Automated response device and automated response method

Abstract

This automated response device is provided with: an automated response unit that, by automatically responding to the statements of a plurality of users including a first user, converses with the plurality of users; a strategy determination unit that determines a strategy for a dialogue with the first user by means of the automated response unit, on the basis of first value information regarding the values of the first user and second value information regarding the values of the automated response unit, such values bringing liveliness to the conversation between the first user and the automated response unit; and a response control unit that, on the basis of the second value information and the dialogue strategy, controls responses of the automated response unit to the statements of the first user.

Description

Automatic Response Apparatus and Automatic Response Method

[0001] The present invention relates to an automatic response apparatus and an automatic response method.

[0002] With the popularization of large-scale language models, it is expected that the demand for chatbots as a substitute for human communication will increase in the future. For example, Patent Document 1 discloses a control apparatus for a dialogue system that generates an answer to a user's question by using QA data generated from sentences having an index with a hierarchical structure.

[0003] Japanese Unexamined Patent Publication No. 2025-000374

[0004] However, the control apparatus for a dialogue system disclosed in Patent Document 1 has a problem in that depending on the compatibility between the user and the dialogue system, the conversation may not become lively, and it may not be possible to extract a large amount of information from the user.

[0005] The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to extract more information from a user.

[0006] An automatic response apparatus according to a preferred aspect of the present invention includes: an automatic response unit that converses with a plurality of users including a first user by automatically responding to statements made by the plurality of users; a strategy determination unit that determines a dialogue strategy for the first user used by the automatic response unit, based on first value information related to values of the first user and second value information related to values of the automatic response unit that make a conversation between the first user and the automatic response unit lively; and a response control unit that controls a response of the automatic response unit to a statement made by the first user based on the second value information and the dialogue strategy.

[0007] An automated response method according to a preferred embodiment of the present invention determines a dialogue strategy for the automated response unit to the first user based on first value information relating to the first user's values ​​and second value information relating to the values ​​of the automated response unit that make the conversation between the first user and the automated response unit more engaging. Based on the second value information and the dialogue strategy, the automated response unit controls its response to the first user's statements. The automated response unit then converses with the multiple users, including the first user, by automatically responding to the statements of the multiple users, and this process is executed by a computer.

[0008] According to the automatic response device and automatic response method of the present invention, it is possible to extract more information from the user.

[0009] This figure shows the overall configuration of the automated response system according to the first embodiment. This is a block diagram showing an example configuration of the terminal device in Figure 1. This is a block diagram showing an example configuration of the automated response server in Figure 1. This figure shows an example of elements included in the first value information. This figure shows an example of elements of a value pair vector. This is a schematic diagram showing the result of clustering multiple value pair vectors. This figure shows an example of elements of intimacy. This figure shows an example of elements of excitement. This figure shows an example of the relationship between the values ​​of the first user, the values ​​of the automated response unit, and the elements of the dialogue. This figure shows an example of a table of dummy variables related to topics. This figure shows an example of a table of dummy variables related to reactions. This figure shows an example of a table of dummy variables related to language use. This is a flowchart showing an example of the operation of the processing unit in Figure 3. This is a flowchart showing an example of the operation of the subroutine in Figure 13.

[0010] 1. The configuration of the question sentence determination device according to the first embodiment of the present invention will be described below with reference to Figures 1 to 14.

[0011] 1.1. Configuration of the First Embodiment 1.1.1. Configuration of the Automatic Response System Figure 1 is a diagram showing the overall configuration of the automatic response system 1 according to the first embodiment. The automatic response system 1 includes a terminal device 10 and an automatic response server 20.

[0012] The automated response system 1 is a system that interacts with user U and presents answers to user U's inquiries, questions, etc.

[0013] A communication network (NET) is a telecommunications line, such as a mobile communication network, managed by a telecommunications carrier providing communication services. A communication network (NET) includes either or both wired and wireless communication networks. For example, a communication network (NET) may be connected via the Internet to other networks (not shown) managed by other telecommunications carriers.

[0014] In the automated response system 1, the terminal device 10 and the automated response server 20 are connected to each other via a communication network NET so that they can communicate with one another.

[0015] Terminal device 10 includes n terminal devices 10[1], 10[2], 10[3], ..., 10[n], where n is any natural number. In this embodiment, the configurations of terminal devices 10[1] to 10[n] are identical to each other. However, terminal device 10 may include terminal devices with different configurations.

[0016] The user using terminal device 10[1] is the first user U1, the user using terminal device 10[2] is user U2, and the user using terminal device 10[n] is user Un. When referring to an unspecified number of users or all users, user is also written as user U.

[0017] The terminal device 10 includes portable devices such as personal computers, tablet terminals, smartphones, and smartwatches.

[0018] The automated response server 20 is a device that receives data related to the user U's statements sent from the terminal device 10 based on the user U's operations, and outputs the response result to the terminal device 10 based on the input data. In this embodiment, the automated response server 20 is assumed to be a device that has functions such as a chatbot that serves as a customer support window, or an AI (Artificial Intelligence) avatar that appears in the metaverse space.

[0019] 1.1.2. Terminal Device Configuration Diagram 2 is a block diagram showing an example configuration of the terminal device 10 in Figure 1. As shown in Figure 2, the terminal device 10 comprises a processing unit 11, a storage device 12, a communication device 13, a display device 14, and an input device 15. Each element of the terminal device 10 is interconnected by one or more buses for communicating information.

[0020] The processing unit 11 is a processor that controls the entire terminal device 10, and is configured, for example, using one or more chips. The processing unit 11 is configured using a central processing unit (CPU) that includes, for example, interfaces with peripheral devices, arithmetic units, registers, etc. Some or all of the functions of the processing unit 11 may be implemented by hardware such as a DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), FPGA (Field Programmable Gate Array). The processing unit 11 executes various processes in parallel or sequentially.

[0021] The storage device 12 is a recording medium that can be read from and written to by the processing device 11. The storage device 12 includes, for example, non-volatile memory and volatile memory. Non-volatile memory includes, for example, ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory), and EEPROM (Electrically Erasable Programmable Read Only Memory). Volatile memory includes, for example, RAM (Random Access Memory).

[0022] The storage device 12 stores multiple programs, including the control program PR1, which is executed by the processing unit 11. The storage device 12 also functions as a work area for the processing unit 11. The control program PR1 is a program that controls the entire processing unit 11.

[0023] The communication device 13 is hardware that acts as a transmitting and receiving device for communicating with other devices. The communication device 13 is also called, for example, a network device, a network controller, a network card, a communication module, etc. The communication device 13 may be equipped with a connector for wired connection and an interface circuit corresponding to the connector. The communication device 13 may also be equipped with a wireless communication interface. Examples of connectors and interface circuits for wired connection include products compliant with wired LAN, IEEE 1394, and USB. Examples of wireless communication interfaces include products compliant with Wi-Fi® and Bluetooth®.

[0024] The display device 14 is a device that displays images and text information. The display device 14 displays various images based on control by the processing device 11. For example, various display panels such as liquid crystal panels and organic EL (Electro-Luminescence) panels are preferably used as the display device 14.

[0025] The input device 15 accepts operations from the user U. For example, the input device 15 is configured to include a pointing device such as a keyboard, touchpad, touch panel, or mouse. If the input device 15 is configured to include a touch panel, it may also function as the display device 14.

[0026] The processing unit 11 functions as a receiving unit 111, a transmitting unit 112, an acquisition unit 113, and a display control unit 114, for example, by reading and executing the control program PR1 from the storage device 12.

[0027] The reception unit 111 receives user U's operations on the input device 15. User U's operations include, for example, tapping on the touch panel which is the input device 15.

[0028] The transmission unit 112 transmits the text input using the input device 15 to the automatic response server 20 via the communication device 13.

[0029] The acquisition unit 113 acquires text information transmitted from the automatic response server 20 via the communication device 13. The text information includes the question and answer from the automatic response server 20.

[0030] The display control unit 114 causes the display device 14 to display various information.

[0031] 1.1.3. Configuration of the Automated Response Server Figure 3 is a block diagram showing an example configuration of the automated response server 20 shown in Figure 1. As shown in Figure 3, the automated response server 20 comprises a processing unit 21, a storage device 22, and a communication device 23. Each element of the automated response server 20 is interconnected by one or more buses for communicating information.

[0032] The processing unit 21 is a processor that controls the entire automatic response server 20, and is configured, for example, using one or more chips. The processing unit 21 is configured, for example, using a central processing unit (CPU) that includes interfaces with peripheral devices, an arithmetic unit, and registers. Some or all of the functions of the processing unit 21 may be implemented by hardware such as a DSP, ASIC, PLD, FPGA, etc. The processing unit 21 executes various processes in parallel or sequentially.

[0033] The storage device 22 is a recording medium that can be read from and written to by the processing device 21. The storage device 22 includes, for example, non-volatile memory and volatile memory. The non-volatile memory is, for example, ROM, EPROM, and EEPROM. The volatile memory is, for example, RAM.

[0034] The storage device 22 stores multiple programs, including the control program PR2 for execution by the processing device 21, the first learning model LM1, the second learning model LM2, the third learning model LM3, the large-scale language model LLM, and the first value database DB1. The storage device 22 also functions as a work area for the processing device 21. Examples of large-scale language models LLM include BERT (Bidirectional Encoder Representations from Transformers), GPT-4o, Gemini, and Llama2.

[0035] The communication device 23 is hardware that acts as a transmitting and receiving device for communicating with other devices. The communication device 23 is also called, for example, a network device, network controller, network card, communication module, etc. The communication device 23 may be equipped with a connector for wired connection and an interface circuit corresponding to the connector. The communication device 23 may also be equipped with a wireless communication interface. Examples of connectors and interface circuits for wired connection include products compliant with wired LAN, IEEE 1394, and USB. Examples of wireless communication interfaces include products compliant with Wi-Fi® and Bluetooth®.

[0036] The processing unit 21 functions as an acquisition unit 2101, an automatic response unit 2102, a first value determination unit 2103, an extraction unit 2104, a value pair determination unit 2105, a clustering unit 2106, a representative value determination unit 2107, a closeness determination unit 2108, a second value determination unit 2109, an excitement level determination unit 2110, a strategy determination unit 2111, a response control unit 2112, and a display control unit 2113, for example, by reading and executing the control program PR2 from the storage device 22.

[0037] The acquisition unit 2101 acquires statements from multiple users U, including the first user U1, from the terminal device 10. In this example, the statements of users U are acquired as text information. However, the statements of users U may also be voice information. If it is voice information, the processing unit 21 may further include a voice recognition unit and be configured to convert the voice information into text information.

[0038] The automated response unit 2102 converses with multiple users U by automatically responding to statements made by multiple users U, including the first user U1. The automated response unit 2102 inputs the statements of multiple users U into the large-scale language model LLM, obtains the responses to the statements of multiple users U output from the large-scale language model LLM, and presents the responses to the users U. The automated response unit 2102 may respond to the statements of users U using predefined templates or in a rule-based manner.

[0039] Hereinafter, a case where the first user U1 among the plurality of users U uses the automatic response system 1 will be described as an example.

[0040] A first value determination unit 2103 determines first value information based on attribute information relating to attributes of the first user U1 and first conversation information relating to a conversation between the first user U1 and an automatic response unit 2102. The first value information is information relating to values of the first user U1.

[0041] The attribute information includes information relating to the age, gender, address, workplace, and hobbies of the first user U1. The first conversation information includes a history of conversations between the first user U1 and the automatic response unit 2102.

[0042] FIG. 4 is a diagram showing an example of elements included in the first value information. The values of the first user U1 include a plurality of value elements. For example, the plurality of value elements include cooperativeness, extroversion, frugality orientation, conformity effect, and the like. A score is assigned to each of the plurality of value elements. The first value information is a first value vector that represents the values of the first user U1 as a vector. The first value vector V1 is represented by the following formula (1).

[0043] V1=(v11,v12,v13,・・・,v1n) ・・・(1)

[0044] v11, v12, v13, ..., v1n each represent a score of an element of the first value. For example, v11 represents a cooperativeness score, v12 represents an extroversion score, and v13 represents a frugality orientation score.

[0045] In FIG. 4, a user ID indicates a unique ID that identifies each of the plurality of users U. As shown in FIG. 4, the user with user ID: 0001 has a cooperativeness score, an extroversion score, a frugality orientation score, and a conformity effect score of 20, 5, 6, and 3, respectively. The user with user ID: 0002 has a cooperativeness score, an extroversion score, a frugality orientation score, and a conformity effect score of 5, 10, 3, and 1, respectively. The first value information of each user is stored in a first value database DB1.

[0046] Referring again to FIG. 3, the extraction unit 2104 extracts a plurality of pairs of users whose conversation frequency is equal to or higher than a threshold, based on conversation information related to conversations of a plurality of users U. Each of the plurality of pairs includes User A and User B. User A and User B may refer to two unspecified users. User A is an example of a second user, and User B is an example of a third user. The conversation information is stored in a conversation information storage device (not shown). The conversation information is, for example, information related to chats on an SNS (Social Network Service). The conversation frequency is determined, for example, based on the amount of text per day, the number of times stamps are sent, and the like in a chat between the second user and the third user.

[0047] The plurality of pairs extracted by the extraction unit 2104 are combinations whose conversation frequency is equal to or higher than the threshold, and thus can be said to be pairs that have lively conversations. It can also be said that the plurality of pairs extracted by the extraction unit 2104 are pairs that have mutually compatible values.

[0048] A values pair determination unit 2105 determines, for each of the plurality of pairs extracted by the extraction unit 2104, a values pair vector Vp, which is a pair of the values vector Va of User A and the values vector Vb of User B. Here, the values pair vector Vp is represented by the following formula (2).

[0049] Vp = (Va, Vb) ・・・(2)

[0050] The values vector Va of User A indicates the values of User A. The values vector Vb of User B indicates the values of User B. Similar to the first values vector V1, the values vector Va of User A and the values vector Vb of User B are n-dimensional vectors representing scores of n values elements. Accordingly, the values vector Va of User A and the values vector Vb of User B are represented by the following formulas (3) and (4), respectively. The values vector Va of User A is an example of a second user values vector. The values vector of User B is an example of a third user values vector.

[0051] Va=(va1, va2,..., van)...(3) Vb=(vb1, vb2,..., vbn)...(4)

[0052] Figure 5 shows an example of the elements of a values ​​pair vector Vp. Since the values ​​pair vector Vp is a combination of user A's values ​​vector Va and user B's values ​​vector Vb, it is a 2n-dimensional vector. The values ​​pair vector Vp can also be called values ​​pair information. Each of the multiple values ​​pair information entries contains the values ​​of two people belonging to the corresponding user pair. In other words, values ​​pair information can be said to be a combination of the values ​​vectors of each user pair. Note that for clustering purposes, the values ​​pair vector Vp can be considered as a vector in which a pair of values ​​vectors is conveniently treated as a single vector.

[0053] The clustering unit 2106 classifies multiple value pair vectors Vp, which correspond one-to-one to multiple pairs, into multiple clusters through clustering. The pairs of multiple value vectors that correspond one-to-one to multiple clusters constitute multiple value pair information.

[0054] Figure 6 is a schematic diagram showing the results of clustering multiple value pair vectors Vp. Although Figure 6 schematically shows a two-dimensional vector space, the actual vector space corresponds to the dimension of the value pair vectors Vp.

[0055] Figure 6 shows that multiple value pair vectors Vp were classified into three clusters. The three clusters are the first cluster C1, the second cluster C2, and the third cluster C3. Classification into three clusters is merely an example, and the number of clusters is not limited to three. Furthermore, while it is not necessary to specifically define the number of clusters to be classified in clustering, the number of clusters to be classified may be predetermined.

[0056] Referring again to Figure 2, the representative values ​​determination unit 2107 determines a pair of value vectors that represent the corresponding cluster for each of the multiple clusters classified by the clustering unit 2106. More specifically, the representative values ​​determination unit 2107 averages the scores of each element represented by one or more value pair vectors Vp belonging to each cluster, and determines the vector having the average score of each element as the value pair vector that represents each cluster. Hereafter, the value pair vector that represents each cluster will be referred to as the representative values ​​pair vector. In this case, the representative values ​​pair vector can be said to be the central vector of the cluster.

[0057] For example, the representative values ​​determination unit 2107 averages the scores of each element of the value pair vector Vp belonging to the first cluster C1. This allows the representative values ​​determination unit 2107 to determine the central vector Vc1 of the first cluster C1. The central vector Vc1 is a pair of user A's value vector Va and user B's value vector Vb, and is expressed as Vc1 = (Vca1, Vcb1). The combination of vector Vca1 and vector Vcb1 is a compatible combination that leads to engaging conversation.

[0058] Similarly, the representative values ​​determination unit 2107 determines the center vector Vc2 = (Vca2, Vcb2) of the second cluster C2 and the center vector Vc3 = (Vca3, Vcb3) of the third cluster C3. Figure 6 illustrates the center vectors Vc1, Vc2, and Vc3.

[0059] Furthermore, the method for calculating the representative value pair vector is not limited to averaging the scores of each element represented by one or more value pair vectors Vp. For example, by finding the median of the scores of each element represented by one or more value pair vectors Vp, a vector containing the median of each element can be obtained, and this vector containing the median of each element may be used as the representative value pair vector.

[0060] The intimacy determination unit 2108 determines the intimacy level between the first user U1 and the automatic response unit 2102 by inputting the conversation history between the first user U1 and the automatic response unit 2102 into the first learning model LM1.

[0061] The first learning model LM1 is a model that has learned the relationship between conversational information about conversations between multiple users U and first evaluation values ​​in which multiple users U evaluate the level of familiarity with those conversations. The first evaluation values ​​are obtained, for example, by conducting a questionnaire with multiple users U that corresponds to the conversational information of multiple users U. The first learning model LM1 is trained using supervised learning with the first evaluation values ​​as the ground truth labels. The first learning model LM1 is trained using methods such as neural networks, logistic regression, and support vector machines.

[0062] Figure 7 shows an example of the elements of intimacy. As shown in Figure 7, intimacy includes elements such as topic similarity, the proportion of questions from User A, the proportion of questions from User B, the proportion of polite language used by User A, the proportion of polite language used by User B, the number of stamps, and the level of self-disclosure.

[0063] Topic similarity indicates the degree to which a topic provided by user A and a topic provided by user B are similar. For example, if user A's topic is about good lunch spots and user B's topic is about convenience store sweets, then the two topics can be said to be similar. A topic is just one example of a subject.

[0064] The question rate for User A represents the proportion of questions among all of User A's utterances. The question rate for User B represents the proportion of questions among all of User B's utterances. The polite language rate for User A represents the proportion of utterances in which polite language is used among all of User A's utterances. The polite language rate for User B represents the proportion of utterances in which polite language is used among all of User B's utterances.

[0065] The stamp usage rate is the percentage of all utterances by User A and User B in which stamps are used. The self-disclosure level indicates the extent to which User A and User B disclose personal information to each other. If User A and User B are close enough to confide their secrets to each other, the self-disclosure level can be considered high.

[0066] User A and User B assign scores to each element shown in Figure 7 in their conversation. The level of intimacy is calculated from the scores for each element. This calculated level of intimacy, along with the conversation information between User A and User B, is used as the first evaluation value in training the first learning model LM1.

[0067] Referring again to Figure 3, the second value determination unit 2109 determines second value information relating to the values ​​of the automatic response unit 2102 that will make the conversation between the first user U1 and the automatic response unit 2102 more engaging. More specifically, the second value determination unit 2109 determines the second value information based on the first value information, multiple value pair information, and intimacy information. The second value information is a second value vector that represents the values ​​of the automatic response unit 2102 as a vector. The second value vector V2 is expressed by the following equation (5).

[0068] V2=(v21,v22,v23,...,v2n)...(5)

[0069] v21, v22, v23, ..., v2n each represent the score for an element of the second value. For example, v21 represents the agreeableness score, v22 represents the extraversion score, and v23 represents the frugality score.

[0070] Multiple value pair information is information that corresponds one-to-one with multiple user pairs among multiple users U that engage in engaging conversations. Whether a conversation is engaging or not is based on whether the level of engagement, which indicates the degree to which the conversation is engaging between the user pairs, is above a threshold. The level of engagement may include the frequency of conversation as an objective indicator and an evaluation value, which is a subjective indicator, given by the user pairs who participated in the conversation and how they assessed the level of engagement.

[0071] The intimacy information concerns the level of intimacy between the first user U1 and the automated response unit 2102. Intimacy is the degree of closeness determined by the emotional connection, trust, shared experiences, shared values, etc., between the two people conversing. Intimacy can be evaluated by the frequency of conversations between the two people, their language use, etc.

[0072] The second value determination unit 2109 determines the value vector that is most similar to the first value vector V1 among the multiple value vectors based on the similarity between the multiple value vector pairs and the first value vector V1, in order to obtain multiple value pair information. The second value determination unit 2109 determines the value vector that is paired with the identified value vector among the multiple value vector pairs as the second value vector V2.

[0073] More specifically, for example, the second value determination unit 2109 identifies the value vector that is most similar to the first value vector V1 from among the pairs of value vectors included in the central vector Vc1 (Vca1, Vcb1), the pairs of value vectors included in the central vector Vc2 (Vca2, Vcb2), and the pairs of value vectors included in the central vector Vc3 (Vca3, Vcb3).

[0074] The second value determination unit 2109 determines the value vector that is paired with the identified value vector from among a plurality of value vector pairs as the second value vector V2. For example, if the value vector most similar to the first value vector V1 is Vca2, the second value determination unit 2109 determines the value vector Vcb2 that is paired with the value vector Vca2 as the second value vector V2. The identified value vector pair represents the values ​​of people who are compatible with each other and who can have an engaging conversation.

[0075] The excitement level determination unit 2110 determines an excitement level indicating the degree to which the conversation between the first user U1 and the automatic response unit 2102 is exciting. More specifically, the excitement level determination unit 2110 determines a first excitement level by inputting a first value vector V1, a second value vector V2, and the level of intimacy between the first user U1 and the automatic response unit 2102 into the second learning model LM2.

[0076] The second learning model LM2 is a model that has already learned the relationship between a second evaluation value, which assesses the degree of conversational engagement, a value vector representing the values ​​of the two people involved in the conversation, and the level of intimacy between the two people involved in the conversation. The second learning model LM2 is trained using supervised learning with the second evaluation value as the ground truth label. The second learning model LM2 is trained using methods such as neural networks, logistic regression, and support vector machines.

[0077] Figure 8 shows an example of the elements of excitement level. As shown in Figure 8, excitement level includes the elements of User A's values, User B's values, and the degree of intimacy between User A and User B. User A's values ​​include multiple elements. User B's values ​​include multiple elements.

[0078] User A and User B score each element of their conversation as shown in Figure 8. The level of engagement is determined from the scores for each element. This determined level of engagement is used as a second evaluation value, along with the pairs of User A's value vector Va and User B's value vector Vb, and the level of intimacy between User A and User B, to train the second learning model LM2.

[0079] Furthermore, the excitement level determination unit 2110 determines a second excitement level by inputting the first value vector V1, the third value vector V3, and the level of intimacy between the first user U1 and the automatic response unit 2102 into the second learning model LM2.

[0080] The third value vector V3 is a vector that is different from the second value vector V2 and whose similarity to the second value vector V2 is greater than or equal to a threshold. Here, if we assume that the similarity is 1 when the two vectors coincide, then the threshold is a value greater than 0 and less than 1. More preferably, the threshold is a value greater than 0.8 and less than 1.

[0081] The second value determination unit 2109 determines the third value vector V3 as the vector representing the values ​​of the automatic response unit 2102, instead of the second value vector V2, if the second level of excitement is greater than the first level of excitement.

[0082] If the second level of excitement is less than or equal to the first level of excitement, the excitement level determination unit 2110 determines the third level of excitement by inputting the first value vector V1, the fourth value vector V4, and the level of intimacy between the first user U1 and the automatic response unit 2102 to the second learning model LM2. The fourth value vector V4 is a vector that is different from the second value vector V2 and the third value vector V3, and whose similarity to the second value vector V2 is greater than or equal to a threshold.

[0083] The second value determination unit 2109 determines the fourth value vector V4 as the vector representing the values ​​of the automatic response unit 2102, instead of the second value vector V2, if the third level of excitement is greater than the first level of excitement.

[0084] If the third level of excitement is less than or equal to the first level of excitement, the excitement level calculation is repeated until the level of excitement determined using a value vector that is different from the second value vector V2, the third value vector V3, and the fourth value vector V4, but similar to the second value vector V2, exceeds the first level of excitement. In this way, the values ​​of the automatic response unit 2102 can be determined so that the conversation between the first user U1 and the automatic response unit 2102 is the most exciting.

[0085] Referring again to Figure 3, the strategy decision unit 2111 determines a dialogue strategy for the first user U1 by the automated response unit 2102 based on the first value information and the second value information. More specifically, the strategy decision unit 2111 determines the type of dialogue element that maximizes the level of engagement obtained by inputting the first value vector V1, the second value vector V2, and the dialogue elements into the third learning model LM3 as a dialogue strategy that the automated response unit 2102 can adopt.

[0086] The third learning model LM3 is a model that has already learned the relationship between a value vector representing the values ​​of two people involved in a conversation, one or more elements of the conversation, and a second evaluation value that assesses the degree of conversational engagement. The third learning model LM3 is trained using methods such as neural networks, logistic regression, and support vector machines. When the first value vector V1, the second value vector V2, and the elements of the conversation are input to the third learning model LM3, the third learning model LM3 outputs the type of conversational element that maximizes the level of engagement.

[0087] Figure 9 shows an example of the relationship between the values ​​of the first user U1, the values ​​of the automated response unit 2102, and the elements of the dialogue. As shown in Figure 9, the elements of the dialogue include the topic of conversation, reactions, and language. As shown in Figure 9, the values ​​of the first user U1 include multiple elements, namely value α, value β, ... and the values ​​of the automated response unit 2102 include multiple elements, namely value α, value β, ... For example, the number of multiple elements is n. Given the values ​​of the first user U1 and the values ​​of the automated response unit 2102 as shown in Figure 9, the strategy decision unit 2111 determines the topic type to be "weather", the reaction type to be "parroting", and the language type to be "formal". In practice, the topic type, reaction type, and language type are each represented by dummy variables.

[0088] Figure 10 shows an example of a table of dummy variables related to topics. As shown in the first table TBL1 in Figure 10, examples of topic types include weather, news, culture, sports, music, travel, and food. For example, the dummy variables "001", "002", "003", "004", "005", "006", and "007" are associated with "weather", "news", "culture", "sports", "music", "travel", and "food", respectively.

[0089] Figure 11 shows an example of a table of dummy variables related to reactions. As shown in the second table TBL2 in Figure 11, examples of reaction types include echoing, empathy, surprise, questioning, encouragement, and humor. For example, the dummy variables "101", "102", "103", "104", "105", and "106" correspond to "echoing", "empathy", "surprise", "questioning", "encouragement", and "humor", respectively.

[0090] A "parrot" reaction is simply repeating what the other person said. For example, if someone says, "I recently went on a trip," a parrot reaction would be, "Oh, you went on a trip." By using a parrot reaction, you can make the other person feel that you are listening to them with interest, and you can keep the conversation going naturally.

[0091] An "empathetic" reaction is, for example, nodding along with "Uh-huh" or "I see," or saying things like "I understand" or "I think so too." A "surprised" reaction is saying things like "Wow, that's amazing!" or "Really?" A "questioning" reaction is showing a desire to hear more about a topic that interests you, such as "And then what happened?" A "encouraging" reaction is saying things like "You're doing great" or "I'm rooting for you." A "humorous" reaction is sharing laughter, for example, by saying something like "That's funny!"

[0092] Figure 12 shows an example of a table of dummy variables related to language use. As shown in the third table TBL3 in Figure 12, the types of language use are classified, for example, into formal, semi-formal, slightly casual, casual, and very casual. For example, "formal," "semi-formal," "slightly casual," "casual," and "very casual" are associated with the dummy variables "201," "202," "203," "204," and "205," respectively.

[0093] The language used will vary depending on the level of familiarity between the parties and the frequency of conversations. "Formal" is intended for conversations with someone you've just met or someone you respect. Polite and respectful language is used. "Semi-formal" is intended for conversations with new friends or acquaintances. While polite language is used, casual elements are incorporated, and respectful language is used relatively frequently. "Slightly casual" incorporates more casual elements than "semi-formal," and respectful language is used less frequently. "Casual" is intended for everyday conversations with friends. You should be natural and not worry about formality; respectful language is rarely used. "Very casual" is intended for conversations with close friends or family. The atmosphere is relaxed, and respectful language is not used. Abbreviations and slang are frequently used.

[0094] The first table TBL1, the second table TBL2, and the third table TBL3 are stored in the memory device 22. Furthermore, the elements of a dialogue only need to include at least one of the following: a conversation topic, a reaction, or word choice.

[0095] Referring again to Figure 3, the response control unit 2112 controls the response of the automatic response unit 2102 to the statements of the first user U1 based on the second value information and the dialogue strategy. For example, when the response control unit 2112 causes the large-scale language model LLM to generate a response for the automatic response unit 2102, it can obtain a response from the large-scale language model LLM that reflects those values ​​by defining the values ​​and dialogue strategy of the automatic response unit 2102 in the prompt.

[0096] The display control unit 2113 displays the result of the response from the automatic response unit 2102 on the display device 14 of the first user U1.

[0097] 1.2. Operation of the Automatic Response Server According to the First Embodiment 1.2.1. Operation of the Processing Unit 21 Figure 13 is a flowchart illustrating an example of the operation of the processing unit 21 in Figure 3. The operation of the processing unit 21 will be described below with reference to Figure 13. The routine in Figure 13 is started, for example, when the processing unit 21 is started, and is executed at regular intervals.

[0098] In step S11, the processing unit 21 functions as a first value determination unit 2103 to determine the values ​​of the first user U1 from the attribute information of the first user U1, the behavioral history of the first user U1, and the past conversation data of the first user U1.

[0099] In step S12, the processing unit 21 functions as an extraction unit 2104, a value pair determination unit 2105, a clustering unit 2106, and a representative value determination unit 2107 to determine compatible combinations of value pairs. The detailed processing in step S12 will be described later with reference to Figure 14.

[0100] In step S13, the processing unit 21, by functioning as a closeness determination unit 2108, determines the closeness between the first user U1 and the automatic response unit 2102 based on the conversation history between the first user U1 and the automatic response unit 2102. More specifically, the processing unit 21 determines the closeness between the first user U1 and the automatic response unit 2102 by inputting the conversation history between the first user U1 and the automatic response unit 2102 into the first learning model LM1.

[0101] In step S14, the processing unit 21, by functioning as a second value determination unit 2109, determines second value information related to the values ​​of the automatic response unit 2102. More specifically, the processing unit 21 determines the second value information based on the first value information, a plurality of value pair information, and intimacy information.

[0102] In step S15, the processing unit 21 determines a dialogue strategy by functioning as a strategy decision unit 2111. More specifically, the processing unit 21 determines one or more dialogue elements obtained by inputting the first value vector V1 and the second value vector V2 into the third learning model LM3 as a dialogue strategy that the automatic response unit 2102 can adopt.

[0103] In step S16, the processing unit 21 functions as a response control unit 2112 and controls the response of the automatic response unit 2102 to the statements of the first user U1 based on the second value information and the determined dialogue strategy.

[0104] In step S17, the processing unit 21 functions as a display control unit 2113 and displays the response from the automatic response unit 2102 on the display device 14 of the first user U1, thereby terminating this routine.

[0105] Note that the processes in steps S11 and S12 do not need to be performed before the process in step S13 is executed. Therefore, the processes in steps S11 and S12 may be performed offline.

[0106] 1.2.2. Operation of Processing Unit 21 (Subroutine Operation) Figure 14 is a flowchart showing an example of the operation of the subroutine in Figure 13. The operation of the subroutine in step S12 will be explained below with reference to Figure 14.

[0107] In step S121, the processing unit 21 functions as an extraction unit 2104 to extract multiple pairs of users from the conversation data whose conversation frequency is above a threshold.

[0108] In step S122, the processing unit 21 functions as a value pair determination unit 2105 and determines a value pair vector Vp for each of the multiple pairs extracted by the extraction unit 2104.

[0109] In step S123, the processing unit 21 functions as a clustering unit 2106 to classify multiple value pair vectors Vp into multiple clusters by clustering.

[0110] In step S124, the processing unit 21 functions as a representative value determination unit 2107 to determine a pair of value vectors that represent the corresponding cluster for each of the multiple clusters classified by the clustering unit 2106, and then terminates this subroutine.

[0111] Although steps S121 to S124 were described as a subroutine, it is not necessarily required that they be processed as a subroutine.

[0112] 1.3. Effects of the First Embodiment According to the above description, the automatic response server 20 according to the first embodiment comprises an automatic response unit 2102, a strategy decision unit 2111, and a response control unit 2112. The automatic response unit 2102 converses with multiple users U by automatically responding to statements made by multiple users U, including a first user U1. The strategy decision unit 2111 determines a dialogue strategy for the automatic response unit 2102 to the first user U1 based on first value information and second value information. The first value information is information relating to the values ​​of the first user U1. The second value information is information relating to the values ​​of the automatic response unit 2102 that will make the conversation between the first user U1 and the automatic response unit 2102 more engaging. The response control unit 2112 controls the response of the automatic response unit 2102 to statements made by the first user U1 based on the second value information and the dialogue strategy.

[0113] In this embodiment, the values ​​of the automated response unit 2102 are determined to be compatible with the values ​​of the first user U1, and a dialogue strategy that will make the conversation between the first user U1 and the automated response unit 2102 more engaging is determined. Therefore, it is expected that the conversation between the first user U1 and the automated response unit 2102 will be more engaging, and more information can be extracted from the first user U1.

[0114] Furthermore, the automated response server 20 according to the first embodiment further comprises a first value determination unit 2103. The first value determination unit 2103 determines first value information based on attribute information and first conversation information. Attribute information is information relating to the attributes of the first user U1. First conversation information is information relating to the conversation between the first user U1 and the automated response unit 2102.

[0115] In this embodiment, the values ​​of the first user U1 are determined not only based on the attribute information of the first user U1, but also based on information regarding the conversation between the first user U1 and the automated response unit 2102. Therefore, the more opportunities there are for the first user U1 to converse with the automated response unit 2102, the more information becomes available to determine the values ​​of the first user U1, and the more accurately the values ​​of the first user U1 can be determined.

[0116] Furthermore, the automated response server 20 according to the first embodiment further includes a second value determination unit 2109. The second value determination unit 2109 determines the second value information based on the first value information, a plurality of value pair information, and intimacy information. The plurality of value pair information is information that corresponds one-to-one with a plurality of user pairs among a plurality of users U in which conversations are engaging. The intimacy information is information relating to the intimacy between the first user U1 and the automated response unit 2102. Each of the plurality of value pair information includes two sets of value information relating to the values ​​of the two people belonging to the corresponding user pair.

[0117] In this embodiment, the values ​​of the automated response unit 2102 are determined to be those that are compatible with the values ​​of the first user U1. Therefore, it is expected that the conversation between the first user U1 and the automated response unit 2102 will be more engaging, and more information can be extracted from the first user U1.

[0118] Furthermore, the automated response server 20 according to the first embodiment further comprises an extraction unit 2104, a value pair determination unit 2105, a clustering unit 2106, and a representative value determination unit 2107. The extraction unit 2104 extracts pairs of users whose conversation frequency is above a threshold based on conversation information relating to conversations between multiple users U. Each of the multiple pairs belongs to user A and user B.

[0119] The value pair determination unit 2105 determines a value pair vector Vp for each of the extracted pairs. The value pair vector Vp is a pair consisting of User A's value vector Va, which represents User A's values, and User B's value vector Vb, which represents User B's values. The clustering unit 2106 classifies the multiple value pair vectors Vp, which correspond one-to-one to the multiple pairs, into multiple clusters by clustering.

[0120] The representative value determination unit 2107 determines a pair of value vectors that represent the corresponding cluster for each of the classified clusters. The first value information is a first value vector that represents the values ​​of the first user U1 as a vector. The second value information is a second value vector that represents the values ​​of the automatic response unit 2102 as a vector. Multiple pairs of value vectors that correspond one-to-one to multiple clusters constitute multiple value pair information.

[0121] The second value determination unit 2109 identifies the value vector that is most similar to the first value vector among the multiple value vectors based on the similarity between the multiple value vector pairs and the first value vector. The second value determination unit 2109 then determines the value vector that is paired with the identified value vector among the multiple value vector pairs as the second value vector.

[0122] In this embodiment, clustering determines a pair of two compatible value vectors, and among the multiple value vectors, the value vector most similar to the value vector of the first user U1 is identified. Furthermore, the value vector that is paired with the identified value vector is determined to be the value vector of the automatic response unit 2102. Therefore, values ​​that are compatible with the values ​​of the first user U1 can be determined as the values ​​of the automatic response unit 2102.

[0123] Furthermore, the automated response server 20 according to the first embodiment further includes a closeness determination unit 2108 that determines the closeness level. The closeness determination unit 2108 determines the closeness level between the first user U1 and the automated response unit 2102 by inputting the history of conversations between the first user U1 and the automated response unit 2102 into a first learning model LM1 that has already learned the relationship between conversation information and a first evaluation value. The conversation information is information about conversations between multiple users U. The first evaluation value is a value in which multiple users U evaluate the closeness level of conversations between multiple users U.

[0124] In this embodiment, the level of intimacy between the first user U1 and the automated response unit 2102 is determined using a learning model that has learned the relationship between conversational information and a first evaluation value regarding the level of intimacy evaluated by the user about the conversational information. Therefore, the level of intimacy between the first user U1 and the automated response unit 2102 can be determined more accurately.

[0125] Furthermore, the automatic response server 20 according to the first embodiment further includes an excitement level determination unit 2110. The excitement level determination unit 2110 determines the excitement level. The excitement level indicates the degree to which the conversation between the first user U1 and the automatic response unit 2102 is exciting.

[0126] The excitement level determination unit 2110 determines a first excitement level by inputting the first value vector V1, the second value vector V2, and the level of intimacy between the first user U1 and the automatic response unit 2102 to the second learning model LM2. The second learning model LM2 is a model that has already learned the relationship between the second evaluation value, the value vectors that represent the values ​​of the two people who are the subject of the conversation, and the level of intimacy between the two people who are the subject of the conversation.

[0127] The excitement level determination unit 2110 determines a second excitement level by inputting a first value vector V1, a third value vector V3, and the intimacy level between the first user U1 and the automatic response unit 2102 into the second learning model LM2. The third value vector V3 is a vector that is different from the second value vector V2 and whose similarity to the second value vector V2 is greater than or equal to a threshold. If the second excitement level is greater than the first excitement level, the second value determination unit 2109 determines the third value vector V3 as the vector representing the values ​​of the automatic response unit 2102 instead of the second value vector V2.

[0128] In this embodiment, the value vector of the automatic response unit 2102 determined by clustering is further fine-tuned to maximize the level of excitement. Therefore, the values ​​of the automatic response unit 2102 can be determined to be more compatible with the values ​​of the first user U1.

[0129] Furthermore, the strategy decision unit 2111 determines the type of dialogue element that maximizes the level of excitement obtained by inputting the first value vector V1, the second value vector V2, and the dialogue element into the third learning model LM3, as a dialogue strategy that the automatic response unit 2102 can adopt. The third learning model LM3 is a model that has learned the relationship between the combination of each value vector of multiple users, one or more dialogue elements, and a second evaluation value that evaluates the degree of excitement in the conversation.

[0130] In this embodiment, a learning model is used that has learned the relationship between the combination of value vectors of multiple users, one or more dialogue elements, and a second evaluation value regarding the level of engagement evaluated by the user for one or more dialogue elements, to determine the dialogue element that maximizes the level of engagement in the conversation. Therefore, a dialogue strategy that makes the conversation between the first user U1 and the automated response unit 2102 more engaging can be easily determined.

[0131] Furthermore, one or more elements of a dialogue must include at least one of the following: a topic of conversation, a reaction, or language.

[0132] According to this embodiment, the dialogue strategy involves at least one of the following: determining a topic that will make the conversation between the first user U1 and the automated response unit 2102 more engaging; taking reactions that will make the conversation more engaging; and using language that will make the conversation more engaging.

[0133] Furthermore, the automated response method according to the first embodiment is executed by a computer, which determines a dialogue strategy for the automated response unit 2102 to the first user U1 based on first value information and second value information, and controls the automated response unit 2102's response to the statements of the first user U1 based on the second value information and the dialogue strategy. The first value information is information relating to the values ​​of the first user U1. The second value information is information relating to the values ​​of the automated response unit 2102 that make the conversation between the first user U1 and the automated response unit 2102 more engaging. The automated response unit 2102 converses with multiple users U by automatically responding to the statements of multiple users U, including the first user U1.

[0134] In this embodiment, the values ​​of the automated response unit 2102 are determined to be compatible with the values ​​of the first user U1, and a dialogue strategy that will make the conversation between the first user U1 and the automated response unit 2102 more engaging is determined. Therefore, it is expected that the conversation between the first user U1 and the automated response unit 2102 will be more engaging, and more information can be extracted from the first user U1.

[0135] 2. Modifications This disclosure is not limited to the embodiments illustrated above. Specific examples of modifications are given below. Two or more embodiments may be arbitrarily selected from the following examples and combined. Furthermore, the embodiments of the above embodiments and the modifications below can be combined arbitrarily as long as they do not contradict each other.

[0136] 2.1. Modification 1 In the first embodiment, the elements of dialogue include the topic of conversation, reactions, and word choice, but the elements of dialogue may also include the frequency of emoji use. Emoji use tends to be higher the closer the relationship between the two parties is. Also, differences in emoji use can be observed depending on the generation of user U. Therefore, by implementing a dialogue strategy that takes emoji use into consideration, the conversation between the first user U1 and the automated response unit 2102 may become more engaging.

[0137] 2.2. Modification 2 In the first embodiment, the automatic response unit 2102 obtains a response to user U's statement from the large-scale language model LLM by inputting user U's statement into the large-scale language model LLM. However, the automatic response server 20 may be applied to a RAG (Retrieval-Augmented Generation) system that has an external knowledge database and obtains a response to user U's statement from the large-scale language model LLM by inputting the search results obtained from the knowledge database, which include documents related to user U's statement, along with user U's statement, into the large-scale language model LLM.

[0138] 3. Other (1) In the embodiments described above, the storage devices 12 and 22 are exemplified by ROM and RAM, but they can also be flexible disks, magneto-optical disks (e.g., compact disks, digital multipurpose disks, Blu-ray® disks), smart cards, flash memory devices (e.g., cards, sticks, key drives), CD-ROMs (Compact Disc-ROMs), registers, removable disks, hard disks, floppy® disks, magnetic strips, databases, servers, and other suitable storage media. The program may also be transmitted from a network via a telecommunications line. The program may also be transmitted from a communication network NET via a telecommunications line.

[0139] (2) In the embodiments described above, the information, signals, etc. may be represented using any of the various different techniques. For example, the data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0140] (3) In the embodiments described above, the input and output information may be stored in a specific location (e.g., memory) or managed using a management table. The input and output information may be overwritten, updated, or appended to. The output information may be deleted. The input information may be transmitted to other devices.

[0141] (4) In the embodiments described above, the determination may be made by a value represented using one bit (0 or 1), by a boolean value (true or false), or by a numerical comparison (for example, a comparison with a predetermined value).

[0142] (5) The processing procedures, sequences, flowcharts, etc., exemplified in the embodiments described above may be rearranged in order, as long as there is no contradiction. For example, the methods described in this disclosure present various step elements using an exemplary order and are not limited to the specific order presented.

[0143] (6) Each function illustrated in Figures 1 to 14 is realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each function block is not particularly limited. That is, each function block may be realized using one device that is physically or logically coupled, or it may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wired or wireless connections). A function block may also be realized by combining software with the one or more devices described above.

[0144] (7) The programs illustrated in the embodiments described above should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, functions, etc., whether they are called software, firmware, middleware, microcode, hardware description languages ​​or by other names.

[0145] Furthermore, software, instructions, information, etc., may be transmitted and received via a transmission medium. For example, if software is transmitted from a website, server, or other remote source using at least one of wired technology (such as coaxial cable, fiber optic cable, twisted pair, or digital subscriber line (DSL)) and wireless technology (such as infrared or microwave), then at least one of these wired and wireless technologies is included in the definition of a transmission medium.

[0146] (8) In each of the above-mentioned forms, the terms “system” and “network” shall be used interchangeably.

[0147] (9) The information, parameters, etc. described in this disclosure may be expressed using absolute values, relative values ​​from a given value, or other corresponding information.

[0148] (10) In the embodiments described above, the terminal device 10 may be a mobile station (MS). 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 several other appropriate terms. In this disclosure, terms such as “mobile station,” “user terminal,” “user equipment (UE),” and “terminal” may be used interchangeably.

[0149] (11) In the embodiments described above, the terms “connected,” “coupled,” or any variation thereof, mean 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” with each other. The coupling or connection between elements may be a physical coupling or connection, a logical coupling or connection, or a combination thereof. For example, “connection” may be reinterpreted as “access.” As used in this disclosure, two elements may be considered to be “connected” or “coupled” with each other using at least one of one or more wires, cables and printed electrical connections, and, in some non-limiting and non-exclusive examples, electromagnetic energy having wavelengths in the radio frequency domain, microwave domain and optical (both visible and invisible) domain.

[0150] (12) In the embodiments described above, the phrase "based on" does not mean "based solely on" unless otherwise specified. In other words, the phrase "based on" means both "based solely on" and "based at least on".

[0151] (13) The terms “determining” and “determining” as used in this disclosure may encompass a wide variety of actions. “Determining” may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, or inquiring (e.g., searching in a table, database, or other data structure), or ascertaining. “Determining” may also include receiving (e.g., receiving information), transmitting (e.g., sending information), inputting, outputting, or accessing (e.g., accessing data in memory). Furthermore, "judgment" and "decision" can include considering something as having been "judged" or "decided" after resolving, selecting, choosing, establishing, comparing, etc. In other words, "judgment" and "decision" can include considering something as having been "judged" or "decided" after some action. Also, "judgment (decision)" can be reinterpreted as "assuming," "expecting," or "considering."

[0152] (14) In the embodiments described above, where “include,” “including,” and variations thereof are used, these terms are intended to be inclusive, as is the term “comprising.” Furthermore, the term “or” as used in this disclosure is not intended to be exclusive OR.

[0153] (15) In the present disclosure, if articles are added by translation, such as a, an, and the in English, the present disclosure may include the fact that the noun following these articles is plural.

[0154] (16) In this 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 “combine” may be interpreted in the same way as “different.”

[0155] (17) Each aspect / embodiment described herein may be used individually, in combination, or switched between as needed during implementation. Furthermore, notification of the specified information (e.g., notification that "it is X") is not limited to explicit notification, but may also be implicit (e.g., by not providing such notification).

[0156] Although the present disclosure has been described in detail above, it will be 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 intent and scope of the present disclosure as defined by the claims. Accordingly, the descriptions in the present disclosure are illustrative and not restrictive in any way.

[0157] 10...Terminal device, 14...Display device, 20...Automatic response server, 2102...Automatic response unit, 2103...First value determination unit, 2104...Extraction unit, 2105...Value pair determination unit, 2106...Clustering unit, 2107...Representative value determination unit, 2108...Intimacy determination unit, 2109...Second value determination unit, 2110...Excitement level determination unit, 2111...Strategy determination unit, 2112...Response control unit, 2113...Display control unit, C1, C2, C3...Cluster, DB1...Knowledge database, LLM...Large-scale language model, LM1...First learning model, LM2...Second learning model, LM3...Third learning model, U...User, U1...First user, V1...First value vector, V2...Second value vector, V3...Third value vector, Va...Value vector of user A. Vb...Value vector of user B.

Claims

1. An automated response device comprising: an automated response unit that converses with multiple users by automatically responding to statements made by multiple users, including a first user; a strategy determination unit that determines a dialogue strategy for the automated response unit to the first user based on first value information relating to the first user's values ​​and second value information relating to the values ​​of the automated response unit that enhance the conversation between the first user and the automated response unit; and a response control unit that controls the automated response unit's response to the first user's statements based on the second value information and the dialogue strategy.

2. The automated response device according to claim 1, further comprising a first value determination unit that determines the first value information based on attribute information relating to the attributes of the first user and first conversation information relating to a conversation between the first user and the automated response unit.

3. The automatic response device according to claim 1, further comprising a second value determination unit for determining the second value information, wherein the second value determination unit determines the second value information based on the first value information, a plurality of value pair information corresponding one-to-one to a plurality of user pairs among the plurality of users with whom conversation is engaging, and intimacy information relating to the intimacy between the first user and the automatic response unit, and each of the plurality of value pair information includes two value information relating to the values ​​of two people belonging to the corresponding user pair.

4. Based on conversation information relating to the conversations of the multiple users, the system extracts multiple pairs of users whose conversation frequency is above a threshold, and each of the multiple pairs comprises a second user and a third user; a value pair determination unit determines a value pair vector for each of the extracted multiple pairs, which is a pair of a second user value vector representing the values ​​of the second user and a third user value vector representing the values ​​of the third user; a clustering unit classifies the multiple value pair vectors corresponding one-to-one to the multiple pairs into multiple clusters by clustering; and a representative value determination unit determines a pair of value vectors representing the corresponding cluster for each of the classified multiple clusters, wherein the first value information is a first value vector representing the values ​​of the first user as a vector, the second value information is a second value vector representing the values ​​of the automatic response unit as a vector, the pairs of multiple value vectors corresponding one-to-one to the multiple clusters are the multiple value pair information, and the second value determination unit is The automatic response device according to claim 3, wherein, based on the similarity between the pair of the plurality of value vectors and the first value vector, the value vector that is most similar to the first value vector among the plurality of value vectors is identified, and the value vector that is paired with the identified value vector among the pair of the plurality of value vectors is determined to be the second value vector.

5. The automatic response device according to claim 3, further comprising a closeness determination unit for determining the closeness, wherein the closeness determination unit determines the closeness between the first user and the automatic response unit by inputting the history of conversations between the first user and the automatic response unit into a first learning model that has learned the relationship between conversation information relating to conversations between the plurality of users and first evaluation values ​​(closeness labels) in which the plurality of users evaluated the closeness of conversations between the plurality of users.

6. The automatic response device according to claim 4, further comprising an excitement level determination unit for determining an excitement level indicating the degree to which the conversation between the first user and the automatic response unit is engaging, wherein the excitement level determination unit determines a first excitement level as the excitement level by inputting the first value vector, the second value vector, and the level of intimacy between the first user and the automatic response unit into a second learning model that has learned the relationship between a second evaluation value that evaluates the degree of excitement of the conversation, a value vector indicating the values ​​of the two people who are the subject of the conversation, and the level of intimacy between the two people who are the subject of the conversation; and determines a second excitement level as the excitement level by inputting the first value vector, a third value vector that is different from the second value vector and whose similarity to the second value vector is greater than or equal to a threshold, and the level of intimacy between the first user and the automatic response unit into the second learning model; and if the second excitement level is greater than the first excitement level, the second value level determination unit determines the third value vector as the vector representing the values ​​of the automatic response unit in place of the second value vector.

7. The automatic response device according to claim 4, wherein the strategy determination unit determines, as a dialogue strategy that the automatic response unit can adopt, the type of dialogue element that maximizes the level of excitement obtained by inputting the first value vector, the second value vector, and the dialogue element into a third learning model that has learned the relationship between the combination of value vectors of the plurality of users, one or more dialogue elements, and a second evaluation value that evaluates the degree of excitement of the conversation.

8. The automatic response device according to claim 7, wherein the one or more elements of the dialogue include at least one of the following: a topic of conversation, a reaction, and word choice.

9. An automated response method performed by a computer, comprising: first value information relating to the values ​​of a first user and second value information relating to the values ​​of the automated response unit that make the conversation between the first user and the automated response unit more engaging, determining a dialogue strategy for the automated response unit to the first user; controlling the automated response unit's response to the first user's statements based on the second value information and the dialogue strategy; and the automated response unit conversing with multiple users, including the first user, by automatically responding to the statements of multiple users.

Images