Generation device

Abstract

Proper hearing can be difficult to achieve.SOLUTION: The generation apparatus includes an answer acquisition unit configured to acquire an answer to a predetermined question from a subject of hearing, and a question generation unit configured to generate and output an additional question corresponding to the answer in response to input of information corresponding to the answer acquired by the answer acquisition unit to a learned model.SELECTED DRAWING: Figure 8

Description

[Technical Field]

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

[0002] 2. Description of the Related Art Techniques used when conducting interviews such as questionnaires with customers are known.

[0003] For example, Patent Document 1 describes a survey management server having a selection means, a determination means, a transmission means, and an execution means. According to Patent Document 1, the selection means selects one or a group of questions included in a survey to be administered to a user. The determination means determines a benefit to be granted to the user in exchange for answering the questions, based on the user's purchase history of one or more products or services related to the questions. Thereafter, the transmission means transmits information regarding the questions selected by the selection means and the benefit determined by the determination means to a user terminal operated by the user. Then, upon receiving the answers to the questions transmitted by the transmission means from the user terminal, the execution means executes a process of granting the benefit determined by the determination means to the user. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-71931 Summary of the Invention [Problem to be solved by the invention]

[0005] When conducting hearings such as questionnaires using the technology described in Patent Document 1, there are cases where it is not possible to obtain sufficient answers to the matters that are to be heard, due to the small amount of answers or abstract content, etc. In this way, there has been a problem that it is sometimes difficult to obtain appropriate answers when conducting hearings.

[0006] Therefore, one object of the present disclosure is to provide a generation device, a generation method, and a program that can solve the above-mentioned problems. [Means for solving the problem]

[0007] To achieve this purpose, the generating device in the present disclosure comprises: an answer acquisition unit that acquires answers from the interviewee to predetermined questions; a question generation unit that generates and outputs an additional question corresponding to the answer in response to inputting information corresponding to the answer acquired by the answer acquisition unit into a trained model; have The structure is as follows.

[0008] In addition, the production method in the present disclosure includes: The information processing device Obtaining responses from interviewees to predetermined questions; By inputting information corresponding to the obtained answer into the trained model, an additional question corresponding to the answer is generated and output.

[0009] In addition, the program in this disclosure In the information processing device, Obtaining responses from interviewees to predetermined questions; By inputting information corresponding to the obtained answer into the trained model, an additional question corresponding to the answer is generated and output. It is a program for realizing the processing. [Effects of the Invention]

[0010] According to the above-mentioned configurations, it becomes possible to obtain more appropriate answers when conducting a hearing. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram illustrating an overview of a hearing system. [Figure 2]FIG. 2 is a block diagram illustrating an example of the configuration of a generating device. [Figure 3] FIG. 10 is a diagram illustrating an example of a process for generating a question. [Figure 4] FIG. 10 is a diagram illustrating an example of processing when a response is obtained. [Figure 5] 10 is a flowchart illustrating an example of the operation of the generating device. [Figure 6] FIG. 10 is a block diagram illustrating another example configuration of the generation device. [Figure 7] FIG. 2 is a diagram illustrating an example of a hardware configuration of a second generation device according to the present disclosure. [Figure 8] FIG. 2 is a block diagram illustrating an example of the configuration of a generating device. [Figure 9] 10 is a flowchart illustrating an example of the operation of the generating device. DETAILED DESCRIPTION OF THE INVENTION

[0012] [First embodiment] An example configuration of a hearing system 100 in the present disclosure will be described with reference to Figs. 1 to 6. Fig. 1 is a diagram illustrating an overview of the hearing system 100. Fig. 2 is a block diagram illustrating an example configuration of a generation device 200. Fig. 3 is a diagram illustrating an example process when a question is generated. Fig. 4 is a diagram illustrating an example process when an answer is obtained. Fig. 5 is a flowchart illustrating an example operation of the generation device 200. Fig. 6 is a block diagram illustrating an example configuration of a generation device 300, which is another example configuration. Note that in the present disclosure, the drawings may be associated with one or more embodiments.

[0013] This disclosure describes a hearing system 100 that can generate follow-up questions for an acquired answer as needed by using a trained model such as an LLM (Large Language Model). As will be described later, the hearing system 100 acquires an answer from a hearing subject to a predetermined question. Then, the hearing system 100 inputs information corresponding to the acquired answer into the LLM, which is a trained model. For example, the hearing system 100 inputs the results of a predetermined language analysis of the answer into the LLM as information corresponding to the answer. This allows the hearing system 100 to generate follow-up questions corresponding to the answer.

[0014] Furthermore, when the hearing system 100 acquires an answer, it evaluates the answer using the acquired answer. When generating the follow-up question, the hearing system 100 can determine whether to generate the follow-up question based on the evaluation result of the answer. For example, the hearing system 100 can determine whether to generate the follow-up question based on an evaluation that the answer is insufficient, such as that the answer is ambiguous or a more detailed answer is required. When generating the follow-up question, the hearing system 100 may input the evaluation result, etc. into the LLM, thereby generating a question corresponding to the evaluation result, etc. For example, the hearing system 100 can generate a follow-up question based on the answer, the evaluation result of the answer, etc., by utilizing the LLM, such as generating a follow-up question to dig deeper into a part of the answer that is determined to be ambiguous.

[0015] Furthermore, the hearing system 100 can calculate an incentive for the answer based on the obtained answer, and perform predetermined analysis processing on the answer, such as evaluating consistency or concentration. The hearing system 100 can also output the results of the incentive calculation and analysis. The hearing system 100 may use at least a part of the results of the incentive calculation and analysis when evaluating the answer or generating additional questions.

[0016] In the hearing system 100, hearing content information 241 is defined in advance as hearing matter information indicating the content to be heard from the hearing target. The hearing content information 241 may include, as hearing matter information, information indicating the content to be heard, as well as at least a part of an expected customer usage scenario, desired functions, values, budget, and other arbitrary information.

[0017] As will be described later, the hearing system 100 can be configured to generate questions based on customer attribute information, etc., using LLM or the like. Here, attribute information refers to information indicating the attributes of the customer to be heard, such as information based on the customer's profile. The attribute information includes at least some of the following: age group, information indicating interests and concerns, information indicating purchasing power, information indicating the area of ​​residence, etc. The attribute information may also include at least some of gender, family structure, and other optional information. Note that the hearing system 100 may be configured to generate questions without using attribute information, or predetermined questions, such as the initial question, may be prepared in advance.

[0018] FIG. 1 shows an overview of a hearing system 100. Referring to FIG. 1, the hearing system 100 includes a generating device 200, which is an information processing device capable of generating additional questions as needed, and a trained model such as an LLM. For example, in the case of FIG. 1, an information processing device having a trained model such as an LLM exists outside the generating device 200. However, the trained model such as an LLM may be included in the generating device 200.

[0019] The generating device 200 is an information processing device that acquires answers to questions from interview subjects such as customers and generates additional questions as necessary. Fig. 2 shows an example of the main configuration of the generating device 200. Referring to Fig. 2, the generating device 200 has, as main components, for example, an operation input unit 210, a screen display unit 220, a communication interface unit 230, a storage unit 240, and an arithmetic processing unit 250.

[0020] 2 illustrates an example in which the functions of generating device 200 are realized using one information processing device. However, at least some of the functions of generating device 200 may be realized using multiple information processing devices, for example, on the cloud. Furthermore, generating device 200 may not include some of the components exemplified above, such as not having operation input unit 210 or screen display unit 220, or may have a component other than those exemplified above.

[0021] The operation input unit 210 is composed of operation input devices such as a keyboard, a mouse, etc. The operation input unit 210 detects operations of a customer or any other operator who operates the generation device 200 and outputs the operations to the calculation processing unit 250.

[0022] The screen display unit 220 is composed of a screen display device such as a liquid crystal display, an organic EL (electro-luminescence) display, etc. The screen display unit 220 can display various information stored in the storage unit 240 on the screen in response to instructions from the arithmetic processing unit 250.

[0023] The communication interface unit 230 is composed of a data communication circuit, etc. The communication interface unit 230 performs data communication with an external device connected via a communication line.

[0024] The storage unit 240 is a storage device such as a hard disk or memory. The storage unit 240 stores processing information and programs 243 required for various processes in the arithmetic processing unit 250. The programs 243 are read into the arithmetic processing unit 250 and executed to realize various processing units. The programs 243 are read in advance from an external device or recording medium via a data input / output function such as the communication interface unit 230, and are stored in the storage unit 240. Main information stored in the storage unit 240 includes, for example, hearing content information 241, question and answer information 242, etc.

[0025] The hearing content information 241 includes hearing matter information indicating the content of the hearing to be conducted with the hearing subject. The hearing content information 241 is acquired in advance by means of accepting input using the operation input unit 210, accepting information from an external device via the communication interface unit 230, or the like, and is stored in the storage unit 240.

[0026] As described above, the hearing content information 241 includes information indicating the content of the hearing as hearing matter information. In addition to the information exemplified above, the hearing content information may include at least a part of the expected customer usage scenario, desired functions, values, budget, and other arbitrary information as hearing matter information.

[0027] The question and answer information 242 includes information such as generated questions and answers received in response to questions. The question and answer information 242 may also include example answers generated together with the questions, attribute information used when generating the questions, etc. The question and answer information 242 can be updated as the question generation unit 252 generates questions, etc., or as the answer acquisition unit 253 acquires answers, etc.

[0028] As described above, the generating device 200 can generate additional questions depending on the evaluation results of the obtained answers. When multiple questions are generated for the same hearing target in this way, the question and answer information 242 may store the questions, answers, etc. so that it is possible to distinguish that the questions are a series of questions generated for the same hearing target. As an example, the question and answer information 242 may associate identification information, etc., given to each hearing target with the questions, answers, etc.

[0029] The arithmetic processing unit 250 has an arithmetic device such as a CPU (Central Processing Unit) and its peripheral circuits. The arithmetic processing unit 250 reads and executes a program 243 from the storage unit 240, thereby causing the above hardware and the program 243 to work together to realize various processing units. Major processing units realized by the arithmetic processing unit 250 include, for example, an information acquisition unit 251, a question generation unit 252, an answer acquisition unit 253, an answer analysis unit 254, an answer evaluation unit 255, an incentive calculation unit 256, and an output unit 257.

[0030] In addition, the arithmetic processing unit 250 may have a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating point number Processing Unit), a PPU (Physics Processing Unit), a TPU (Tensor Processing Unit), a quantum processor, a microcontroller, or a combination thereof, instead of the above-mentioned CPU.

[0031] The information acquisition unit 251 acquires attribute information from a hearing subject such as a customer. The information acquisition unit 251 can acquire the attribute information by receiving an input using the operation input unit 210, by receiving information from an external device via the communication interface unit 230, or the like.

[0032] For example, the information acquisition unit 251 can acquire, as attribute information, at least a part of the age group of the person being interviewed, information indicating their interests and concerns, information indicating their purchasing power, information indicating the area where they live, etc. The information acquisition unit 251 may acquire, as attribute information, at least a part of gender, family structure, other arbitrary information, etc., in addition to or instead of the information exemplified above.

[0033] The question generation unit 252 generates a question to be sent to the person being interviewed, according to the attribute information and the like acquired by the information acquisition unit 251. For example, the question generation unit 252 inputs the attribute information and the like into a trained model such as an LLM. Then, the question generation unit 252 acquires a question as an output from the model. In this way, the question generation unit 252 can generate a question according to inputting the attribute information and the like into the trained model. Furthermore, the question generation unit 252 can store the generated question and the like in the storage unit 240 as question and answer information 242.

[0034] As an example, the question generation unit 252 inputs the acquired attribute information and at least a portion of the hearing matter information included in the hearing content information 241 into the trained model. At this time, the question generation unit 252 can select the hearing matters to be input into the trained model according to predetermined conditions. For example, the question generation unit 252 may select the hearing matter information to be input by checking whether each piece of information included in the attribute information satisfies a predetermined condition. Furthermore, the question generation unit 252 acquires a question according to the attribute information output from the model as a result of the above input.

[0035] The question generation unit 252 may be configured to generate a question to be sent to the interviewee in accordance with the attribute information acquired by the information acquisition unit 251, and to generate example answers to the generated question. As described above, the question generation unit 252 can generate a question and example answers by inputting attribute information into the trained model. For example, FIG. 3 illustrates an example in which the question generation unit 252 generates a question and example answers. As illustrated in FIG. 3, the question generation unit 252 may generate a question to be sent to the interviewee, as well as expected and unexpected example answers to the generated question, in accordance with the attribute information acquired by the information acquisition unit 251. In this manner, the example answers generated by the question generation unit 252 may include expected and unexpected example answers to the question. The question generation unit 252 may generate information indicating the format of the answer in addition to specific content as the example answer. For example, the question generation unit 252 may generate information indicating formal matters that are not dependent on the content of the sentence, such as the length of the sentence, as the example answer. Furthermore, the question generator 252 may generate keywords that are expected or not expected to be included as specific content shown as an example answer.

[0036] Furthermore, in cases such as when the answer evaluation unit 255 (described later) evaluates an answer as insufficient, the question generation unit 252 can generate an additional question. In this case, the question generation unit 252 may generate the additional question by inputting, to the trained model, at least a part of information indicating that the question is an additional question, information indicating the evaluation result by the answer evaluation unit 255, the answer that was the subject of evaluation, and other arbitrary information, together with or instead of attribute information. In this case, the question generation unit 252 may generate an additional question according to the answer, the evaluation result, and the like. For example, the question generation unit 252 can generate an additional question to encourage improvement or to dig deeper into an answer evaluated as insufficient, such as "Could you please explain part XX in more detail?"

[0037] The answer acquisition unit 253 acquires answers to questions and additional questions generated by the question generation unit 252. The answer acquisition unit 253 can acquire answers by receiving input using the operation input unit 210, by receiving information from an external device via the communication interface unit 230, or the like. The answer acquisition unit 253 can also store the acquired answers in the storage unit 240 as question and answer information 242.

[0038] The response analysis unit 254 executes a predetermined analysis process on the response acquired by the response acquisition unit 253.

[0039] For example, the answer analysis unit 254 analyzes the answer acquired by the answer acquisition unit 253 using natural language processing technology. As an example, the answer analysis unit 254 performs at least some of morphological analysis, syntactic analysis, semantic analysis, contextual analysis, etc. on the acquired answer. Furthermore, the answer analysis unit 254 uses the results of the above analysis to extract keywords and measure the length of the answer. The answer analysis unit 254 may also perform processing other than those exemplified above using the analysis results, such as understanding the context of the answer.

[0040] The answer evaluation unit 255 evaluates the answer acquired by the answer acquisition unit 253 using the analysis result by the answer analysis unit 254. For example, the answer evaluation unit 255 evaluates the answer by comparing the answer acquired by the answer acquisition unit 253 with the example answers generated by the question generation unit 252. Note that when making the comparison, the answer evaluation unit 255 may make the comparison using example answers stored in advance in the storage unit 240, etc. Furthermore, instead of or in addition to the comparison, the answer evaluation unit 255 may evaluate the answer by checking whether the answer acquired by the answer acquisition unit 253 satisfies a criterion stored in advance.

[0041] FIG. 4 shows an example of the answer evaluation unit 255 evaluating an answer. Referring to FIG. 4, the answer evaluation unit 255 can evaluate an answer by comparing the analysis result by the answer analysis unit 254 with expected answer examples and unexpected answer examples generated by the question generation unit 252. For example, the answer evaluation unit 255 compares the analysis result with the answer examples generated by the question generation unit 252 to check to what extent the acquired answer matches the expected answer examples and unexpected answer examples. As an example, the answer evaluation unit 255 may check at least one of whether the answer is of appropriate length, whether the content does not deviate from the intent of the question, whether the answer contains keywords necessary for the answer, and the like. Furthermore, the answer evaluation unit 255 can evaluate the answer according to the results of the above checks. For example, the more the acquired answer matches the expected answer, the higher the evaluation, indicating that the answer is sufficient. On the other hand, the more the acquired answer matches an unexpected answer, the more the answer matches the answer, indicating that the answer is insufficient. In this way, the answer evaluation unit 255 can evaluate the answer based on the results of checking how much the acquired answer matches expected answer examples and unexpected answer examples. It may be arbitrarily determined whether the answer evaluation unit 255 places more importance on the match to the expected answer or the match to the unexpected answer. When performing the above evaluation, the answer evaluation unit 255 may also evaluate whether the answer is sufficient by calculating a match rate or the like and comparing the calculated match rate with a predetermined threshold value. In this case, the threshold value used when performing the above evaluation may be arbitrarily determined.

[0042] The answer evaluation unit 255 may evaluate the answer using only either an expected answer example or an unexpected answer example. The answer evaluation unit 255 may also evaluate the answer by inputting at least one of the answer acquired by the answer acquisition unit 253 and the analysis result by the answer analysis unit 254 into a trained evaluation model. The answer evaluation unit 255 may evaluate the answer by inputting the analysis result by the answer analysis unit 254, the answer example generated by the question generation unit 252, etc. into a trained evaluation model. In this case, the trained evaluation model may be an LLM or the like, or may be something else.

[0043] Furthermore, the answer evaluation unit 255 can determine whether to generate an additional question depending on the evaluation result. For example, the answer evaluation unit 255 can determine to generate an additional question when the evaluation result indicates that the answer is evaluated as insufficient or otherwise determines that a predetermined condition is met. In this case, the answer evaluation unit 255 can instruct the question generation unit 252 to generate an additional question depending on the result of the above determination. Note that the above determination by the answer evaluation unit 255 may also be made using a trained model such as LLM.

[0044] The incentive calculation unit 256 calculates an incentive to encourage an answer to a question or according to the evaluation of the answer. For example, the incentive calculation unit 256 can calculate an incentive according to the quality or quantity of the answer determined according to the analysis result by the answer analysis unit 254 or the evaluation result by the answer evaluation unit 255. Note that the incentive calculated by the incentive calculation unit 256 may be various points or any other arbitrary currency, or may be a discount on a predetermined product, an arbitrary evaluation, or any other arbitrary value.

[0045] As an example, the incentive calculation unit 256 checks criteria such as the length of the answer, the completeness of the content, the suitability of the answer, and the originality of the answer, using the answer analysis results from the answer analysis unit 254. Then, the incentive calculation unit 256 calculates a score by applying an arbitrary calculation formula to each of the above checked values. Thereafter, the incentive calculation unit 256 can calculate the amount of remuneration by multiplying the calculated score by the upper limit of the incentive.

[0046] The calculation formula used by the incentive calculation unit 256 to calculate the score may be designed arbitrarily. Furthermore, the incentive calculation unit 256 may calculate the incentive using a method other than the above-mentioned examples. For example, the incentive calculation unit 256 may calculate the remuneration amount according to the evaluation result by the answer evaluation unit 255. As an example, the incentive calculation unit 256 may calculate the remuneration amount so that the higher the evaluation value by the answer evaluation unit 255, the closer it is to an upper limit value. Furthermore, the incentive calculation unit 256 may calculate the incentive using a trained model such as LLM. The incentive calculation unit 256 may calculate the score by calculating the degree of deviation of each of the above-mentioned reference values ​​between the answer to be calculated and an expected answer example.

[0047] The output unit 257 outputs the question, the additional question, etc. generated by the question generation unit 252. For example, the output unit 257 may display the generated question, the additional question, etc. on the screen display unit 220, or transmit it to an external device via the communication interface unit 230.

[0048] Furthermore, the output unit 257 can output information indicating the incentive calculated by the incentive calculation unit 256. For example, the output unit 257 may output information according to the calculation result by the incentive calculation unit 256 when outputting an additional question.

[0049] The output unit 257 may be configured to output, in addition to the output related to the implementation of the hearing as described above, the hearing result or information corresponding to the hearing result to the person who executes the hearing. For example, the output unit 257 may be configured to output information included in the question and answer information 242. The output unit 257 may be configured to output any statistical information that can be calculated from the information included in the question and answer information 242.

[0050] The above is an example of the configuration of the generation device 200. The generation device 200 may store in advance the initial question to be asked to the customer. The generation device 200 may also generate questions without using attribute information. In such a configuration, the generation device 200 may not include the information acquisition unit 251. The question generation unit 252 may also be configured to generate only additional questions. In this way, the generation device 200 may be configured with at least some of the functions described in this disclosure. Next, an example of the operation of the generation device 200 will be described with reference to FIG. 5.

[0051] Fig. 5 is a flowchart showing an example of the operation of generation device 200. Referring to Fig. 5, information acquisition unit 251 acquires attribute information from a hearing subject such as a customer (step S101). Information acquisition unit 251 can acquire the attribute information by accepting input using operation input unit 210, accepting information from an external device via communication interface unit 230, or the like.

[0052] The question generation unit 252 generates a question and an example answer to be sent to the person being interviewed, according to the attribute information etc. acquired by the information acquisition unit 251 (step S102). The question generation unit 252 may generate a question, an expected example answer, and an unexpected example answer, according to the attribute information etc.

[0053] The output unit 257 outputs the question generated by the question generation unit 252 to the person being heard (step S103).

[0054] Answer acquisition unit 253 acquires an answer to the question generated by question generation unit 252 (step S104). Answer acquisition unit 253 can acquire an answer by receiving an input using operation input unit 210, by receiving information from an external device via communication interface unit 230, or the like.

[0055] The answer analysis unit 254 executes a predetermined analysis process on the answer acquired by the answer acquisition unit 253 (step S105). The answer analysis unit 254 may perform an analysis using a natural language processing technique on the answer acquired by the answer acquisition unit 253.

[0056] The answer evaluation unit 255 evaluates the answer acquired by the answer acquisition unit 253 using the analysis result by the answer analysis unit 254. For example, the answer evaluation unit 255 can evaluate the answer by comparing the analysis result by the answer analysis unit 254 with expected answer examples and unexpected answer examples generated by the question generation unit 252 (step S106).

[0057] The incentive calculation unit 256 calculates the incentive (step S107). The incentive calculation unit 256 can calculate the incentive according to the quality and quantity of the answers determined according to the analysis result by the answer analysis unit 254 and the evaluation result by the answer evaluation unit 255.

[0058] If the evaluation by the answer evaluation unit 255 does not satisfy the condition (step S108, NO), the answer evaluation unit 255 determines that the acquired answer is insufficient and that an additional question should be generated. In response to this, the question generation unit 252 generates an additional question and an example answer (step S102). On the other hand, if the evaluation by the answer evaluation unit 255 satisfies the condition (step S108, YES), the generation device 200 ends the process.

[0059] The above is an example of the operation of generating device 200. Note that the operation of generating device 200 may be other than that exemplified in Fig. 5. For example, the process of step S101 may be omitted. Furthermore, the process of step S107 may be performed after the process of step S108.

[0060] As described above, the generation device 200 includes the question generation unit 252 and the answer acquisition unit 253. With this configuration, the question generation unit 252 can generate an additional question for the answer acquired by the answer acquisition unit 253. As a result, the generation device 200 can generate a question for digging deeper into the acquired answer. This allows the generation device 200 to ask an additional question even when the initially acquired answer is insufficient, thereby enabling the generation device 200 to obtain a more appropriate answer when conducting a hearing.

[0061] Furthermore, generation device 200 has answer evaluation unit 255. With this configuration, question generation unit 252 can generate additional questions depending on the results of evaluation by answer evaluation unit 255. As a result, if an answer is evaluated as being insufficient, additional questions can be asked, and if necessary, more appropriate questions can be asked regarding matters that the user wishes to hear about.

[0062] The configuration of the generation device 200 is not limited to the example shown in Fig. 2. For example, Fig. 6 shows an example configuration of a generation device 300, which is a modified example of the generation device 200. Referring to Fig. 6, the calculation processing unit 250 can implement at least some of the components shown in Fig. 2 as an example, including a sentiment analysis unit 351, a consistency evaluation unit 352, a concentration evaluation unit 353, and a report generation unit 354, by reading and executing the program 243.

[0063] The emotion analysis unit 351 analyzes the emotions of the interviewee when, for example, the interviewee answers a question, using the answers acquired by the answer acquisition unit 253. For example, the emotion analysis unit 351 performs emotion analysis, such as positive or negative, by identifying words and expressions used in sentences included in the answers in accordance with the analysis results by the answer analysis unit 254. The emotion analysis unit 351 may analyze emotions by inputting the answers acquired by the answer acquisition unit 253, the analysis results by the answer analysis unit 254, and the like into a trained emotion analysis model.

[0064] If there are questions and answers in chronological order, such as when additional questions are generated, the emotion analysis unit 351 may be configured to evaluate the transition and movement of emotions according to the chronological answers, etc. Furthermore, the emotion analysis unit 351 may perform emotion analysis by utilizing a predetermined dedicated library, etc.

[0065] The consistency evaluation unit 352 evaluates the consistency of the answers acquired by the answer acquisition unit 253. For example, when the answer acquisition unit 253 acquires multiple answers due to the generation of additional questions, the consistency evaluation unit 352 can evaluate the consistency of the acquired multiple answers. In other words, the consistency evaluation unit 352 can evaluate how consistent or contradictory the information is between the multiple answers.

[0066] For example, the consistency evaluation unit 352 performs predetermined preprocessing on each answer. For example, the consistency evaluation unit 352 performs at least some of the following preprocessing: removing noise such as typos, extra spaces, and special characters; tokenizing the text; removing stop words and stemming or lemmatizing the text; and extracting keywords and phrases. The consistency evaluation unit 352 can then evaluate the consistency by evaluating the similarity between each answer. For example, the consistency evaluation unit 352 may determine the similarity by vectorizing each answer using the extracted keywords and calculating cosine similarity. The consistency evaluation unit 352 may also determine the similarity using any method other than cosine similarity. Furthermore, the consistency evaluation unit 352 may identify pairs of answers whose similarity is lower than a predetermined value and identify content that is likely to be contradictory by checking the words and contexts contained in the answers of the identified pairs. The consistency evaluation unit 352 may then output the identified content as contradictory content.

[0067] The concentration level evaluation unit 353 evaluates the level of concentration when making an answer based on the answer acquired by the answer acquisition unit 253. For example, the concentration level evaluation unit 353 can evaluate the continuity of concentration or focus on important points in the answer based on the frequency of appearance or relevance of important keywords. Furthermore, the concentration level evaluation unit 353 can evaluate the level of concentration as low when there are scattered answers or a lack of focus.

[0068] For example, the concentration evaluation unit 353 extracts important keywords and phrases from each answer. As an example, the concentration evaluation unit 353 can prioritize extraction of themes and keywords that are expected to be particularly important, taking into account the purpose of the hearing and the context of the questions. The concentration evaluation unit 353 can also evaluate the concentration level by, for example, analyzing the degree to which the same keywords or themes repeatedly appear in the answers. For example, the concentration evaluation unit 353 may evaluate the concentration level as low when it can be evaluated that the focus is not consistent, such as when the appearance of keywords or themes is irregular. The concentration evaluation unit 353 may also evaluate the concentration level as low when, for example, many different themes appear in the answers, the answers are scattered. On the other hand, the concentration evaluation unit 353 may evaluate the concentration level as high when it can be evaluated that the appearance of keywords or themes is regular. In this way, the concentration evaluation unit 353 can evaluate the concentration level by, for example, analyzing the frequency and distribution of keywords in the answers and measuring the degree of consistency of focus.

[0069] The report generation unit 354 generates a report showing the analysis results by the sentiment analysis unit 351, the evaluation results by the consistency evaluation unit 352, the evaluation results by the concentration evaluation unit 353, etc. The report generation unit 354 may generate a report using the results of processing by each of the above-mentioned processing units and information included in the question and answer information 242. The report generated by the report generation unit 354 can be output by the output unit 257 to the person conducting the hearing, etc.

[0070] The analysis result by the sentiment analysis unit 351, the evaluation result by the consistency evaluation unit 352, the evaluation result by the concentration evaluation unit 353, etc. may be used by the report generation unit 354 when generating a report, and may also be used when the answer evaluation unit 255 makes an evaluation. For example, the answer evaluation unit 255 may be configured to give a higher evaluation when the consistency evaluation unit 352 evaluates that there is consistency or when the concentration evaluation unit 353 evaluates that there is a high level of concentration. The analysis result by the sentiment analysis unit 351, the evaluation result by the consistency evaluation unit 352, the evaluation result by the concentration evaluation unit 353, etc. may also be used by the incentive calculation unit 256 when calculating an incentive. For example, the incentive calculation unit 256 may calculate an incentive so as to provide a higher reward amount when the consistency evaluation unit 352 evaluates that there is consistency or when the concentration evaluation unit 353 evaluates that there is a high level of concentration, etc. Furthermore, the analysis result by the sentiment analysis unit 351, the evaluation result by the consistency evaluation unit 352, the evaluation result by the concentration evaluation unit 353, etc. may be input to an LLM or the like and used when generating additional questions. In this case, the question generation unit 252 may generate additional questions that reflect the analysis result by the sentiment analysis unit 351, the evaluation result by the consistency evaluation unit 352, the evaluation result by the concentration evaluation unit 353, etc. For example, the question generation unit 252 may be configured to ask an additional question such as which answer should be given more weight when it is determined that there is low consistency between the first answer and the second answer.

[0071] For example, as described above, the hearing system 100 may have a generating device 300 having at least some of the functions described above, instead of the generating device 200. The generating device 200 or the generating device 300 may have an answer analysis function other than those exemplified.

[0072] [Second embodiment] Next, a generating device 400, which is a modified example of the generating device 200 or the generating device 300, will be described with reference to Fig. 7 to Fig. 9. Fig. 7 is a diagram showing an example of the hardware configuration of the generating device 400. Fig. 8 is a block diagram showing an example of the configuration of the generating device 400. Fig. 9 is a flowchart showing an example of the operation of the generating device 400.

[0073] The generating device 400 is an information processing device that can generate follow-up questions for answers as needed. Fig. 7 shows an example of the hardware configuration of the generating device 400. Referring to Fig. 7, the generating device 400 has, as an example, the following hardware configuration. ·CPU(Central Processing Unit)401(Arithmetic unit) ROM (Read Only Memory) 402 (storage device) RAM (Random Access Memory) 403 (storage device) Programs 404 loaded into RAM 403 A storage device 405 for storing the program group 404 A drive device 406 that reads and writes data from a recording medium 410 outside the information processing device A communication interface 407 for connecting to a communication network 411 outside the information processing device Input / output interface 408 for inputting and outputting data Bus 409 connecting each component

[0074] 8 by the CPU 401 acquiring and executing the program group 404. The program group 404 is stored in advance in the storage device 405 or the ROM 402, for example, and is loaded into the RAM 403 or the like by the CPU 401 for execution as needed. The program group 404 may be supplied to the CPU 401 via the communication network 411, or may be stored in advance in the recording medium 410, with the drive device 406 reading out the programs and supplying them to the CPU 401.

[0075] 7 shows an example of the hardware configuration of the generating device 400. The hardware configuration of the generating device 400 is not limited to the above-described case. For example, the generating device 400 may be configured with only a part of the above-described configuration, such as not including the drive device 406. Furthermore, the CPU 401 may be a GPU or the like exemplified in the first embodiment.

[0076] The answer acquisition unit 421 acquires answers from the interviewees to predetermined questions.

[0077] The question generation unit 422 inputs information corresponding to the answer acquired by the answer acquisition unit 421 into a Large Language Model (LLM), which is a trained model. In response to this, the question generation unit 422 generates and outputs an additional question corresponding to the answer.

[0078] The above is an example of the configuration of generation device 400. Next, an example of the operation of generation device 400 will be described with reference to FIG.

[0079] Fig. 9 is a flowchart showing an example of the operation of the generating device 400. Referring to Fig. 9, the answer acquiring unit 421 acquires an answer from a hearing target to a predetermined question (step S201).

[0080] The question generation unit 422 inputs information corresponding to the answer acquired by the answer acquisition unit 421 into the trained model. In response to this, the question generation unit 422 generates and outputs an additional question corresponding to the answer (step S202).

[0081] The above is an example of the operation of the generating device 400.

[0082] As described above, generation device 400 has answer acquisition unit 421 and question generation unit 422. With this configuration, question generation unit 422 can generate an additional question corresponding to the answer acquired by answer acquisition unit 421. As a result, it becomes possible to ask additional questions as needed, and to acquire more appropriate answers when conducting a hearing.

[0083] The above-described generating device 400 can be realized by incorporating a predetermined program into an information processing device such as the generating device 400. Specifically, a program according to another aspect of the present disclosure is a program for causing an information processing device such as the generating device 400 to realize a process of acquiring an answer to a predetermined question from a hearing subject, inputting information corresponding to the acquired answer into a trained model, and generating and outputting an additional question corresponding to the answer.

[0084] Furthermore, the generation method executed by an information processing device such as the above-mentioned generation device 400 is a method in which the information processing device such as the generation device 400 acquires an answer from a hearing subject to a predetermined question, and inputs information corresponding to the acquired answer into a trained model, thereby generating and outputting an additional question corresponding to the answer.

[0085] A program having the above-described configuration, a computer-readable recording medium having a program recorded thereon, or a generation method, etc., can achieve the same functions and effects as the above-described generation device 400, and therefore can achieve the above-described objective of the present disclosure.

[0086] <Additional Notes> A part or all of the above-described embodiments can be described as follows: The generating device and other components of the present disclosure are outlined below. However, the present disclosure is not limited to the following configuration.

[0087] (Appendix 1) an answer acquisition unit that acquires answers from the interviewee to predetermined questions; a question generation unit that generates and outputs an additional question corresponding to the answer in response to inputting information corresponding to the answer acquired by the answer acquisition unit into a trained model; have generator. (Appendix 2) an answer evaluation unit that evaluates the answer acquired by the answer acquisition unit using a predetermined analysis result for the answer acquired by the answer acquisition unit; The question generation unit generates an additional question in response to inputting the answer and a result of the evaluation by the answer evaluation unit as information corresponding to the answer to the model. 10. The generating device of claim 1. (Appendix 3) an incentive calculation unit that calculates an incentive for the answer using a value according to a predetermined analysis result for the answer acquired by the answer acquisition unit; 10. The generating device of claim 1 or 2. (Appendix 4) a consistency evaluation unit that evaluates the consistency of the answers acquired by the answer acquisition unit and outputs an evaluation result; The consistency evaluation unit evaluates the consistency of the multiple answers acquired in accordance with the answers acquired multiple times by the answer acquisition unit. 10. The generating device of claim 1, wherein the generating device is a (Appendix 5) A concentration evaluation unit evaluates the concentration level when answering based on the answer acquired by the answer acquisition unit and outputs the evaluation result. 10. The generating device of claim 1. (Appendix 6) A sentiment analysis unit analyzes the sentiment of the customer when making a reply based on the reply acquired by the reply acquisition unit and outputs the analysis result. 10. The generating device of any one of Supplementary Notes 1 to 5. (Appendix 7) the answer acquisition unit is configured to acquire the answer to a question generated using the model; the model is configured to, when generating the question, generate the question together with an example answer corresponding to the question; The answer evaluation unit evaluates the answer using a predetermined analysis result for the answer acquired by the answer acquisition unit and an example answer generated by the model. 10. The generating device of claim 2. (Appendix 8) The question generation unit generates an additional question when the answer evaluation unit evaluates the answer as insufficient. 10. The generating device of claim 2. (Appendix 9) The information processing device Obtaining responses from interviewees to predetermined questions; By inputting information corresponding to the obtained answer into the trained model, an additional question corresponding to the answer is generated and output. Generation method. (Appendix 10) In the information processing device, Obtaining responses from interviewees to predetermined questions; By inputting information corresponding to the obtained answer into the trained model, an additional question corresponding to the answer is generated and output. A program to realize the processing.

[0088] Note that some or all of the configurations described in Supplementary Notes 2 to 8 that are dependent on the generating device described in Supplementary Note 1 may also be dependent in a similar dependent relationship on the generating method described in Supplementary Note 9 and the program described in Supplementary Note 10. Furthermore, not limited to Supplementary Notes 9 and 10, some or all of the configurations described as Supplements may also be dependent on various hardware, software, and various recording means, methods, or systems for recording software, within the scope of each of the above-mentioned embodiments.

[0089] The programs described in the above embodiments and appendices can be stored in various types of non-transitory computer-readable media and supplied to a computer. Non-transitory computer-readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)). The programs may also be supplied to a computer by various types of transitory computer-readable media. Examples of transitory computer-readable media include electrical signals, optical signals, and electromagnetic waves. The transitory computer-readable media can supply the programs to a computer via wired communication paths such as electric wires and optical fibers, or via wireless communication paths.

[0090] Although the present disclosure has been described above with reference to the above-described embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, each embodiment can be combined with other embodiments as appropriate. [Explanation of symbols]

[0091] 100 Hearing System 200 generator 210 Operation input section 220 Screen display section 230 Communication interface unit 240 Storage section 241 Hearing Content Information 242 Question / Answer Information 243 Programs 250 Processing Unit 251 Information Acquisition Department 252 Question generation part 253 Answer acquisition part 254 Answer Analysis Department 255 Answer Evaluation Section 256 Incentive Calculation Department 257 output section 300 generator 351 Emotion Analysis Department 352 Consistency Evaluation Unit 353 Concentration Evaluation Department 354 Report Generation Unit 400 generator 401 CPU 402 ROM 403 RAM 404 Programs 405 Storage device 406 Drive Unit 407 Communication Interface 408 Input / Output Interface 409 Bus 410 Recording Media 411 Communication Network 421 Answer acquisition part 422 Question generation part

Claims

1. an answer acquisition unit that acquires answers from the interviewee to predetermined questions; a question generation unit that generates and outputs an additional question corresponding to the answer in response to inputting information corresponding to the answer acquired by the answer acquisition unit into a trained model; have generator.

2. an answer evaluation unit that evaluates the answer acquired by the answer acquisition unit using a predetermined analysis result for the answer acquired by the answer acquisition unit; The question generation unit generates an additional question in response to inputting the answer and a result of the evaluation by the answer evaluation unit as information corresponding to the answer to the model. The generating device of claim 1 .

3. an incentive calculation unit that calculates an incentive for the answer using a value according to a predetermined analysis result for the answer acquired by the answer acquisition unit; The generating device of claim 1 .

4. a consistency evaluation unit that evaluates the consistency of the answers acquired by the answer acquisition unit and outputs an evaluation result; The consistency evaluation unit evaluates the consistency of the multiple answers acquired in accordance with the answers acquired multiple times by the answer acquisition unit. The generating device of claim 1 .

5. A concentration evaluation unit evaluates the concentration level when answering based on the answer acquired by the answer acquisition unit and outputs the evaluation result. The generating device of claim 1 .

6. A sentiment analysis unit analyzes the sentiment of the customer when making a reply based on the reply acquired by the reply acquisition unit and outputs the analysis result. The generating device of claim 1 .

7. the answer acquisition unit is configured to acquire the answer to a question generated using the model; the model is configured to, when generating the question, generate the question together with an example answer corresponding to the question; The answer evaluation unit evaluates the answer using a predetermined analysis result for the answer acquired by the answer acquisition unit and an example answer generated by the model. The generating device of claim 2 .

8. The question generation unit generates an additional question when the answer evaluation unit evaluates the answer as insufficient. The generating device of claim 2 .

9. The information processing device Obtaining responses from interviewees to predetermined questions; By inputting information corresponding to the obtained answer into the trained model, an additional question corresponding to the answer is generated and output. Generation method.

10. In the information processing device, Obtaining responses from interviewees to predetermined questions; By inputting information corresponding to the obtained answer into the trained model, an additional question corresponding to the answer is generated and output. A program to realize the processing.

Images