Information processing system, information processing method, and information processing program

Abstract

An information processing system, information processing method, and information processing program are provided that can support the creation of requirements definitions for the type of person a recruiting company is looking for by using a first generative model that can interact with the user. [Solution] In the recruitment activity support system 10, the server 20 acquires the content of the dialogue between the first user and the first generation model, inputs predetermined instructions regarding the recruitment requirements and the dialogue content regarding the recruitment requirements into the first generation model to generate proposed recruitment requirements, and if the first user agrees to the proposed recruitment requirements, inputs predetermined instructions regarding evaluation viewpoints and the dialogue content regarding the evaluation viewpoints into the first generation model to generate proposed evaluation viewpoints, and after the first user agrees to the proposed evaluation viewpoints, inputs the proposed evaluation viewpoints and predetermined instructions regarding questions to be asked in recruitment interviews into the first generation model to generate example questions, and outputs the proposed evaluation viewpoints and the example questions.

Description

[Technical Field]

[0001] The present disclosure relates to an information processing system, an information processing method, and an information processing program. [Background technology]

[0002] Patent Document 1 discloses a technology that allows interview participants who have taken part in an online interview to be evaluated at low cost with a high degree of scientific objectivity. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6042015 Summary of the Invention [Problem to be solved by the invention]

[0004] In recruitment activities, a job interview is generally conducted between a company and a job seeker, and the hiring decision is made based on the interview. However, it is known that job interviews conducted by human interviewers are easily influenced by subjective factors such as appearance and impression, resulting in variations in evaluation criteria.

[0005] In recent years, support tools that utilize artificial intelligence (AI) have been provided to support manual recruitment activities, such as those described in Patent Document 1. However, conventional support tools are unable to create a profile of a job seeker to hire, including the recruitment requirements and evaluation criteria, in accordance with the job type and corporate culture of each recruiting company, and there is still the issue that recruitment activities are operated on a personal basis.

[0006] Therefore, the present disclosure aims to provide an information processing system, an information processing method, and an information processing program that can assist in creating a requirements definition for the type of person a recruiting company is looking for by using a first generation model that can interact with the user. [Means for solving the problem]

[0007] The information processing system of the first aspect includes a processor, and the processor acquires dialogue information indicating dialogue content between a first user and a first generative model that can interact with the first user regarding the hiring requirements for job seekers that a recruiting company is looking for and evaluation points in a hiring interview, and includes predetermined instruction content regarding the hiring requirements and dialogue content regarding the hiring requirements among the dialogue information. first A prompt is input to the first generative model, first obtaining a proposed hiring requirement as an output result of the first generative model in response to the prompt; outputting the proposed employment requirements generated by the first generative model, accepting input of a response from the first user of an opinion or a request for modification regarding the proposed employment requirements, extracting the speech content indicated in the response of the opinion or request for modification as a dialogue content regarding the new employment requirements, and inputting the updated first prompt including the dialogue content regarding the new employment requirements into the first generative model to obtain the revised proposed employment requirements; When the first user approves the proposed employment requirements, the proposed employment requirements include predetermined instruction content regarding the evaluation criteria and dialogue content regarding the evaluation criteria among the dialogue information. second A prompt is input to the first generative model, second obtaining an evaluation viewpoint proposal corresponding to the employment requirement proposal as an output result of the first generative model in response to the prompt; outputting the proposed evaluation viewpoint generated by the first generative model, accepting input of a corrective or supplemental response to the proposed evaluation viewpoint from the first user, extracting the utterance content indicated in the corrective or supplemental response as dialogue content related to the new evaluation viewpoint, inputting the updated second prompt including the dialogue content related to the new evaluation viewpoint into the first generative model, and obtaining the revised proposed evaluation viewpoint; After obtaining the approval of the first user for the evaluation viewpoint proposal, a job interview is conducted, which includes the evaluation viewpoint proposal and predetermined instructions regarding questions to be asked in the job interview. third A prompt is input to the first generative model, third As an output result of the first generative model in response to the prompt, an example question corresponding to the proposed evaluation viewpoint is obtained, and the proposed evaluation viewpoint and the example question corresponding to the proposed employment requirement generated by the first generative model are output. inputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements into a second generative model in which behavior as an interviewer or a job seeker is predefined; outputting voice data of a question included in the example questions selected by the second generative model from a voice output means on a user terminal of a second user conducting the interview; inputting a fourth prompt based on voice data indicating the second user's answer to the question into the second generative model; causing the second generative model to generate a next response in response to the fourth prompt based on the predefined behavior as an interviewer or a job seeker; inputting, during the progress of the interview between the second user and the second generative model, voice data indicating the second user's answer in the interview and a sixth prompt based on predetermined instructions for evaluating whether the second user's answer contains a template expression or a sign of intervention by a third party, in order to determine signs of misconduct; and outputting, as an output result of the second generative model in response to the sixth prompt, a scale indicating the possibility of misconduct by the second user in multiple stages. and obtain a score, and if the score exceeds a predetermined threshold, consider that the second user has a sign of misconduct, and store the second user in association with the answer content in which the possibility of misconduct was detected; after an interview between the second user and the second generative model is completed, input audio data indicating the answer voice of the second user in the interview and a fifth prompt based on the proposed evaluation viewpoint into the second generative model; obtain an evaluation result for the second user as an output result of the second generative model in response to the fifth prompt; extract multiple answer contents from an answer log including answer contents in which the possibility of misconduct by the second user was detected; input a prompt based on predetermined instructions for detecting contradictions in content between previous and next answer contents included in the multiple extracted answer contents into the second generative model; obtain contradictions in content as an output result of the second generative model in response to the prompt; add a correction value to the evaluation result for the second user according to the number of flags assigned to the contradictions; and output the evaluation result after adding the correction value. .

[0008] With the above configuration, the information processing system of the first aspect can use a first generation model that can interact with the user to assist in creating a requirements definition for the type of person a recruiting company is looking for.

[0010] With the above configuration, oneAccording to the information processing system of this aspect, by having a second generation model that can act as an interviewer or a job seeker handle the interview, a flexible interview environment can be realized that does not require face-to-face real-time contact, making it possible to conduct advance training through mock interviews and reducing the burden of scheduling during the actual interview.

[0012] With the above configuration, one According to the information processing system of this aspect, by having the second generative model generate evaluation results after the interview is completed, it is possible to unify the evaluation perspective and achieve consistent interview evaluation that is not dependent on subjectivity or experience.

[0014] With the above configuration, one According to the information processing system of this aspect, signs of fraudulent activity are detected using the second generative model while the interview is in progress, and information on the relevant second user is recorded, thereby contributing to the detection of fraud and enabling highly reliable employment evaluation.

[0015] No. two The information processing system of the present embodiment is one In an information processing system of the above aspect, in an employment support service that handles a plurality of job information items and a plurality of job seeker information items, the processor notifies a recruiting company to which the second user has applied of identification information regarding the second user who has been determined to have signs of fraudulent activity, along with a notice that the signs of fraudulent activity have been determined.

[0016] With the above configuration, two According to the information processing system of this aspect, by notifying the company to which the applicant has applied of the identification information of the second user in whom signs of fraudulent activity have been detected, the company can take into consideration the presence or absence of fraud when making a hiring decision, thereby reducing the risk of the company hiring inappropriate personnel.

[0017] No. three The information processing system of the present embodiment is oneIn the information processing system of the aspect, the processor includes a plurality of job information registered in an employment support service that handles a plurality of job information and a plurality of job seeker information, and an evaluation result for the second user. Seventh inputting a prompt into the second generative model; Seventh As an output result of the second generative model in response to the prompt, a list of companies hiring that are a good match for the second user is obtained, the second user is notified of the list of companies hiring that are a good match, and each company hiring is notified of identification information regarding the second user.

[0018] With the above configuration, three According to the information processing system of this aspect, by using the second generation model to select compatible recruiting companies based on the evaluation results and job information, it is possible to improve the matching accuracy between the second user and recruiting companies and reduce mismatches after hiring.

[0019] No. four The information processing method of the aspect acquires dialogue information indicating dialogue content between a first user and a first generative model that can interact with the first user regarding the hiring requirements for job seekers that a recruiting company is looking for and evaluation points in a hiring interview, and includes predetermined instruction content regarding the hiring requirements and dialogue content regarding the hiring requirements among the dialogue information. first A prompt is input to the first generative model, first obtaining a proposed hiring requirement as an output result of the first generative model in response to the prompt; outputting the proposed employment requirements generated by the first generative model, accepting input of a response from the first user of an opinion or a request for modification regarding the proposed employment requirements, extracting the speech content indicated in the response of the opinion or request for modification as a dialogue content regarding the new employment requirements, and inputting the updated first prompt including the dialogue content regarding the new employment requirements into the first generative model to obtain the revised proposed employment requirements; When the first user approves the proposed employment requirements, the proposed employment requirements include predetermined instruction content regarding the evaluation criteria and dialogue content regarding the evaluation criteria among the dialogue information. second A prompt is input to the first generative model, second obtaining an evaluation viewpoint proposal corresponding to the employment requirement proposal as an output result of the first generative model in response to the prompt; outputting the proposed evaluation viewpoint generated by the first generative model, accepting input of a corrective or supplemental response to the proposed evaluation viewpoint from the first user, extracting the utterance content indicated in the corrective or supplemental response as dialogue content related to the new evaluation viewpoint, inputting the updated second prompt including the dialogue content related to the new evaluation viewpoint into the first generative model, and obtaining the revised proposed evaluation viewpoint; After obtaining the approval of the first user for the evaluation viewpoint proposal, a job interview is conducted, which includes the evaluation viewpoint proposal and predetermined instructions regarding questions to be asked in the job interview. third A prompt is input to the first generative model, third As an output result of the first generative model in response to the prompt, an example question corresponding to the proposed evaluation viewpoint is obtained, and the proposed evaluation viewpoint and the example question corresponding to the proposed employment requirement generated by the first generative model are output. inputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements into a second generative model in which behavior as an interviewer or a job seeker is predefined; outputting voice data of a question included in the example questions selected by the second generative model from a voice output means on a user terminal of a second user conducting the interview; inputting a fourth prompt based on voice data indicating the second user's answer to the question into the second generative model; causing the second generative model to generate a next response in response to the fourth prompt based on the predefined behavior as an interviewer or a job seeker; inputting, during the progress of the interview between the second user and the second generative model, voice data indicating the second user's answer in the interview and a sixth prompt based on predetermined instructions for evaluating whether the second user's answer contains a template expression or a sign of intervention by a third party, in order to determine signs of misconduct; and outputting, as an output result of the second generative model in response to the sixth prompt, a scale indicating the possibility of misconduct by the second user in multiple stages. and obtain a score, and if the score exceeds a predetermined threshold, consider that the second user has a sign of misconduct, and store the second user in association with the answer content in which the possibility of misconduct was detected; after an interview between the second user and the second generative model is completed, input audio data indicating the answer voice of the second user in the interview and a fifth prompt based on the proposed evaluation viewpoint into the second generative model; obtain an evaluation result for the second user as an output result of the second generative model in response to the fifth prompt; extract multiple answer contents from an answer log including answer contents in which the possibility of misconduct by the second user was detected; input a prompt based on predetermined instructions for detecting contradictions in content between previous and next answer contents included in the multiple extracted answer contents into the second generative model; obtain contradictions in content as an output result of the second generative model in response to the prompt; add a correction value to the evaluation result for the second user according to the number of flags assigned to the contradictions; and output the evaluation result after adding the correction value. The processing is performed by a computer.

[0020] With the above configuration, four According to the information processing method of this aspect, by using a first generative model that is capable of interacting with a user, it is possible to support the creation of a requirements definition for the type of person a recruiting company is looking for.

[0021] No. Five The information processing program of the aspect acquires dialogue information indicating dialogue content regarding the hiring requirements of job seekers sought by a recruiting company and evaluation points in a hiring interview between a first user and a first generative model that can interact with the first user, and includes predetermined instruction content regarding the hiring requirements and dialogue content regarding the hiring requirements among the dialogue information. first A prompt is input to the first generative model, first obtaining a proposed hiring requirement as an output result of the first generative model in response to the prompt; outputting the proposed employment requirements generated by the first generative model, accepting input of a response from the first user of an opinion or a request for modification regarding the proposed employment requirements, extracting the speech content indicated in the response of the opinion or request for modification as a dialogue content regarding the new employment requirements, and inputting the updated first prompt including the dialogue content regarding the new employment requirements into the first generative model to obtain the revised proposed employment requirements; When the first user approves the proposed employment requirements, the proposed employment requirements include predetermined instruction content regarding the evaluation criteria and dialogue content regarding the evaluation criteria among the dialogue information. second A prompt is input to the first generative model, second obtaining an evaluation viewpoint proposal corresponding to the employment requirement proposal as an output result of the first generative model in response to the prompt; outputting the proposed evaluation viewpoint generated by the first generative model, accepting input of a corrective or supplemental response to the proposed evaluation viewpoint from the first user, extracting the utterance content indicated in the corrective or supplemental response as dialogue content related to the new evaluation viewpoint, inputting the updated second prompt including the dialogue content related to the new evaluation viewpoint into the first generative model, and obtaining the revised proposed evaluation viewpoint; After obtaining the approval of the first user for the evaluation viewpoint proposal, a job interview is conducted, which includes the evaluation viewpoint proposal and predetermined instructions regarding questions to be asked in the job interview. third A prompt is input to the first generative model, third As an output result of the first generative model in response to the prompt, an example question corresponding to the proposed evaluation viewpoint is obtained, and the proposed evaluation viewpoint and the example question corresponding to the proposed employment requirement generated by the first generative model are output. inputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements into a second generative model in which behavior as an interviewer or a job seeker is predefined; outputting voice data of a question included in the example questions selected by the second generative model from a voice output means on a user terminal of a second user conducting the interview; inputting a fourth prompt based on voice data indicating the second user's answer to the question into the second generative model; causing the second generative model to generate a next response in response to the fourth prompt based on the predefined behavior as an interviewer or a job seeker; inputting, during the progress of the interview between the second user and the second generative model, voice data indicating the second user's answer in the interview and a sixth prompt based on predetermined instructions for evaluating whether the second user's answer contains a template expression or a sign of intervention by a third party, in order to determine signs of misconduct; and outputting, as an output result of the second generative model in response to the sixth prompt, a scale indicating the possibility of misconduct by the second user in multiple stages. and obtain a score, and if the score exceeds a predetermined threshold, consider that the second user has a sign of misconduct, and store the second user in association with the answer content in which the possibility of misconduct was detected; after an interview between the second user and the second generative model is completed, input audio data indicating the answer voice of the second user in the interview and a fifth prompt based on the proposed evaluation viewpoint into the second generative model; obtain an evaluation result for the second user as an output result of the second generative model in response to the fifth prompt; extract multiple answer contents from an answer log including answer contents in which the possibility of misconduct by the second user was detected; input a prompt based on predetermined instructions for detecting contradictions in content between previous and next answer contents included in the multiple extracted answer contents into the second generative model; obtain contradictions in content as an output result of the second generative model in response to the prompt; add a correction value to the evaluation result for the second user according to the number of flags assigned to the contradictions; and output the evaluation result after adding the correction value. , and have the computer execute the processing.

[0022] With the above configuration, Five According to the information processing program of this aspect, by using a first generative model that is capable of interacting with a user, it is possible to support the creation of a requirements definition for the type of person a recruiting company is looking for. [Effects of the Invention]

[0023] As described above, the information processing system, information processing method, and information processing program disclosed herein can assist in creating a requirements definition for the type of person a recruiting company is looking for by using a first generation model that can interact with the user. [Brief explanation of the drawings]

[0024] [Figure 1] FIG. 1 is a diagram illustrating an example of a schematic configuration of a recruitment activity support system. [Figure 2] FIG. 2 is a block diagram showing the hardware configuration of a server, a hiring company terminal, and a job seeker terminal. [Figure 3] FIG. 2 is a block diagram showing an example of various information stored in the storage of the server. [Figure 4] 10 is a flowchart showing the flow of a first support process for supporting recruitment activities before a recruitment interview. [Figure 5] FIG. 10 is a diagram showing an example of a talk screen displayed on the display unit of the hiring company terminal. [Figure 6] 10 is a flowchart showing the flow of a second support process for supporting recruitment activities during and after a recruitment interview. [Figure 7] FIG. 10 is a diagram showing an example of an interview screen displayed on the display unit of the job seeker terminal. [Figure 8] FIG. 10 is a diagram showing an example of a result screen displayed on a display unit of a job seeker terminal. [Figure 9] 10 is a flowchart showing the flow of a determination process for determining signs of misconduct by a job seeker during an interview. [Figure 10] 10 is a flowchart showing the flow of a generation process for generating a list of hiring companies that are a good match for a job seeker. DETAILED DESCRIPTION OF THE INVENTION

[0025] The following describes the recruitment activity support system 10 according to this embodiment. The recruitment activity support system 10 is a system that has functions to support the creation of a requirement definition for the type of person a recruiting company is looking for, and to conduct mock interviews or actual interviews with interviewers or job seekers.

[0026] (First embodiment) First, a first embodiment of the recruitment activity support system 10 will be described.

[0027] FIG. 1 is a diagram showing an example of a schematic configuration of a recruitment activity support system 10. As shown in FIG. As shown in Fig. 1, the recruitment activity support system 10 includes a server 20, a hiring company terminal 40, and a job seeker terminal 60. The server 20, the hiring company terminal 40, and the job seeker terminal 60 are connected in a state where they can communicate with each other via a network N. The network N may be, for example, the Internet, a LAN (Local Area Network), or a WAN (Wide Area Network).

[0028] Server 20 is a server computer that executes various processes related to the recruitment activity support system 10. The recruitment activity support system 10 is configured to provide an employment support service that handles multiple job information items and multiple job seeker information items, and users, such as recruiting companies or job seekers, can use various functions by accessing the employment support service. Note that server 20 may be configured as a single server computer, or may have a distributed configuration in which multiple server computers are linked together.

[0029] Recruiting company terminal 40 is a terminal used by a user (hereinafter referred to as "recruiting user") who belongs to a recruiting company and is one of the users of the employment support service. Job seeker terminal 60 is a terminal used by a job seeker and is one of the users of the employment support service. As an example, in this embodiment, the recruiting company terminal 40 and the job seeker terminal 60 are both configured as "PCs (Personal Computers)." Note that while the figure shows one recruiting company terminal 40 and one job seeker terminal 60, in reality there may be multiple recruiting company terminals 40 and multiple job seeker terminals 60 depending on the number of users of the employment support service.

[0030] 2 is a block diagram showing the hardware configuration of server 20, hiring company terminal 40, and job seeker terminal 60. Note that since server 20, hiring company terminal 40, and job seeker terminal 60 basically have a general computer configuration, server 20 will be used as a representative for the explanation.

[0031] 2, the server 20 includes a CPU (Central Processing Unit) 21, a ROM (Read Only Memory) 22, a RAM (Random Access Memory) 23, a storage 24, an input unit 25, a display unit 26, and a communication unit 27. Each component is connected to each other via a bus 28 so as to be able to communicate with each other.

[0032] The CPU 21 is a central processing unit that executes various programs and controls each part. That is, the CPU 21 reads programs from the ROM 22 or the storage 24 and executes the programs using the RAM 23 as a work area. The CPU 21 controls each of the above components and performs various arithmetic processing in accordance with the programs stored in the ROM 22 or the storage 24. The CPU 21 is an example of a "processor" in the present disclosure.

[0033] The ROM 22 stores various programs and various data. The RAM 23 serves as a working area for temporarily storing programs or data.

[0034] The storage 24 is configured by a storage device such as a hard disk drive (HDD), a solid state drive (SSD), or a flash memory, and stores various programs and various data.

[0035] The input unit 25 includes, for example, a keyboard, a mouse, various buttons, a microphone, a camera, and the like, and is used to perform various inputs.

[0036] The display unit 26 is, for example, a liquid crystal display, and displays various information. The display unit 26 may function as the input unit 25 by adopting a touch panel system.

[0037] The communication unit 27 is an interface for communicating with other devices. For this communication, a wired communication standard such as Ethernet (registered trademark) or FDDI, or a wireless communication standard such as 4G, 5G, or Wi-Fi (registered trademark) is used.

[0038] The functions of the CPU 41, ROM 42, RAM 43, storage 44, input unit 45, display unit 46, communication unit 47, and bus 48 of the recruiting company terminal 40, and the CPU 61, ROM 62, RAM 63, storage 64, input unit 65, display unit 66, communication unit 67, and bus 68 of the job seeker terminal 60 have the same functional configuration as the CPU 21, ROM 22, RAM 23, storage 24, input unit 25, display unit 26, communication unit 27, and bus 28 of the server 20 described above.

[0039] Fig. 3 is a block diagram showing an example of various information stored in the storage 24 of the server 20. As shown in Fig. 3, the storage 24 stores an information processing program 30 and a generative model 31.

[0040] The information processing program 30 is a program for causing the CPU 21 to execute various processes described below. When executing the information processing program 30, the server 20 executes processing based on the information processing program 30 using the hardware resources shown in Fig. 2. The information processing program 30 is an example of an "information processing program" in the present disclosure.

[0041] The generative model 31 is a so-called generative AI. An example of the generative model 31 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ) and Gemini (Internet Search <url: https: gemini.google.com ?hl="ja">) and other generative AIs. The generative model 31 is configured to realize processing that can be executed by various known generative AIs. The generative model 31 is obtained by causing a neural network to perform deep learning. A prompt containing an instruction is input to the generative model 31, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The generative model 31 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data, graph data, table data, image data, and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0042] In this embodiment, two types of generative models 31 are provided: a first generative model 31A and a second generative model 31B.

[0043] The first generative model 31A, among the various functions provided by the recruitment activity support system 10, performs processing to support the creation of a requirement definition for the type of person a recruiting company is seeking. The second generative model 31B, which has predefined behaviors as an interviewer or a job seeker, performs processing to conduct mock interviews or actual interviews with an interviewer or a job seeker, among the various functions. Here, "interviewer behavior" in the second generative model 31B refers to style information such as "asking logical, probing questions," "speaking in a calm tone," or "responding concisely and without emotion." These are specified using the initial prompt or system role of the generation AI, and the response behavior is controlled. Furthermore, "job seeker behavior" in the second generative model 31B refers to style information such as "answering frankly based on experience," "stating a consistent career history and reasons for applying," or "responding in a polite tone." Similarly, these are specified using the initial prompt or system role. Details of the processes performed by the first generative model 31A and the second generative model 31B will be described later. The first generative model 31A is an example of the "first generative model" of the present disclosure, and the second generative model 31B is an example of the "second generative model" of the present disclosure.

[0044] 4 is a flowchart showing the flow of a first support process that supports recruitment activities before a job interview using a generative model 31. The first support process is performed by the CPU 21 reading the information processing program 30 from the storage 24, expanding it into the RAM 23, and executing it. As an example, the first support process is started when a recruiting user logs in to the employment support service and performs a predetermined operation for conducting recruitment activities before a job interview, such as creating a requirement definition for a hiring profile.

[0045] In step S10 shown in FIG. 4, the CPU 21 acquires dialogue information indicating the content of a dialogue between the recruiting user and the first generative model 31A regarding the hiring requirements for job seekers sought by the recruiting company and the evaluation criteria for job interviews. Here, "dialogue information" refers to information consisting of text data input by the recruiting user and response text output by the first generative model 31A on a chat screen provided by the employment support service. The dialogue information is structured, for example, as log information that chronologically records the input of the recruiting user and the output of the first generative model 31A, and is managed in JSON format or the like. The CPU 21 acquires, as dialogue information, structured dialogue data including the order of utterances made on the chat screen, speaker identification (the recruiting user or the first generative model 31A), and the content of each utterance. The CPU 21 then proceeds to step S11. The recruiting user is an example of a "first user" in the present disclosure.

[0046] In step S11, CPU 21 generates a first prompt including predetermined instruction content related to the employment requirements (hereinafter, "employment requirement instruction") and the dialogue content related to the employment requirements from the dialogue information acquired in step S10. Then, CPU 21 proceeds to step S12.

[0047] Here, the "recruitment requirement instructions" are, for example, instructions such as "Please specify the desired person profile" or "Please list the skills and experience required for performing the job," and are guide information that indicates the format and perspective of the output content desired from the first generative model 31A. Furthermore, the CPU 21, for example, extracts "dialogue content related to the recruitment requirements" from the dialogue information based on a context analysis process and a keyword extraction process. As a result, the CPU 21 generates, for example, a first prompt such as the following:

[0048] <First prompt> "You are an advisor knowledgeable in recruiting. Below is a dialogue with a recruiting user. From this dialogue, please organize the requirements for the type of person you want to hire. [Dialogue content] User: For this position, it would be ideal if you have experience in corporate sales. User: Self-motivated, preferably with experience in the IT industry. →Please clarify the type of person you are looking for in terms of skills, experience, and personality.

[0049] In step S12, the CPU 21 inputs the first prompt generated in step S11 into the first generative model 31A and obtains proposed hiring requirements as an output result of the first generative model 31A in response to the first prompt. Here, the "proposed hiring requirements" refers to output information that systematically organizes and extracts the profile and selection criteria sought by the hiring company from the perspectives of skills, experience, personality, etc., based on a dialogue between the hiring user and the first generative model 31A. The proposed hiring requirements include content that clearly expresses the profile of the job seeker that the hiring company values, such as "three or more years of corporate sales experience," "business communication skills in English," or "a person who can be expected to be both proactive and cooperative." The CPU 21 then proceeds to step S13.

[0050] In step S13, CPU 21 outputs the proposed hiring requirements acquired in step S12 to hiring company terminal 40. In hiring company terminal 40, CPU 41 displays the proposed hiring requirements on display unit 46. CPU 21 then proceeds to step S14.

[0051] In step S14, the CPU 21 determines whether consent has been obtained from the recruiting user for the proposed hiring requirements. If the CPU 21 determines that consent has been obtained (step S14: YES), the process proceeds to step S15. On the other hand, if the CPU 21 determines that consent has not been obtained (step S14: NO), the process returns to step S10. As an example, the CPU 21 determines that consent has been obtained from the recruiting user based on the operation of the consent button 56 (see FIG. 5) on the talk screen described below.

[0052] In step S15, the CPU 21 acquires the conversation information, and then the CPU 21 proceeds to step S16.

[0053] In step S16, CPU 21 generates a second prompt including predetermined instruction content related to the evaluation viewpoint (hereinafter, "evaluation viewpoint instruction") and the dialogue content related to the evaluation viewpoint from the dialogue information acquired in step S15. Then, CPU 21 proceeds to step S17.

[0054] Here, the "evaluation viewpoint instruction" is an instruction such as "Please list the viewpoints for evaluating the job seeker's aptitude" or "Please set evaluation axes related to the job seeker's abilities, personality, and inclinations," and is guide information that indicates the evaluation axes and configuration guidelines for the output content desired from the first generative model 31A. Furthermore, the CPU 21 extracts, as an example, "dialogue content related to evaluation viewpoints" from the dialogue information based on a context analysis process and a keyword extraction process. As a result, the CPU 21 generates, for example, a second prompt such as the following:

[0055] <Second prompt> "You are a human resources consultant with expertise in recruitment. Below is a dialogue with a recruiting user. From this dialogue, please organize the evaluation criteria for evaluating job applicants. [Dialogue content] User: Sales skills are important, of course, but at our company, it's important to be able to identify problems and take action on your own. User: There is a lot of teamwork involved, so cooperation is also essential. →Please list the criteria for evaluating job applicants in terms of skills, experience, and personality.

[0056] In step S17, the CPU 21 inputs the second prompt generated in step S16 into the first generation model 31A and obtains a proposed evaluation viewpoint corresponding to the proposed hiring requirements as an output result of the first generation model 31A in response to the second prompt. Here, the "proposed evaluation viewpoint" refers to information that specifies and structures viewpoints and criteria for multifaceted evaluation of a job seeker's aptitude from the perspectives of skills, personality, orientation, etc., based on the proposed hiring requirements approved by the hiring user. The proposed evaluation viewpoint includes evaluation criteria such as "logical thinking ability," "teamwork orientation," "problem-finding ability," "interpersonal negotiation ability," or "motivation for self-improvement." Then, the CPU 21 proceeds to step S18.

[0057] In step S18, CPU 21 outputs the proposed evaluation viewpoints acquired in step S17 to hiring company terminal 40. In said hiring company terminal 40, CPU 41 causes said proposed evaluation viewpoints to be displayed on display unit 46. CPU 21 then proceeds to step S19.

[0058] In step S19, the CPU 21 determines whether consent has been obtained from the recruiting user for the proposed evaluation viewpoint. If the CPU 21 determines that consent has been obtained (step S19: YES), the process proceeds to step S20. On the other hand, if the CPU 21 determines that consent has not been obtained (step S19: NO), the process returns to step S15. As an example, the CPU 21 determines that consent has been obtained from the recruiting user based on the operation of the consent button 56 on the talk screen.

[0059] In step S20, CPU 21 generates a third prompt including the proposed evaluation viewpoints approved by the recruiting user and predetermined instructions regarding questions to be asked in the job interview (hereinafter, "question instructions"). Then, CPU 21 proceeds to step S21.

[0060] Here, the "question content instruction" is an instruction such as "Please create specific questions to confirm the abilities and aptitudes of the job seeker based on each evaluation perspective" or "Please format the questions so that behavioral characteristics can be grasped," and is guide information that specifies the structure and purpose of the question to be output to the first generation model 31A. As a result, the CPU 21 generates a third prompt such as the following:

[0061] <Third prompt> "You are a human resources advisor who is knowledgeable about creating questions for job interviews. Below are some suggested evaluation criteria. For each of these suggested evaluation criteria, please create questions that will bring out the abilities and behavioral characteristics of job seekers. [Evaluation criteria] ·Logical thinking ability Problem-solving ability ·Cooperation ·Willingness for self-growth →Please output one question for each evaluation perspective.

[0062] In step S21, the CPU 21 inputs the third prompt generated in step S20 into the first generation model 31A and obtains example questions corresponding to the proposed evaluation viewpoints as output results of the first generation model 31A in response to the third prompt. Here, the "example questions" are natural language questions generated for the purpose of grasping the job seeker's abilities, personality, inclinations, etc., corresponding to each item in the proposed evaluation viewpoints, and include questions in the form of prompting behavioral examples, such as "Tell us about an experience in the past where you found and solved a problem on your own" or "How did you act when there was a conflict of opinion in a team?" Then, the CPU 21 proceeds to step S22.

[0063] In step S22, the CPU 21 outputs the proposed evaluation criteria and sample questions corresponding to the proposed hiring requirements acquired in step S21 in a predetermined data structure suitable for subsequent processing or display. Here, the predetermined data structure is a hierarchically structurable format, such as JSON, that allows each evaluation criteria and its corresponding question to be stored in a one-to-one association. The CPU 21 then terminates the first support process. The CPU 21 then saves the proposed evaluation criteria and sample questions corresponding to the output hiring requirements in storage 24 in a structure that maintains their mutual association. In the subsequent second support process, these saved data are compared with information on the job seeker or interviewer to obtain proposed evaluation criteria and sample questions suitable for the progress of the interview, and are used, for example, as questions to be presented to the job seeker.

[0064] 5 is a diagram showing an example of a talk screen displayed on the display unit 46 of the hiring company terminal 40. On the talk screen, a dialogue regarding the creation of hiring requirements and evaluation perspectives takes place between the hiring company user and the first generation model 31A.

[0065] In the figure, speech bubbles 50, 52, and 54 showing the speech content of the first generation model 31A and speech bubbles 51 and 53 showing the speech content of the recruiting user are displayed. For the sake of convenience, the specific speech content in each of the speech bubbles 50 to 54 is omitted.

[0066] In addition, at the bottom of the talk screen, there are provided an input field 55 where the recruiting user can input the content of the utterance via the input unit 45, and an approval button 56 for performing an approval operation on the generated proposed hiring requirements, etc. Below, we will explain the displayed content when creating the proposed hiring requirements, followed by the displayed content when creating the proposed evaluation criteria.

[0067] <Contents to be displayed when drafting recruitment requirements> When drafting employment requirements, the first speech bubble 50 displays instructions on employment requirements, such as "Please specify the type of person you are looking for."

[0068] The recruiting user inputs a response to the utterance content shown in the balloon 50 in the input field 55 and confirms the input content. Based on this confirmation operation, the response content is displayed in the balloon 51.

[0069] The CPU 21 of the server 20 generates a first prompt based on the speech content shown in the speech bubbles 50 and 51, and inputs the first prompt to the first generative model 31A to acquire a proposed employment requirement. The acquired proposed employment requirement is displayed as the speech content of the speech bubble 52.

[0070] Furthermore, when the recruiting user enters a response such as an opinion or a request for correction in response to the proposed employment requirements displayed in speech bubble 52 in input field 55 and performs a confirmation operation, the response content is displayed in speech bubble 53.

[0071] The CPU 21 extracts the utterance content shown in the speech bubble 53 as a new “dialogue content regarding the hiring requirements,” and inputs the updated first prompt again into the first generative model 31A to obtain a revised proposed hiring requirements. This revised proposed hiring requirements is displayed as the utterance content in the speech bubble 54.

[0072] Thereafter, the recruiting user operates the accept button 56, thereby accepting the revised recruitment requirement proposal. In this manner, until the accept button 56 is operated, a dialogue is continuously carried out between the recruiting user and the first generative model 31A through the content of utterances, and the recruitment requirement proposal is gradually refined.

[0073] <Operations when creating evaluation perspective proposals> When creating a draft of evaluation criteria, an evaluation criteria instruction such as "Please list the criteria for evaluating the job seeker's aptitude" is displayed in the first speech bubble 50.

[0074] The recruiting user inputs a response to the utterance content shown in the balloon 50 in the input field 55 and confirms the input content. Based on this confirmation operation, the response content is displayed in the balloon 51.

[0075] The CPU 21 generates a second prompt based on the utterance content shown in the speech bubbles 50 and 51, and inputs the second prompt to the first generative model 31A to acquire a proposed evaluation viewpoint. The acquired proposed evaluation viewpoint is displayed as the utterance content of the speech bubble 52.

[0076] Furthermore, if the recruiting user wishes to make corrections or additions to the proposed evaluation criteria shown in speech bubble 52, they can enter the relevant content in input field 55 and perform a confirmation operation, and the response content will be displayed in speech bubble 53.

[0077] The CPU 21 extracts the utterance content shown in the speech bubble 53 as a new "dialogue content related to the evaluation viewpoint," and inputs the updated second prompt again into the first generative model 31A to obtain a revised proposed evaluation viewpoint. This revised proposed evaluation viewpoint is displayed as the utterance content in the speech bubble 54.

[0078] Thereafter, the revised evaluation viewpoint proposal is deemed to have been accepted by the recruiting user operating the accept button 56. In this manner, until the accept button 56 is operated, a dialogue is continuously carried out between the recruiting user and the first generative model 31A through the content of the utterances, and the evaluation viewpoint proposal is gradually refined.

[0079] The CPU 21 inputs the generated third prompt into the first generation model 31A based on the acceptance operation by the employing user, and acquires example questions corresponding to the proposed evaluation viewpoints accepted by the employing user. Furthermore, the CPU 21 saves the example questions corresponding to the acquired proposed evaluation viewpoints in the storage 24 in a structure that maintains their mutual relevance.

[0080] FIG. 6 is a flowchart showing the flow of a second support process that uses a generative model 31 to support recruiting activities during and after a job interview. The second support process is performed by the CPU 21 reading the information processing program 30 from the storage 24, expanding it into the RAM 23, and executing it. As an example, the second support process is initiated when a job seeker who is the interviewee or an interviewer belonging to a recruiting company logs in to the employment support service and performs a predetermined operation to conduct a mock interview or a real interview. When the predetermined operation is performed, the CPU 21 distributes a uniform resource locator (URL) for the interview to the terminal used by the interviewee (i.e., the recruiting company terminal 40 or the job seeker terminal 60). In the following, as an example, the interviewee is referred to as a "job seeker," and a case where the job seeker conducts a real interview will be described. In other words, in the following, the job seeker is an example of a "second user" in the present disclosure.

[0081] In step S30 shown in Fig. 6, the CPU 21 acquires job seeker information managed by the employment support service, corresponding to the job seeker to be the subject of the actual interview. The job seeker information includes, for example, the job seeker's name, ID, the position for which they are applying, interview history, desired job type, and aptitude test results. The employment support service also holds job information registered by companies seeking employees, and the job information includes the job content of the position, the desired profile, the selection flow, the number of job openings, and the work location. The CPU 21 then proceeds to step S31.

[0082] In step S31, the CPU 21 identifies proposed employment requirements suitable for the job seeker based on the job seeker information acquired in step S30, and acquires proposed evaluation viewpoints and example questions associated with the proposed employment requirements from the storage 24. For example, the CPU 21 may use a search key corresponding to the job position or job type for which the job seeker is applying to identify a corresponding set of proposed evaluation viewpoints and example questions from among multiple proposed employment requirements stored in the storage 24. The CPU 21 inputs the identified proposed evaluation viewpoints and example questions into the second generation model 31B for use in the subsequent interview dialogue processing. The CPU 21 then proceeds to step S32.

[0083] In step S32, the CPU 21 outputs audio data of the question included in the example question selected by the second generation model 31B from a speaker on the job seeker terminal 60 of the job seeker. Here, the second generation model 31B selects a specific question from the acquired example questions based on predefined interviewer behavior. The speaker on the job seeker terminal 60 may be a speaker built into the job seeker terminal 60 or an external speaker connected to the job seeker terminal 60. Note that since the actual interview is conducted as an online interview via a web conference system, the CPU 21 may output the audio data of the question to the job seeker terminal 60 using, for example, a streaming method that transmits audio in real time, or a method that transmits an audio file in advance and issues a playback instruction at a predetermined timing. Then, the CPU 21 proceeds to step S33. The job seeker terminal 60 is an example of a "user terminal" in the present disclosure, and the speaker is an example of an "audio output means" in the present disclosure.

[0084] In step S33, the CPU 21 acquires voice data indicating the job seeker's voice response to the question output in step S32. The voice data may be collected by a microphone provided in the job seeker terminal 60 or an externally connected microphone. The CPU 21 generates a fourth prompt including the acquired voice data. The CPU 21 then proceeds to step S34.

[0085] For example, the CPU 21 generates a fourth prompt such as the following: <Fourth prompt> You are the interviewer conducting the actual interview. Below are the responses the job seeker gave to your questions. [Question]: Tell us about an experience where you solved a difficult problem as a team. [Answer]: In my previous job, when a problem occurred within the team, we would have multiple discussions with the team members... (omitted) Please pose the following appropriate questions based on this answer to assess logical thinking and leadership skills.

[0086] In step S34, the CPU 21 inputs the fourth prompt generated in step S33 into the second generation model 31B and obtains the next response as the output result of the second generation model 31B in response to the fourth prompt. Here, the "next response" refers to content including feedback utterances, confirmation questions, or probing questions generated in accordance with the interviewer's style in response to the job seeker's answer. The "next response" refers to utterance content constituting a continuous dialogue, such as "What specific role did you play at that time?" or "Please tell us what you learned from that experience." Then, the CPU 21 proceeds to step S35.

[0087] In step S35, CPU 21 generates voice data for the next response acquired in step S34, applying a voice style based on the predefined behavior of an interviewer, and outputs the voice data from the speaker on the job seeker terminal 60. Then, CPU 21 proceeds to step S36.

[0088] In step S36, the CPU 21 determines whether the interview has ended. If the CPU 21 determines that the interview has ended (step S36: YES), the process proceeds to step S37. On the other hand, if the CPU 21 determines that the interview has not ended (step S36: NO), the process returns to step S33. As an example, the CPU 21 determines that the interview has ended when the job seeker operates the end button 75 (see FIG. 7) on an interview-in-progress screen, which will be described later, or when a predetermined interview time has elapsed.

[0089] In step S37, the CPU 21 generates a fifth prompt based on speech data representing the job seeker's response speech in the actual interview conducted between the job seeker and the second generation model 31B and the proposed evaluation viewpoints used in the actual interview. Then, the CPU 21 proceeds to step S38.

[0090] For example, CPU 21 generates the following fifth prompt:

[0091] <Fifth prompt> You are a recruiting expert. Below is a log of a conversation from a real interview with a job seeker. Please rate each response in this conversation based on the following criteria: [Evaluation criteria] ·Logical thinking ability ·Team collaboration ·Willingness for self-growth [Answer 1]: "At my previous job, when I was unsure about the direction of a project... (omitted)" [Answer 2]: "It's especially during difficult times that I try to speak up to my teammates... (omitted)" Please use this information to evaluate the job applicant for each evaluation criteria.

[0092] In step S38, the CPU 21 inputs the fifth prompt generated in step S37 into the second generation model 31B and obtains an evaluation result for the job seeker as the output result of the second generation model 31B in response to the fifth prompt. Here, the "evaluation result" refers to the result of evaluating the job seeker for each of the multiple evaluation criteria set based on the job seeker's answers in the actual interview. The "evaluation result" includes, for example, score information for each evaluation criteria, a judgment rank, comments for improvement, or comments regarding overall suitability for employment. The CPU 21 then proceeds to step S39.

[0093] In step S39, CPU 21 outputs the evaluation results acquired in step S38 to hiring company terminal 40, which will make a hiring decision on the job seeker based on the evaluation results, and to job seeker terminal 60 of the job seeker. This allows the hiring company to check the interview results and make a hiring decision, and allows the job seeker to check their own evaluation results. CPU 21 then ends the second support process.

[0094] Fig. 7 is a diagram showing an example of an interview screen displayed on the display unit 66 of the job seeker terminal 60. As an example, on the interview screen shown in Fig. 7, a live interview is being conducted in real time between the job seeker and the second generation model 31B.

[0095] On the interview screen, an interviewer image 70, a job seeker image 71, elapsed time information 72, end time information 73, a group of buttons 74, and an end button 75 are displayed.

[0096] The interviewer image 70 is a visual representation of the interviewer, and may be, for example, a facial image of a person virtually generated by the second generative model 31B. This allows the job seeker to have the experience of interviewing with a real person.

[0097] The job applicant image 71 is a facial image of the job applicant captured in real time by a camera (not shown), and may be configured to be used for feedback to the second generative model 31B or for recording.

[0098] The elapsed time information 72 indicates the time elapsed since the start of the interview, and the end time information 73 indicates the scheduled end time of the actual interview. This information allows the job seeker to understand the progress of the interview.

[0099] The button group 74 is made up of multiple buttons for operating a GUI (Graphical User Interface) that is generally provided in a Web conference system. For example, the button group 74 includes a button for switching the camera on / off, a button for switching the audio on / off, etc.

[0100] The end button 75 is an operation button for ending the actual interview at a timing of the job seeker's choosing, and the CPU 21 determines the end of the interview based on the operation of the end button 75.

[0101] Fig. 8 is a diagram showing an example of a results screen displayed on display unit 46 of hiring company terminal 40. As an example, the results screen shown in Fig. 8 displays feedback information including the evaluation results for the job seeker as a result of the actual interview carried out on the interviewing screen shown in Fig. 7.

[0102] The results screen displays summary information 80, audio information 81, results information 85, and a memo field 86.

[0103] Summary information 80 displays the main points of the entire actual interview. Specifically, summary information 80 includes basic information such as the name of the job seeker who is being interviewed and the job type applied for, an overall evaluation of the evaluation results, and the job seeker's strengths and areas for improvement. For ease of explanation, the specific displayed content has been omitted from the figure.

[0104] The voice information 81 indicates information related to the job seeker's voice responses during the interview. The audio information 81 includes an audio playback unit 82, and by using a slider and various operation buttons provided in the audio playback unit 82, it is possible to play back the answer audio at any point during the actual interview.

[0105] Furthermore, questions and answers corresponding to each scene in the interview are organized in text format and displayed in the audio information 81. In Fig. 8, scene information 83 and scene information 84 are examples of such information.

[0106] The corresponding questions and answers are displayed in text form in the scene information 83 and 84. For ease of explanation, the specific questions and answers displayed in the scene information 83 and 84 are not shown in the drawings.

[0107] Here, tag 83A is displayed on the right edge of scene information 83, and tag 84A is displayed on the right edge of scene information 84. Tags 83A and 84A are unique tag information assigned based on the question content and answer content corresponding to each of scene information 83 and 84. These tags are generated by CPU 21, for example, by extracting specific keywords contained in the answer content or by performing a categorization process based on the context of the dialogue. The unique tags assigned in this manner visualize the content of utterances during the interview by topic and function as important auxiliary information for subsequent evaluation and analysis. For example, tag 83A displays "Java (registered trademark)," indicating that "Java" was mentioned in the answer content corresponding to scene information 83. Similarly, tag 84A displays "remote work," indicating that "remote work" was a major topic in the answer content corresponding to scene information 84.

[0108] The result information 85 displays a summary of the evaluation results of the actual interviews for the job seeker. The result information 85 includes score information 85A, judgment information 85B, and behavioral information 85C. For ease of explanation, the specific content displayed in the score information 85A, judgment information 85B, and behavioral information 85C is not shown in the figures.

[0109] The score information 85A displays the job seeker's overall score as well as scores for each evaluation aspect (for example, "Leadership: 85 points," "Flexibility: 90 points," etc.).

[0110] The judgment information 85B displays the employment judgment result such as "passed," "failed," or "re-interview recommended," and also includes a comment on the judgment.

[0111] The behavioral information 85C presents the next recommended action based on the results of the actual interview, such as "share with other recruiters" or "set up a re-interview."

[0112] The employer user can determine whether to hire or reject the job seeker based on the score information 85A, the judgment information 85B, and the behavioral information 85C. However, without being limited to this, in some actual interviews, such as the first interview, the hiring or rejection decision of the job seeker may be automatically made based on the evaluation results generated by the second generation model 31B without manual judgment.

[0113] The memo field 86 is an input area where the recruiting user can freely record their own impressions and observations while viewing the results screen.

[0114] The memo field 86 consists of an input field 86A and a save button 86B, and after the recruiting user enters a memo in the input field 86A, the recruiting user can save the memo information in a designated storage area by pressing the save button 86B.

[0115] Note that, while an example of a result screen displayed on the display unit 46 of the hiring company terminal 40 has been described above with reference to Figure 8, the evaluation results generated by the second generation model 31B are also output to the job seeker's job seeker terminal 60, and therefore the same evaluation results are also displayed on the display unit 66 of the job seeker terminal 60. In this case, the content of the evaluation results displayed on the display unit 46 and the display unit 66 may be the same for both, or may differ in part depending on the role. For example, a detailed breakdown of the score and evaluation comments may be displayed for the hiring company, and a summary of the main points may be displayed as feedback for the job seeker.

[0116] As described above, in the recruitment activity support system 10, the CPU 21 inputs a first prompt including instructions for hiring requirements and dialogue content related to the hiring requirements into the first generation model 31A to acquire proposed hiring requirements. Furthermore, if the CPU 21 obtains consent for the proposed hiring requirements from the recruiting user, it inputs a second prompt including instructions for evaluation viewpoints and dialogue content related to the evaluation viewpoints into the first generation model 31A to acquire proposed evaluation viewpoints corresponding to the proposed hiring requirements. Furthermore, after obtaining consent for the proposed evaluation viewpoints from the recruiting user, the CPU 21 inputs a third prompt including the proposed evaluation viewpoints and instructions for question content into the first generation model 31A to acquire example questions corresponding to the proposed evaluation viewpoints. The CPU 21 then outputs the proposed evaluation viewpoints and example questions corresponding to the proposed hiring requirements generated by the first generation model 31A. This enables the recruitment activity support system 10 to comprehensively support processes from defining the profile of the hiring candidate desired by the recruiting company to specifying evaluation axes and automatically generating example questions through natural dialogue with the recruiting user. Therefore, the recruitment activity support system 10 can support the creation of a requirement definition for the type of person to be hired that a recruiting company is looking for by using the first generative model 31A that is capable of dialogue with recruiting users.

[0117] In addition, in the recruitment activity support system 10, the CPU 21 inputs the proposed evaluation perspectives and example questions corresponding to the proposed hiring requirements, generated by the first generative model 31A, to the second generative model 31B. Furthermore, the CPU 21 outputs audio data of the question included in the example question selected by the second generative model 31B from a speaker on the job seeker terminal 60 of the job seeker, who is the interviewee. The CPU 21 then inputs a fourth prompt based on audio data indicating the job seeker's response to the question to the second generative model 31B, and causes the second generative model 31B to generate the next response corresponding to the fourth prompt based on predefined interviewer behavior. Thus, according to the recruitment activity support system 10, by having the second generative model 31B, which can act as an interviewer, handle the interview, a flexible interview environment that does not require face-to-face contact in real time can be realized, thereby reducing the burden of scheduling actual interviews. In particular, the recruitment activity support system 10 eliminates the need to coordinate schedules with employed job seekers based on their working hours, and allows actual interviews to be conducted at any time that suits the job seeker's convenience, thereby contributing to reducing opportunity loss.

[0118] Here, in the above embodiment, we have described a case where the job seeker, who is the interviewee, conducts a real interview, but if the interviewee conducts a mock interview with the second generation model 31B, the following effects are achieved.

[0119] That is, the recruitment activity support system 10 enables advance training through mock interviews by having the second generative model 31B, which can act as an interviewer or a job seeker, respond to an interview. This allows the job seeker to effectively improve their response skills and the interviewer to effectively improve their evaluation skills prior to the actual interview.

[0120] Furthermore, in the recruitment activity support system 10, after an interview between a job seeker and the second generative model 31B is completed, the CPU 21 inputs voice data indicating the job seeker's response voice in the interview and a fifth prompt based on the proposed evaluation viewpoint into the second generative model 31B, and obtains an evaluation result for the job seeker. The CPU 21 then outputs the evaluation result generated by the second generative model 31B. As a result, according to the recruitment activity support system 10, by having the second generative model 31B generate an evaluation result after the interview is completed, it is possible to unify the evaluation viewpoints and achieve consistent interview evaluation that is not dependent on subjectivity or experience.

[0121] (Second embodiment) Next, a second embodiment of the recruitment activity support system 10 will be described while omitting or simplifying parts that overlap with the above embodiment.

[0122] The recruitment activity support system 10 according to the second embodiment differs from the first embodiment in that it performs a judgment process to determine signs of fraudulent behavior by the job seeker while an interview is in progress between the job seeker and the second generation model 31B.

[0123] 9 is a flowchart showing the flow of a determination process for determining signs of fraudulent behavior by a job seeker during an interview using the generative model 31. The determination process is performed by the CPU 21 reading the information processing program 30 from the storage 24, expanding it into the RAM 23, and executing it. As an example, the determination process is started when voice data indicating the job seeker's response voice is acquired in step S33 of the second support process shown in FIG.

[0124] In step S40 shown in Fig. 9, CPU 21 generates a sixth prompt based on the voice data indicating the job seeker's response voice acquired in step S33 of Fig. 6 and predetermined instructions for determining signs of fraudulent activity (hereinafter, "fraud determination instructions"). CPU 21 then proceeds to step S41.

[0125] Here, the "fraud determination instruction" is guide information for specifying the viewpoint and format of the determination result to be output to the second generation model 31B, such as "Please determine whether the following answer content is likely to have been given by a third party or based on a cue card," or "Please evaluate whether the following answer content is likely to have been automatically generated by a generation AI." Based on such a fraud determination instruction, the CPU 21 generates, for example, a sixth prompt such as the following:

[0126] <Sixth prompt> You are an experienced professional interviewer. Below are some responses from a job candidate during an interview. Please rate the responses based on the following criteria to determine whether they are unnatural, typical template expressions, or contain signs of third-party intervention. <Evaluation criteria> Is the content consistent? -Lack of detail or excessive use of vague phrases Is there specificity based on real-life experience? Are there any similar syntax or vocabulary trends in AI-generated sentences? <Answer content> I have demonstrated leadership in every team, taken a bird's-eye view of the whole picture, identified issues, and motivated my team members. I can adapt flexibly to any environment.

[0127] In step S41, the CPU 21 inputs the sixth prompt generated in step S40 into the second generation model 31B, and obtains the possibility of misconduct by the job seeker as an output result of the second generation model 31B in response to the sixth prompt. Here, the possibility of misconduct is a numerical representation of the possibility that the job seeker's answers were answered by a third party, referenced a cue card, or automatically generated by a generation AI. For example, the possibility of misconduct is output as a score from 0 to 100. The CPU 21 then proceeds to step S42.

[0128] In step S42, the CPU 21 determines whether the job seeker shows signs of misconduct. If the CPU 21 determines that there are signs of misconduct (step S42: YES), the process proceeds to step S43. On the other hand, if the CPU 21 determines that there are no signs of misconduct (step S42: NO), the process returns to step S40. As an example, the CPU 21 determines that the job seeker shows signs of misconduct if the likelihood of misconduct acquired in step S41 exceeds a predetermined threshold.

[0129] In step S43, CPU 21 associates the job seeker with the answer content for which the possibility of fraudulent activity was detected, and stores the association information. For example, CPU 21 may be configured to associate identification information (e.g., ID, name, etc.) related to the job seeker with the answer content for which the possibility of fraudulent activity was detected in response to a predetermined question, and store the association information in a fraud detection log area provided in storage 24. CPU 21 then proceeds to step S44.

[0130] In step S44, CPU 21 determines whether the interview has ended. If CPU 21 determines that the interview has ended (step S44: YES), CPU 21 proceeds to step S45. On the other hand, if CPU 21 determines that the interview has not ended (step S44: NO), CPU 21 returns to step S40.

[0131] In step S45, CPU 21 notifies the hiring company to which the job seeker applied of the identification information related to the job seeker determined to have signs of fraudulent activity, along with a notice that signs of fraudulent activity have been determined. The hiring company to which the job seeker applied can be identified, for example, based on the job information and application history registered with the employment support service. As a result, CPU 21 sends notification information to the hiring company, such as "Signs of fraudulent activity have been detected for the job seeker with entry number 123456." This notification information may be provided to the hiring company terminal 40 of the hiring company's staff member by means of a screen display, email transmission, or an alert display on the administrator dashboard. CPU 21 then terminates the determination process.

[0132] As described above, in the recruitment activity support system 10, the CPU 21 inputs audio data indicating the job seeker's responses during an interview and a sixth prompt based on the fraud determination instruction into the second generative model 31B while an interview between the job seeker and the second generative model 31B is in progress, and acquires the possibility of fraudulent activity by the job seeker. If the possibility of fraudulent activity by the job seeker exceeds a predetermined threshold, the CPU 21 associates the job seeker with the response content in which the possibility of fraudulent activity was detected and saves it. As a result, the recruitment activity support system 10 detects signs of fraudulent activity using the second generative model 31B while the interview is in progress and records information about the job seeker in question, which contributes to the detection of fraud and enables highly reliable recruitment evaluations.

[0133] Furthermore, in the recruitment activity support system 10, the CPU 21 notifies the hiring company to which the job seeker applied of the identification information of the job seeker who has been determined to have signs of fraudulent activity in the job hunting support service, along with a notice that the signs of fraudulent activity have been determined. Thus, according to the recruitment activity support system 10, by notifying the hiring company of the identification information of the job seeker who has been detected to have signs of fraudulent activity, the hiring company can take into consideration the presence or absence of fraud when making hiring decisions, thereby reducing the risk of the company hiring inappropriate personnel.

[0134] (Third embodiment) Next, a third embodiment of the recruitment activity support system 10 will be described while omitting or simplifying parts that overlap with the above-described embodiments.

[0135] The recruitment activity support system 10 according to the third embodiment differs from the first and second embodiments in that, after an interview between a job seeker and the second generation model 31B is completed, the second generation model 31B performs a generation process to generate a list of recruiting companies that are a good match for the job seeker from among multiple recruiting companies registered with the employment support service.

[0136] 10 is a flowchart showing the flow of a generation process for generating a list of hiring companies that are compatible with a job seeker using a generation model 31. The generation process is performed by the CPU 21 reading the information processing program 30 from the storage 24, expanding it in the RAM 23, and executing it. As an example, the generation process is started when an evaluation result for the job seeker is obtained in step S38 of the second support process shown in FIG. 6.

[0137] In step S50 shown in Fig. 10, CPU 21 generates a seventh prompt including a plurality of pieces of job information registered with the employment support service and the evaluation result for the job seeker acquired in step S38 shown in Fig. 6. Then, CPU 21 proceeds to step S51.

[0138] Here, the job information includes information such as the job description of the position, the desired profile, the selection process, the number of people being recruited, and the work location, and the CPU 21 combines this information with the evaluation results to generate a seventh prompt, for example, as shown below.

[0139] <Seventh prompt> "You are a career advisor. Below are the results of the job seeker's evaluation and several job openings. Please select five companies that you think would be a good fit for this job seeker and briefly explain why. [Evaluation results]: Logical thinking: 90 points, teamwork: 85 points, leadership: 80 points, preferred work location: Kanto region [Job listing]: (Job A: XX industry, Work location: Tokyo, Desired profile: Leader type)..."

[0140] In step S51, CPU 21 inputs the seventh prompt generated in step S50 into second generation model 31B, and obtains a list of hiring companies that are a good match for the job seeker as an output result of second generation model 31B in response to the seventh prompt. Here, the "list of hiring companies that are a good match for the job seeker" is a group of information on multiple hiring companies that have been scored based on the degree of match between the evaluation results and the job information, and the list may include company names, job types, reasons for matching, recommendation scores, etc. Then, CPU 21 proceeds to step S52.

[0141] In step S52, CPU 21 notifies the job seeker of a list of hiring companies that are a good match for the job seeker. For example, CPU 21 transmits list information indicating the list to job seeker terminal 60 so that the job seeker can check it. CPU 21 also notifies each hiring company included in the list of identification information related to the job seeker (for example, an entry number, etc.), thereby enabling each hiring company to list the job seeker. For example, CPU 21 may be configured to send a notification message including the job seeker's entry number, etc. to hiring company terminal 40 of each hiring company. CPU 21 then terminates the generation process.

[0142] As described above, in the recruitment activity support system 10, the CPU 21 inputs a seventh prompt, which includes multiple pieces of job information registered with the employment support service and the evaluation results for the job seeker, into the second generation model 31B, and obtains a list of hiring companies that are a good fit for the job seeker. The CPU 21 then notifies the job seeker of the list of hiring companies that are a good fit, and notifies each hiring company of identification information related to the job seeker. As a result, the recruitment activity support system 10 can improve the matching accuracy between job seekers and hiring companies and reduce mismatches after hiring by using the second generation model 31B to select hiring companies that are a good fit based on the evaluation results and the job information.

[0143] (others) Although the embodiments of the present disclosure have been described in detail above with reference to the accompanying drawings, the technical scope of the present disclosure is not limited to such examples. It is clear that a person skilled in the art of the present disclosure can conceive of various modifications or alterations within the scope of the technical idea described in the claims, and it is understood that these modifications or alterations also naturally fall within the technical scope of the present disclosure.

[0144] Furthermore, the effects described in the above embodiments are explanatory or exemplary and are not limited to those described in the above embodiments. In other words, the technology according to the present disclosure may achieve other effects that are obvious to a person skilled in the art of the present disclosure from the description in the above embodiments, in addition to or instead of the effects described in the above embodiments.

[0145] The processes described in the above embodiments can also be realized by dedicated hardware circuits, in which case they may be executed by a single piece of hardware or by multiple pieces of hardware.

[0146] The above embodiment may also employ a configuration for analyzing the consistency of evaluation results based on multiple types of generative models 31. For example, a configuration may be adopted in which multiple second generative models 31B having different initial parameters or learning characteristics are stored in the storage 24. In this case, the CPU 21 may input a prompt related to the same evaluation perspective to each second generative model 31B, analyze the degree of agreement or divergence of the multiple output evaluation results, and, if the analyzed consistency index is below a predetermined threshold, output a risk index indicating low reliability in association with the evaluation result.

[0147] In the above embodiment, the CPU 21 may be configured to analyze the causal relationships and contextual consistency between multiple answers of a job seeker using the generative model 31. For example, the CPU 21 may be configured to extract and analyze multiple answer contents from the answer log of the job seeker, and if a content contradiction is detected between previous and subsequent answer contents, to assign a flag to that part. Furthermore, the CPU 21 may be configured to add a correction value to the evaluation result for the job seeker in accordance with the number of flags detected.

[0148] The above embodiment may be configured such that the CPU 21 uses the generative model 31 to evaluate the consistency between the resume information submitted in advance by the job seeker and the answers given during the interview. For example, the CPU 21 may compare the job content in the resume submitted as the resume information with the response information related to the work history included in the job seeker's answers, and assign a flag to any portion where an inconsistency in the content is detected. Furthermore, the CPU 21 may be configured to visibly display the detected portion indicated by the flag on the result screen to issue a warning.

[0149] The above embodiment may be configured such that the CPU 21 uses the generative model 31 to evaluate the likelihood of imitation based on a relative comparison of the answers of the job seeker and other job seekers. For example, the CPU 21 may cluster the answers of multiple job seekers and determine that there is a tendency for imitation if the answers belong to a group in which the vocabulary, syntax, or emotional expression patterns are extremely similar. Furthermore, the CPU 21 may flag job seekers determined to have a tendency for imitation and notify hiring companies to which the job seekers have applied of warning information. The warning information may include the job seeker's identification information, the detected cluster ID, the answer similarity score, and comment information indicating the likelihood of imitation.

[0150] The above embodiment may be configured to have the CPU 21 use the generative model 31 to determine signs of misconduct based on non-verbal characteristics of the job seeker other than their voice. For example, the CPU 21 may use the camera and microphone of the job seeker terminal 60 to acquire eye movements, facial micro-expressions, body swaying, unnatural movements, and the like, and determine whether the behavior during the answer contains any abnormal patterns. Examples of abnormal patterns include frequent deviations of the gaze, stiff facial expressions in response to specific questions, or excessive swaying of posture, and the like, which may be flagged as signs of misconduct.

[0151] In this disclosure, the term "information processing system" is a concept that encompasses both a system that is configured with a single device and a system that is configured with a combination of multiple devices. For example, the information processing system of this disclosure may be configured with a server 20 alone, or may be configured with a combination of the server 20 and the hiring company terminal 40, or the server 20 and the job seeker terminal 60, etc.

[0152] In this disclosure, the term "processor" refers to a processor in a broad sense, including general-purpose processors (e.g., CPU: Central Processing Unit, etc.) and dedicated processors (e.g., GPU: Graphics Processing Unit, ASIC: Application Specific Integrated Circuit, FPGA: Field Programmable Gate Array, programmable logic device, etc.).

[0153] The operations of the processor in the present disclosure may be performed not only by a single processor but also by multiple processors located in physically separate locations working together. Furthermore, the order of the operations of the processor is not limited to the order described in the present disclosure and may be changed as appropriate.

[0154] In the above embodiment, the information processing program 30 is pre-stored (installed) in the storage 24, but the present invention is not limited to this. The information processing program 30 may be provided in a form recorded on a recording medium such as a CD-ROM (Compact Disk Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), or a USB (Universal Serial Bus) memory. The information processing program 30 may also be downloaded from an external device via a network N. The technology disclosed herein may also be applied to programs and program products.

[0155] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference. [Explanation of symbols]

[0156] 21 CPU (processor) 30 Information Processing Program 31A First Generative Model 31B Second Generative Model 60 Job Seeker Terminal (User Terminal)< / url:>

Claims

1. a processor; The processor: acquire dialogue information indicating dialogue content between a first user and a first generative model capable of dialogue with the first user regarding the hiring requirements for job seekers sought by a recruiting company and evaluation points in a hiring interview; inputting a first prompt, which includes predetermined instruction content related to the hiring requirements and dialogue content related to the hiring requirements from among the dialogue information, into the first generative model, and obtaining a draft hiring requirement as an output result of the first generative model in response to the first prompt; outputting the proposed employment requirements generated by the first generative model, and receiving input of a response from the first user regarding an opinion or a request for modification regarding the proposed employment requirements; extracting the utterance content indicated in the response to the opinion or the request for revision as a dialogue content related to the new hiring requirements, inputting the updated first prompt including the dialogue content related to the new hiring requirements into the first generative model, and obtaining the revised draft hiring requirements; When the first user approves the proposed adoption requirement, a second prompt including predetermined instruction content regarding the evaluation viewpoint and dialogue content regarding the evaluation viewpoint from among the dialogue information is input to the first generation model, and an evaluation viewpoint proposal corresponding to the proposed adoption requirement is obtained as an output result of the first generation model in response to the second prompt; outputting the proposed evaluation viewpoint generated by the first generative model, and receiving input of a correction or supplementary response to the proposed evaluation viewpoint from the first user; extracting the utterance content indicated in the corrected or supplemented response as dialogue content related to the new evaluation perspective, inputting the updated second prompt including the dialogue content related to the new evaluation perspective into the first generative model, and obtaining the revised evaluation perspective proposal; After obtaining consent from the first user regarding the proposed evaluation viewpoint, inputting a third prompt including the proposed evaluation viewpoint and predetermined instructions regarding questions to be asked in the job interview into the first generative model, and obtaining example questions corresponding to the proposed evaluation viewpoint as an output result of the first generative model in response to the third prompt; outputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements, which are generated by the first generative model; inputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements into a second generative model in which behavior as an interviewer or a job seeker is predefined; outputting voice data of the question sentence included in the example question selected by the second generation model from a voice output means on a user terminal of a second user who is conducting an interview; inputting a fourth prompt based on voice data indicating the second user's response voice to the question sentence into the second generative model, and causing the second generative model to generate a next response in response to the fourth prompt based on predefined behavior as an interviewer or a job seeker; During the progress of an interview between the second user and the second generative model, audio data indicating the second user's answer voice in the interview and a sixth prompt based on predetermined instructions for evaluating whether the second user's answer content includes a template expression or signs of intervention by a third party in order to determine signs of misconduct are input to the second generative model, and a score indicating the possibility of misconduct by the second user is obtained as an output result of the second generative model in response to the sixth prompt, If the score exceeds a predetermined threshold, it is determined that the second user has a sign of cheating, and the second user and the answer content in which the possibility of cheating was detected are associated with each other and saved; After the interview between the second user and the second generative model is completed, voice data indicating the response voice of the second user in the interview and a fifth prompt based on the proposed evaluation viewpoint are input to the second generative model, and an evaluation result for the second user is obtained as an output result of the second generative model in response to the fifth prompt; extracting a plurality of answer contents from an answer log including answer contents for which the possibility of fraudulent activity by the second user has been detected, inputting a prompt based on predetermined instructions for detecting inconsistencies in content between preceding and succeeding answer contents included in the extracted plurality of answer contents into the second generative model, and acquiring the inconsistencies in content as an output result of the second generative model in response to the prompt; adding a correction value to the evaluation result for the second user according to the number of flags assigned to the contradictory portions, and outputting the evaluation result after adding the correction value; Information processing system.

2. The processor: In an employment support service that handles a plurality of job information and a plurality of job seeker information, identification information regarding the second user who has been determined to have signs of fraudulent activity is notified to a recruiting company to which the second user has applied, along with a notice that the second user has been determined to have signs of fraudulent activity. The information processing system according to claim 1 .

3. The processor: inputting a seventh prompt including the plurality of job information items registered in an employment support service that handles a plurality of job information items and a plurality of job seeker information items and an evaluation result for the second user into the second generation model, and obtaining a list of companies that are hiring and compatible with the second user as an output result of the second generation model in response to the seventh prompt; notifying the second user of a list of companies hiring that are compatible with the second user, and notifying each company hiring that is hiring of the second user with identification information; The information processing system according to claim 1 .

4. acquire dialogue information indicating dialogue content between a first user and a first generative model capable of dialogue with the first user regarding the hiring requirements for job seekers sought by a recruiting company and evaluation points in a hiring interview; inputting a first prompt, which includes predetermined instruction content related to the hiring requirements and dialogue content related to the hiring requirements from among the dialogue information, into the first generative model, and obtaining a draft hiring requirement as an output result of the first generative model in response to the first prompt; outputting the proposed employment requirements generated by the first generative model, and receiving input of a response from the first user regarding an opinion or a request for modification regarding the proposed employment requirements; extracting the utterance content indicated in the response to the opinion or the request for revision as a dialogue content related to the new hiring requirements, inputting the updated first prompt including the dialogue content related to the new hiring requirements into the first generative model, and obtaining the revised draft hiring requirements; When the first user approves the proposed adoption requirement, a second prompt including predetermined instruction content regarding the evaluation viewpoint and dialogue content regarding the evaluation viewpoint from among the dialogue information is input to the first generation model, and an evaluation viewpoint proposal corresponding to the proposed adoption requirement is obtained as an output result of the first generation model in response to the second prompt; outputting the proposed evaluation viewpoint generated by the first generative model, and receiving input of a correction or supplementary response to the proposed evaluation viewpoint from the first user; extracting the utterance content indicated in the corrected or supplemented response as dialogue content related to the new evaluation perspective, inputting the updated second prompt including the dialogue content related to the new evaluation perspective into the first generative model, and obtaining the revised evaluation perspective proposal; After obtaining consent from the first user regarding the proposed evaluation viewpoint, inputting a third prompt including the proposed evaluation viewpoint and predetermined instructions regarding questions to be asked in the job interview into the first generative model, and obtaining example questions corresponding to the proposed evaluation viewpoint as an output result of the first generative model in response to the third prompt; outputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements, which are generated by the first generative model; inputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements into a second generative model in which behavior as an interviewer or a job seeker is predefined; outputting voice data of the question sentence included in the example question selected by the second generation model from a voice output means on a user terminal of a second user who is conducting an interview; inputting a fourth prompt based on voice data indicating the second user's response voice to the question sentence into the second generative model, and causing the second generative model to generate a next response in response to the fourth prompt based on predefined behavior as an interviewer or a job seeker; During the progress of an interview between the second user and the second generative model, audio data indicating the second user's answer voice in the interview and a sixth prompt based on predetermined instructions for evaluating whether the second user's answer content includes a template expression or signs of intervention by a third party in order to determine signs of misconduct are input to the second generative model, and a score indicating the possibility of misconduct by the second user is obtained as an output result of the second generative model in response to the sixth prompt, If the score exceeds a predetermined threshold, it is determined that the second user has a sign of cheating, and the second user and the answer content in which the possibility of cheating was detected are associated and saved; After the interview between the second user and the second generative model is completed, voice data indicating the response voice of the second user in the interview and a fifth prompt based on the proposed evaluation viewpoint are input to the second generative model, and an evaluation result for the second user is obtained as an output result of the second generative model in response to the fifth prompt; extracting a plurality of answer contents from an answer log including answer contents for which the possibility of fraudulent activity by the second user has been detected, inputting a prompt based on predetermined instructions for detecting inconsistencies in content between preceding and succeeding answer contents included in the extracted plurality of answer contents into the second generative model, and acquiring the inconsistencies in content as an output result of the second generative model in response to the prompt; adding a correction value to the evaluation result for the second user according to the number of flags assigned to the contradictory portions, and outputting the evaluation result after adding the correction value; An information processing method in which processing is performed by a computer.

5. acquire dialogue information indicating dialogue content between a first user and a first generative model capable of dialogue with the first user regarding the hiring requirements for job seekers sought by a recruiting company and evaluation points in a hiring interview; inputting a first prompt, which includes predetermined instruction content related to the hiring requirements and dialogue content related to the hiring requirements from among the dialogue information, into the first generative model, and obtaining a draft hiring requirement as an output result of the first generative model in response to the first prompt; outputting the proposed employment requirements generated by the first generative model, and receiving input of a response from the first user regarding an opinion or a request for modification regarding the proposed employment requirements; extracting the utterance content indicated in the response to the opinion or the request for revision as a dialogue content related to the new hiring requirements, inputting the updated first prompt including the dialogue content related to the new hiring requirements into the first generative model, and obtaining the revised draft hiring requirements; When the first user approves the proposed adoption requirement, a second prompt including predetermined instruction content regarding the evaluation viewpoint and dialogue content regarding the evaluation viewpoint from among the dialogue information is input to the first generation model, and an evaluation viewpoint proposal corresponding to the proposed adoption requirement is obtained as an output result of the first generation model in response to the second prompt; outputting the proposed evaluation viewpoint generated by the first generative model, and receiving input of a correction or supplementary response to the proposed evaluation viewpoint from the first user; extracting the utterance content indicated in the corrected or supplemented response as dialogue content related to the new evaluation perspective, inputting the updated second prompt including the dialogue content related to the new evaluation perspective into the first generative model, and obtaining the revised evaluation perspective proposal; After obtaining consent from the first user regarding the proposed evaluation viewpoint, inputting a third prompt including the proposed evaluation viewpoint and predetermined instructions regarding questions to be asked in the job interview into the first generative model, and obtaining example questions corresponding to the proposed evaluation viewpoint as an output result of the first generative model in response to the third prompt; outputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements, which are generated by the first generative model; inputting the proposed evaluation viewpoints and the example questions corresponding to the proposed hiring requirements into a second generative model in which behavior as an interviewer or a job seeker is predefined; outputting voice data of the question sentence included in the example question selected by the second generation model from a voice output means on a user terminal of a second user who is conducting an interview; inputting a fourth prompt based on voice data indicating the second user's response voice to the question sentence into the second generative model, and causing the second generative model to generate a next response in response to the fourth prompt based on predefined behavior as an interviewer or a job seeker; During the progress of an interview between the second user and the second generative model, audio data indicating the second user's answer voice in the interview and a sixth prompt based on predetermined instructions for evaluating whether the second user's answer content includes a template expression or signs of intervention by a third party in order to determine signs of misconduct are input to the second generative model, and a score indicating the possibility of misconduct by the second user is obtained as an output result of the second generative model in response to the sixth prompt, If the score exceeds a predetermined threshold, it is determined that the second user has a sign of cheating, and the second user and the answer content in which the possibility of cheating was detected are associated with each other and saved; After the interview between the second user and the second generative model is completed, voice data indicating the response voice of the second user in the interview and a fifth prompt based on the proposed evaluation viewpoint are input to the second generative model, and an evaluation result for the second user is obtained as an output result of the second generative model in response to the fifth prompt; extracting a plurality of answer contents from an answer log including answer contents for which the possibility of fraudulent activity by the second user has been detected, inputting a prompt based on predetermined instructions for detecting inconsistencies in content between preceding and succeeding answer contents included in the extracted plurality of answer contents into the second generative model, and acquiring the inconsistencies in content as an output result of the second generative model in response to the prompt; adding a correction value to the evaluation result for the second user according to the number of flags assigned to the contradictory portions, and outputting the evaluation result after adding the correction value; An information processing program that causes a computer to execute a process.

Images