Device and method

Abstract

A RAG system 20 (device) according to the present disclosure comprises: a reception unit 21 that receives a message from a customer and information about a customer support person; a control unit 22 that determines information about a correction policy for correcting the message, on the basis of stress-related information acquired from the message; a generation unit 23 that generates a prompt for instructing generation of a correction message obtained by correcting the message, such prompt generated on the basis of the message, the information about the correction policy, and the information about the customer support person; and an acquisition unit 24 that inputs the prompt to a generation AI model 31 for generating the correction message and acquires the correction message from the generation AI model 31.

Description

Apparatus and method

[0001] One aspect of the present disclosure relates to an apparatus and a method.

[0002] In conventional customer support, when a chatbot cannot handle a particular issue, a customer support person handles the issue via chat or telephone. The customer support person may feel stressed due to a customer's harsh words or tone of voice. Patent Documents 1 and 2 disclose technologies for reducing the stress of customer support people.

[0003] Patent Literature 1 describes a technology that converts a customer's speech containing specific words into synthetic speech and outputs it to an operator. The technology described in Patent Literature 1 controls whether or not to convert the speech into synthetic speech depending on the operator's stress level.

[0004] Patent Document 2 describes a technology that analyzes the content of a customer's speech, converts the customer's abusive language using a speech converter, and outputs the converted phrases to an operator. The technology described in Patent Document 2 controls the conversion of phrases depending on the operator's tolerance for customer harassment.

[0005] JP 2023-164770 A JP 2021-196462 A

[0006] For example, the voice replacement process according to Patent Documents 1 and 2 is performed within a range of predetermined expressions. However, since the sensitivity to stress varies depending on the customer service representative, there may be cases where the predetermined expressions do not provide sufficient stress relief.

[0007] Therefore, an object of the present disclosure is to provide a technology that reduces the stress of customer service personnel.

[0008] An apparatus according to one aspect of the present disclosure includes a reception unit that receives a message from a customer and information about a customer service representative; a control unit that determines information about a correction policy for correcting the message based on stress information obtained from the message; a generation unit that generates a prompt to instruct the generation of a corrected message by correcting the message based on the message, information about the correction policy, and information about the customer service representative; and an acquisition unit that inputs the prompt to a generation AI model that generates the correction message and acquires the correction message from the generation AI model.

[0009] In a device according to one aspect of the present disclosure, information regarding a correction policy is determined based on stress information acquired from a message, a prompt for instructing the generation of a correction message is generated based on the message, information regarding the correction policy, and information regarding the customer service representative, and the prompt is input to a generative AI model. Because the generative AI model is controlled by the prompt that takes into account the information regarding the correction policy and the information regarding the customer service representative, a correction message is generated that corresponds to the information regarding the correction policy and the information regarding the customer service representative. As a result, the stress of the customer service representative can be reduced.

[0010] According to one aspect of the present disclosure, a technology for reducing stress on customer service personnel can be provided.

[0011] FIG. 1 is a block diagram showing the device configuration of a generation system. FIG. 2 is a diagram showing an example of a correction proposal database. FIG. 3 is a diagram showing an example of an advice proposal database. FIG. 4 is a diagram showing an example of a prompt. FIG. 5 is a flowchart showing an example of the operation of the RAG system. FIG. 6 is a block diagram showing another example of the configuration of a generation system. FIG. 7 is a diagram showing an example of a hardware configuration.

[0012] The present disclosure will be described with reference to the accompanying drawings. Whenever possible, the same parts are designated by the same reference numerals and redundant description will be omitted.

[0013] The generation system of the present disclosure is applied to, for example, customer support. In customer support, a customer sends an inquiry or the like to a customer support person via a message. The customer support person is also called an operator, a supporter, a staff member, a help desk, or a person in charge. The generation system corrects the message received from the customer and provides the corrected message to the customer support person.

[0014] Fig. 1 is a diagram showing the device configuration of a generation system according to this embodiment. The generation system shown in Fig. 1 includes a terminal 10, a Retrieval-Augmented Generation (RAG) system 20, and a server device 30, which are configured to be able to communicate with each other via networks including a wireless communication network and a fixed communication network. The RAG system 20 constitutes a device (generation device) that generates a prompt for correcting a message received from the terminal 10. A prompt is information indicating an instruction or question input to an AI (Artificial Intelligence) model in an interactive system such as a dialogue with an AI model or a command line interface (CLI).

[0015] The terminals 10 are devices used by customers or users who handle customers. The terminals 10 are, for example, personal computers, smartphones, tablet terminals, feature phones, server devices, game consoles, etc. Although only two terminals 10 are illustrated in FIG. 1 , the generation system may include any number of terminals 10 greater than or equal to two.

[0016] The storage device 15 is a non-transitory storage device that stores various information used in the generation system. The storage device 15 may be a component of the RAG system 20. The storage device 15 may include multiple databases. For example, the storage device 15 may include a customer service representative database, a stress expression database, a correction suggestion database, an advice suggestion database, a character database, and a voice database.

[0017] The customer service representative database may store information about the customer service representative (user information). The information about the customer service representative may be information associated with the user account of the customer service representative, or may be information received from the customer service representative. The information about the customer service representative may be information registered when the user account is registered, or may be information obtained from application log information, etc.

[0018] Information about the customer service person includes, but is not limited to, the customer service person's attributes, personality, preference information, and compatibility with stress. Attributes may be, for example, at least one of gender, age, generation, and region (such as address or workplace location). Personality may be input in advance from the customer service person, or may be estimated from a questionnaire or the like. Preference information may be estimated based on the application's behavioral history (such as historical information about a specific product or service). Preference information may be expressed as, for example, liking anime or sports. Compatibility with stress may be expressed as, for example, anger expressions and overbearing expressions are OK (tolerance exists), and aggressive expressions are NG (not tolerance exists).

[0019] The stress expression database may store stress expressions that are expressions that may cause stress. In one example, the stress expression database may store stress expressions such as "hurry up."

[0020] The correction suggestion database may store correction suggestions for sentences for each stress expression. The correction suggestion database stores stress expressions before correction (correction) and expressions after correction. The correction suggestion database may store stress expressions before correction and expressions after correction in association with a stress level, which is the degree of stress felt. The stress level is information indicating the degree of stress felt by the customer service representative. For example, the stress level may be a score or a rank. The correction suggestion database may store stress expressions before correction and expressions after correction in association with information about the customer service representative.

[0021] The advice suggestion database may store advice suggestions regarding customer service. The advice suggestions may be advice suggestions for reducing a customer's anger. Examples of advice suggestions include, but are not limited to, example replies to messages, wording, and apologies. The advice suggestion database may store advice suggestions in association with the customer's age. The advice suggestion database may store advice suggestions in association with keywords included in messages.

[0022] The character database may store information about characters. The information about characters may be information indicating a specific character. The information about characters may also be a display method for the character, etc. The display method for the character is a method for displaying the character on the terminal 10 of the customer service representative, and may be, for example, specifying an image or video. The character database may store stress levels and information about characters in association with each other. The character database may store information about customer service representatives and information about characters in association with each other.

[0023] The voice database may store information about voice. The information about voice may be, for example, voice data of a calm tone. The information about voice may be, for example, voice data of a voice actor. The voice database may store stress levels and information about voice in association with each other.

[0024] The server device 30 is a device that stores a generative AI model 31 and enables the provision of questions or answers to a user using the generative AI model 31. The generative AI model is a model that, in response to the input of a prompt including input information, generates content according to any one or a combination of the instructions, context, question, and output format indicated by the prompt and returns the content as response information. The prompt may also include input information, in which case the generative AI model generates response information targeted at the input information. The generative AI model may be, for example, an interactive AI model that includes a large-scale language model (LLM) and a user interface (UI) for interacting with the user, enabling text or voice chat with the user. Examples of such generative AI models include ChatGPT, GPT (registered trademark)-3.5, GPT-4V, PaLM2, etc. In this embodiment, an example is described in which the server device 30 provides questions, etc. using one generative AI model 31. However, the server device 30 may also provide questions, etc. using multiple generative AI models. 1 illustrates only one server device 30, the generation system may include multiple server devices 30. Furthermore, although the above describes an example of a large-scale language model, other AI models may also be used.

[0025] The RAG system 20 is configured to include, as functional components, a reception unit 21, a control unit 22, a generation unit 23, and an acquisition unit 24. The RAG system 20 relays prompts corresponding to input information from the customer's terminal 10 to the server device 30, and relays response information from the server device 30 to the prompts to the customer service representative's terminal 10. The RAG system 20 also has a function to generate prompts based on input information from the terminal 10. The functions of each functional unit of the RAG system 20 will be described in detail below.

[0026] The reception unit 21 receives a message from a customer. For example, the reception unit 21 receives a message from the customer's terminal 10. The reception unit 21 also receives information about a customer service representative. For example, the reception unit 21 obtains information about the customer service representative from the storage device 15.

[0027] For example, the message may be text data of a chat or email, or may be text data obtained by performing voice recognition on the contents of a telephone utterance.

[0028] The content of the message is not limited. For example, the message may be an angry sentence such as "Hurry up." The message may also be a stressful sentence such as a nagging. In one example, the message may be a phrase pointing out a mistake, such as "You forgot to fill out XX. Why can't you do that?" In another example, the message may be "Has your application for XX been processed yet?"

[0029] The control unit 22 acquires information about stress from the message received by the receiving unit 21. The information about stress may be a stress level. For example, the control unit 22 may determine the stress level using a learning model that has previously learned stress expressions and stress levels. The control unit 22 determines the stress level by inputting the message into the learning model.

[0030] For example, the learning model may learn features obtained by vectorizing stress expressions and the results of a questionnaire on whether the stress expressions cause stress. The questionnaire may be conducted in advance for a large number of customer service personnel. In one example, the questionnaire may be in a format in which the level of stress felt in response to the stress expression "hurry up" is rated on a five-point scale (e.g., 1: not felt, 2: hardly felt, 3: depending on the situation, 4: felt somewhat, and 5: felt).

[0031] The control unit 22 may determine the stress level as a score. For example, the control unit 22 may determine the stress level as a range of 0 to 100 points. The control unit 22 may classify the stress level as a rank such as "high," "normal," or "low." For example, the higher the stress level value indicated as a score, the more angrier the customer is presumed to be. Furthermore, the more the stress level indicated as a classification is ranked, such as "high," the more angrier the customer is presumed to be.

[0032] The stress level may be information indicating the degree of stress expressed by the customer service representative. The stressful expression may be an expression including anger from the customer, an expression indicating an attack on the customer service representative, or an overbearing expression. The stress level may be at least one of information indicating an anger level indicating the degree of anger from the customer, an aggression level indicating the degree of attack on the customer service representative, and an overbearing level indicating the degree of overbearing behavior.

[0033] For example, the learning model may learn messages received from customers and information indicating the anger level, the aggression level, and the pressure level. The learning model may output scores or classifications regarding the anger level, the aggression level, and the pressure level for the input message. In one example, the learning model may output an anger level of 78 (or "high"), an aggression level of 68 (or "high"), and a pressure level of 73 (or "high") for the stress expression "hurry up."

[0034] The control unit 22 determines information regarding a correction policy for correcting the message based on the information regarding stress. The information regarding the correction policy is information regarding a policy for correcting the message to change the impression of the message recipient. For example, the information regarding the correction policy is information regarding a policy for correcting a sentence received from a customer to a sentence that reduces the stress level. For example, the information regarding the correction policy may be information indicating a policy for changing the sentence. The control unit 22 may determine whether or not to correct the message based on the stress level.

[0035] In one example, the control unit 22 may determine whether or not correction of the message is "necessary" when the stress level indicated as the score is higher than a predetermined threshold (or equal to or greater than the threshold). The control unit 22 may determine whether or not correction of the message is "not necessary" when the stress level indicated as the score is lower than a predetermined threshold (or equal to or less than the threshold).

[0036] In another example, when the stress level indicated as a classification satisfies a predetermined condition (for example, the classification is "high"), the control unit 22 may determine whether or not correction of the message is "necessary." When the stress level indicated as a classification does not satisfy a predetermined condition (for example, the classification is "low" or "normal"), the control unit 22 may determine whether or not correction of the message is "not necessary."

[0037] The control unit 22 may determine information on a correction policy for each stress expression included in the message. For example, the control unit 22 may extract stress expressions included in the message and obtain a correction plan for each stress expression from the correction plan database in the storage device 15.

[0038] For example, the control unit 22 may extract stress expressions by dividing and matching the message. For example, the control unit 22 may divide the message by morphological analysis, syntactic analysis, etc. In one example, the message is "Hurry up, don't make me wait any longer, you idiot." The control unit 22 may divide the message into "Hurry up, don't make me wait any longer, you idiot." Here, " / " indicates the division point. The control unit 22 may extract stress expressions from the message by matching the divided message with stress expressions obtained from the storage device 15.

[0039] The control unit 22 may determine information regarding the correction policy for the entire message. For example, the control unit 22 may determine, as information regarding the correction policy, that at least one of an exclamation mark “!”, an emoticon, and an emoji be added to the end of every sentence.

[0040] The control unit 22 may determine a presentation policy for presenting a correction message to the customer service person based on at least one of information about the customer service person and information about stress.

[0041] The control unit 22 may determine a policy for presenting advice regarding how to deal with a customer as the presentation policy. Advice is information regarding a response to a customer's message. Advice can also be considered additional information for a correction message. The control unit 22 may determine whether or not to present advice based on the stress level. The control unit 22 may use the customer's age to obtain suggested advice from the advice proposal database in the storage device 15. The control unit 22 may use keywords included in the message to obtain suggested advice from the advice proposal database in the storage device 15. The control unit 22 may decide to use suggested advice as the advice presentation policy.

[0042] For example, the control unit 22 may determine whether or not it is necessary to present advice as "necessary" when the stress level indicated as a score is higher than a predetermined threshold (or equal to or higher than the threshold). The control unit 22 may determine whether or not it is necessary to present advice as "not necessary" when the stress level indicated as a score is lower than a predetermined threshold (or equal to or lower than the threshold).

[0043] The control unit 22 may determine a presentation policy for characters representing customers as the presentation policy. The control unit 22 may acquire information about characters from a character database in the storage device 15. The control unit 22 may acquire information about characters from the storage device 15 using information about customer service personnel. The control unit 22 may determine to use information about characters as the character presentation policy.

[0044] The control unit 22 may determine a voice presentation policy as the presentation policy. When a customer service representative is handling a phone call, the control unit 22 may determine a policy of converting the customer's voice into another voice and presenting it to the customer service representative. The control unit 22 may use the stress level to acquire information about the voice from the voice database in the storage device 15. The control unit 22 may determine to use information about the voice as the voice presentation policy. The control unit 22 may determine to create synthetic voice as the voice presentation policy.

[0045] The generation unit 23 generates a prompt for instructing the generation of a corrected message by correcting the message based on the message, information about the correction policy, and information about the customer service representative. The message can be considered a generation request to the generation AI. The correction message can be considered a generation result of the generation AI.

[0046] The generation unit 23 may generate the prompt further based on the presentation policy. For example, the generation unit 23 may generate the prompt further based on at least one of an advice presentation policy, a character presentation policy, and a voice presentation policy.

[0047] The prompt may include, for example, text expressions representing commands to be executed by the generative AI model (interactive AI model), tasks to be executed by the generative AI model, background / context to be considered by the generative AI model (e.g., roles, conditions), questions to be answered by the generative AI model, and output formats of response information from the generative AI model. The prompt may also include input information to be used as the target of commands / tasks to be executed by the generative AI model. Examples of such input information include data files with file names that include a predetermined extension, such as text data, image data, application-related data, audio data, video data, and still image data. Application-related data includes document data, table data, graph data, and other data that can be processed by a default application program.

[0048] The acquisition unit 24 inputs a prompt to the generation AI model 31 that generates the correction message, and acquires the correction message from the generation AI model 31. The acquisition unit 24 may relay the correction message, which is a response result to the input of the prompt, to the terminal 10 of the customer service representative.

[0049] 2 is a diagram showing an example of a correction proposal database stored in the storage device 15. For example, the correction proposal database may store stress expressions before correction and expressions after correction in association with information about a customer service person. The correction proposal database may store stress expressions before correction and expressions after correction in association with a stress level (for example, a range of stress levels indicated as a score). In one example, the correction proposal database stores a stress expression before correction "Hurry up" and a corrected expression "I'd be happy if you could respond quickly!" in association with information about a customer service person, "I like anime," and a stress level of "40 to 60 points."

[0050] 3 is a diagram showing an example of an advice suggestion database stored in the storage device 15. For example, the advice suggestion database may store advice suggestions in association with the customer's age and keywords. Keywords are specific words contained in messages and may be the same as stress expressions. In one example, the advice suggestion database stores the advice suggestion "Use more polite language" in association with the customer's age of "60s" and the keyword "irritation."

[0051] Fig. 4 is a diagram showing an example of a prompt generated by the RAG system 20. The prompt shown in Fig. 4 includes a role, a task, a condition, and input information.

[0052] The role may correspond to, for example, a presentation policy for a character. For example, a role may be assigned to a specific character as a reference for correcting a message. An example of a role is "You are character X."

[0053] The task is information that indicates an outline of the command content to the generative AI model 31. An example of a task is "Please become character X and revise the following sentence to an expression that does not cause stress to the user."

[0054] The condition is, for example, a condition that the corrected message must satisfy. An example of the condition is, "Please revise the message to a non-stressful expression." The non-stressful expression may be a sentence that does not contain a stressful expression.

[0055] The input information is information used to create the correction message. For example, the input information includes the sentence to be corrected and the sentence to be used as reference. The sentence to be corrected includes the customer's message and stress level. The customer's message is a message received by the reception unit 21. The stress level is a stress level indicated as a score determined by the control unit 22. An example of a customer's message is, "Hurry up! How many minutes are you going to keep me waiting? Don't be ridiculous." An example of a stress level is "stress level 76."

[0056] The reference sentences include stress expressions before correction, expressions after correction, information about the customer service person, and stress levels. For example, the reference sentences may be content obtained from a correction proposal database. An example of a stress expression before correction is "hurry up." An example of an expression after correction is "please hurry up! I can't wait any longer~(>_<)." An example of information about the customer service person is "I like anime." An example of a stress level is "80 to 100 points."

[0057] The prompt may include a presentation policy. For example, the prompt may include at least one of an advice presentation policy, a character presentation policy, and a voice presentation policy.

[0058] For example, the prompt may include a designation to associate a correction message with a character. The prompt may include a designation to display an icon (image) of character X speaking the correction message. The prompt may include a designation to change the display manner of the character depending on the stress level. In one example, the prompt may include a designation to change the facial expression, posture, etc. of character X depending on the anger level.

[0059] The prompt may include a designation to associate a corrective message with the advice. The prompt may include a designation to vary the wording of the advice depending on customer information (such as the customer's age).

[0060] The prompt may include instructions to output a correction message by voice, and may include instructions to change the voice depending on the stress level.

[0061] An example of a method (generation method) according to this embodiment will be described with reference to Fig. 5. Fig. 5 is a flowchart showing an example of the operation of the RAG system 20.

[0062] In step S1, the reception unit 21 receives a message from a customer. For example, the reception unit 21 receives a message from the customer's terminal 10. The reception unit 21 also receives information about a customer service representative. For example, the reception unit 21 obtains information about the customer service representative from the storage device 15.

[0063] In step S2, the control unit 22 acquires information about stress from the message received by the receiving unit 21. The information about stress may be a stress level. For example, the control unit 22 may determine the stress level using a learning model that has previously learned stress expressions and stress levels. The control unit 22 determines the stress level by inputting the message into the learning model.

[0064] In step S3, the control unit 22 determines information regarding a correction policy for correcting the message based on the information regarding stress. The control unit 22 may determine whether or not the message needs to be corrected based on the stress level.

[0065] The control unit 22 may determine information on a correction policy for each stress expression included in the message. For example, the control unit 22 may extract stress expressions included in the message and obtain a correction plan for each stress expression from the correction plan database in the storage device 15.

[0066] The control unit 22 may determine information regarding the correction policy for the entire message. For example, the control unit 22 may determine, as information regarding the correction policy, that at least one of an exclamation mark “!”, an emoticon, and an emoji be added to the end of every sentence.

[0067] The control unit 22 may determine a presentation policy for presenting a correction message to the customer service person based on at least one of information about the customer service person and information about stress.

[0068] The control unit 22 may determine a policy for presenting advice regarding how to deal with the customer as the presentation policy. The control unit 22 may determine whether or not to present advice based on the stress level. The control unit 22 may use the age of the customer to acquire suggested advice from the advice proposal database in the storage device 15. The control unit 22 may use keywords included in the message to acquire suggested advice from the advice proposal database in the storage device 15. The control unit 22 may decide to use suggested advice as the advice presentation policy.

[0069] The control unit 22 may determine a presentation policy for characters representing customers as the presentation policy. The control unit 22 may acquire information about characters from a character database in the storage device 15. The control unit 22 may acquire information about characters from the storage device 15 using information about customer service personnel. The control unit 22 may determine to use information about characters as the character presentation policy.

[0070] The control unit 22 may determine a voice presentation policy as the presentation policy. When a customer service representative is handling a phone call, the control unit 22 may determine a policy of converting the customer's voice into another voice and presenting it to the customer service representative. The control unit 22 may use the stress level to acquire information about the voice from the voice database in the storage device 15. The control unit 22 may determine to use information about the voice as the voice presentation policy. The control unit 22 may determine to create synthetic voice as the voice presentation policy.

[0071] In step S4, the generation unit 23 generates a prompt for instructing the generation of a corrected message by correcting the message based on the message, the information on the correction policy, and the information on the customer service representative. In one example, the generation unit 23 may generate the prompt shown in FIG.

[0072] The generation unit 23 may generate the prompt further based on the presentation policy. For example, the generation unit 23 may generate the prompt further based on at least one of an advice presentation policy, a character presentation policy, and a voice presentation policy.

[0073] In step S5, the acquisition unit 24 inputs a prompt to the generation AI model 31 that generates the correction message, and acquires the correction message from the generation AI model 31. The acquisition unit 24 may relay the correction message, which is a response result to the input of the prompt, to the terminal 10 of the customer service representative.

[0074] Below is an example of a customer message: Message from a customer in their 60s: "Because of the problem, I was unable to receive service at the time I wanted. I was annoyed because I had to wait an hour when I called."

[0075] An example of the generated advice is shown below: First advice: "To calm the other person's anger, empathize with them, understand their feelings, and offer a sincere apology. It is effective to first calm the customer's emotions by repeating what they said or proceeding at the customer's pace. Also, when dealing with an elderly person who is irritated, it is effective to use more polite language. Below are some example sentences for when you should respond. Please modify them as necessary."

[0076] Another example of generated advice is shown below: Second advice: "I see you were unable to receive the service at the time you wanted. We deeply apologize for the inconvenience caused to you due to a defect in our service. We will take your opinion seriously and strive to consider measures to prevent recurrence."

[0077] As described above, the RAG system 20 (device) according to one aspect of the present disclosure comprises a reception unit 21 that receives messages from customers and information about customer service personnel, a control unit 22 that determines information about a correction policy for correcting the message based on stress information obtained from the message, a generation unit 23 that generates a prompt to instruct the generation of a corrected message that corrects the message based on the message, information about the correction policy, and information about the customer service personnel, and an acquisition unit 24 that inputs a prompt to a generation AI model 31 that generates the correction message and acquires the correction message from the generation AI model 31.

[0078] A method according to one aspect of the present disclosure includes the steps of receiving a message from a customer and information about a customer service representative; determining information about a correction policy for correcting the message based on information about stress obtained from the message; generating a prompt to instruct the generation of a corrected message by correcting the message based on the message, information about the correction policy, and information about the customer service representative; and inputting the prompt to a generation AI model 31 that generates the correction message and obtaining the correction message from the generation AI model 31.

[0079] In an apparatus and method according to one aspect of the present disclosure, information regarding a correction policy is determined based on stress information acquired from a message, a prompt for instructing the generation of a correction message is generated based on the message, information regarding the correction policy, and information regarding the customer service person, and the prompt is input to the generation AI model 31. Because the generation AI model 31 is controlled by the prompt that takes into account the information regarding the correction policy and the information regarding the customer service person, a correction message is generated that corresponds to the information regarding the correction policy and the information regarding the customer service person. As a result, the stress of the customer service person can be reduced.

[0080] The control unit 22 determines information on a correction policy for each stress expression included in the message. By determining information on a correction policy for each stress expression, a more detailed correction message can be generated.

[0081] The control unit 22 determines information regarding the correction policy for the entire message. By determining information regarding the correction policy for the entire message, it is possible to improve the sense of unity of the correction message.

[0082] The control unit 22 determines a presentation policy for presenting a correction message to the customer service person based on at least one of information about the customer service person and information about stress. The generation unit 23 generates a prompt based further on the presentation policy. The presentation policy based on at least one of information about the customer service person and information about stress is reflected in the prompt. By taking the presentation policy into consideration when presenting the correction message, stress of the customer service person can be further reduced.

[0083] The control unit 22 determines the advice presentation policy for dealing with customers as the presentation policy. By reflecting the advice presentation policy in the prompt, the customer service representative can quickly find an appropriate approach to the customer. As a result, the customer service representative's stress can be further reduced.

[0084] The control unit 22 determines a presentation policy for a character representing a customer as the presentation policy. By reflecting the character presentation policy in the prompt, the customer service staff can reduce stress caused by imagining the customer's facial expression, etc.

[0085] The control unit 22 determines a voice presentation policy as the presentation policy. By reflecting the voice presentation policy in the prompt, stress can be reduced when responding to a voice call or the like.

[0086] The generation system is not limited to the configuration shown in FIG. 1 , but may be configured as shown in FIG. 6 in which the generative AI model 31 is implemented in the RAG system 20. This configuration can be realized by installing an application that executes the functions of the generative AI model 31 in the RAG system 20. Also, while FIGS. 1 and 6 show an example in which the storage device 15 is implemented outside the RAG system 20 (for example, on a network), the storage device 15 may also be implemented in the RAG system 20.

[0087] The generative AI model 31 may be implemented in a terminal or on a network. The RAG system 20 may be implemented in a terminal or on a network. The storage device 15 may be implemented in a terminal or on a network.

[0088] The present disclosure has the following configuration. [1] An apparatus including: a receiving unit that receives a message from a customer and information about a customer service representative; a control unit that determines information about a correction policy for correcting the message based on stress information acquired from the message; a generation unit that generates a prompt for instructing the generation of a corrected message obtained by correcting the message based on the message, information about the correction policy, and information about the customer service representative; and an acquisition unit that inputs the prompt to a generative AI model that generates the correction message and acquires the correction message from the generative AI model. [2] The apparatus described in [1], wherein the control unit determines information about the correction policy for each stress expression included in the message. [3] The apparatus described in [1] or [2], wherein the control unit determines information about the correction policy for the entire message. [4] The apparatus described in any of [1] to [3], wherein the control unit determines a presentation policy for presenting the correction message to the customer service representative based on at least one of information about the customer service representative and information about stress, and the generation unit generates the prompt further based on the presentation policy. [5] The device according to [4], wherein the control unit determines, as the presentation policy, a presentation policy for advice regarding how to deal with the customer. [6] The device according to [4] or [5], wherein the control unit determines, as the presentation policy, a presentation policy for a character representing the customer. [7] The device according to any of [4] to [6], wherein the control unit determines, as the presentation policy, a presentation policy for audio.[8] A method comprising: a step of receiving a message from a customer and information about a customer service representative; a step of determining information about a correction policy for correcting the message based on information about stress obtained from the message; a step of generating a prompt to instruct the generation of a corrected message that corrects the message based on the message, information about the correction policy, and information about the customer service representative; and a step of inputting the prompt to a generative AI model that generates the correction message and obtaining the correction message from the generative AI model.

[0089] The block diagrams used to explain the above embodiments show functional blocks. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method for realizing each functional block is not particularly limited. That is, each functional block may be realized using a single device that is physically or logically coupled, or may be realized using two or more physically or logically separated devices that are directly or indirectly connected (e.g., wired, wireless, etc.) and these multiple devices. The functional block may also be realized by combining software with the single device or multiple devices.

[0090] Functions include, but are not limited to, judgment, determination, assessment, calculation, computation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, consideration, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assignment. For example, a functional block (component) that performs transmission is called a transmitting unit or transmitter. As mentioned above, there are no particular limitations on how these functions are implemented.

[0091] For example, the RAG system 20 constituting the generation system according to an embodiment of the present disclosure may function as a computer that performs processing of the control method of the present disclosure. Figure 7 is a diagram illustrating an example of the hardware configuration of the RAG system 20 according to an embodiment of the present disclosure. The above-described RAG system 20 may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, and the like. The RAG system 20 may be configured as a computer device including at least one processor such as a CPU or GPU, or may be configured as a computer device including multiple processors or may be configured to include multiple computer devices. The terminal 10 and the server device 30 may also have a similar hardware configuration.

[0092] In the following description, the term "device" can be interpreted as a circuit, a device, a unit, etc. The hardware configuration of the RAG system 20 may be configured to include one or more of the devices shown in the figure, or may be configured to exclude some of the devices.

[0093] Each function in the RAG system 20 is realized by loading specified software (programs) onto hardware such as a processor 1001 and memory 1002, causing the processor 1001 to perform calculations, control communication via a communication device 1004, and control at least one of reading and writing data in the memory 1002 and storage 1003.

[0094] The processor 1001 controls the entire computer by running, for example, an operating system. The processor 1001 may be configured by a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic unit, a register, etc. For example, the above-mentioned reception unit 21, control unit 22, generation unit 23, acquisition unit 24, etc. may be realized by the processor 1001.

[0095] The processor 1001 also reads programs (program codes), software modules, data, etc. from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes in accordance with the programs. The programs used are programs that cause a computer to execute at least some of the operations described in the above-described embodiments. For example, the reception unit 21, the control unit 22, the generation unit 23, and the acquisition unit 24 may be implemented by a control program stored in the memory 1002 and running on the processor 1001, and similar implementations may be used for other functional blocks. While the above-described various processes have been described as being executed by a single processor 1001, they may also be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The programs may also be transmitted from a network via a telecommunications line.

[0096] The memory 1002 is a computer-readable recording medium and may be configured, for example, by at least one of a read-only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a random access memory (RAM), etc. The memory 1002 may also be called a register, a cache, a main memory (primary storage device), etc. The memory 1002 can store executable programs (program codes), software modules, etc. for implementing a control method according to an embodiment of the present disclosure.

[0097] Storage 1003 is a computer-readable recording medium, and may be composed of at least one of, for example, an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disk), a smart card, a flash memory (e.g., a card, a stick, a key drive), a floppy (registered trademark) disk, a magnetic strip, etc. Storage 1003 may also be referred to as an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, a server, or other appropriate medium including at least one of memory 1002 and storage 1003.

[0098] The communication device 1004 is hardware (transmission / reception device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as, for example, a network device, a network controller, a network card, a communication module, etc. The communication device 1004 may be configured to include a high-frequency switch, a duplexer, a filter, a frequency synthesizer, etc. to realize at least one of frequency division duplex (FDD) and time division duplex (TDD). For example, the above-mentioned reception unit 21, acquisition unit 24, etc. may be realized by the communication device 1004.

[0099] The input device 1005 is an input device (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that accepts input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, an LED lamp, etc.) that outputs to the outside. Note that the input device 1005 and the output device 1006 may be integrated into one device (e.g., a touch panel).

[0100] Furthermore, each device, such as the processor 1001 and the memory 1002, is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or may be configured using different buses between each device.

[0101] The RAG system 20 may also be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a field programmable gate array (FPGA), and some or all of the functional blocks may be realized by the hardware. For example, the processor 1001 may be implemented using at least one of these pieces of hardware.

[0102] The notification of information is not limited to the aspects / embodiments described in the present disclosure and may be performed using other methods. For example, the notification of information may be performed by physical layer signaling (e.g., Downlink Control Information (DCI) and Uplink Control Information (UCI)), higher layer signaling (e.g., Radio Resource Control (RRC) signaling, Medium Access Control (MAC) signaling, broadcast information (Master Information Block (MIB) and System Information Block (SIB))), other signals, or a combination thereof. Furthermore, the RRC signaling may be referred to as an RRC message, and may be, for example, an RRC Connection Setup message, an RRC Connection Reconfiguration message, or the like.

[0103] The order of the procedures, sequences, flowcharts, etc. of each aspect / embodiment described in this disclosure may be changed unless it is consistent. For example, the methods described in this disclosure present elements of various steps using an example order, and are not limited to the particular order presented.

[0104] Input and output information may be stored in a specific location (for example, memory) or may be managed using a management table. Input and output information may be overwritten, updated, or added to. Output information may be deleted. Input information may be sent to another device.

[0105] The determination may be made based on a value represented by one bit (0 or 1), a Boolean value (true or false), or a numerical comparison (e.g., comparison with a predetermined value).

[0106] The aspects / embodiments described in this disclosure may be used alone, in combination, or switched depending on the implementation. Notification of predetermined information (e.g., notification that "X is true") is not limited to explicit notification, but may be implicit (e.g., not notifying the predetermined information).

[0107] Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the description of the present disclosure is intended to be illustrative and does not have any limiting meaning on the present disclosure.

[0108] Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0109] Software, instructions, information, etc. may also be transmitted or received over a transmission medium. For example, if software is transmitted from a website, server, or other remote source using wired technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave), then these wired and / or wireless technologies are included within the definition of transmission media.

[0110] The information, signals, etc. described in this disclosure may be represented using any of a variety of different technologies. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0111] Note that terms described in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of a channel and a symbol may be a signal (signaling). Furthermore, a signal may be a message. Furthermore, a component carrier (CC) may be called a carrier frequency, a cell, a frequency carrier, etc.

[0112] Furthermore, the information, parameters, etc. described in the present disclosure may be expressed using absolute values, relative values ​​from a predetermined value, or other corresponding information. For example, a radio resource may be indicated by an index.

[0113] The names used for the above-described parameters are not intended to be limiting in any way. Furthermore, the mathematical expressions using these parameters may differ from those explicitly disclosed in this disclosure. The various channels (e.g., PUCCH, PDCCH, etc.) and information elements may be identified by any suitable names, and therefore the various names assigned to these various channels and information elements are not intended to be limiting in any way.

[0114] In this disclosure, the terms "Mobile Station (MS)," "user terminal," "User Equipment (UE)," "terminal," and the like may be used interchangeably.

[0115] A mobile station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or some other suitable terminology.

[0116] As used in this disclosure, the terms "determining" and "determining" may encompass a wide variety of actions. "Determining" and "determining" may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, inquiring (e.g., searching in a table, database, or other data structure), ascertaining, and the like. "Determining" and "determining" may also include receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in memory), and the like. Furthermore, "judgment" and "decision" can include regarding resolving, selecting, choosing, establishing, comparing, etc. as having been "judged" or "decided." In other words, "judgment" and "decision" can include regarding some action as having been "judged" or "decided." Furthermore, "judgment (decision)" can be interpreted as "assuming," "expecting," "considering," etc.

[0117] The terms "connected," "coupled," or any variation thereof, refer to any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are "connected" or "coupled" to each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, "connected" may be read as "access." As used in this disclosure, two elements may be considered to be "connected" or "coupled" to each other using one or more wires, cables, and / or printed electrical connections, as well as electromagnetic energy having wavelengths in the radio frequency range, microwave range, and optical (both visible and invisible) range, as some non-limiting and non-exhaustive examples.

[0118] As used in this disclosure, the phrase "based on" does not mean "based only on," unless expressly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."

[0119] As used in this disclosure, any reference to an element using a designation such as "first," "second," etc. does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient method of distinguishing between two or more elements. Thus, a reference to a first and a second element does not imply that only two elements may be employed or that the first element must in some way precede the second element.

[0120] When the terms "include," "including," and variations thereof are used in this disclosure, these terms are intended to be inclusive, similar to the term "comprising." Furthermore, when the term "or" is used in this disclosure, it is not intended to be an exclusive or.

[0121] In this disclosure, where articles are added by translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural form.

[0122] In the present disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "coupled" may also be interpreted in the same way as "different."

[0123] 10...terminal, 20...RAG system (device), 21...reception unit, 22...control unit, 23...generation unit, 24...acquisition unit, 30...server device, 31...generated AI model

Claims

1. An apparatus comprising: a reception unit that receives a message from a customer and information about a customer service representative; a control unit that determines information about a correction policy for correcting the message based on information about stress obtained from the message; a generation unit that generates a prompt to instruct the generation of a corrected message that corrects the message based on the message, information about the correction policy, and information about the customer service representative; and an acquisition unit that inputs the prompt to a generation AI model that generates the correction message and acquires the correction message from the generation AI model.

2. The device according to claim 1, wherein the control unit determines information relating to the correction policy for each stress expression included in the message.

3. The device according to claim 1, wherein the control unit determines information regarding the correction policy for the entire message.

4. The device described in claim 1, wherein the control unit determines a presentation policy for presenting the correction message to the customer service representative based on at least one of information about the customer service representative and information about stress, and the generation unit generates the prompt further based on the presentation policy.

5. The device according to claim 4, wherein the control unit determines, as the presentation policy, a policy for presenting advice regarding how to deal with the customer.

6. The device according to claim 4, wherein the control unit determines, as the presentation policy, a presentation policy for a character representing the customer.

7. The device according to claim 4, wherein the control unit determines a presentation policy for audio as the presentation policy.

8. A method comprising the steps of: accepting a message from a customer and information about a customer service representative; determining information about a correction policy for correcting the message based on stress information obtained from the message; generating a prompt to instruct the generation of a corrected message that corrects the message based on the message, information about the correction policy, and information about the customer service representative; and inputting the prompt into a generative AI model that generates the correction message and obtaining the correction message from the generative AI model.

Images