Interaction system, interaction method, and interaction program

Abstract

To provide more smooth communication.SOLUTION: An interaction system includes: a first character string acquisition unit which acquires a first character string generated based on a first speech of a user using voice or texts; a first reply unit which provides, to the user, a first reply character string for the first speech using voice or texts, the first reply character string being generated based on the first character string; a character string generation request generation unit which generates character string generation request information for generating a second reply character string for the first speech in an external information processing system; and a second reply unit which provides, to the user, a second character string generated based on the character string generation request information, using voice or texts.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] 2. Description of the Related Art Conventionally, there is known a technique for diagnosing a user's symptoms using a dialogue system and providing medical care.

[0003] For example, the communication robot described in Patent Document 1 recognizes the user's voice input from a microphone, estimates the user's biological state based on the acoustic characteristics of the voice input from the microphone, and outputs from a speaker a voice determined by a conversation engine in accordance with the estimation result. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-100221 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the communication robot described in Patent Document 1 is premised on responding to a user's voice with a voice determined by a single conversation engine. Therefore, if the conversation engine's processing is delayed, for example, an interval will increase between the end of the user's speech and the communication robot's response, which will reduce the user's satisfaction with the communication and prevent the robot from providing smooth communication to the user.

[0006] Therefore, an object of the present invention is to provide smoother communication. [Means for solving the problem]

[0007] A dialogue system according to one embodiment of the present invention includes a first string acquisition unit that acquires a first string generated based on a user's first utterance in the form of voice or text, a first reply unit that provides the user with a first reply string to the first utterance, generated based on the first string, in the form of voice or text, a string generation request generation unit that generates string generation request information for generating a second reply string to the first utterance in an external information processing system, and a second reply unit that provides the user with the second reply string generated based on the string generation request information in the form of voice or text.

[0008] In this invention, a "unit" does not simply mean a physical means, but also includes cases where the functions of the "unit" are realized by software. Furthermore, the functions of one "unit" or device may be realized by two or more physical means, devices, or software, and the functions of two or more "units" or devices may be realized by one physical means, device, or software. [Effects of the Invention]

[0009] According to the present invention, smoother communication can be provided. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing an overview of processing in a dialogue system 100 according to an embodiment of the present invention. [Figure 2] 1 is a diagram showing the configuration of a dialogue system 100 according to an embodiment of the present invention. [Figure 3] 10 is a diagram showing an example of first character string information stored in storage unit 105. FIG. [Figure 4] 10 is a diagram showing an example of a response character string selection process performed by a group selection unit 120 and a character string selection unit 125. FIG. [Figure 5] FIG. 10 is a diagram showing an example of first reply string information stored in the storage unit 105. [Figure 6] 10 is a diagram showing an example of prompt information stored in the storage unit 105. FIG. [Figure 7] 10 is a diagram showing an example of second reply string information stored in the storage unit 105. FIG. [Figure 8] 10 is a flowchart showing an example of processing in the dialogue system 100. [Figure 9] 10 is a flowchart showing an example of processing in the dialogue system 100. [Figure 10] 10 is a flowchart showing an example of processing in the dialogue system 100. [Figure 11] 10 is a flowchart showing an example of processing in the dialogue system 100. [Figure 12] 10 is a flowchart showing an example of processing in the dialogue system 100. DETAILED DESCRIPTION OF THE INVENTION

[0011] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will now be described with reference to the accompanying drawings, in which: Figure 1 is a diagram showing an overview of processing in a dialogue system 100 according to an embodiment of the present invention;

[0012] The dialogue system 100 is an information processing system implemented by a dialogue program. The dialogue system 100 dialogues with a user through voice or text.

[0013] First, the dialogue system 100 acquires, for example, from a user, a first character string generated based on a first utterance of the user (S101). The dialogue system 100 generates a first reply character string based on the first character string (S102) and provides the first reply character string to the user (S103).

[0014] Next, the dialogue system 100 generates string generation request information (e.g., a prompt) for generating a second reply string to the first utterance in an external information processing system (e.g., a string generation system 200 described later) (S104). The dialogue system 100 outputs the string generation request information (e.g., the prompt) to the string generation system 200 (S105).

[0015] Then, the dialogue system 100 acquires the second reply string generated by the string generation system 200 based on the string generation request information (for example, a prompt) (S106), and provides it to the user (S107).

[0016] In this way, the dialogue system 100 replies to the user using a first reply string generated within the dialogue system 100, and further replies to the user using a second reply string generated in an external information processing system (for example, the string generation system 200). In other words, the dialogue system 100 can quickly reply to the user using a first reply string in response to a user's utterance, and after replying with the first reply string, can further reply using the second string.

[0017] This allows the dialogue system 100 to communicate smoothly with the user and realize an uninterrupted dialogue with the user. That is, the dialogue system 100, for example, obtains the second reply string from an external information processing system in parallel with the process of providing the first reply string. This allows the dialogue system 100 to provide a dialogue that is closer to a human dialogue rather than a mechanical dialogue, without making the user feel like they are waiting for a reply from the dialogue system 100.

[0018] The user may be, for example, a patient suffering from cancer. In this regard, mental care has traditionally been provided to patients with diseases that are difficult to cure, such as terminal cancer, patients with mental illnesses, the elderly, and the like. In mental care, it is important for the user (e.g., a cancer patient) to feel that others (e.g., a caregiver or system) are there for them. For example, it is important for the user (e.g., a patient) to feel that their statements are being listened to. Furthermore, mental care using information processing technology has been attracting attention in recent years. However, in cases of mental care using information processing technology, if the user feels that they are interacting with a machine, the user will not feel that their statements are being listened to. Therefore, the dialogue system 100 responds to the user using a first reply string generated within the dialogue system 100 and then responds to the user using a second reply string generated in an external information processing system (e.g., the string generation system 200), thereby providing the user with a dialogue that is closer to a human dialogue than a mechanical dialogue.

[0019] 2 is a diagram showing the configuration of a dialogue system 100 according to one embodiment of the present invention. The dialogue system 100 is communicably connected to a character string generation system 200 via a network such as the Internet. Details of the dialogue system 100 will be described later.

[0020] The string generation system 200 is an information processing system that generates a second reply string based on string generation request information (e.g., a prompt) acquired from the dialogue system 100. The string generation system 200 can generate the second reply string based on machine learning technology or natural language processing technology, more specifically, based on, for example, a large-scale language model. Note that, although the following specifically describes a case where the string generation request information is a prompt, the string generation request information is not limited to a prompt and may be, for example, any information that requests the string generation system 200 to generate a second reply string.

[0021] The large-scale language model used by the string generation system 200 may be, for example, a Generative Pre-trained Transformer (GPT), particularly, for example, GPT-4. The string generation system 200 may be, for example, an information processing system that inputs prompts and outputs strings in a chat format, particularly, for example, ChatGPT.

[0022] Also, although only one character string generation system 200 is shown in FIG. 2, multiple character string generation systems 200 may be provided.

[0023] Next, details of the dialogue system 100 will be described. The dialogue system 100 includes a storage unit 105, a user information acquisition unit 110, a character string acquisition unit 115, a group selection unit 120, a character string selection unit 125, a first reply unit 130, a prompt generation unit 135, a prompt output unit 140, a second reply character string acquisition unit 145, a second reply unit 150, a cancellation processing unit 155, a determination unit 160, and a notification unit 165. Each unit shown in Fig. 2 can be realized, for example, by using a memory area or by having a processor execute a program stored in the memory area.

[0024] The storage unit 105 stores information to be processed in the dialogue system 100. The storage unit 105 can store, for example, user information, first string information, second string information, prompt information, first reply string information, second reply string information, and cancellation request information, which will be described later.

[0025] The user information acquisition unit 110 acquires user information and stores it in the storage unit 105. The user information acquisition unit 110 may acquire the user information, for example, from a user device used by the user, or may acquire the user information from a predetermined database (for example, a database in which information related to users is managed).

[0026] Here, the user information includes, for example, information about the user's attributes, information about the user's medical history, information about the user's utterance history, or information about the user's response history. The information about the user's attributes includes, for example, information about the user's age, sex, or family structure. The information about the user's medical history includes, for example, information about the user's medical examination records, surgery records, drug prescription records, or drug administration records.

[0027] The character string acquisition unit 115 acquires a character string generated based on a user's speech in the form of voice or text, and stores character string information relating to the character string in the storage unit 105.

[0028] When the user's utterance is a voice utterance, the character string acquisition unit 115 may acquire a character string generated by recognizing a voice acquired through a predetermined microphone, for example. When the user's utterance is input as text, the character string acquisition unit 115 may acquire text corresponding to the user's utterance from a user device used by the user that is communicably connected to the dialogue system 100 via a network such as the Internet.

[0029] The character string obtaining unit 115 can obtain a first character string generated based on a first utterance of a user. The character string obtaining unit 115 can also obtain a second character string generated based on a second utterance of a user in the form of voice or text, corresponding to a first reply character string described below. The character string obtaining unit 115 that obtains a first character string and the character string obtaining unit 115 that obtains a second character string may be distinguished from each other as a first character string obtaining unit and a second character string obtaining unit.

[0030] Here, the first character string may be, for example, a character string related to the user's answer to a question provided to the user by the dialogue system 100. The question may be, for example, a question related to the user's symptoms, such as "Are you in pain?" or "How are you feeling?" The first character string may be, for example, a response related to the user's symptoms, such as "I'm not in pain," "I feel unwell," or "I've had a fever of about 39°C since yesterday."

[0031] The first character string may include, for example, symptom information regarding the user's symptoms. Here, the user's symptoms may be, for example, symptoms of a disease the user has or side effects of a medicine used to treat the disease. The disease the user has is not particularly limited, and may be, for example, cancer, an incurable disease, or a mental illness. Furthermore, the symptoms may be, for example, pain, nausea, fatigue, nausea, fever, or depression.

[0032] The symptom information may include, for example, degree information regarding the degree of the symptom. The symptom degree information may be qualitative or quantitative information consisting of multiple levels, such as "no symptoms," "slight symptoms," "present symptoms," etc., or "symptom level 0," "symptom level 1," etc. The symptom degree information may particularly be, for example, pain degree information regarding the degree of pain, and may be qualitative or quantitative information consisting of multiple levels, such as "no pain," "slight pain," "severe pain," etc., or "pain level 0," "pain level 1," etc. The information regarding the degree of the symptom may be, for example, information regarding the degree of a symptom such as pain, nausea, nausea, fever, or depression.

[0033] Furthermore, the symptom information may be, for example, information about the site where the symptom occurs, or information about the time period when the symptom occurs.

[0034] The second character string is a character string of a user's utterance corresponding to the first reply character string to the first character string. The second character string may be, for example, a character string correcting the first character string or a character string correcting the first reply character string. Specifically, if the first character string is "It hurts a little," and the first reply character string is "It hurts a little, huh?", the second character string may be "It hurts a lot after all.", or if the first character string is "It hurts a little," and the first reply character string is "It doesn't hurt, huh?", the second character string may be "No, it hurts a little."

[0035] 3 is a diagram showing an example of first string information stored in storage unit 105. The first string information stored in storage unit 105 includes, for example, a string ID and comment content information. The string ID is information that identifies a string (first string or second string) stored in storage unit 105. The comment content information is information that indicates the content of a string generated based on a user's comment.

[0036] The group selecting unit 120 selects at least one group from a plurality of group candidates each including a plurality of reply string candidates, based on the symptom information included in the first string.

[0037] Here, in the dialogue system 100, for example, a plurality of group candidates are set in advance according to the content of the symptom (for example, the level of the symptom, the part of the body where the symptom is present, and the time period when the symptom is present). Each of the plurality of group candidates includes, for example, a plurality of reply string candidates according to the content of the symptom. More specifically, for example, a group candidate corresponding to the level of pain "none" includes reply string candidates such as "There's no pain, huh?" and "So you're saying there's no pain, huh?", and a group candidate corresponding to the level of pain "There's a little pain, huh?" and "There's only a little pain, huh?", etc.

[0038] The plurality of group candidates may also include group candidates including multiple reply string candidates that include symptom-related content and group candidates including reply string candidates that do not include symptom-related content. More specifically, for example, a group candidate corresponding to the pain level of "none" and including multiple reply string candidates that include symptom-related content may include reply string candidates such as "It's not painful, is it?" and "So it's not painful, then?"; a group candidate corresponding to the pain level of "none" and including multiple reply string candidates that do not include symptom-related content may include reply string candidates such as "That's good" and "I see."; and a group candidate corresponding to the pain level of "a little" and including multiple reply string candidates that do not include symptom-related content may include reply string candidates such as "That's tough."

[0039] For example, based on the symptom information included in the first character string, the group selection unit 120 can select at least one group according to the content of the symptom indicated by the symptom information. In this case, the group selection unit 120 may select multiple groups according to the content of the symptom indicated by the symptom information, in particular, for example, a group including multiple reply string candidates including content related to the symptom and a group including reply string candidates not including content related to the symptom.

[0040] The groups are not limited to groups based on the severity of the symptoms, but may be groups based on, for example, a disease thought to be the cause of the symptoms, a part of the body where the symptoms occur, or a time period when the symptoms occur.

[0041] Furthermore, the group selection unit 120 may select at least one group further based on the user information. Specifically, the group selection unit 120 may select a group corresponding to a disease suffered by the user based on information about a medical history included in the user information.

[0042] The string selection unit 125 selects at least one reply string from among the multiple reply string candidates included in at least one selected group.

[0043] The character string selection unit 125 selects at least one reply character string from among a plurality of reply character string candidates included in the selected group, such as "It doesn't hurt, does it?", "So it doesn't hurt?", "That's terrible.", etc.

[0044] When the group selecting unit 120 selects multiple groups, the string selecting unit 125 may select multiple reply strings by selecting at least one reply string from each of the multiple groups.

[0045] The string selection unit 125 may, for example, randomly select at least one reply string, or may select at least one reply string based on user information. Specifically, when the string selection unit 125 selects at least one reply string based on user information, it may select at least one reply string that has not been used as a reply string within a predetermined period of time based on information about the reply history included in the user information.

[0046] 4 is a diagram showing an example of a response string selection process performed by the group selection unit 120 and the string selection unit 125. An example of a response string selection process performed by the group selection unit 120 and the string selection unit 125 will be described with reference to FIG.

[0047] For example, in the dialogue system 100, a plurality of group candidates (e.g., group candidate 410 and group candidate 420) are set, and each of the plurality of group candidates includes a plurality of reply string candidates (e.g., reply string candidate 411 and reply string candidate 421). Here, group candidate 410 is, for example, a group candidate corresponding to the pain level of "none" and includes a plurality of reply string candidates that include content related to symptoms, and group candidate 420 is, for example, a group candidate corresponding to the pain level of "none" and includes a plurality of reply string candidates that do not include content related to symptoms.

[0048] The group selecting unit 120 selects the group candidates 410 and 420 based on, for example, symptom information corresponding to the pain level "none." The string selecting unit 125 selects, for example, at least one reply string candidate (e.g., reply string candidate 411 and reply string candidate 421) from each of the group candidates 410 and 420.

[0049] The response character string selection process performed by the group selection unit 120 and the character string selection unit 125 shown in FIG. 4 is an example, and the selection process method is not limited to this.

[0050] The first reply unit 130 generates a first reply string based on the first string and provides the first reply string to the first comment to the user by voice or text. At this time, first reply string information about the provided first reply string may be stored in the storage unit 105.

[0051] The first reply unit 130 may provide the user with a first reply string selected from a plurality of preset reply string candidates, for example. In this case, the first reply unit 130 may select a reply string at random, for example. Alternatively, the first reply unit 130 may select a reply string based on user information, for example. Specifically, the first reply unit 130 may select a reply string corresponding to the illness the user suffers from from a plurality of preset reply string candidates based on medical history information included in the user information. Alternatively, the first reply unit 130 may select a reply string that has not been used as a reply string within a predetermined period from a plurality of preset reply string candidates based on information about the reply history included in the user information.

[0052] Furthermore, the first reply section 130 can provide the user with at least one reply string selected by the string selection section 125 as a first reply string, for example.

[0053] Furthermore, the first reply unit 130 can generate a first reply string based on, for example, the multiple reply strings selected by the string selection unit 125, and provide it to the user. In this case, the first reply unit 130 can provide the user with a first reply string that is, for example, a combination of the selected multiple reply strings. Here, the first reply unit 130 can provide the user with a first reply string that is, for example, a concatenation of the selected multiple reply strings. This makes it possible to provide the user with a variety of first reply strings, and the dialogue system 100 can provide the user with smoother communication.

[0054] The first reply string may be a string that does not include a string of advice for the user. In other words, the first reply string may be a string of responses to the user's first utterance. This allows the user interacting with the dialogue system 100 to have the impression that their first utterance is being listened to attentively, and the dialogue system 100 can provide the user with smoother communication.

[0055] Here, the interjection may be, for example, a character string that repeats the content of the user's first utterance, a character string that confirms the content of the first utterance, a character string that indicates agreement with the user, or a character string that indicates concern for the user. Note that the character string that repeats the content of the user's first utterance may be the same as the first utterance, or may be a character string that is different from the first utterance but has the same or similar semantic content as the first utterance.

[0056] The first response string may also be a string that does not include content related to treatment for the user's symptoms. For example, the first response string may be a string that does not include content related to medical advice for the user.

[0057] Furthermore, when the first reply unit 130 provides a first reply string that repeats the content of the user's first utterance by voice, the first reply unit 130 may provide the first reply string in a tone that is different from the tone used when the user uttered the utterance.

[0058] 5 is a diagram showing an example of first reply string information stored in the storage unit 105. The first reply string information stored in the storage unit 105 includes, for example, a reply string ID and reply content information. The reply string ID is information that identifies the reply string (in this case, the first reply string) stored in the storage unit 105. The reply content information is information that indicates the content of the reply string.

[0059] The prompt generation unit 135 generates a prompt for generating a second reply string to the first comment in an external information processing system, and stores prompt information related to the prompt in the storage unit 105. The prompt generation unit 135 can generate a prompt for generating a second reply string to the first comment in the string generation system 200, for example.

[0060] The prompt generator 135 generates a prompt for generating a second reply string, which will be described later. That is, the prompt generator 135 can generate a prompt for generating a second reply string, which includes a string of characters related to advice for the user, a string of characters following a backchannel response to the user's first utterance, content related to treatment for the user's symptoms, or content related to medical advice for the user.

[0061] The prompt is, for example, information in a format that allows the string generation system 200 to generate a second response string based on the prompt. The string generation system 200 can generate the second response string by inputting the prompt into a large-scale language model, for example. The prompt may be in a natural language, or may be a string in a format other than a natural language, such as a programming language or JSON (JavaScript Object Notation).

[0062] The prompt generator 135 may generate a prompt further based on the first character string. That is, the prompt may include, for example, information about the first character string. The information about the first character string may be, for example, the first character string itself, or a character string indicating the same or similar meaning as the meaning indicated by the first character string. The prompt may include, for example, the character string "The patient says, 'I'm not in pain.'"

[0063] The prompt may include, for example, information regarding instructions to the string generation system 200. The information regarding instructions to the string generation system 200 may be, for example, a string of characters regarding instructions to generate an appropriate string as a response to the first string. The prompt may include, for example, a string such as, "You are a doctor. The patient says, 'I'm not in pain.' What would you say to a patient who is in trouble?"

[0064] The prompt generation unit 135 may generate a prompt further based on user information. That is, the prompt may include user information. Here, the user information may be, for example, medical history information. In this case, for example, the prompt may include a character string such as "You are an oncologist." The prompt may also include a character string such as "The cancer patient says, 'I'm not in pain.' This cancer patient has been taking prescription drug A for one year."

[0065] The prompt may include information about a first reply string for the user. Here, the information about the first reply string may be, for example, the content of the string provided to the user as the first reply string and information indicating the fact that the first reply string has been provided or will be provided to the user. Specifically, for example, the prompt may include a string such as "I said, 'That's good. It doesn't hurt, does it?'" or "I plan to say, 'That's good. It doesn't hurt, does it?'"

[0066] Furthermore, the prompt generating unit 135 may generate a prompt based on the result of a necessity determination made by the determining unit 160, which will be described later.

[0067] 6 is a diagram showing an example of prompt information stored in the storage unit 105. The prompt information stored in the storage unit 105 includes, for example, a prompt ID and prompt content information. The prompt ID is information that identifies a prompt stored in the storage unit 105. The prompt content information is information that indicates the content of a prompt.

[0068] The prompt output unit 140 outputs the prompt to an external information processing system, for example, the character string generation system 200.

[0069] The string generation system 200 generates a second reply string based on the acquired prompt. At this time, the string generation system 200 may, for example, input the prompt into a large-scale language model to generate the second reply string.

[0070] The prompt output unit 140 may generate a prompt based on the result of a necessity determination made by the determination unit 160, which will be described later.

[0071] The second reply string acquisition unit 145 acquires a second reply string from the string generation system 200 and stores second reply string information relating to the second reply string in the storage unit 105.

[0072] The second reply string may be, for example, a string following a response to the user's first utterance.

[0073] The second reply string may be a string containing a string of advice for the user. The second reply string may be a string containing content related to treatment for the user's symptoms. The second reply string may be, for example, a string containing content related to medical advice for the user, such as "If you have a prolonged fever, you are likely to become dehydrated, so be sure to drink plenty of fluids."

[0074] The second reply string acquisition unit 145 may also generate a prompt based on the result of a necessity determination made by the determination unit 160, which will be described later.

[0075] 7 is a diagram showing an example of second reply string information stored in the storage unit 105. The second reply string information stored in the storage unit 105 includes a reply string ID and reply content information.

[0076] The second reply section 150 generates a second reply string based on the prompt and provides it to the user by voice or text.

[0077] In this way, the first reply unit 130 provides the user with at least one reply string selected from a plurality of preset reply string candidates, for example, as the first reply string, and the second reply unit 150 provides the user with a string generated by the string generation system 200, for example. Here, the process from the user's first utterance to the completion of the reply by the second reply unit 150 includes the process of generating the second reply string, which may take some time. As a result, the user may feel like they are waiting for a reply and may get the impression that they are interacting with a machine.

[0078] In this respect, the response processing by the first response unit 130 is faster than the response processing by the second response unit 150. By using both the response processing by the first response unit 130 and the response processing by the second response unit 150, the dialogue system 100 can first provide the user with, for example, a string of characters such as a backchannel response through the response processing by the first response unit 130, and then provide the user with, for example, a string of characters of specific advice through the response processing by the second response unit 150. This allows the user to receive a response including more specific advice, as if they were conversing with a human, without feeling any waiting time for the response.

[0079] The series of second reply processes performed by the prompt generation unit 135, prompt output unit 140, second reply string acquisition unit 145, and second reply unit 150 may be performed in parallel with the series of first reply processes performed by the group selection unit 120, string selection unit 125, and first reply unit 130. In this case, the first reply unit 130 may provide the user with a first reply string, followed by the second reply unit 150 providing the user with a second reply string. This allows the dialogue system 100 to provide the user with a second reply string containing more specific advice, as if they were conversing with a human, without making the user feel like they had to wait for a reply.

[0080] The cancellation processing unit 155 generates cancellation request information based on the second string to request cancellation of the process of generating a second response string based on a prompt in an external information processing system (e.g., the string generation system 200), and stores the information in the memory unit 105.

[0081] The second string is a string of a user's comment corresponding to the first response string to the first string, such as a string correcting the first string or a string correcting the first response string. In this case, the second response string based on the first string is not appropriate as a response string to be provided to the user. Therefore, the cancellation processor 155 generates cancellation request information based on the second string.

[0082] The cancellation request information may include, for example, information about an API (Application Programming Interface) for requesting cancellation of the process of generating a second reply string based on the prompt. The cancellation request information may also include, for example, a string indicating a prompt (cancellation prompt) for requesting cancellation of the process of generating a second reply string based on the prompt.

[0083] Furthermore, the cancellation processing unit 155 may output the cancellation request information to the string generation system 200. This makes it possible to cancel the process of generating the second response string in the string generation system 200. Note that the cancellation processing unit 155 may output the cancellation request information to an information processing system different from the string generation system 200, thereby outputting the cancellation request information to the string generation system 200 via the different information processing system.

[0084] The determining unit 160 determines whether a reply using a second reply string is necessary based on the first string.

[0085] If the determination unit 160 determines that a reply using a second reply string is necessary, the prompt generation unit 135 may perform prompt generation processing, the prompt output unit 140 may perform prompt output processing, the second reply string acquisition unit 145 may perform second reply string acquisition processing, or the second reply unit 150 may perform reply processing. This allows the dialogue system 100 to execute the second reply processing when a reply using a second reply string is necessary.

[0086] For example, the determination unit 160 may determine the need for a reply using a second reply string when the severity of the symptom indicated by the symptom severity information included in the first string is equal to or greater than a predetermined severity. Specifically, the determination unit 160 may determine that a reply using a second reply string is necessary when the first string is, for example, "I am in very severe pain," and may determine that a reply using a second reply string is not necessary when the first string is, for example, "I am not in pain."

[0087] The determination unit 160 may determine the necessity of replying with a second reply string further based on the user information. Specifically, for example, based on information about the user's medical history included in the user information, the determination unit 160 may determine that the user is suffering from a predetermined disease, that the second reply string is necessary.

[0088] Furthermore, the determining unit 160 determines the urgency of the action to be taken for the user based on the first character string.

[0089] For example, the determination unit 160 may determine that treatment is urgent when the degree of the symptom indicated by the symptom degree information included in the first character string is equal to or greater than a predetermined degree. Specifically, for example, the determination unit 160 may determine that treatment is urgent when the first character string is "I am in very severe pain," and may determine that treatment is not urgent when the first character string is "I am not in pain."

[0090] Furthermore, the determination unit 160 may determine that there is an urgent need for treatment of the user when the user does not respond to a call made by the dialogue system 100.

[0091] At this time, the judgment unit 160 may determine that there is a first urgency for action against the user, for example, when the first reply string or the second reply string is in the form of a question to the user, and the string acquisition unit 115 does not acquire a string based on the user's utterance (i.e., there is no response from the user) even after a predetermined time has passed since the first reply string was provided by the first reply unit 130 or the second reply string was provided by the second reply unit 150.

[0092] Next, the dialogue system 100 (e.g., the third reply unit) provides the user with a third reply string to check the user's reaction if there is no reply, for example, in accordance with the result of the first urgency determination. At this time, the third reply string may be a string in the form of a question to the user, such as "What's wrong?" or "Are you okay?". The third reply string may be a string selected from at least one string set in advance, or may be a string obtained from the string generation system 200 in response to a predetermined prompt generated by the prompt generation unit 135.

[0093] Then, when the character string acquisition unit 115 does not acquire a character string based on the user's utterance (i.e., there is no response from the user) even after a predetermined time has elapsed since the provision of the third response character string, the determination unit 160 determines that there is a second urgency of treatment for the user. The notification unit 165, which will be described later, notifies an external information processing system (for example, an information processing system used by a family doctor) according to the result of the determination of the second urgency. At this time, the notification unit 165 notifies, for example, that there is no response from the user.

[0094] The determination unit 160 may determine the urgency of treatment for the user based on the user information. Specifically, for example, based on information about the user's medical history included in the user information, if the user is suffering from a predetermined disease, the determination unit 160 may determine that treatment for the user is urgent.

[0095] If the determining unit 160 determines that the treatment is urgent, the notifying unit 165 may perform a notification process, which will be described later.

[0096] The notification unit 165 notifies an external information processing system of information related to the first message based on the result of the urgency determination.

[0097] The information regarding the first statement notified by the notification unit 165 may be, for example, information indicating the content of the first statement (for example, a first character string), and may also include user information of the user who made the first statement.

[0098] Here, the external information processing system may be, for example, an information processing system used by someone in a position to provide treatment to the user, particularly, for example, a medical institution, doctor, family doctor, nurse, pharmacist, caregiver, fire department, etc., or it may also be, for example, an information processing system used by someone who should be informed of the user's situation, particularly, for example, a family member or other person who has a close relationship with the user.

[0099] By the notification by notification unit 165, the external information processing system notifies, for example, a person who is in a position to provide medical treatment to the user or a person who should be informed of the user's condition, of information related to the first utterance. This allows the person who is in a position to provide medical treatment to the user to make preparations to quickly provide medical treatment to the user, rush to the user's location, or contact the person who should be informed of the user's condition, and the person who should be informed of the user's condition to rush to the user's location or contact the person who is in a position to provide medical treatment to the user.

[0100] FIG. 8 is a flowchart showing an example of processing in the dialogue system 100.

[0101] First, the user information acquiring unit 110 acquires user information (S801). The string acquiring unit 115 (first string acquiring unit) acquires a first string (e.g., "I've had a fever of about 39°C since yesterday.") generated based on the user's first utterance (S802). The first replying unit 130 generates a first reply string (e.g., "That's terrible.") and provides it to the user (S803). At this time, the group selecting unit 120 may select at least one group from a plurality of group candidates, the string selecting unit 125 may select at least one reply string from reply string candidates included in the at least one group, and the first replying unit 130 may generate the first reply string based on the at least one selected reply string.

[0102] Next, the prompt generation unit 135 generates a prompt for generating a second reply string and outputs it to, for example, the string generation system 200 (S804). The second reply string acquisition unit 145 acquires the second reply string (for example, "If the fever continues, you are likely to become dehydrated, so be sure to drink plenty of fluids.") from the string generation system 200 and provides it to the user (S805).

[0103] Fig. 9 is a flowchart showing an example of processing in the dialogue system 100. The flowchart shown in Fig. 9 is a flowchart showing processing for requesting cancellation of the processing for generating a second reply string based on a prompt.

[0104] First, the first reply unit 130 provides a first reply string to the user (S901). Next, the string acquisition unit 115 (second string acquisition unit) acquires a second string (S902). Then, the cancellation processing unit 155 generates cancellation request information and outputs it to, for example, the string generation system 200 (S903).

[0105] Fig. 10 is a flowchart showing an example of processing in the dialogue system 100. The flowchart shown in Fig. 10 is a flowchart showing the processing for determining the necessity of a reply using a second reply string.

[0106] First, the string acquisition unit 115 (first string acquisition unit) acquires a first string generated based on the user's first utterance (S1001). Next, the determination unit 160 determines, based on the first string, whether a reply using a second reply string is necessary (S1002). Then, based on the result of the determination, the prompt generation unit 135 generates a prompt for generating a second reply string and outputs it to, for example, the string generation system 200 (S1003). The second reply string acquisition unit 145 acquires the second reply string from the string generation system 200 and provides it to the user (S1004).

[0107] Fig. 11 is a flowchart showing an example of processing in the dialogue system 100. The flowchart shown in Fig. 11 is a flowchart relating to processing for determining the urgency of treatment for the user.

[0108] First, the character string acquisition unit 115 (first character string acquisition unit) acquires a first character string generated based on a first utterance of a user (S1101). Next, the determination unit 160 determines the urgency of an action to be taken against the user based on the first character string (S1102). Then, the notification unit 165 notifies an external information processing system of information related to the first utterance based on the result of the urgency determination (S1103).

[0109] Fig. 12 is a flowchart showing an example of processing in the dialogue system 100. The flowchart shown in Fig. 12 is a flowchart showing processing for determining the urgency of treatment for a user when the user does not respond to a call made by the dialogue system 100.

[0110] First, the determination unit 160 determines that there is a first urgency of action for the user (S1201). At this time, the determination unit 160 may determine that there is a first urgency of action for the user, for example, when the first reply string or the second reply string is in the form of a question for the user, and the string acquisition unit 115 does not acquire a string based on the user's utterance (i.e., there is no response from the user) even after a predetermined time has passed since the first reply string was provided by the first reply unit 130 or the second reply string was provided by the second reply unit 150.

[0111] Next, the dialogue system 100 provides the user with a third reply string to check the user's reaction to the absence of a reply, for example, depending on the result of the first urgency determination (S1202).

[0112] Then, if the character string acquisition unit 115 does not acquire a character string based on the user's utterance (i.e., there is no response from the user) even after a predetermined time has elapsed since the provision of the third response character string, the determination unit 160 determines that there is a second urgency of treatment for the user (S1203).The notification unit 165 notifies an external information processing system (for example, an information processing system used by the primary care doctor) in accordance with the result of the determination of the second urgency (S1204).

[0113] The above describes one embodiment of the present invention. The dialogue system 100 can acquire a first string generated based on a first utterance of a user, provide a first reply string to the user, generate a prompt for generating a second reply string, and provide the user with the second reply string generated based on the prompt. This allows the dialogue system 100 to provide smoother communication with the user.

[0114] Furthermore, the dialogue system 100 can generate a prompt that includes information about the first reply string, which allows the dialogue system 100 to perform a second reply process based on the first reply process, thereby providing smoother communication with the user.

[0115] Furthermore, the dialogue system 100 can generate a prompt based on the user information, thereby enabling the dialogue system 100 to provide the user with a second response string that is suitable for the user, thereby enabling smoother communication with the user.

[0116] Furthermore, the dialogue system 100 can acquire a second string generated based on the user's second utterance, and generate cancellation request information based on the second string. This allows the dialogue system 100 to cancel the process of generating a second reply string based on the first utterance in the string generation system 200 when the dialogue system 100 acquires the second utterance.

[0117] Furthermore, the dialogue system 100 can determine the necessity of replying with a second reply string based on the first string, and generate a prompt based on the result of the determination of the necessity, thereby enabling the dialogue system 100 to execute a process of obtaining a second reply string in response to the user's first utterance.

[0118] Furthermore, the dialogue system 100 can determine the urgency of the action to be taken against the user based on the first character string, and can notify an external information processing system of information related to the first utterance based on the result of the determination of the urgency. In this way, the dialogue system 100 can notify the user that there is an urgency of the action to be taken against the user in accordance with the first utterance of the user, and can cooperate with an appropriate person or institution.

[0119] It should be noted that the present embodiment is provided to facilitate understanding of the present invention and is not intended to limit the present invention. The present invention may be modified or improved without departing from the spirit thereof, and equivalents thereof are also included in the present invention. [Explanation of symbols]

[0120] 100 Dialogue system, 105 Memory unit, 110 User information acquisition unit, 115 Character string acquisition unit, 120 Group selection unit, 125 Character string selection unit, 130 First reply unit, 135 Prompt generation unit, 140 Prompt output unit, 145 Second reply character string acquisition unit, 150 Second reply unit, 155 Cancellation processing unit, 160 Determination unit, 165 Notification unit, 200 Character string generation system

Claims

1. a first character string acquisition unit that acquires a first character string generated based on a first utterance of a user in the form of voice or text; a first reply unit that provides the user with a first reply string to the first comment by voice or text, the first reply string being generated based on the first string; a string generation request generating unit that generates string generation request information for generating a second reply string to the first message in an external information processing system; a second response unit that provides the user with a second response string, which is generated based on the string generation request information, by voice or text; A dialogue system comprising:

2. The dialogue system according to claim 1 , wherein the string generation request information includes information regarding the first response string to the user.

3. The dialogue system according to claim 1 , wherein the character string generation request generating unit generates the character string generation request information further based on user information relating to the user.

4. The dialogue system according to claim 3 , wherein the user information includes at least one of an attribute of the user, a medical history of the user, and a history of utterances by the user.

5. a second character string acquisition unit that acquires a second character string generated based on a second utterance by a user in the form of voice or text corresponding to the first response character string; a cancellation processing unit that generates, based on the second character string, cancellation request information for requesting cancellation of a process of generating the second response character string based on the character string generation request information in the external information processing system; The dialogue system according to claim 1 or 2, further comprising:

6. a determination unit that determines the necessity of a reply using the second reply string based on the first string, The character string generation request generating unit generates the character string generation request information based on a result of the necessity determination.

3. A dialogue system according to claim 1 or 2.

7. a determination unit that determines an urgency of a treatment for the user based on the first character string; a notification unit that notifies an external information processing system of information related to the first message based on a result of the urgency determination; The dialogue system according to claim 1 or 2, further comprising:

8. a group selection unit that selects at least one group from a plurality of group candidates each including a plurality of response string candidates, based on symptom information related to a symptom of the user that is included in the first string; and a string selection unit that selects at least one response string from among a plurality of response string candidates included in the at least one selected group; Further equipped with the first response unit generates the first response string based on the selected at least one response string.

3. A dialogue system according to claim 1 or 2.

9. the group selection unit selects a plurality of groups from the plurality of group candidates; the string selection unit selects at least one response string from among a plurality of response string candidates included in each of the selected groups to select a plurality of response strings; the first reply unit generates the first reply string based on the plurality of reply strings.

9. The dialogue system according to claim 8.

10. the first response string does not include a string related to advice to the user; the second response string includes a string related to advice for the user; 3. A dialogue system according to claim 1 or 2.

11. The computer obtaining a first character string generated based on a first utterance of a user by voice or text; providing the user with a first response string to the first comment by voice or text, the first response string being generated based on the first string; generating character string generation request information for generating a second reply character string to the first message in an external information processing system; providing the user with a second response string, which is generated based on the string generation request information, by voice or text; A method of interaction with computer software.

12. On the computer, a first character string acquisition unit that acquires a first character string generated based on a first utterance of a user in the form of voice or text; a first reply unit that provides the user with a first reply string to the first comment by voice or text, the first reply string being generated based on the first string; a string generation request generating unit that generates string generation request information for generating a second reply string to the first message in an external information processing system; a second response unit that provides the user with a second response string, which is generated based on the string generation request information, by voice or text; An interactive program to achieve this.

Images