in step S108 (step S109). Specifically, the business support unit 102 creates the first prompt by setting the utterance text of the customer and the operator from the start of the voice call to the present for the "#input sentence" included in the template acquired in step S108. The utterance text of the customer and the operator from the start of the voice call to the present is stored in the utterance text storage unit 104.

[0054] The information generation unit 103 of the business support system 10 receives the first prompt created in step S109 as an input and generates input information according to the first prompt (step S110). That is, the information generation unit 103 inputs the first prompt created in step S109 to the generation AI, and the generation AI generates input information according to the first prompt.

[0055] The task support unit 102 of the task support system 10 transmits the input information generated in the above step S110 to the operator terminal 20 (step S111).

[0056] When the UI control unit 201 of the operator terminal 20 receives the input information transmitted in step S111, it sets the input information for the item corresponding to the automatic input component selected by the selection operation received in step S106 (step S112). In other words, the UI control unit 201 sets the input information for the item corresponding to the automatic input component identified by the automatic input component identification information included in the operation information transmitted in step S107. This automatically sets the input information for the item corresponding to the automatic input component selected by the operator. This eliminates the need for the operator to input information for the item, thereby improving the efficiency of the operator's customer service work.

[0057] In the following steps S113 to S116, a case where the voice call with the customer has ended will be described. When the voice call with the customer has ended, the operator selects the registration button 2300 on the call content registration screen 2000 shown in FIG.

[0058] The UI control unit 201 of the operator terminal 20 accepts the selection operation of the registration button 2300 (step S113).

[0059] When the UI control unit 201 of the operator terminal 20 accepts the selection operation of the registration button 2300, it transmits the identification information of the operator and the call content registration information to the business support system 10 (step S114). Here, the call content registration information includes identification information (hereinafter also referred to as item identification information) that identifies each of the items 2111 to 2120 on the call content registration screen 2000 shown in FIG. 5 and input information set for each of the items 2111 to 2120.

[0060] Upon receiving the operator identification information and call content registration information transmitted in step S114, the business support unit 102 of the business support system 10 creates response content information (step S115). Here, the response content information includes, for example, identification information of the response content information (hereinafter also referred to as response content identification information), the operator identification information, the text of the utterances between the operator and the customer during the voice call (i.e., the text of the utterances between the operator and the customer from the start to the end of the voice call), and the call content registration information. Various information can be used as the response content identification information, such as an ID uniquely assigned to the response content information. The response content information may also include various other information, such as the audio data of the voice call.

[0061] The business support unit 102 of the business support system 10 stores the response content information created in step S115 in the response content information storage unit 106 (step S116). Note that the response content information represents the content of past calls and the history of response content, and therefore may be called, for example, "response history information" or "call history information."

[0062] <Modification of Customer Service Support Process> A modification of the above-described customer service support process will now be described.

[0063] Variation 1-1 In step S109 above, all of the utterance texts of the customer and the operator from the start of the voice call to the present are set for the "#input sentence" included in the template of the first prompt, but this is not limiting, and some of these utterance texts may be set. For example, the utterance texts shown in (a) to (d) below may be set for the "#input sentence" included in the template of the first prompt.

[0064] (a) The spoken text of either the customer or the operator in the voice call is set as "#input sentence."

[0065] (b) Only the text spoken by the customer and the operator within the last few seconds to tens of seconds of the voice call is set as "#input text."

[0066] (c) Only the most recently uttered text of the voice call is set as "#input sentence."

[0067] (d) Of the texts spoken by the operator during the voice call, only the texts spoken by the operator when the operator repeats the texts are set as "#input text."

[0068] The above (a) to (d) make it possible to reduce the cost of generating input information compared to when all of the customer and operator utterance text from the start of the voice call to the present is set. Furthermore, the above (d) is expected to achieve both cost reduction and maintaining the accuracy of the input information. The above (a) and (b) may also be combined.

[0069] Variation 1-2 In the above-described customer service support process, the operator selects an automatic input component, and input information is set for the item corresponding to that automatic input component. However, for example, input information may be automatically generated during a voice call with a customer, triggered by the satisfaction of some condition, and the input information may be set for the item corresponding to that condition.

[0070] For example, the appearance of the keyword "address change" in the customer's spoken text (or the operator's spoken text) may be used as a trigger to generate and set input information indicating that the address change item 4001 has been checked, as shown in Figure 7.

[0071] Additionally, for example, if item 2113 on the call content registration screen 2000 shown in FIG. 5 has not been entered, input information for item 2113 may be generated and set when the voice call with the customer ends as a trigger.

[0072] Note that trigger conditions can include various conditions other than the appearance of a specific keyword or key phrase during a voice call or the end of the voice call, such as whether the call is inbound or outbound, the call reaching a specific scene, the identification of a specific talk script, or the transition to a specific scenario in the talk script.

[0073] Modification 1-3 In the above modification 1-2, the satisfaction of some condition may be used as a trigger to cause the UI control unit 201 to perform some kind of display control.

[0074] For example, the appearance of the keyword "change in contract content" in the customer's spoken text (or the operator's spoken text) may trigger the display (expansion) of each item included in the contract content change details field 4002, as shown in Figure 7.

[0075] - Variation 1-4 For example, by using the satisfaction of some condition as a trigger, the "#input sentence" included in the template of the second prompt described below can be set to the text spoken by the customer and the operator from the start of the voice call to the present, and the operator's response from the start of the voice call to the present can be evaluated, and the evaluation information can be sent to the operator terminal 20. This allows the operator to know the evaluation up to the present during the response with the customer, which can be used as a reference for future responses. The trigger condition can be the same as in Variation 1-2 above.

[0076] Modification 1-5 In the above modification 1-4, in addition to the evaluation information, evaluation information of other operators who received high evaluations for similar inquiries, etc., and the text of their utterances at that time, etc. may be transmitted to the operator terminal 20. This allows the operator to refer to the evaluation information of other operators who received high evaluations for similar inquiries, etc., and the text of their utterances at that time, etc.

[0077] Variation 1-6: In the above embodiment, an automatic input component corresponding to each item exists, but for example, an automatic input component (hereinafter also referred to as a batch automatic input component) may exist that can collectively generate and set input information for all items (or all items for which input information has not been set). If there are n items corresponding to the batch automatic input component, when the batch automatic input component is selected, input information may be generated by n first prompts corresponding to the n items, respectively, or input information may be generated by a first prompt that generates input information in a format common to the n items.

[0078] <Summary of the support process for customer service operations> As described above, the support process for customer service operations can significantly reduce the burden on operators of the task of recording call content, thereby improving the efficiency of customer service operations for operators.

[0079] <Evaluation Work Support Process> The process of supporting the evaluation work of a certain supervisor will be described below with reference to Fig. 8. Hereinafter, it is assumed that a certain response content identification information is selected by the supervisor from the list of response content identification information, for example.

[0080] The UI control unit 301 of the supervisor terminal 30 accepts a selection operation of response content identification information (step S201).

[0081] The UI control unit 301 of the supervisor terminal 30 transmits the response content identification information selected by the selection operation accepted in the above step S201 to the business support system 10 (step S202).

[0082] When the business support unit 102 of the business support system 10 receives the response content identification information transmitted in step S202, it acquires a template for the second prompt stored in the template storage unit 105 (step S203). Here, the template for the second prompt includes at least a "#instruction" for instructing the generation AI to generate evaluation information for the response content information, and a "#input statement" in which data used for the generation is set.

[0083] An example of a template for a second prompt is shown in FIG. 9 . Template 5000 shown in FIG. 9 is a template for a second prompt for generating evaluation information representing an evaluation of the response content information, and includes "# command," "# constraint," and "# input sentence." Template 5000 shown in FIG. 9 is a template for a second prompt for evaluating the response content represented by the response content information from five perspectives: "basic greeting," "basic response," "listening skills," "speaking skills," and "problem-solving ability." Furthermore, "basic greeting" includes four items: "initial greeting and greeting," "confirmation of the purpose," "confirmation of unclear points," and "greeting at the end of the call." "basic response" includes two items: "honorific language" and "cushion words." "listening skills" includes four items: "block," "discrepancy," "repetition," and "empathy." "speaking skills" includes three items: "clarity of explanation," "method of expression," and "affirmative expression." "problem-solving ability" includes four items: "understanding the current situation," "understanding needs," "proposal," and "appeal." However, these five viewpoints and the items included in each viewpoint are merely examples, and various viewpoints and items for evaluating the content of a call can be set.

[0084] The business support unit 102 of the business support system 10 creates a second prompt from the template acquired in step S203 (step S204). Specifically, the business support unit 102 acquires the response content information identified by the response content identification information from the response content information storage unit 106, and then creates the second prompt by setting the utterance text included in the response content information to "#input sentence" included in the template acquired in step S203.

[0085] The information generation unit 103 of the business support system 10 receives the second prompt created in step S204 and generates evaluation information according to the second prompt (step S205). That is, the information generation unit 103 inputs the second prompt created in step S204 to the generation AI, and the generation AI generates evaluation information according to the second prompt.

[0086] An example of the evaluation information is shown in Figure 10. The evaluation information 6000 shown in Figure 10 is evaluation information created by inputting the second prompt created from the template 5000 shown in Figure 9 into the generation AI. In the evaluation information 6000 shown in Figure 10, four items of "basic greetings" ("initial greeting and greeting," "confirming the purpose of the call," "confirming any unclear points," and "greeting at the end of the call"), two items of "basic response" ("honorific language" and "cushion words"), four items of "listening skills" ("blocking," "discrepancies," "repetition," and "empathy"), three items of "speaking skills" ("clarity of explanation," "method of expression," and "affirmative expression"), and four items of "problem-solving ability" ("understanding the current situation," "understanding needs," "proposals," and "appeals") are each evaluated on a five-point scale.

[0087] The business support unit 102 of the business support system 10 transmits the evaluation information generated in the above step S205 to the supervisor terminal 30 (step S206).

[0088] When the UI control unit 301 of the supervisor terminal 30 receives the evaluation information transmitted in step S206, it displays the evaluation information (step S207), allowing the supervisor to check the evaluation of the call handling content indicated by the call handling content information selected by the supervisor.

[0089] <Modification of Evaluation Work Support Process> A modification of the evaluation work support process described above will be described below.

[0090] In the above evaluation support process, the supervisor selects the information on the details of the customer service call to be evaluated, but the information on the details of the customer service call to be evaluated may be selected automatically. For example, information on the details of the customer service call to be evaluated daily may be selected, and evaluation information for that information on the details of the customer service call may be created.

[0091] Modification 2-2 In the evaluation work support process described above, the service content was evaluated from the utterance text included in the service content information, but if the service content information includes voice data, for example, the voice data may also be input to the generation AI. This makes it possible to evaluate the service content by taking into account not only the utterance text but also, for example, the intonation of the customer or operator when they speak.

[0092] Variation 2-3: In addition to generating evaluation information for one piece of information on the content of responses, evaluation information may be generated for multiple pieces of information on the content of responses corresponding to multiple responses within a predetermined period. In other words, evaluation information may be generated by collectively evaluating multiple pieces of information on the content of responses within a certain predetermined period.

[0093] Variation 2-4: For example, the operator's end of a call may be used as a trigger to evaluate the response content information related to that call, and the evaluation information (especially evaluation information including whether there were any problems with the response content, etc.) may be sent and notified to the supervisor terminal 30.

[0094] <Summary of the evaluation work support process> As described above, the evaluation work support process can significantly reduce the workload of supervisors evaluating the content of the agent's responses. Moreover, because the content of the agent's responses is evaluated by the generation AI, it is possible to provide an objective evaluation that eliminates the supervisor's subjectivity. This not only improves the efficiency of the supervisor's evaluation work, but also enables fair evaluation.

[0095] The present invention is not limited to the above-described specifically disclosed embodiments, and various modifications, changes, and combinations with known technologies are possible without departing from the scope of the claims.

[0096] REFERENCE SIGNS LIST 1 Contact center system 10 Business support system 20 Operator terminal 30 Supervisor terminal 40 Telephone 50 PBX 60 NW switch 70 Customer terminal 80 Communication network 101 Speech recognition unit 102 Business support unit 103 Information generation unit 104 Speech text storage unit 105 Template storage unit 106 Response content information storage unit 201 UI control unit 301 UI control unit

Claims

1. A support device that supports a task in which a first person interacts with a second person, the support device comprising: a creation unit that creates, based on a template created in advance, a text representing speech by at least one of the first person and the second person and an instruction sheet that includes at least the text and an instruction to generate information to support the task from the text; and a generation unit that generates the information using a trained machine learning model based on the instruction sheet.

2. The support device described in claim 1, wherein the creation unit creates the instruction sheet including at least an instruction to generate the information to be set in an item corresponding to the target selected by the selection operation of the first person in accordance with the selection operation of the first person.

3. The assistance device according to claim 1, wherein the creation unit creates the instruction sheet including at least an instruction to generate the information to be set in the item corresponding to a predetermined condition when the predetermined condition is met.

4. A support method for supporting a task in which a first person interacts with a second person, the support method comprising: a creation step of creating, based on a template created in advance, a text representing speech by at least one of the first person and the second person, an instruction manual including at least the text and an instruction for generating information for supporting the task from the text; and a generation step of generating the information using a trained machine learning model based on the instruction manual.

5. A program for supporting a task in which a first person interacts with a second person, the program causing a computer to execute the following steps: a creation procedure for creating, based on a template created in advance, an instruction sheet containing at least text representing speech by at least one of the first and second persons and an instruction for generating information to support the task from the text; and a generation procedure for generating the information using a trained machine learning model based on the instruction sheet.

Images