Problem consultation device, problem consultation system, control program, and control method

Abstract

To make it possible to estimate a schema being a cause of an anguish of a user by interaction with the user, and to make advice for the estimated schema.SOLUTION: An information processing device 10 that functions as an anguish consultation device includes a computer 12, and a microphone 14 and an output device 16 which are connected with the computer. The information processing device outputs a plurality of questions regarding the anguish of the user by voice sound from the output device according to a column rule. The information processing device detects a response voice sound of the user to each question, and estimates the schema being the cause of the anguish of the user on the basis of the detected voice sound for each question. When the schema is estimated, the information processing device outputs an advice by voice sound for the estimated schema from the output device.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a problem consultation device, a problem consultation system, a control program, and a control method, and more particularly to a problem consultation device, a problem consultation system, a control program, and a control method that provide advice on a schema that is the cause of a user's problem, for example. [Background technology]

[0002] An example of this type of conventional counseling device is disclosed in Patent Document 1. The automated cognitive behavioral therapy system disclosed in Patent Document 1 is a device that automatically performs cognitive behavioral therapy by algorithmizing it to improve cognitive distortions and behaviors by having a subject self-assess the intensity of their mood at each time or when problem behavior occurs, or evaluate the progress of task achievement, and then sending the status and evaluation score to a database from a smartphone or the like, categorizing it by keyword, recording, calculating, comparing, and analyzing the evaluation score for each category, and automatically providing counseling based on the results. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2015-153413 Summary of the Invention [Problem to be solved by the invention]

[0004] The cognitive behavioral therapy automation system disclosed in the above-mentioned Patent Document 1 only discloses that cognitive behavioral therapy is algorithmized, but does not disclose any specific algorithm.

[0005] Therefore, a primary object of the present invention is to provide a novel problem consultation device, problem consultation system, control program, and control method.

[0006] Another object of the present invention is to provide a problem consultation device, a problem consultation system, a control program, and a control method that can infer the schema that is the cause of a user's problem through dialogue with the user and provide advice regarding the inferred schema. [Means for solving the problem]

[0007] The first invention includes an output unit that outputs each of a predetermined plurality of first questions by voice, a voice detection unit that detects a user's voice, and a keyword detection unit that detects a keyword for each of the predetermined plurality of first questions based on the user's voice detected by the voice detection unit. ,tree The keyword detection unit detects the cause of the user's trouble based on the detected keyword. These are thinking habits that are classified into three types: high achievement orientation, other-dependent evaluation, and fear of failure. Infer the schema to be used a related keyword storage unit that stores related keywords corresponding to each of the three types; and a score adding unit that adds a score for the type corresponding to the related keyword when the detected keyword matches the related keyword. Equipped with the estimation unit estimates a schema based on the addition result of the score addition unit; The output unit is a problem consultation device that outputs advice about the schema estimated by the estimation unit in the form of voice.

[0008] A second invention is according to the first invention, and the predetermined plurality of first questions include 5W1H questions about the cause of the worry.

[0010] No. 3 The invention is 1st or 2nd The estimation unit estimates, as a schema, the type with the highest score among the three types for which the scores have been added by the score adding unit.

[0011] No. 4 The invention is 3 The invention is based on the above, and when there are multiple types with the highest score, the output unit outputs a second question by voice, which is a question for narrowing down the types to one, the voice detection unit detects the voice of the user's answer to the second question, and the estimation unit estimates the one narrowed-down type as a schema based on the user's answer to the second question.

[0012] No. 5 The invention is 4The invention is based on the above, and when there is a category for which no score has been added by the score adding section, the output section outputs a third question by voice, which is a question as to whether the user corresponds to the category for which no score has been added, and the voice detection section detects the voice of the user's answer to the third question, and when the user's answer to the third question indicates that the user corresponds to a category for which no score has been added, the voice detection section corrects the score for the category for which no score has been added to the maximum score.

[0013] No. 6 The invention includes an output device that outputs each of a predetermined number of questions by voice, a voice detection unit that detects the voice of a user, Quality a keyword detection unit that detects a keyword based on the user's voice detected by the voice detection unit for each question; ,tree The keyword detection unit detects the cause of the user's trouble based on the detected keyword. These are thinking habits that are classified into three types: high achievement orientation, other-dependent evaluation, and fear of failure. Infer the schema to be used a related keyword storage unit that stores related keywords corresponding to each of the three types; and a score adding unit that adds a score for the type corresponding to the related keyword when the detected keyword matches the related keyword. Equipped with the estimation unit estimates a schema based on the addition result of the score addition unit; The output device is a problem consultation system that outputs advice about the schema estimated by the estimation unit in the form of voice.

[0014] No. 7 The invention of The system is provided with a related keyword storage unit that stores related keywords corresponding to each of the three types of thinking habits: high achievement orientation, other-dependent evaluation, and failure anxiety. A control program executed by a computer of a problem consultation device, the control program including: a voice output step of outputting each of a predetermined plurality of questions by voice to an output means; a voice detection step of detecting a user's voice; and a keyword detection step of detecting a keyword for each of the predetermined plurality of questions based on the user's voice detected in the voice detection step. ,tree The cause of the user's trouble is identified based on the detected keywords, which are keywords detected in the keyword detection step. These are thinking habits that are classified into three types. Inference step to infer the schema and a score adding step of adding a score for a type corresponding to a related keyword when the detected keyword matches the related keyword. Execute The estimation step estimates a schema based on the addition result of the score addition step, The voice output step is a control program that outputs advice about the schema estimated in the estimation step to the output means in the form of voice.

[0015] No. 8 The invention of The system is provided with a related keyword storage unit that stores related keywords corresponding to each of the three types of thinking habits: high achievement orientation, other-dependent evaluation, and failure anxiety. A method for controlling a problem consultation device, comprising: (a) a step of outputting each of a plurality of predetermined questions by voice to an output means; (b) a step of detecting a user's voice; and (c) a step of detecting a keyword for each of the plurality of predetermined questions based on the user's voice detected in step (b). 、( d) determining the cause of the user's trouble based on the detected keywords, which are the keywords detected in step (c); These are thinking habits that are classified into three types. Steps to infer a schema and (e) adding a score to the type corresponding to the related keyword when the detected keyword matches the related keyword. Including, Step (d) infers a schema based on the summation result of step (e); Step (a) is a control method in which advice about the schema estimated in step (d) is outputted by voice to an output means. [Effects of the Invention]

[0016] According to this invention, the schema that is the cause of the user's worries is inferred through dialogue with the user, and therefore advice can be given regarding the inferred schema.

[0017] The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description of the preferred embodiments with reference to the drawings. [Brief explanation of the drawings]

[0018] [Figure 1] FIG. 1 is a block diagram showing the electrical configuration of an information processing apparatus according to this embodiment of the present invention. [Figure 2] Figure 2(A) shows an example of a user's response to a question about the situation, Figure 2(B) shows an example of detected keywords obtained from the response shown in Figure 2(A), Figure 2(C) shows an example of the first table of expected keywords for the worry category, and Figure 2(D) shows an example of the agent's speech content in response to the user's response. [Figure 3] Figure 3(A) shows an example of a user's response to the first question about triggers, Figure 3(B) shows an example of a user's response to the second question about triggers, and Figure 3(C) shows an example of detected keywords obtained from the responses to the questions about triggers. [Figure 4] FIG. 4 shows an example of the detection keywords stored in the memory. [Figure 5] Figure 5(A) shows an example of a second table listing expected keywords for each category based on the four perspectives in the 5Ws, and Figure 5(B) shows an example of the agent's speech using the names of categories based on the perspectives when the expected keywords include detected keywords obtained from the user's answers to questions about the 5Ws. [Figure 6] Figure 6(A) shows an example of a user's response to a question about mood, Figure 6(B) shows an example of detected keywords obtained from the response to the question about mood, Figure 6(C) shows an example of a third table of expected keywords corresponding to mood categories, and Figure 6(D) shows examples of the agent's speech when the detected keywords obtained from the user's response to the question about mood are included in the expected keywords and when the detected keywords are not included in the expected keywords. [Figure 7] FIG. 7 shows an example of a fourth table of related keywords corresponding to schema types. [Figure 8] FIG. 8(A) shows an example of a question for a schema type, and FIG. 8(B) shows an example of a question for narrowing down the search results for two schema types. [Figure 9] FIG. 9(A) shows an example of an automatic thought question about high achievement orientation, and FIG. 9(B) shows an example of an automatic thought question about other-dependent evaluation. [Figure 10] FIG. 10(A) shows an example of an automatic thought question about anxiety about failure, and FIG. 10(B) shows an example of a question when a schema has not yet been estimated. [Figure 11] FIG. 11(A) is a diagram showing an example of a detailed schema of high achievement orientation, and FIG. 11(B) is a diagram showing an example of a detailed schema of other-dependent evaluation. [Figure 12] Figure 12 shows an example of a detailed schema of failure anxiety. [Figure 13]FIG. 13(A) shows an example of a fourth table of keywords corresponding to the detailed schema of high achievement orientation, and FIG. 13(B) shows an example of a fifth table of keywords corresponding to the detailed schema of other-dependent evaluation. [Figure 14] FIG. 14 shows an example of the sixth table of corresponding keywords in the detailed schema of the fear of failure. [Figure 15] FIG. 15(A) shows an example of the explanation content for each schema type, and FIG. 15(B) shows an example of the advice content for each schema type. [Figure 16] FIG. 16 is a diagram showing an example of a memory map of a RAM built into the computer shown in FIG. [Figure 17] FIG. 17 is a flowchart showing an example of information processing by the CPU built in the computer shown in FIG. [Figure 18] FIG. 18 is a flowchart showing an example of a process for asking questions about the status of the CPU built in the computer shown in FIG. [Figure 19] FIG. 19 is a flowchart showing an example of question processing regarding a trigger by the CPU built in the computer shown in FIG. [Figure 20] FIG. 20 is a flowchart showing an example of a query process regarding 5W of the CPU built into the computer shown in FIG. [Figure 21] FIG. 21 is a flowchart showing an example of mood question processing by the CPU built into the computer shown in FIG. [Figure 22] FIG. 22 is a flow chart showing an example of estimating a schema of a CPU built in the computer shown in FIG. [Figure 23] FIG. 23 is a flow chart showing an example of a query process for a schema of the CPU built in the computer shown in FIG. [Figure 24] FIG. 24 is a flowchart showing an example of a query process for a zero-point schema by the CPU built in the computer shown in FIG. [Figure 25] FIG. 25 is a flowchart showing an example of a query process for narrowing down the search results by the CPU built in the computer shown in FIG. [Figure 26] FIG. 26 is a flowchart showing the automatic thought question processing of the CPU built in the computer shown in FIG. [Figure 27] FIG. 27 is a flowchart showing a part of an example of detailed schema estimation processing by the CPU built in the computer shown in FIG. [Figure 28] FIG. 27 is a flowchart showing another part of the detailed schema estimation processing of the CPU built in the computer shown in FIG. 1, and follows FIG. [Figure 29] FIG. 29 is a flowchart showing advice processing by the CPU built in the computer shown in FIG. DETAILED DESCRIPTION OF THE INVENTION

[0019] Referring to FIG. 1, an information processing apparatus 10 of this embodiment includes a computer 12 , a microphone 14 and an output device 16 , and the microphone 14 and the output device 16 are connected to the computer 12 .

[0020] The computer 12 is a general-purpose server or PC and includes components such as a CPU 12a, RAM 12b, a communication device 12c, and an input / output interface. Although not shown, the computer 12 also includes other storage units configured with nonvolatile memory such as an HDD, flash memory, or EEPROM, or semiconductor memory such as an SSD.

[0021] The CPU 12a is a processor that controls the overall operation of the computer 12. The RAM 12b is the main storage device of the computer 12 and functions as a buffer area and a work area for the CPU 12a. The communication device 12c is a communication module for communicating with other computers on a network via wired or wireless communication using a communication method such as Ethernet or Wi-Fi. However, as will be described later, if the output device 16 is a robot, the computer 12 communicates with the robot using the communication device 12c.

[0022] The microphone 14 is a general-purpose sound-collecting microphone, and functions as a sensor for inputting or detecting the user's voice.

[0023] The output device 16 is, for example, a general-purpose speaker that outputs the voice of the agent. However, the output device 16 may also include a display in addition to the speaker, and a still image or a moving image (animation) of the agent (character object) may be displayed on the display. Alternatively, the output device 16 may be a robot equipped with a speaker. In such a case, a problem consultation system is configured in which the computer 12 and the robot, i.e., the output device 16, are communicatively connected. The robot may be a communication robot that performs communication behavior using at least one of physical movement and voice, such as Robovie (registered trademark) developed and sold by the applicant or Sota (registered trademark) manufactured by Vstone Co., Ltd. In this case, the microphone equipped on the robot may be used as the microphone 14. The computer 12 may also be built into the robot, or the robot's control device or computer may function as the computer 12.

[0024] Although detailed explanations will be omitted, when using a robot, when outputting voice, the robot expresses the appearance of speaking through body movements, and when listening to the user's speech, the direction of the robot's eyes or face (or head) is controlled so that it focuses on the user.

[0025] The information processing device 10 configured in this manner also functions as a problem consultation device, in which a user using this problem consultation device interacts with an agent, the agent asks a number of questions about the problem, and by obtaining the user's answers to the number of questions, the information processing device infers the schema that is the cause of the user's problem and provides the user with advice according to the inferred schema.

[0026] However, in this embodiment, the agent is a predetermined character or robot. The agent may be a predetermined character displayed on a display as the output device 16, or may include the computer 12. Furthermore, if the output device 16 is a speaker, the agent, which has no physical form and only speaks, interacts with the user.

[0027] The information processing device 10 stores in memory (RAM 12b) the voice data of the speech content required for interacting with the user, and selects an appropriate response according to a scenario created or set in advance, and in response to keywords detected from the speech uttered by the user, and causes the agent to speak it.

[0028] However, the scenario is a dialogue flow set in a flow diagram (see FIG. 12) of information processing (overall processing for problem consultation) to be described later.

[0029] Furthermore, the information processing device 10 detects the voice spoken by the user using the microphone 14, converts the voice data corresponding to the detected voice into character data by performing voice recognition, and detects keywords from the converted character data.

[0030] Hereinafter, the problem consultation services provided by the information processing device 10 will be described, while showing specific examples of dialogues regarding work-related problems and school-related problems.

[0031] In this embodiment, the information processing device 10 uses the column method to ask questions to the user in a consultation session, and infers the schema that is the cause of the user's worries based on the user's answers.

[0032] Here, "schema" is a term used in cognitive psychology, and in this example refers to a human cognitive process or a habit of thinking. In this example, it is classified into three types based on the Depression Schema Scale created for Japanese people. The first is high achievement orientation, the second is other-dependent evaluation, and the third is failure anxiety.

[0033] The classification based on the Depression Schema Scale was made with reference to "Iesetsugu, Tetsuji, and Masahiro Kodama. An attempt to create a new Depression Schema Scale. Health Psychology Research, Vol. 12, No. 2, pp. 37-46, 1999."

[0034] For people whose schema is high achievement orientation, their thinking habits include a tendency to have high ideals and be strict with themselves. For people whose schema is other-dependent evaluation, their thinking habits include a tendency to value what others think of them. For people whose schema is failure anxiety, their thinking habits include a tendency to feel strong anxiety about the possibility of things not going well.

[0035] To give an overview of the problem consultation, the information processing device 10 executes a process of asking the user questions about the situation, then a process of asking questions about triggers, then a process of asking questions about the 5Ws, and further a process of asking questions about mood, and then executes a process of estimating a schema based on the answers to these questions. In the schema estimation process, if a schema cannot be estimated based on the results of the above question processing, the information processing device 10 further executes a question processing to estimate a schema, and estimates a schema based on the answers.

[0036] In the situation question process, the agent asks a question by voice, such as "Is there anything that has been bothering you recently?" The user responds to this question. The information processing device 10 then detects the voice of the user's response, performs speech recognition on the detected voice, and detects keywords from the text sentence generated by the speech recognition.

[0037] FIG. 2(A) shows examples of user responses, and FIG. 2(B) shows examples of keywords detected from the user responses (hereinafter referred to as "detected keywords"). However, in FIGS. 2(A) and (B), when the identification information (ID) is "1", it is a case of work-related worries, and when the ID is "2", it is a case of school-related worries. This is also true for FIGS. 2(B), 2(D), 3(A)-3(C), 4, 5(B), 6(A), 6(B), and 6(D).

[0038] As shown in FIGS. 2A and 2B, in the case of a work-related problem consultation, a response such as "I've made a lot of mistakes at work recently, and I hate myself for it." Keywords such as "recently," "workplace," "many times," "mistake," and "self-loathing" are detected from such a user's response. That is, detected keywords are obtained. In addition, in the case of a school-related problem consultation, a response such as "I can't keep up with the classes, and if this continues, I feel like I'll have to repeat the year" is obtained, and keywords such as "class" and "repeated the year" are detected from such a user's response. That is, detected keywords are obtained. The detected keywords are stored in memory (in this embodiment, RAM 12b). However, the detected keywords are categorized by the 5W perspectives or elements (in this embodiment, "when," "where," "with whom," "what you're doing," and "why") and stored in memory in association with each perspective. The detected keywords are classified using dictionary data (not shown). As an example, the dictionary data is provided to classify the detected keywords for each of the perspectives of "when," "where," "with whom," "what you're doing," and "why," excluding the "why" perspective. Although a detailed explanation will be omitted, the dictionary data contains a large number of keywords corresponding to each aspect of "when," "where," "with whom," and "what you were doing." The same applies to the case where a detected keyword is obtained.

[0039] The information processing device 10 determines whether the detected keyword is an expected keyword (hereinafter referred to as an "expected keyword"). An expected keyword means a keyword that is expected to be included in the user's answer to a question from an agent.

[0040] FIG. 2(C) is a table (hereinafter referred to as "Table 1") showing examples of expected keywords for answers to questions about situations. The expected keywords are classified into categories (here, "worry categories"). As shown in Table 1 of FIG. 2(C), in this example, examples of worry categories are "I failed at something" and "Anxiety about the future."

[0041] In the first table, examples of expected keywords are listed for the worry category "I failed at something," such as "failure," "mistake," and "mistake." In addition, in the first table, examples of expected keywords are listed for the worry category "anxiety about the future," such as "repeating a year," "career path," and "job hunting."

[0042] These are merely examples and are not necessarily limiting. The concern categories and expected keywords may be changed or added as appropriate depending on the concern.

[0043] If the detected keyword is included in the expected keywords, the information processing device 10 checks the content, but if the detected keyword is not included in the expected keywords, it makes a reply.

[0044] FIG. 2(D) shows an example of the content of the agent's utterance when the detected keyword is included in the expected keywords.

[0045] For example, in the above example, in the case of a work-related problem consultation, the expected keyword among the detected keywords is "mistake," and the problem category is "I failed at something." Therefore, as shown in Figure 2(D), the agent utters, "I see, so you failed at something."

[0046] In the above example, when the student is seeking advice about school-related worries, the expected keyword among the detected keywords is "repeating a grade," and the worry category is "anxiety about the future." Therefore, as shown in Figure 2(D), the agent utters, "I see, so you're worried about the future."

[0047] In this way, when the detected keyword is included in the expected keywords, the system confirms whether the worry category corresponding to the expected keyword is correct as a classification of the content (i.e., the worry) answered by the user by speaking the content using the name of the worry category.

[0048] In this embodiment, the worry category itself is not used to estimate the schema, so even if the user answers in response to checking whether the worry category is correct, no special consideration is given to the user's answer.

[0049] On the other hand, if the detected keyword is not included in the expected keywords, the agent will only reply. This is because the content of the user's answer cannot be classified into any of the worry categories. For example, the agent may reply, "I see."

[0050] Next, the information processing device 10 executes a questioning process about the trigger. Here, the agent asks two questions about the trigger that caused the user to have a problem. First, the agent asks the first question about the trigger: "When did this condition start?"

[0051] As an example, as shown in Figure 3(A), in response to the first question about the trigger, if the concern is work-related, the answer will be something like "It's been about the past month," and if the concern is school-related, the answer will be something like "It started in April."

[0052] Once the first question is answered, the agent responds, "I understand," and then asks a second question about the trigger: "Tell me about the trigger."

[0053] As an example, as shown in Figure 3(B), in response to the second question about what triggered the problem, in the case of work-related concerns, responses such as "I think there was a time when my boss got really angry at me" are obtained, and in the case of school-related concerns, responses such as "It was when I started taking classes online at home" are obtained.

[0054] When the answer to the second question is obtained, the agent replies, "I understand." Subsequently, the information processing device 10 executes a question process regarding the 5Ws.

[0055] However, when the user answers each question, the information processing device 10 detects the voice of the user's answer, performs voice recognition on the detected voice, and detects keywords from the text sentence generated by the voice recognition.

[0056] Therefore, as shown in Figure 3(C), the keywords "one month," "boss," and "being scolded" are detected from the answers to the above two questions in the case of work-related worries, while the keywords "April," "class," and "home" are detected in the case of school-related worries.

[0057] In the 5W question processing, questions are asked from the 5W perspectives excluding perspectives corresponding to detected keywords obtained from answers to questions about the situation and answers to questions about the trigger.

[0058] In this embodiment, even if answers related to "why" are given to questions about the situation and questions about the trigger, they are not classified. Therefore, in the 5W question processing, questions are always asked from the perspective of "why" to collect answers (or information).

[0059] FIG. 4 is an example of a classification table in which detected keywords obtained from the answers to the questions about the above-mentioned situations and the questions about triggers are classified by viewpoint.

[0060] As can be seen from FIG. 4, in the case of work-related worries, since detected keywords are stored for the perspectives of "when," "where," and "with whom," it is decided to ask questions from the perspectives of "what you are doing" and "why." When questions are asked from multiple perspectives, the questions are executed in the order of earliest when the 5W perspectives are arranged in the order of "when," "where," "with whom," "what you are doing," and "why." Therefore, the questions about "what you are doing" and "why" are executed in that order. However, this is just an example, and the order of questions may also be decided randomly.

[0061] Therefore, the agent asks the first question, "What were you doing?" To this, an answer such as "I was working" or "I was doing my job" is obtained. Therefore, the detected keyword is "work" or "job." Next, the agent asks the second question, "Why did it happen?" To this, an answer such as "I was absent-minded" is obtained.

[0062] Furthermore, as can be seen from Figure 4, in the case of a school-related problem consultation, the detected keywords are stored for the perspectives of "when," "where," and "what you were doing," so it is decided that questions will be asked in that order for the perspectives of "with whom" and "why."

[0063] So the agent asks the first question, "Who were you with?", which elicits answers like "I was with my family," "I was with my parents," or "I was with my mother." The agent then asks the second question, "Why did it happen?", which elicits answers like "I don't understand."

[0064] When the information processing device 10 receives an answer from the user, it acquires the detected keyword and determines whether the detected keyword is an expected keyword. Fig. 5(A) is an example of a table of expected keywords for questions about the 5Ws (hereinafter referred to as "second table").

[0065] As shown in the second table of FIG. 5(A), the expected keywords are classified into categories (here, "categories by perspective") for each of the 5W perspectives. As shown in the second table of FIG. 5(A), in this embodiment, examples of categories by perspective for the perspective of "when" are "recently" and "long ago." Also, examples of categories by perspective for the perspective of "where" are "home" and "workplace." Furthermore, examples of categories by perspective for the perspective of "with whom" are "family" and "coworkers." Furthermore, examples of categories by perspective for the perspective of "what you are doing" are "study" and "work."

[0066] As shown in Table 2, no expected keywords have been set for the "why" perspective of the 5Ws. This is because there are many different perspectives, or reasons, for "why," and the expected keywords are enormous, and it would be cumbersome to categorize them by perspective.

[0067] In the second table, examples of expected keywords are listed for the "recently" category of the "when" perspective, such as "recently," "this week," and "one month." In addition, examples of expected keywords are listed for the "long ago" category of the "when" perspective, such as "long ago" and "when I was a child."

[0068] Similarly, in the second table, examples of expected keywords are listed for "home" in the category by perspective regarding "where." Also, examples of expected keywords are listed for "workplace" in the category by perspective regarding "where."

[0069] Similarly, in the second table, examples of expected keywords are listed for "family" in the category by perspective regarding "with whom." Also, examples of expected keywords are listed for "coworkers" in the category by perspective regarding "with whom."

[0070] Similarly, in the second table, examples of expected keywords are listed for "study," "homework," and "class," corresponding to the category "study" by perspective regarding the perspective of "what you are doing." Also, examples of expected keywords are listed for "work" by perspective regarding the perspective of "what you are doing."

[0071] These are merely examples and are not intended to be limiting. Here, the categories and expected keywords by perspective are only a partial list, and may be changed or added as appropriate depending on the problem. For example, as a category by perspective for the perspective of "where," "school" is further provided, and "school" and "classroom" are listed as expected keywords. Furthermore, as a category by perspective for the perspective of "with whom," "acquaintance" is further provided, and "acquaintance," "friend," and "acquaintance" are listed as expected keywords.

[0072] When the information processing device 10 receives a response from the user, it refers to the second table to determine whether the detected keyword is an expected keyword, and if the detected keyword is an expected keyword, it checks the user's response using categories by viewpoint. On the other hand, if the detected keyword is not an expected keyword, it simply replies, "I see."

[0073] FIG. 5(B) shows the content of utterances using the names of categories by viewpoint when the detected keywords obtained from the user's answers to questions about the 5Ws are included in the expected keywords.

[0074] For example, in the case of a consultation regarding work-related worries, if the detected keywords are "work" or "tasks," the agent will say, "So you were at work," using "work," a category based on that perspective, as shown in Figure 5(B), since the detected keywords are expected keywords.

[0075] Also, in the case of a consultation about school-related worries, if the detected keywords are "family," "parents," "mother," or "father," the agent will utter, "You were with your family," as shown in Figure 5(B), using the perspective-specific category "family," since the detected keywords are expected keywords.

[0076] In this way, when the detected keyword is included in the expected keywords, the content using the name of the perspective category is spoken to confirm whether the perspective category corresponding to the expected keyword is correct as a classification of the content answered by the user (i.e., the 5W perspective).

[0077] This questioning process for the 5Ws is repeated until all the determined questions have been asked. However, in this embodiment, even if sufficient information is not obtained in response to a question, the same question is not repeated. However, the same question may be repeated until sufficient information is obtained in response to the question.

[0078] When the information processing device 10 finishes the question process about 5W, it then executes a question process about mood, that is, a question process about 1H.

[0079] In the mood question process, the agent asks a question by voice, such as "Tell me how you're feeling at that time." In response to this, the user answers. The information processing device 10 then detects the voice of the user's answer, performs speech recognition on the detected voice, and detects keywords from the text sentence generated by the speech recognition.

[0080] As shown in FIGS. 6(A) and 6(B), in the case of a work-related problem consultation, the user answers something like, "I felt sad and self-loathing." From such an answer, the keywords "sad" and "self-loathing" are detected. In other words, detected keywords are obtained. Also, in the case of a school-related problem consultation, an answer like, "I thought I might have to repeat the year if I continued like this," is obtained, and the keyword "repeating the year" is detected from such an answer. In other words, detected keywords are obtained.

[0081] The information processing device 10 determines whether the detected keyword is an expected keyword. Here, an expected keyword means a keyword that is expected to be included in the user's answer to a question about mood from an agent.

[0082] FIG. 6(C) is a table (hereinafter referred to as "Table 3") showing examples of expected keywords that are expected to be included in answers to questions about mood. The expected keywords are classified into categories (here, "mood categories"). As shown in Table 3 of FIG. 6(C), in this example, examples of mood categories are "sad" and "self-loathing."

[0083] In the third table, examples of expected keywords are listed for the mood category "sad," such as "sad" and "was sad." Also, in the third table, examples of expected keywords are listed for the mood category "self-loathing," such as "self-loathing" and "hate myself."

[0084] These are merely examples and are not necessarily limiting. Mood categories and assumed keywords may be changed or added as appropriate depending on the worries.

[0085] When the detected keyword is included in the expected keywords, the information processing device 10 checks the content, but when the detected keyword is not included in the expected keywords, it points out that the answer about mood has not been given and asks about mood again.

[0086] FIG. 6(D) shows an example of the agent's utterance content when the detected keyword is included in the expected keywords, and an example of the agent's utterance content when the detected keyword is not included in the expected keywords.

[0087] For example, in the above example, when the user is seeking advice about work-related worries, the detected keywords "sad" and "self-loathing" are included in the expected keywords, and the mood categories are "sad" and "self-loathing." Therefore, as shown in Figure 6(D), the agent utters, "I see, your mood in that situation can be summarized as 'sad', 'self-loathing', etc."

[0088] Also, in the above example, when the consultation is about school-related worries, the detected keyword "repeating a grade" is not an expected keyword. Therefore, as shown in Figure 6(D), the agent utters, "I'm sorry if I missed it, but what you just said was more of a thought than a mood. Moods can often be expressed in one word, such as 'happy' or 'sad'. So, how do you feel about the thought you just said?" In this way, by asking about mood again, an answer about mood can be obtained.

[0089] In this embodiment, the mood category itself is not used to estimate the schema, so even if the user answers in response to checking whether the mood category is correct, no special consideration is given to the user's answer.

[0090] Furthermore, the detection keywords for mood are stored in the memory in correspondence with the viewpoint of mood (1H), similar to the detection keywords for 5W.

[0091] After completing the mood question processing, the information processing device 10 executes a schema estimation process. In the schema estimation process, a table of related keywords written corresponding to the schema (hereinafter referred to as the "fourth table") is used. As described above, the schema types are high achievement orientation, other-dependent evaluation, and failure anxiety. Each related keyword is a keyword associated with the corresponding schema. More specifically, the related keywords are keywords contained in the answers of users who have the corresponding schema, and are empirically determined based on the answers of a large number of users obtained through experiments. However, the user's answers are the user's answers to each of the above-mentioned situation question processing, trigger question processing, 5W question processing, and mood question processing.

[0092] As shown in Figure 7, the fourth table lists related keywords corresponding to the high-achievement-oriented schema, such as compromise, unyielding, complex, impatience, irritation, not making an effort, lack of effort, lack of practice, average, in comparison, and failure.

[0093] Additionally, the fourth table lists related keywords corresponding to the other-dependent evaluation schema, such as wanting to be recognized, wanting to be understood, wanting to be understood, isolated, feeling out of place, peer pressure, not being able to express one's opinion, not being able to express oneself well, being ridiculed, being rejected, and being scolded.

[0094] Furthermore, in the fourth table, related keywords such as failure, anxiety, anxiety about the future, true feelings, weakness, self-loathing, distrust of people, cannot be trusted, betrayal, and being deceived are listed in correspondence with the failure anxiety schema.

[0095] In this embodiment, to estimate the schema that is the cause of the user's worries, the number (score) of keywords related to or associated with high achievement orientation, other-dependent evaluation, and failure anxiety is calculated. That is, for each of the detected keywords obtained from the user's answers to the above-mentioned question processing about the situation, question processing about triggers, question processing about the 5Ws, and question processing about mood, it is determined whether it is a related keyword (i.e., whether it matches a related keyword). If the detected keyword is a related keyword, a score is added for the schema corresponding to the related keyword.

[0096] After the process of adding up the scores for all detected keywords is executed, the scores of each schema are compared, and the schema that is the cause of the user's worries is inferred based on the comparison results. In this embodiment, if there is no schema with a score of 0 and there is one schema with the maximum score (i.e., the maximum score), the schema with the maximum score is inferred as the schema that is the cause of the user's worries.

[0097] However, if there is a schema with a score of 0 or / and there are two or more schemas with the maximum score, the information processing device 10 does not infer a schema based on the results of the question processing about mood, but further executes question processing about schemas, narrows down two or three candidate schemas to one schema, and infers the schema that is the cause of the user's trouble. However, even in the question processing about schemas, there are cases where a schema cannot be inferred.

[0098] Furthermore, if there is a schema with a score of 0, it is impossible to determine whether the keyword corresponding to that schema cannot be detected because it is completely unsuitable for the user, or whether the keyword corresponding to that schema cannot be detected because the question is poorly worded even though the schema actually does suit the user. For this reason, if there is a schema with a score of 0, a process is executed to determine whether the user does not actually have any schemas with a score of 0 (hereinafter referred to as "question process for 0 points") before a process is executed to narrow down the schemas to one (hereinafter referred to as "narrowing question process").

[0099] If there are two or more schemas with a score of 0, a question process for the 0 points is executed for each of the schemas.

[0100] In the process of processing questions about schemas, first, how to answer the questions will be explained. In this example, the agent utters, "Please answer the questions with 'yes' or 'no'."

[0101] Next, if there is a schema with a score of 0, the query process for the schema with a score of 0 is executed, and then the refinement query process is executed. On the other hand, if there is no schema with a score of 0, the query process for the schema with a score of 0 is not executed, and the refinement query process is executed.

[0102] As a result of executing the schema estimation process, if there is a schema with a score of 0, the agent asks whether that schema is relevant. In other words, the information processing device 10 executes a question process for the schema with a score of 0. Fig. 8(A) shows an example of the question content for each schema.

[0103] In this embodiment, as shown in FIG. 8(A), if the schema with a score of 0 is high-achievement oriented, the agent asks the user by voice, "Is it a problem that you had ideals but things didn't turn out as you expected?"

[0104] If the schema with a score of 0 is an other-dependent evaluation, the agent asks the user by voice, "Is it a problem that you didn't receive good evaluations or reactions from others?"

[0105] Furthermore, if the schema with a score of 0 is anxiety about failure, the agent asks the user aloud, "Is it a problem that you feel anxious about what will happen if things don't go well?"

[0106] For these questions, if the user answers "yes," it is determined that the question corresponds to the schema, and the score for that schema is corrected from 0 to the maximum score. However, if the result of executing the schema estimation process is that all schemas have scores of 0, there is no maximum score, so the score for the schema of the question to which the user answered "yes" is corrected from 0 to 1.

[0107] On the other hand, if the user answers "No" to the above question, it is determined that the schema of the question does not apply, and the score for that schema is maintained. In other words, the score remains 0 points.

[0108] Furthermore, in the schema question processing, a narrowing down process is executed. When the information processing device 10 asks a question about schemas with 0 points, and if the schemas with 0 points also apply, the information processing device 10 (agent) executes a narrowing down process to narrow down the two or three schemas with the maximum points to one schema. Furthermore, if the schemas with 0 points do not apply, the information processing device 10 (agent) executes a narrowing down process to narrow down the two schemas other than those with 0 points to one schema.

[0109] Three questions are provided for narrowing down the schema. The first narrowing down question is a question for determining whether the user corresponds to a high achievement orientation or an other-dependent evaluation. In this embodiment, the agent asks aloud, "Do you strongly feel that the evaluation of others is an issue because you have high ideals?" If the user answers "yes," it is determined that the user corresponds to a high achievement orientation, and if the user answers "no," it is determined that the user corresponds to an other-dependent evaluation.

[0110] The second narrowing question is a question to determine whether the user is high achievement-oriented or has a fear of failure. In this embodiment, the agent asks aloud, "Do you strongly believe that having high ideals causes a fear of failure?" If the user answers "yes," it is determined that the user has high achievement-oriented behavior, and if the user answers "no," it is determined that the user has a fear of failure.

[0111] The third narrowing question is a question for determining whether the problem corresponds to other-dependent evaluation or anxiety about failure. In this embodiment, the agent asks aloud, "Do you strongly feel that anxiety about failure is a bigger problem than the evaluation of others?" If the user answers "yes," it is determined that the problem corresponds to other-dependent evaluation, and if the user answers "no," it is determined that the problem corresponds to anxiety about failure.

[0112] When narrowing down three schemata to one, three narrowing-down questions are asked, and the information processing device 10 narrows down the schemata to one based on the answers to all the questions.

[0113] For example, if the answers to both the first and second filtering questions are "yes," the schema causing the user's distress is inferred (i.e., narrowed) to be high-achievement orientation.

[0114] Also, for example, if the answer to the first narrowing-down question is "No" and the answer to the third narrowing-down question is "Yes," the schema that is the cause of the user's worries is estimated to be an other-dependent evaluation.

[0115] Furthermore, for example, if the answers to the second and third narrowing down questions are both "No," the schema causing the user's worries is estimated to be failure anxiety.

[0116] Furthermore, when narrowing down two schemas to one, one of the three narrowing questions is asked about the two schemas. As described above, the schema that is the cause of the user's trouble is inferred based on the user's answer to the narrowing question.

[0117] However, in the query process for schemas with a score of 0, even if the query process for 0 points is executed, if the score for all schemas is 0 points, the narrowing process is not executed. In other words, as mentioned above, it is not possible to infer the schema that is the cause of the user's trouble.

[0118] Once a schema has been estimated, the information processing device 10 executes an automatic thought question process. After a schema has been estimated, the automatic thought question process, detailed schema estimation process, and advice process including an explanation of the schema are executed for the estimated schema, regardless of the content of the consultation. As shown in FIGS. 9(A), 9(B), and 10(A), four automatic thought questions are prepared in advance for each schema. Similarly, as shown in FIG. 10(B), four automatic thought questions are also prepared in advance for when a schema cannot be estimated (when it has not been estimated).

[0119] Figure 9(A) shows questions about automatic thoughts regarding high achievement orientation. Question 1 is, "What came to mind at that time?" Question 2 is, "What did you think about yourself when you were unable to do something?" Question 3 is, "What came to mind when you were being strict with yourself and trying to accomplish something?" Question 4 is, "If a close friend of yours was in a similar situation or had the same thoughts, what would you say to them?"

[0120] Figure 9(B) shows questions about automatic thoughts regarding other-dependent evaluation. Question 1 is, "What did you think about yourself in that situation?" Question 2 is, "What did you think about the other person in that situation?" Question 3 is, "What comes to mind when other people do not accept you or are negative about you?" Question 4 is, "If a close friend were in a similar situation or thinking the same way, what would you say to them?"

[0121] Figure 10(A) shows questions about automatic thoughts regarding failure anxiety. Question 1 is, "How do you feel about yourself when things don't go well or you fail?" Question 2 is, "What do you think will happen to you if you fail at something?" Question 3 is, "If others knew your true feelings or weaknesses, what would come to mind?" Question 4 is, "If a close friend of yours was in a similar situation or thinking the same way, what would you say to them?"

[0122] Figure 10(B) shows questions about automatic thoughts when a schema has not been estimated. Question 1 is, "What came to your mind at that time?" Question 2 is, "What did you think about yourself?" Question 3 is, "If the situation continues, what do you think is the worst that could happen?" Question 4 is, "If a close friend of yours was in a similar situation or had the same thoughts, what would you say to them?"

[0123] When the voices of the answers to questions 1 to 4 are detected, a voice recognition process is executed to detect keywords from the text of the answers, i.e., the detected keywords are acquired.

[0124] After completing the automatic thought question process, the information processing device 10 executes a detailed schema estimation process. Figures 11(A), 11(B), and 12 show detailed schemas for each schema. In this example, the detailed schemas are detailed features for each schema, and eight detailed schemas are set for each schema.

[0125] As shown in Figure 11(A), when the schema is high-achievement oriented, the first detailed schema is "I will not compromise on anything," the second detailed schema is "I should not be satisfied with a mediocre life," the third detailed schema is "I should not let a day go by without doing anything," the fourth detailed schema is "I should not be satisfied with average results," the fifth detailed schema is "If I do not be strict with myself, I will become a second-rate person," the sixth detailed schema is "I must solve my problems quickly," the seventh detailed schema is "I should not make the same mistakes twice," and the eighth detailed schema is "I should not have any regrets in life."

[0126] As shown in Figure 11(B), when the schema is other-dependent evaluation, the first detailed schema is, "It is important how others evaluate me," the second detailed schema is, "I must feel lonely if others do not pay attention to me," the third detailed schema is, "If I am isolated from others, I will definitely be unhappy," the fourth detailed schema is, "If others rate me as unattractive, I will do the same," the fifth detailed schema is, "If others dislike me, I will not be happy," the sixth detailed schema is, "I will be miserable if I do not have anyone to support me," the seventh detailed schema is, "I must be a good person," and the eighth detailed schema is, "I need the approval of others to be happy."

[0127] As shown in Figure 12, when the schema is failure anxiety, the first detailed schema is, "If others knew about my weaknesses, I would be rejected," the second detailed schema is, "If others knew who I really am, they would despise me," the third detailed schema is, "It is shameful to show my weaknesses," the fourth detailed schema is, "If I make a big mistake once, I cannot recover," the fifth detailed schema is, "I cannot trust others because I might betray them," the sixth detailed schema is, "If I ask questions about things I don't understand, I will surely be ridiculed," the seventh detailed schema is, "I shouldn't try because I don't know what the outcome will be," and the eighth detailed schema is, "If I fail at work, I will be a failure in life."

[0128] In the examples shown in Figures 11(A), 11(B) and 12, identification information (ID) is assigned to identify each detailed schema individually. This is also true for the cases shown in Figures 13(A), 13(B) and 14.

[0129] As described above, after completing the automatic thought question process, the information processing device 10 executes the detailed schema estimation process. In the detailed schema estimation process, a table of related keywords written corresponding to each detailed schema is used. The fifth table shown in FIG. 13(A) is a table of related keywords written corresponding to each detailed schema of high achievement orientation. The sixth table shown in FIG. 13(B) is a table of related keywords written corresponding to each detailed schema of other-dependent evaluation. The seventh table shown in FIG. 14 is a table of related keywords written corresponding to each detailed schema of failure anxiety.

[0130] The related keywords are keywords contained in the answers of users who have corresponding detailed schemas, and are empirically determined based on the answers of many users obtained through experiments, where the user answers are the answers given by the users in the above-mentioned automatic thought question processing.

[0131] As shown in Figure 13(A), in the fifth table, keywords related to the first detailed schema of high achievement orientation, such as compromise, cutting corners, concessions, non-compromise, and compromise, are listed. Keywords related to the second detailed schema of high achievement orientation, such as comparison, disparity, complex, and concern, are listed. Keywords related to the third detailed schema of high achievement orientation, such as impatience, impatience, feeling of irritation, and busy, are listed. Keywords related to the fourth detailed schema of high achievement orientation, such as losing, defeat, lack of effort, and grades, are listed. Keywords related to the fifth detailed schema of high achievement orientation, such as good person, strict with oneself, and must work hard, are listed. Keywords related to the sixth detailed schema of high achievement orientation, such as impatience, irritation, and hurry, are listed. Keywords related to the seventh detailed schema of high achievement orientation, such as failure, mistake, and many times, are listed. Keywords related to the eighth detailed schema of high achievement orientation, such as regret, should have done, and should have done, are listed.

[0132] As shown in Figure 13(B), in the sixth table, keywords related to the first detailed schema of other-dependent evaluation, such as "I want to be recognized," "I want to be understood," and "My career," are listed. Keywords related to the second detailed schema of other-dependent evaluation, such as "I'm alone," "I feel lonely," and "I feel empty," are listed. Keywords related to the third detailed schema of other-dependent evaluation, such as "I'm isolated," "I can't join in the conversation," and "I can't speak," are listed. Keywords related to the fourth detailed schema of other-dependent evaluation, such as "I'm upset," "I'm shocked," and "I'm ridiculed," are listed. Keywords related to the fifth detailed schema of other-dependent evaluation, such as "I'm disliked" and "I'm scolded," are listed. Keywords related to the sixth detailed schema of other-dependent evaluation, such as "I'm left out" and "I can't get help," are listed. Keywords related to the seventh detailed schema of other-dependent evaluation, such as "I sacrifice myself," "I'm a good person," and "I'm wonderful person," are listed. Keywords related to the eighth detailed schema of other-dependent evaluation, such as "I want to be recognized" and "I want to be understood," are listed.

[0133] As shown in Figure 14, in the seventh table, keywords related to the first detailed schema of failure anxiety, such as whining, true feelings, genuine intentions, rejection, and being disliked, are listed. Keywords related to the second detailed schema of failure anxiety, such as whining, true feelings, genuine intentions, and contempt, are listed. Keywords related to the third detailed schema of failure anxiety, such as whining, true feelings, genuine intentions, and embarrassment, are listed. Keywords related to the fourth detailed schema of failure anxiety, such as failure, mistake, and self-loathing, are listed. Keywords related to the fifth detailed schema of failure anxiety, such as distrust of people, not being able to trust, and not being able to be trusted, are listed. Keywords related to the sixth detailed schema of failure anxiety, such as not being able to confirm, not being able to ask, and not asking for help, are listed. Keywords related to the seventh detailed schema of failure anxiety, such as anxiety, result, and future, are listed. Keywords related to the eighth detailed schema of failure anxiety, such as work, failure, and loser, are listed.

[0134] In this embodiment, if the schema that is the cause of the user's worries has already been estimated, the system checks whether the user fits each of the eight detailed schemas of the estimated schema. Whether the user fits a detailed schema is determined by determining whether each of all detected keywords obtained from the answers in the automatic thought question processing is a related keyword. It is determined that the user fits a detailed schema in which at least one detected keyword is a related keyword. In other words, a detailed schema that is determined to fit is estimated as the user's detailed schema.

[0135] If the schema that is the cause of the user's distress has not yet been identified, the system checks whether the user fits all of the detailed schemas, i.e., the eight detailed schemas of the high-achievement-orientation schema, the eight detailed schemas of the other-dependent evaluation schema, and the eight detailed schemas of the failure anxiety schema. The method for checking whether the user fits each detailed schema is the same as above.

[0136] If the schema has not been estimated, it is checked whether the user fits all detailed schemas, and the schema to which the user fits the most detailed schemas is estimated as the schema that is causing the user's trouble.

[0137] However, if there is no detailed schema that applies to the user, the information processing device 10 executes exception processing. In this embodiment, in the exception processing, a schema of the user is randomly selected, and further, a detailed schema that applies to the user is randomly selected from the eight detailed schemas of the randomly selected schemas.

[0138] After completing the detailed schema estimation process, the information processing device 10 executes advice processing. In the advice processing, the information processing device 10, i.e., the agent, explains the estimated schema to the user, explains the estimated detailed schema to the user, and further advises the user about the estimated schema.

[0139] FIG. 15(A) shows an example of a schema and an explanation about the schema, and FIG. 15(B) shows an example of a schema and advice about the schema.

[0140] As shown in Figure 15(A), if the estimated schema is high-achievement oriented, the agent will explain, "Your thinking habits show a tendency to have high ideals and be strict with yourself."

[0141] If the inferred schema is an other-dependent evaluation, the agent will explain, "Your way of thinking shows that you tend to care about what others think of you."

[0142] Furthermore, if the inferred schema is failure anxiety, the agent will explain, "Your way of thinking shows that you tend to feel strong anxiety about the possibility of failure."

[0143] As described above, the estimated schema has been described, and next, the estimated detailed schema will be described. In this example, the detailed schemas that have been determined to be applicable will be described. The detailed schemas are as described using FIG. 13(A), FIG. 13(B), and FIG. 14. In this example, each of all the detailed schemas that have been determined to be applicable will be described. However, several of the detailed schemas that have been determined to be applicable will be selected, and each of the selected several detailed schemas will be described. For example, several of the detailed schemas that have been determined to be applicable will be selected at random.

[0144] For example, if a user's schema is estimated to be high achievement-oriented and the user is determined to fit the first detailed schema of high achievement-oriented, the explanation will be, "There is a tendency to not compromise on anything." If the user's schema is estimated to be other-dependent evaluation and the user is determined to fit the third detailed schema of other-dependent evaluation, the explanation will be, "There is a tendency to become unhappy whenever isolated from others." If the user's schema is estimated to be failure anxiety and the user is determined to fit the seventh detailed schema of failure anxiety, the explanation will be, "There is a tendency to not try things because you don't know what the outcome will be." The same applies when explaining other detailed schemas.

[0145] As mentioned above, when the detailed schema is explained, the agent gives advice about the inferred schema. As shown in Figure 15(B), if the inferred schema is high-achievement oriented, the agent gives advice such as, "For example, isn't there something that doesn't go well for everyone?" and "Also, instead of trying to do too many things at once, why not try doing one thing at a time?"

[0146] Furthermore, if the estimated schema is an other-dependent evaluation, the agent will give advice such as, "For example, perhaps you have some good points that others don't have," or "It's okay to have people with whom you don't get along."

[0147] Furthermore, if the estimated schema is anxiety about failure, the agent will advise, "For example, you may have had times when things didn't go well, but you may have also had successes in the past," and "Also, there is no one who doesn't fail."

[0148] Fig. 16 shows an example of a memory map 300 of the RAM 12b of the computer 12 shown in Fig. 1. As shown in Fig. 16, the RAM 12b includes a program storage area 302 and a data storage area 304.

[0149] The program memory area 302 stores control programs for the information processing device 10, and the control programs include an overall control program 302a, a speech program 302b, a voice detection program 302c, a voice recognition program 302d, a keyword detection program 302e, a schema estimation program 302f, a schema question program 302g, a detailed schema estimation program 302h, and an advice execution program 302i.

[0150] The overall control program 302a is a program for executing the overall processing for the problem consultation according to a scenario. The speech program 302b is a program for executing speech processing, i.e., processing for outputting voice data, according to a scenario.

[0151] The voice detection program 302c is a program for detecting voice uttered by a user of the information processing device 10, i.e., a person seeking advice on a problem. The voice recognition program 302d is a program for recognizing the user's voice detected according to the voice detection program 302c and generating a text sentence corresponding to the detected voice.

[0152] The keyword detection program 302e is a program for decomposing the text generated by the speech recognition program 302d into words by morphological analysis and detecting keywords from the decomposed words, i.e., for obtaining detected keywords.

[0153] The schema estimation program 302f is a program for estimating the schema that is the cause of the user's trouble, based on the keywords detected by the keyword detection program 302e. The method for estimating the schema is as described above.

[0154] The schema questioning program 302g is a program for asking questions about schemas and estimating schemas when a schema cannot be estimated by the schema estimation program 302f. The schema questioning program 302g includes a program for processing questions about schemas with zero points and for processing narrowing down questions.

[0155] The detailed schema estimation program 302h is a program for estimating a detailed schema to which the user applies for one estimated schema when a schema has been estimated, and for estimating a detailed schema to which the user applies from all detailed schemas when a schema has not been estimated. The detailed schema estimation program 302h is also a program for estimating an unestimated schema based on the result of estimating a detailed schema to which the user applies from all detailed schemas. However, if the user does not apply to any detailed schema, the detailed schema estimation program 302h executes exception processing to randomly select a schema, and then randomly selects a detailed schema for the selected schema.

[0156] The advice execution program 302i is a program for explaining the estimated schema and the estimated detailed schema, and for executing advice regarding the estimated schema.

[0157] Although not shown, the program storage area 302 also stores other programs for executing information processing (in this embodiment, overall processing for problem consultation).

[0158] The data storage area 304 stores speech data 304a, table data 304b, detected voice data 304c, detected keyword data 304d, score data 304e, estimation result data 304f, and the like.

[0159] The speech data 304a is pre-generated voice data (e.g., synthesized voice data) corresponding to each utterance content of the agent. The table data 304b is data for the above-mentioned first, second, third, fourth, fifth, sixth, and seventh tables.

[0160] The detected voice data 304c is data on the user's voice detected by the voice detection program 302c. The detected keyword data 304d is data on detected keywords obtained from the user's answers to the question processing about the situation and the question processing about the trigger, and the detected keywords obtained from the user's answers to the question processing about the automatic thought.

[0161] Score data 304e is data on the number of detected keywords obtained from the user's answers in each of the question processing about the situation, the question processing about the trigger, the question processing about the 5Ws, and the question processing about the mood that were determined to be related keywords for each schema, i.e., the score.

[0162] The inference result data 304f is data indicating the schema that is the cause of the user's trouble as inferred by the schema inference program 302f and the detailed schema that is inferred by the detailed schema inference program 302h. However, before the schema is inferred, or when the schema inferring program 302f and the schema questioning program 302g cannot infer the schema that is the cause of the user's trouble, information indicating "not inferred" is entered as the inference result data 304e.

[0163] Although not shown, the data storage area 304 stores other data necessary for information processing (processing about advice on worries in this embodiment), and also includes a timer (counter) and flags necessary for information processing.

[0164] Fig. 17 is a flow diagram of information processing (overall processing for problem consultation) by the CPU 12a of the information processing device 10 shown in Fig. 1. When the user instructs the start of information processing, in step S1, a question processing about the situation (see Fig. 18) is executed, in step S3, a question processing about the trigger (see Fig. 19) is executed, in step S5, a question processing about the 5Ws (see Fig. 20) is executed, in step S7, a question processing about the mood (see Fig. 21) is executed, and in step S9, a schema estimation processing (see Fig. 22) is executed.

[0165] In the next step S11, it is determined whether or not a schema has been inferred. If the answer is "YES" in step S11, that is, if a schema has been inferred, the process proceeds to step S15. On the other hand, if the answer is "NO" in step S11, that is, if a schema has not been inferred, the process executes a query process about the schema (see Figures 23 to 25) in step S13, and then proceeds to step S15.

[0166] In step S15, automatic thought question processing (see FIG. 26) is executed. In the next step S17, detailed schema estimation processing (see FIGS. 27 and 28) is executed. Then, in step S19, advice processing (see FIG. 29) is executed, and information processing is terminated.

[0167] Each subroutine shown in FIG. 17 will be explained below, but duplicate explanations of steps with the same content will be omitted.

[0168] Figure 18 is a flow diagram of the question processing about the situation in step S1 shown in Figure 17. As shown in Figure 18, when the CPU 12a starts the question processing about the situation, it asks a question about the situation in step S31. In this embodiment, the CPU 12a reads out voice data of "Is there anything that has been bothering you lately?" from the speech data 304a and outputs it to the output device 16.

[0169] In the next step S33, it is determined whether or not there is an answer. Here, the CPU 12a determines whether or not the user's voice is detected. If "NO" in step S33, that is, if there is no answer, the process returns to step S33.

[0170] If the answer is "YES" in step S33, that is, if there is an answer, keywords are detected in step S35 and stored in step S37. In step S35, the CPU 12a performs speech recognition on the detected speech and morphological analysis on the recognized text to detect keywords. In this embodiment, in the case of a work-related problem consultation, an answer such as "I've made a lot of mistakes at work recently and I hate myself for it" is obtained, and the keywords "recently," "workplace," "many," "mistakes," and "self-hatred" are detected from this answer. In the case of a school-related problem consultation, an answer such as "I can't keep up with the classes, and I feel like I'm going to have to repeat the year if this continues" is obtained, and the keywords "class" and "repeated the year" are detected from this answer. In step S37, the CPU 12a stores detected keyword data 304d for the detected keywords in the RAM 12b.

[0171] Then, in step S39, it is determined whether the detected keyword is an expected keyword. Here, the CPU 12a refers to the first table of the table data 304b and determines whether the detected keyword corresponding to the detected keyword data 304d is included in the expected keywords of the first table.

[0172] If the answer is "YES" in step S39, that is, if the detected keyword is an expected keyword, the content is confirmed in step S41, the question processing about the situation is terminated, and the process returns to the information processing shown in Fig. 17. As described above, in step S41, the CPU 12a makes an utterance to confirm whether the worry category corresponding to the expected keyword is correct as a classification of the content (i.e., worry) answered by the user.

[0173] For example, in the above example, in the case of a work-related problem consultation, the expected keyword among the detected keywords is "mistake," and the problem category is "I failed at something." Therefore, the CPU 12a reads out the voice data "I see, so you failed at something" from the speech data 304a and outputs it to the output device 16.

[0174] In the above example, in the case of a consultation about worries about school, the expected keywords among the detected keywords are “repeated a year,” and the worry category is “anxiety about the future.” Therefore, the CPU 12a reads out the voice data “I see, so you are worried about the future” from the speech data 304a and outputs it to the output device 16.

[0175] On the other hand, if the result of step S39 is "NO," that is, if the detected keyword is not an expected keyword, a reply is made in step S43, and the process returns to information processing. In step S43, the CPU 12a reads out the voice data of "That's right" from the speech data 304a and outputs it to the output device 16.

[0176] Fig. 19 is a flow diagram of the question processing about the trigger in step S3 shown in Fig. 17. As shown in Fig. 19, when the CPU 12a starts the question processing about the trigger, it asks the first question about the trigger in step S61. In this embodiment, the CPU 12a reads out voice data of "When did this state begin?" from the speech data 304a and outputs it to the output device 16.

[0177] In the following step S63, it is determined whether or not there is an answer. If "NO" in step S63, the process returns to step S63. On the other hand, if "YES" in step S63, a keyword is detected in step S65, and the keyword is stored in data storage area 304 in step S67.

[0178] In this example, in response to the first question, if the concern is work-related, an answer such as "for the past month or so" will be obtained, and if the concern is school-related, an answer such as "since April began" will be obtained.

[0179] In the next step S69, the CPU 12a responds by reading out the voice data of "I understand" from the speech data 304a and outputting it to the output device 16.

[0180] Then, in step S71, it is determined whether or not the two questions have been completed. If the answer is "YES" in step S71, that is, if the two questions have been completed, the question processing regarding the trigger is completed, and the process returns to the overall processing shown in FIG.

[0181] On the other hand, if the answer is "NO" in step S71, that is, if the two (second) questions have not been completed, the CPU 12a asks a second question about the trigger in step S73, and then returns to step S63. In this embodiment, in step S73, the CPU 12a reads out the voice data "Please tell me about the trigger" from the speech data 304a and outputs it to the output device 16.

[0182] Then, keywords are detected and stored based on the user's answers. In this embodiment, in response to the second question, if the user is seeking advice on work-related worries, an answer such as "I think there was a time when my boss got really angry with me" is obtained, and in response to the second question, if the user is seeking advice on school-related worries, an answer such as "It started when I started taking online classes at home" is obtained.

[0183] Therefore, from the answers to the two questions, the keywords "one month," "boss," and "being scolded" are detected in the case of work-related worries, while the keywords "April," "class," and "home" are detected in the case of school-related worries.

[0184] Fig. 20 is a flow diagram of the question processing about 5W in step S5 shown in Fig. 17. As shown in Fig. 20, when the CPU 12a starts the question processing about 5W, in step S91, it refers to the stored keyword, i.e., the detected keyword data 304d, and in step S93, it determines the perspective from which to ask a question about 5W.

[0185] In the case of work-related worries, since the detected keywords are stored for the perspectives of "when," "where," and "with whom" in the processing up to step S3, it is decided to ask questions from the perspectives of "what you were doing" and "why." Specifically, the questions are asked in order of "what you were doing" and "why it happened."

[0186] In the case of a school-related problem, since the detected keywords are stored for the aspects of "when," "where," and "what you were doing" in the processing up to step S3, it is decided to ask questions from the aspects of "with whom" and "why." Specifically, the questions are "with whom" and "why it happened," in that order.

[0187] 20 , in the next step S95, a first question is asked regarding the viewpoint determined in step S93. In the above example, in the case of a work-related problem consultation, the CPU 12a reads out the voice data of “What were you doing?” from the utterance data 304a and outputs it to the output device 16. In the above example, in the case of a school-related problem consultation, the CPU 12a reads out the voice data of “Who were you with?” from the utterance data 304a and outputs it to the output device 16.

[0188] In the next step S97, it is determined whether or not there is an answer. If "NO" in step S97, the process returns to step S97. On the other hand, if "YES" in step S97, a keyword is detected in step S99, and the keyword is stored in data storage area 304 in step S101.

[0189] Then, in step S103, it is determined whether the detected keyword is an expected keyword. Here, the CPU 12a refers to the second table included in the table data 304b and determines whether the detected keyword is listed in the second table as an expected keyword.

[0190] It should be noted that the "why" perspective is not included in the second table, so the result of step S103 is "NO."

[0191] If the answer is "YES" in step S103, that is, if the detected keyword is an expected keyword, the process confirms the content in step S105 and proceeds to step S109. In step S105, the CPU 12a confirms the category to which the detected keyword belongs. Therefore, for example, if "work" or "task" is obtained as the detected keyword, the agent utters "You were at work, weren't you?" using the name of a category listed in the second table corresponding to these detected keywords. Also, for example, if "family," "parents," "mother," or "father" is obtained as the detected keyword, the agent utters "You were with your family, weren't you?" using the name of a category listed in the second table corresponding to these detected keywords. However, voice data regarding the content of the utterance is included in the utterance data 304a.

[0192] On the other hand, if the result in step S103 is "NO," that is, if the detected keyword is not an expected keyword, a reply is made in step S107, and the process proceeds to step S109. In step S107, the CPU 12a reads out the voice data of "That's right" from the speech data 304a and outputs it to the output device 16.

[0193] In step S109, it is determined whether all of the 5W viewpoints have been stored. That is, it is determined whether questions have been asked about all of the viewpoints determined in step S93. If the answer is "YES" in step S109, that is, if all of the 5W viewpoints have been stored, the question processing for 5W ends and the process returns to the overall processing shown in FIG.

[0194] On the other hand, if step S109 is "NO," that is, if there is one or more unstored perspectives among the 5W perspectives, that is, if there are any unfilled perspectives among the 5W perspectives, then in step S111, the next question about the 5W is asked, and the process returns to step S97. That is, in step S111, a question is asked about the next perspective among the perspectives determined in step S93.

[0195] Fig. 21 is a flow diagram of the mood question processing in step S7 shown in Fig. 17. As shown in Fig. 21, when the mood question processing starts, the CPU 12a asks a question about the mood at that time in step S131. In this embodiment, the CPU 12a reads out voice data of "Please tell me how you are feeling at that time" from the utterance data 304a and outputs it to the output device 16.

[0196] In the following step S133, it is determined whether or not there is an answer. If "NO" in step S133, the process returns to step S133. On the other hand, if "YES" in step S133, a keyword is detected in step S135, and the keyword is stored in data storage area 304 in step S137.

[0197] In this example, in response to questions about mood, if the person is seeking advice about work-related worries, they may receive a response such as "I felt sad and self-loathing," and if the person is seeking advice about school-related worries, they may receive a response such as "I thought I might have to repeat the year if I continued like this."

[0198] Then, in step S139, it is determined whether the detected keyword is a keyword about mood. Here, the CPU 12a refers to the third table included in the table data 304b and determines whether the detected keyword is listed in the third table as an expected keyword.

[0199] If the answer is "YES" in step S139, that is, if the detected keyword is a keyword about mood, the content is confirmed in step S141, the mood question processing is terminated, and the process returns to the overall processing shown in FIG. 17. In step S141, the CPU 12a confirms the category to which the detected keyword belongs. Therefore, for example, if the keywords "sad" and "self-loathing" are detected, the agent will use the name of the category listed in the third table corresponding to these keywords and utter, "I see, your mood in that situation can be summarized as 'sad', 'self-loathing', etc." However, voice data about the content of the utterance is included in the utterance data 304a.

[0200] On the other hand, if the answer is "NO" in step S139, that is, if the detected keyword is not a keyword about mood, the agent points this out and asks the question again in step S143, and then returns to step S133. In step S143, the agent utters, "I'm sorry if I missed it, but what you just said was more of a thought than a mood. Moods can often be expressed in one word, such as 'happy' or 'sad'. So, how do you feel about the thought you just said?" However, the voice data about the content of the utterance is included in the utterance data 304a.

[0201] Fig. 22 is a flow diagram of the schema estimation process in step S9 shown in Fig. 17. As shown in Fig. 22, when the CPU 12a starts the schema estimation process, in step S161, it refers to the stored keywords, i.e., the detected keywords obtained from the user's answers to the question process about the situation and the question process about the trigger, and in step S163 it refers to the fourth table of the table data 304b to calculate the number (score) of keywords related to high achievement orientation, other-dependent evaluation, and fear of failure (i.e., related keywords), and in step S165 it stores the score data 304e. In step S163, the CPU 12a determines whether each of the detected keywords referred to in step S161 matches the related keywords of each schema, and calculates the score of match for each schema.

[0202] In the next step S167, it is determined whether a schema can be inferred. As described above, if there are no schemas with a score of 0 and there is one schema with the maximum score, the schema with the maximum score can be inferred as the schema that is causing the user's trouble.

[0203] If step S167 is "YES", that is, if the schema can be inferred, then in step S169 the inferred schema is stored, the schema inference process is terminated, and the process returns to the overall process shown in Fig. 17. On the other hand, if step S167 is "NO", that is, if the schema cannot be inferred, then in step S171 the fact that the schema cannot be inferred is stored, and the process returns to the overall process.

[0204] Fig. 23 is a flow diagram of the query processing about the schema in step S13 shown in Fig. 17. As shown in Fig. 23, when the CPU 12a starts the query processing about the schema, in step S181 it explains how to answer the question. Here, the CPU 12a reads out voice data "Please answer the question with 'yes' or 'no'" from the utterance data 304a and outputs it to the output device 16.

[0205] In the next step S183, it is determined whether or not there is a schema with a score of 0. If the answer is "NO" in step S183, that is, if there is no schema with a score of 0, the process proceeds to step S187. On the other hand, if the answer is "YES" in step S183, that is, if there is a schema with a score of 0, in step S185, a question process (see FIG. 24) is executed to ask questions about the schema with a score of 0, which will be described later, and the process proceeds to step S187.

[0206] In step S187, it is determined whether there are multiple schemas with the highest score. If the answer is "NO" in step S187, that is, if there is one schema with the highest score, the process proceeds to step S191. On the other hand, if the answer is "YES" in step S187, that is, if there are multiple schemas with the highest score, the process executes the narrowing down question process (see FIG. 25) described later, and proceeds to step S191.

[0207] In step S191, as described above, a schema is estimated, and in step S193, the estimated schema is stored, the query process regarding the schema is terminated, and the process returns to the overall process shown in FIG.

[0208] Fig. 24 is a flow diagram of the query processing for the schema with 0 points in step S185 shown in Fig. 23. As shown in Fig. 24, when the CPU 12a starts the query processing for the schema with 0 points, it asks a question about the schema with 0 points in step S201.

[0209] In step S201, as described above, if the schema with a score of 0 is high achievement orientation, the agent asks aloud, "Is it a problem that you had ideals, but things didn't turn out as you expected?" If the schema with a score of 0 is other-dependent evaluation, the agent asks aloud, "Is it a problem that you didn't get good evaluations or reactions from other people?" If the schema with a score of 0 is failure anxiety, the agent asks aloud, "Is it a problem that you feel anxious about what will happen if things don't go well?" However, the voice data on the utterance content is included in the utterance data 304a. The same applies to step S215, which will be described later.

[0210] However, if there are multiple schemata with a score of 0, in step S201, a question is asked about the first schema with a score of 0 determined by a predetermined method (for example, randomly). In step S215, which will be described later, a question is asked about the second or third schema with a score of 0 determined by a predetermined method.

[0211] 24, in the next step S203, it is determined whether or not there is an answer. If "NO" in step S203, the process returns to step S203. On the other hand, if "YES" in step S203, it is determined in step S205 whether or not the answer is "YES".

[0212] If step S205 returns "NO," that is, if the answer is "NO," the process proceeds to step S209. On the other hand, if step S205 returns "YES," that is, if the answer is "YES," the score for the schema in question is corrected from 0 to the maximum score in step S207, as described above, and the process proceeds to step S209.

[0213] In step S209, the answer is stored. In step S209, the CPU 12a performs voice recognition on the detected voice, detects the voice-recognized answer of "yes" or "no", and stores it in the data storage area 304 (not shown). The same applies to the case where an answer is stored hereinafter.

[0214] In the next step S211, a reply is made. Here, the CPU 12a reads out the voice data of "I understand" from the speech data 304a and outputs it to the output device 16. The same applies to the case where the reply is "I understand."

[0215] In the next step S213, it is determined whether there are any schemas with a score of 0 that have not been queried. If the answer is "YES" in step S213, that is, if there are any schemas with a score of 0 that have not been queried, then in step S215, as described above, a question is asked about the second or third schema with a score of 0 that have not been queried, and the process returns to step S203. On the other hand, if the answer is "NO" in step S213, that is, if there are no schemas with a score of 0 that have not been queried, the process of querying the schemas with a score of 0 is terminated, and the process returns to the process of querying the schemas shown in FIG. 23.

[0216] Fig. 25 is a flow diagram of the narrowing-down question processing in step S189 shown in Fig. 23. Here, the agent asks three questions in order to narrow down the schemata.

[0217] As described above, in the first narrowing-down question, the agent asks by voice, "Do you strongly feel that what others think of you is more of a problem than having high ideals?" In the second narrowing-down question, the agent asks by voice, "Do you strongly feel that the fear of failure is a problem because of having high ideals?" In the third narrowing-down question, the agent asks by voice, "Do you strongly feel that the fear of failure is a problem more than what others think of you?" However, the voice data about the utterance content is included in the utterance data 304a. Furthermore, the order may be determined randomly each time the narrowing-down question process is executed.

[0218] As shown in Fig. 25, in step S221, the first narrowing question is asked. In the following step S223, it is determined whether or not there is an answer. If "NO" in step S223, the process returns to step S223. On the other hand, if "YES" in step S223, the answer is stored in step S225, and the reply "I understand" is made in step S227.

[0219] Next, in step S229, it is determined whether all narrowing questions have been asked. If the answer is "NO" in step S229, that is, if there are any narrowing questions that have not been asked, in step S231, the next narrowing question (i.e., the second or third narrowing question) is asked, and the process returns to step S223.

[0220] On the other hand, if "YES" in step S229, that is, if all narrowing down questions have been asked, the narrowing down question processing is terminated, and the process returns to the query processing for the schema shown in FIG.

[0221] Figure 26 is a flow diagram of the automatic thought question processing in step S15 shown in Figure 17. As shown in Figure 26, when the CPU 12a starts the automatic thought question processing, it selects a question group in step S241. However, the question group is a question group about automatic thoughts, and as described above, it is determined in advance for each of the three schemata and the unestimated schema. As described above, each question group is made up of four questions. The specific content of the questions is as described above.

[0222] In the following step S243, the first question of the selected question group is asked. In the next step S245, it is determined whether or not an answer is available. If "NO" in step S245, the process returns to step S245. On the other hand, if "YES" in step S245, a keyword is detected in step S247, and the keyword is stored in data storage area 304 in step S249.

[0223] In the next step S251, the user replies "I understand," and in step S253, it is determined whether all questions (i.e., the fourth question) have been answered. If the answer is "NO" in step S253, that is, if there are questions that have not been answered, in step S255, the next question in the selected group of questions is asked, and the process returns to step S245.

[0224] On the other hand, if "YES" in step S253, that is, if all questions have been asked, the question processing regarding automatic thoughts is ended, and the process returns to the overall processing shown in FIG.

[0225] Figures 27 and 28 are flow diagrams of the detailed schema estimation process in step S17 shown in Figure 17. As shown in Figure 27, when the CPU 12a starts the detailed schema estimation process, in step S261, it refers to the stored keywords, that is, the detected keywords obtained from the user's answers in the automatic thought question process.

[0226] In the following step S263, it is determined whether or not the schema has been estimated. If "NO" in step S263, that is, if the schema has not been estimated, the process proceeds to step S279 shown in FIG.

[0227] On the other hand, if the answer is "YES" in step S263, that is, if the schema has been estimated, the variable n is set to 1 (n=1) in step S265. The variable n is a variable for identifying the first to eighth detailed schemata of the estimated schema.

[0228] In the next step S267, it is checked whether the user matches the nth detailed schema of the estimated schema. In this embodiment, for each schema, the detailed schema with the smallest ID number is the first detailed schema, and detailed schemas are identified in order as the ID number increases. In the next step S269, the estimation result, i.e., whether the user matches or does not match the nth detailed schema, is stored.

[0229] Then, in step S271, it is determined whether the variable n is 8. That is, the CPU 12a determines whether or not all of the eight detailed schemata for the estimated schema have been checked to see if they apply.

[0230] If step S271 is "NO", that is, if variable n is less than 8, then in step S273, variable n is incremented by 1 (n=n+1), and the process returns to step S267. On the other hand, if step S271 is "YES", that is, if variable n is 8, the detailed schema estimation process is terminated, and the process returns to the overall process shown in FIG.

[0231] 28, in step S279, a variable m is set to 1 (m=1). The variable m is a variable for identifying the first to 24th detailed schemata of the three schemata.

[0232] In the next step S281, it is checked whether the m-th detailed schema is applicable. In this embodiment, the number set in the variable m matches the ID number assigned to the detailed schema. In the next step S283, the estimation result, i.e., whether the m-th detailed schema is applicable or not, is stored.

[0233] Then, in step S285, it is determined whether the variable m is 24. In other words, the CPU 12a determines whether all of the 24 detailed schemata for the three schemata have been checked to see if they are applicable.

[0234] If the result in step S285 is "NO", that is, if the variable m is less than 24, then in step S287, the variable m is incremented by 1 (m=m+1), and the process returns to step S281. On the other hand, if the result in step S285 is "YES", that is, if the variable m is 24, then in step S289, it is determined whether or not there is an applicable detailed schema. That is, the CPU 12a determines whether or not there is one or more detailed schemas that have been determined to apply among the 24 detailed schemas.

[0235] If the answer is "NO" in step S289, that is, if there is no applicable detailed schema, in step S291, exception processing is executed as described above, the detailed schema estimation processing is terminated, and the process returns to the overall processing shown in Figure 17.

[0236] On the other hand, if "YES" in step S289, in step S293, a schema is estimated as described above, and in step S295, the estimated schema is stored, the detailed schema estimation process is terminated, and the process returns to the overall process shown in Figure 17.

[0237] Fig. 29 is a flow diagram of the advice processing in step S19 shown in Fig. 17. As shown in Fig. 29, when the advice processing starts, the CPU 12a explains the estimated schema in step S301. Here, the CPU 12a reads out voice data explaining the estimated schema from the utterance data 304a and outputs it to the output device 16.

[0238] In the next step S303, detailed schemata applicable to the user, i.e., estimated detailed schemata, are explained. Here, the CPU 12a reads out voice data explaining the detailed schemata applicable to the user from the speech data 304a, and outputs the voice data to the output device 16. However, all detailed schemata applicable to the user may be explained, or several (2 to 3) detailed schemata randomly selected from the applicable detailed schemata may be explained.

[0239] Then, in step S305, advice is given about the estimated schema, the advice processing is terminated, and the process returns to the overall processing shown in Fig. 17. In step S305, the CPU 12a reads out speech data of advice about the estimated schema from the speech data 304a, and outputs it to the output device 16.

[0240] According to this embodiment, it is possible to infer the schema that is the cause of the user's worries through dialogue with the user and to provide advice on the inferred schema, thus providing a novel advice device for worries.

[0241] In this embodiment, the voice data to be spoken by the agent is generated in advance, but this is not limited to this. In another example, data of text (hereinafter referred to as "spoken text") excluding the portion corresponding to the name of each category from the content to be spoken by the agent is generated in advance, and when a detected keyword acquired from the user's answer matches an expected keyword, the text of the name of the category corresponding to the expected keyword is inserted into the spoken text, and the agent reads out the spoken text with the category text inserted.

[0242] Furthermore, in this embodiment, the computer executes the speech recognition function and the schema estimation function, but the speech recognition function may be executed by another computer communicatively connected to the computer. In this case, another problem consultation system is configured. In this case, when the computer detects the user's speech, it transmits speech data corresponding to the detected speech to the other computer and receives the speech recognition processing result (text sentence) from the other computer. As an example, the other computer is a PC or a server having a speech recognition function and installed on a network such as the Internet.

[0243] The specific numerical values and control methods shown in the above embodiments are merely examples and are not limited to these, but can be changed as appropriate depending on the actual product and the environment in which the product is applied. [Explanation of symbols]

[0244] 10... Advice device 12...Computer 12a...CPU 12b...RAM 12c...Communication device 14. Mike 16...Output device

Claims

1. an output unit that outputs each of the predetermined plurality of first questions by voice; a voice detection unit that detects the voice of a user; a keyword detection unit that detects a keyword for each of the predetermined plurality of first questions based on the user's voice detected by the voice detection unit; an estimation unit that estimates a schema, which is a habit of thinking that is the cause of the user's worries and is classified into three types, i.e., high achievement orientation, other-dependent evaluation, and failure anxiety, based on the detected keywords that are keywords detected by the keyword detection unit; a related keyword storage unit that stores related keywords corresponding to each of the three types; a score adding unit that adds a score to a type corresponding to the related keyword when the detected keyword matches the related keyword; the estimation unit estimates the schema based on the addition result of the score addition unit; The output unit outputs advice about the schema estimated by the estimation unit by voice.

2. 2. The problem consultation device according to claim 1, wherein the predetermined plurality of first questions include 5W1H questions about the cause of the problem.

3. 3. The problem consultation device according to claim 1, wherein the estimation unit estimates, as the schema, the type with the highest score among the three types for which the scores have been added by the score adding unit.

4. the output unit outputs, by voice, a second question for narrowing down the types to one when there are a plurality of types with the highest score; the voice detection unit detects a voice of the user's answer to the second question; The problem consultation device according to claim 3 , wherein the estimation unit estimates, as the schema, one type narrowed down based on the user's answer to the second question.

5. the output unit outputs a third question by voice, when there is a category with no score among the categories to which the score has been added by the score adding unit, the third question being a question as to whether the person corresponds to the category with no score; the voice detection unit detects a voice of the user's answer to the third question; 5. The problem consultation device according to claim 4, wherein when the user's answer to the third question indicates that the answer corresponds to the type for which no score is given, the score for the type for which no score is given is corrected to the maximum score.

6. an output device that outputs each of a predetermined plurality of questions by voice; a voice detection unit that detects the voice of a user; a keyword detection unit that detects a keyword for each of the predetermined plurality of questions based on the user's voice detected by the voice detection unit; an estimation unit that estimates a schema, which is a habit of thinking that is the cause of the user's worries and is classified into three types, i.e., high achievement orientation, other-dependent evaluation, and failure anxiety, based on the detected keywords that are keywords detected by the keyword detection unit; a related keyword storage unit that stores related keywords corresponding to each of the three types; a score adding unit that adds a score to a type corresponding to the related keyword when the detected keyword matches the related keyword; the estimation unit estimates the schema based on the addition result of the score addition unit; The output device outputs advice about the schema estimated by the estimation unit by voice.

7. A control program executed by a computer of a problem consultation device having a related keyword memory unit that stores related keywords corresponding to each of three types of thinking habits classified as high achievement orientation, other-dependent evaluation, and failure anxiety, comprising: a processor of the computer; a voice output step of outputting each of a predetermined plurality of questions by voice to an output means; a voice detection step of detecting a user's voice; a keyword detection step of detecting a keyword based on the user's voice detected in the voice detection step for each of the predetermined plurality of questions; an estimation step of estimating a schema, which is a cause of the user's worries and is a habit of thinking classified into the three types, based on the detected keywords, which are keywords detected in the keyword detection step; and a score adding step of adding a score to a type corresponding to the related keyword when the detected keyword matches the related keyword; the estimation step estimates the schema based on the addition result of the score addition step; The speech output step outputs advice about the schema estimated in the estimation step to the output means in the form of speech.

8. A method for controlling a trouble consultation device having a related keyword storage unit that stores related keywords corresponding to each of three types of thinking habits classified as high achievement orientation, other-dependent evaluation, and failure anxiety, comprising: (a) outputting each of a predetermined plurality of questions by voice to an output means; (b) detecting a user's voice; (c) detecting keywords for each of the predetermined plurality of questions based on the user's voice detected in step (b); (d) a step of estimating a schema, which is a cause of the user's trouble and is a thought pattern classified into the three types, based on the detected keywords, which are keywords detected in the step (c); and (e) adding a score to a type corresponding to the related keyword when the detected keyword matches the related keyword; The step (d) estimates the schema based on the sum of the step (e), A control method, wherein the step (a) outputs advice about the schema estimated in the step (d) to the output means by voice.

Images