Information processing device, evaluation method, and program
The information processing device addresses interviewer subjectivity and AI uniformity by generating an evaluation model that aligns with company requirements, effectively identifying suitable candidates.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Filing Date
- 2024-12-25
- Publication Date
- 2026-07-07
Smart Images
Figure 2026113225000001_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a technique for evaluating examinees in an employment interview.
Background Art
[0002] In the employment activities of companies, examinees are evaluated using an employment process including document screening, aptitude tests, multiple-stage interviews, etc., and passers are determined. In recent years, in the creation of entry sheets for document screening, it has become common for examinees to use tools such as AI (artificial intelligence), making it difficult to evaluate examinees in document screening. For this reason, the weight of evaluating examinees in interviews has increased. In such a situation, in order for companies to efficiently conduct interviews in the employment process, it is required to prepare a large number of interviewers. On the other hand, there is a technology in which AI substitutes for interviews to evaluate examinees and provides evaluation results. Patent Document 1 describes a technology in which a computer analyzes images, voices, etc. of interview examinees for evaluation.
Prior Art Documents
Patent Documents
[0003]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0004] When there are a large number of interviewers, there is a possibility that bias based on the interviewers' subjectivity will occur in the evaluation results of examinees. For example, if there are differences among interviewers in the weighting for each evaluation item for evaluating examinees, employment of personnel may be carried out based on the subjectivity of the examiner. Also, when using AI substitution in employment interviews, the evaluation of examinees becomes uniform. For example, while the evaluation of examinees with high basic abilities is high, there may be cases where examinees who meet the requirements sought by the company are not necessarily highly evaluated.
[0005] This invention provides an evaluation technology that efficiently extracts candidates who meet the requirements of a company and possess high basic abilities. [Means for solving the problem]
[0006] An information processing device according to one aspect of the present invention is an information processing device for evaluating applicants in a job interview, comprising: identification means for identifying the weights for each of a predetermined plurality of evaluation items for evaluating the applicant; generation means for generating an evaluation model including the weights for each of the predetermined plurality of evaluation items; first acquisition means for acquiring a set of scores for at least a portion of the predetermined plurality of evaluation items that have been assigned to the applicant based on the job interview conducted with the applicant; and output means for outputting an overall evaluation of the applicant obtained by applying the evaluation model to the set of scores assigned to the applicant. [Effects of the Invention]
[0007] According to the present invention, it is possible to efficiently select test takers who meet the requirements of a company and possess high basic abilities. [Brief explanation of the drawing]
[0008] [Figure 1] This is a diagram showing an example of the configuration of an information processing system. [Figure 2] This figure shows an example of the hardware configuration of the evaluation server. [Figure 3] This figure shows an example of the functional configuration of the evaluation server. [Figure 4] This figure shows an example of the processing flow executed by the evaluation server. [Figure 5] This figure shows an example of setting up a personnel requirements model. [Figure 6] This figure shows an example of setting up a personnel requirements model. [Figure 7] This figure shows an example of the structure of a test taker dataset. [Figure 8] This figure shows an example of how the overall rating is displayed. [Figure 9]This figure shows an example of the processing flow executed by the evaluation server. [Figure 10] This figure shows an example of the structure of the dataset for successful candidates and the dataset for unsuccessful candidates. [Figure 11] This figure shows an example of the processing flow executed by the evaluation server. [Figure 12] This figure shows an example of clustering. [Figure 13] This figure shows an example of how to display a personnel requirements model. [Figure 14] This figure shows an example of the processing flow executed by the evaluation server. [Modes for carrying out the invention]
[0009] The embodiments will be described in detail below with reference to the attached drawings. Note that the following embodiments do not limit the invention as defined in the claims. While the embodiments describe multiple features, not all of these features are essential to the invention, and the features may be combined in any way. Furthermore, in the attached drawings, identical or similar configurations are given the same reference numerals, and redundant descriptions are omitted.
[0010] (System Configuration) Figure 1 shows an example of the configuration of the information processing system 100 according to this embodiment. The information processing system 100 is a system for evaluating applicants who apply for a job at a company and participate in an interview. For example, the information processing system 100 uses the interview results data of the applicants as input, performs an overall evaluation of each applicant, and outputs the overall evaluation.
[0011] The information processing system 100 is comprised of, for example, an evaluation server 101, an interview server 102, a recruitment terminal 111, an examinee terminal 112, a pass / fail information database 121, an examination results database 122, and a network 131. Each of the devices constituting the information processing system 100 may be connected via the network 131. Note that some of these devices constituting the information processing system 100 may not be included, and the information processing system 100 may include other devices. For example, in the information processing system 100, the interview server 102 and the examinee terminal 112 may not be included, and the evaluation server 101 may perform an overall evaluation of examinees using data stored in the examination results database 122. Also, one device included in the information processing system 100 in Figure 1 may be comprised of two or more devices, and two or more devices constituting the information processing system 100 in Figure 1 may be comprised of one device. For example, the pass / fail information database 121 and the examination results database 122 may be configured as a single database. Alternatively, the interview server 102 and the examination results database 122 may be configured as a single device, and the results of interviews conducted by the interview server may be reflected in the examination results database in real time.
[0012] The examinee terminal 112 is a device used by the examinee when taking the interview. For example, the examinee terminal 112 may be a personal computer (PC), smartphone, tablet, etc. As an example, the examinee may take the interview using a web conferencing application installed on the examinee terminal 112. For example, the examinee terminal 112 may have a display and speakers and be able to display to the examinee video, text, audio, etc. from the interviewer's side received from the interview server 102 or the recruitment terminal 111. In addition, the examinee terminal 112 may have a camera, microphone, keyboard, mouse, etc. and be able to transmit video, audio, text, etc. from the examinee's side to the interview server 102 or the recruitment terminal 111. The examinee terminal 112 may also have a touchscreen or the like that serves as both an input and output means.
[0013] The interview server 102 is a server that conducts an interview with an examinee. For example, the interview server 102 can be realized by software operating on an information processing device such as a computer connected to the network 131. The interview server 102 has one or more unique URLs (Uniform Resource Locators), and the interview server 102 and the examinee terminal 112 can be connected via the network 131 by the web conference application installed on the examinee terminal 112 accessing the specified URL. The interview server 102 asks the examinee one or more questions and evaluates the examinee based on the examinee's answers to the questions. The interview server 102 can evaluate the examinee for a predetermined evaluation item. The predetermined evaluation item can include, for example, initiative, influencing power, execution ability, problem discovery ability, planning ability, creativity, communication ability, listening ability, flexibility, situation grasping ability, discipline, stress control ability, motivation from a forward-looking perspective, motivation from a work perspective, motivation from a contribution perspective, and motivation from an environmental perspective. The predetermined evaluation item may include evaluation items other than these, and may not include some of these evaluation items. The predetermined evaluation item may be composed of evaluation items different from these. The interview server 102 assigns points within a predetermined range to each of the predetermined evaluation items for each examinee. For example, the predetermined range can be 1 to 3, 1 to 5, 1 to 10, 1 to 25, 1 to 100, etc. The minimum value of the predetermined range may be 0 or a negative value. Also, the points are not limited to integers and may include decimal places. The predetermined range may be different for each evaluation item. Also, instead of points, symbols, characters, etc. may be assigned, as long as a value indicating the evaluation result of the examinee for each evaluation item is assigned. Hereinafter, the description will be made assuming that points within a predetermined range are assigned to each evaluation item as the evaluation result of the examinee. The interview server 102 stores the result of the interview in the examination result database 122.
[0014] Note that the interview server 102 can receive various settings from the recruitment staff of the target company (hereinafter referred to as "recruitment staff") via the employment terminal 111. For example, the interview server 102 can receive settings for the time to conduct interviews with each of the examinees. Also, the interview server 102 can receive selections such as one or more questions to be presented to the examinees and methods for analyzing the examinees' answers to the questions. For example, the interview server 102 can perform natural language processing on the examinees' answers and select the next question to be presented based on the keywords contained therein. Also, the interview server 102 can assign scores to each of the predetermined evaluation items based on the keywords and their combinations contained in the examinees' answers. The interview server 102 can present questions and analyze the answers within the set time range for the interview, evaluate the examinees, and generate the examination results. For example, the examination results can include information for identifying the examinee and the scores assigned to each of the predetermined evaluation items.
[0015] The examination result database 122 stores the examination results of the examinees. For example, the examination result database 122 can store the examination results of the examinees generated by the interview server 102. The examination result database 122 can store, as the examination results of the examinees, each examinee and the scores assigned to each of the above-mentioned predetermined evaluation items for that examinee in association with each other. The set of scores assigned to each of the predetermined evaluation items associated with each examinee is referred to as an examinee dataset. For example, the examinee dataset includes information for identifying each examinee (name, examination number, etc.) and at least a part of the scores assigned to each of the predetermined evaluation items. The examination result database 122 can store one or more examinee datasets.
[0016] Furthermore, the examination results stored in the examination results database 122 are not limited to those generated by the interview server 102. For example, a recruiter may conduct an interview with an applicant and store the evaluated interview results in the examination results database 122. For example, a recruiter may use a recruiter terminal 111 to conduct an online interview with an applicant using an applicant terminal 112 via the network 131. Alternatively, a recruiter may conduct an offline interview with an applicant. In these cases, the recruiter may use the recruiter terminal 111 to store an applicant dataset created based on the content of the interview with the applicant in the examination results database 122. Furthermore, the examination results stored in the examination results database 122 may also be an applicant dataset generated based on an interview conducted outside of the information processing system shown in Figure 1. For example, an interview with an applicant may be conducted by an agency or an AI other than the interview server 102 included in the information processing system 100, and the examination results may be provided as an applicant dataset. In this case, the examination results database 122 can store, for example, a data set of examinees provided via the recruitment terminal 111 or the like.
[0017] The Passer Information Database 121 stores information about test takers hired by the target company. Test takers hired by the target company are sometimes referred to as passers. Information about passers is sometimes referred to as passer information. For example, the Passer Information Database 121 may store the test results of test takers hired by the target company, among the test results of test takers stored in the Test Results Database 122, as passer information. The Passer Information Database 121 may also receive passer information as data from, for example, the hiring terminal 111. The set of scores assigned to each predetermined evaluation item associated with each passer is referred to as the passer dataset. For example, the passer dataset includes information that identifies each passer (name, employee number, etc.) and at least a portion of the scores assigned to each predetermined evaluation item. The Passer Information Database 121 may store one or more passer datasets.
[0018] The recruitment terminal 111 is a terminal used by the recruitment officer. For example, the recruitment terminal 111 may be a personal computer (PC), smartphone, tablet, etc. The recruitment terminal 111 may have a display, speaker, camera, microphone, keyboard, mouse, touchscreen, etc. For example, the recruitment officer may use a web conferencing application installed on the recruitment terminal 111 to conduct interviews with applicants using the applicant terminal 112. The recruitment officer may also operate the recruitment terminal 111 to store the examination results in the examination results database 122 and the information of successful candidates in the successful candidate information database 121. Furthermore, the recruitment officer may operate the recruitment terminal 111 to configure or select the personnel requirement model used by the evaluation server 101. The personnel requirement model will be described later. The recruitment terminal 111 may present the overall evaluation of the applicant output by the evaluation server 101 to the recruitment officer. For example, the recruitment terminal 111 may display the overall evaluations of applicants in a ranking format, ordered from those with the highest overall evaluation values for each talent requirement model. The recruitment terminal 111 may also display the scores assigned to each applicant for each predetermined evaluation item output by the evaluation server 101. These scores may be displayed in the form of a radar chart, bar graph, or similar. The methods by which the recruitment terminal 111 presents the applicant's overall evaluation are not limited to these; any method that allows the recruitment officer to understand the applicant's overall evaluation may be applied.
[0019] The evaluation server 101 performs an overall evaluation of the test takers. For example, the evaluation server 101 applies a talent requirements model selected by the hiring manager to each test taker's test results stored in the test result database 122 to perform an overall evaluation of each test taker. The evaluation server 101 can be implemented, for example, by software running on an information processing device connected to the network 131. The configuration and operation of the evaluation server 101 are described below.
[0020] (Device configuration) Figure 2 shows an example of the hardware configuration of the evaluation server 101. In this example, the evaluation server 101 includes a processor 201, ROM 202, RAM 203, storage device 204, and communication circuit 205. The processor 201 is a computer that includes one or more processing circuits, such as a general-purpose CPU (Central Processing Unit) or ASIC (Application-Specific Integrated Circuit). The processor 201 performs the overall processing of the device and the aforementioned processing by reading and executing programs stored in the ROM 202 and storage device 204. The ROM 202 is a read-only memory that stores information such as programs and various parameters related to the processing performed by the evaluation server 101. The RAM 203 functions as a workspace when the processor 201 executes programs and is a random-access memory that stores temporary information. The storage device 204 is composed of, for example, a removable external storage device. The communication circuit 205 is composed of, for example, a circuit for communicating with other devices.
[0021] (Functional Configuration) Figure 3 shows an example of the functional configuration of the evaluation server 101. The evaluation server 101 includes, for example, a personnel requirements identification unit 301, a passer data acquisition unit 302, an evaluation model generation unit 303, a test taker data acquisition unit 304, a comprehensive evaluation execution unit 305, a user instruction reception unit 306, a comprehensive evaluation output unit 307, and a passer data update unit 308. Figure 3 shows the functional configuration of the evaluation server 101 in this embodiment, and omits, for example, the general configuration of the information processing device on which the evaluation server is implemented. These functional units can be realized, for example, by the processor 201 executing a program stored in the ROM 202 or storage device 204 and controlling the communication circuit 205 as needed. However, it is not limited to this, and for example, dedicated hardware for realizing each function may be provided.
[0022] The personnel requirements identification unit 301 identifies the weights for each of the predetermined evaluation items used to evaluate applicants in the employment interview. For example, the personnel requirements identification unit 301 can identify the weights for each of the predetermined evaluation items using a first set, which includes a set of scores for each of the predetermined evaluation items assigned to each of the successful applicants hired by the target company and stored in the successful applicant information database 121. The personnel requirements identification unit 301 can generate a plurality of second sets using at least a portion of the first set, and identify the weights for each of the predetermined evaluation items for each of the second sets. Each of the second sets may represent the result of a classification based on the similarity of the sets of scores for each of the predetermined evaluation items with respect to the first set. The personnel requirements identification unit 301 can also identify a set of weights for each of the predetermined evaluation items set by the hiring manager as the weights for each of the predetermined evaluation items. The personnel requirements identification unit 301 can identify a plurality of sets of the weights for each of the predetermined evaluation items that differ in at least a portion of their weights.
[0023] The applicant data acquisition unit 302 acquires information on applicants who have passed the recruitment process of the target company. For example, the applicant data acquisition unit 302 acquires a set of scores assigned to each of several predetermined evaluation items for each applicant. A first set can be formed from the sets of scores assigned to each of the multiple applicants.
[0024] The evaluation model generation unit 303 generates an evaluation model for the overall evaluation of the examinee. For example, the evaluation model generation unit 303 can generate an evaluation model that includes weights for each of the predetermined evaluation items. Furthermore, if the personnel requirement identification unit 301 identifies multiple sets of weights, at least some of which are different, for multiple predetermined evaluation items corresponding to multiple personnel requirements, the evaluation model generation unit 303 can generate multiple evaluation models corresponding to each of the multiple sets of weights. For example, the evaluation model generation unit 303 can generate an evaluation model for each of the second sets. In addition, the evaluation model generation unit 303 can generate an evaluation model for each set of weights for each of the predetermined evaluation items set by the recruiter.
[0025] The examinee data acquisition unit 304 acquires information about the examinee. For example, the examinee data acquisition unit 304 may acquire a set of scores for at least some of a predetermined set of evaluation items that were assigned to the examinee based on the employment interview conducted with that examinee. The employment interview that generates the examinee data may be an interview with the examinee conducted by artificial intelligence (AI).
[0026] The comprehensive evaluation execution unit 305 performs a comprehensive evaluation for each examinee. For example, the comprehensive evaluation execution unit 305 can perform a comprehensive evaluation by applying an evaluation model to each set of points for predetermined evaluation items assigned to the examinee. If there are multiple evaluation models, the comprehensive evaluation execution unit 305 can perform a comprehensive evaluation of the examinee by applying each of the multiple evaluation models to the set of points assigned to the examinee.
[0027] The user instruction receiving unit 306 accepts selections and settings from operators and recruiters. For example, the user instruction receiving unit 306 can accept the selection of an evaluation model to be used for the overall evaluation. As an example, if the user instruction receiving unit 306 has generated weights and evaluation models for each of the predetermined evaluation items for each second set, it can accept the selection from the operator or recruiter of the weights and evaluation models to be used for the overall evaluation. The user instruction receiving unit 306 can also accept the setting of weights for each of the predetermined evaluation items from the operator or recruiter. If there are multiple sets of weights for each of the predetermined evaluation items that differ in at least some respects, the user instruction receiving unit 306 can accept the setting of each set of weights.
[0028] The overall evaluation output unit 307 outputs the results of the overall evaluation. For example, the overall evaluation output unit 307 outputs the results of the overall evaluation to the recruitment terminal 111 or the terminal used by the operator of the evaluation server 101. The results of the overall evaluation may be displayed in the form of a table, radar chart, bar graph, etc. on the recruitment terminal 111 or the terminal used by the operator. The overall evaluations of multiple applicants or the overall evaluations using multiple evaluation models may be displayed together.
[0029] The passer data update unit 308 updates the passer information stored in the passer information database 121. For example, the passer data update unit 308 may update the first set using a set of scores for each of the predetermined evaluation items assigned to applicants who have passed the recruitment process of the target company. The updated first set may be used by the personnel requirement identification means 301 when it uses the first set to identify the weight for each of the predetermined evaluation items.
[0030] (Process flow) (Example of processing in the first case) In this example, the evaluation server 101 performs an overall evaluation of the applicant using a talent requirements model set by the recruiter. The talent requirements model may consist of information indicating the weight of each evaluation item used in the overall evaluation. For example, the evaluation items used in the overall evaluation may be the predetermined evaluation items described above. However, the evaluation items used in the overall evaluation may differ from the predetermined evaluation items described above. For example, the talent requirements model may be created based on the talent requirements sought by the target company.
[0031] The evaluation server 101 can perform an overall evaluation of a test-taker by assigning weights to the scores given to each of the test-taker's evaluation items, based on the talent requirements model assigned to those items, and then calculating the sum of these weights. This configuration makes it possible to perform an overall evaluation that aligns with the talent requirements sought by the target company, compared to simply calculating the sum of the scores given to each of the test-taker's evaluation items. As a result, candidates who have high scores in basic abilities and who match the talent requirements sought by the target company are given high overall evaluation values. Therefore, by identifying candidates in order of their overall evaluation values, it is possible to efficiently identify candidates who should be hired.
[0032] Figure 4 shows an example of the processing flow executed by the evaluation server 101 in this processing example. This processing flow can be started when the evaluation server 101 receives a processing start instruction on the operation screen provided to the operator. Alternatively, if a recruiter operates the evaluation server via the recruiter terminal 111, the processing flow can be started when the evaluation server receives a processing start operation performed by the recruiter on the recruiter terminal 111. First, the evaluation server 101 receives the personnel requirements (S401). In this processing example, the recruiter sets the personnel requirements model via the recruiter terminal 111. An example of setting the personnel requirements model is shown in Figure 5. Figure 5 shows an example of setting the personnel requirements model in the form of a radar chart. Each of items 1 to N corresponds to each of the evaluation items used in the overall evaluation. Here, each evaluation item is assumed to correspond to the predetermined evaluation items mentioned above. For example, item 1 may correspond to initiative, item 2 to initiative, item 3 to execution ability, ..., and item N may correspond to motivation from an environmental perspective. The evaluation items that can be set as part of the talent requirements model may be only a part of the predetermined evaluation items. This reduces the number of items to display, making it easier for even inexperienced personnel to set up the talent requirements model. In the example setting in Figure 5, for example, a recruiter can set the weight of each evaluation item in accordance with the talent requirements sought by the target company by moving each vertex that forms the radar chart. For example, in the radar chart in Figure 5, the larger the radius of the circle on which the vertices are placed, the greater the weight of the item corresponding to that vertex.
[0033] The method for setting up a talent requirements model is not limited to radar charts. For example, as shown in Figure 6, the weight of each evaluation item may be set using sliders corresponding to each evaluation item. In Figure 6, one slider is associated with each evaluation item. Moving the slider to the right assigns a higher priority. Alternatively, the weight of each evaluation item can be changed by changing the number displayed to the right of each evaluation item. Furthermore, groups containing multiple evaluation items may be assigned additional weights. In Figure 6, the weights can be adjusted between the "Motivation" group, which contains four evaluation items, and the "Ability" group, which contains twelve evaluation items. The set up talent requirements model can be saved with an identification name. The method for setting up a talent requirements model is not limited to these methods; any method that allows the recruiter to set the weight of each evaluation item can be applied.
[0034] The evaluation server 101 determines an overall evaluation model based on the received personnel requirements model (S402). For example, the evaluation server 101 may determine an overall evaluation model that calculates the weighted sum of the scores assigned to each evaluation item of the examinee, using the weights of each evaluation item set as the personnel requirements model. The calculation formula may be expressed as shown in Equation 1 below.
[0035] Overall evaluation = ΣAi × Pi - (Equation 1) Here, Ai is a weighting coefficient indicating the weight assigned to each evaluation item, Pi is the score assigned to each evaluation item, and i is an identifier that identifies the evaluation item (for example, in Figure 5, i = 1 to N).
[0036] Next, the evaluation server 101 acquires the examinee data subject to the overall evaluation (S403). For example, the evaluation server 101 may acquire the examinee dataset for each examinee stored in the examination result database 122. An example of an examinee dataset is shown in Figure 7. The examinee dataset may consist of the examinee's identification information and the scores assigned to each evaluation item. The evaluation server 101 performs an overall evaluation of the examinee using an overall evaluation model (S404). For example, the evaluation server 101 may calculate the overall evaluation by applying Equation 1 to the examinee dataset for each examinee.
[0037] When the evaluation server 101 has completed the overall evaluation for all applicants, it outputs the results (S405). For example, the evaluation server 101 may display the overall evaluation on the operation screen that the evaluation server 101 provides to the operator. Also, when a recruiter operates the evaluation server via the recruiter terminal 111, the evaluation server 101 may display the overall evaluation on the recruiter terminal 111. For example, the evaluation server 101 may display the overall evaluation in a ranking format that shows the applicant's name or identification information and the result of the overall evaluation, so that applicants with high overall evaluations are displayed at the top. The evaluation server 101 may also further indicate the ranking of the applicants' overall evaluations. This makes it possible to determine which applicants should be selected for the next step in the recruitment process, depending on the number of people to be hired and the number of people to advance to the next step in the recruitment process. The evaluation server 101 may also further display the values assigned to each applicant's evaluation items. This makes it possible for recruiters to determine which applicants should be selected for the next step in the recruitment process based not only on the overall evaluation but also on the values of each evaluation item.
[0038] The evaluation server 101 can use multiple models as an overall evaluation model. For example, if a target company has multiple requirements for the personnel it seeks, an overall evaluation model can be created that corresponds to each of the personnel requirements. As an example, a recruiter can set up multiple personnel requirement models via the recruiter terminal 111 (S401). For example, in Figure 6, the recruiter sets the weights of each evaluation item based on the requirements of the first personnel, assigns a setting name (first setting name), and saves it as the first personnel requirement model. Then, the recruiter sets the weights of each evaluation item based on the requirements of the second personnel, assigns a setting name (second setting name), and saves it as the second personnel requirement model. The recruiter repeats this process to create personnel requirement models that correspond to the requirements of each personnel. The evaluation server 101 generates an overall evaluation model for each created personnel requirement model (S402), and can perform an overall evaluation of the applicant data using each overall evaluation model (S404). For example, the evaluation server 101 reads each of the saved talent requirement models, generates an overall evaluation model, and performs an overall evaluation. The evaluation server 101 repeats this process for each talent requirement model. In this case, the evaluation server 101 can present a summary of the overall evaluation of the applicant data for each talent requirement model (S405). Figure 8 shows an example of outputting a summary of the overall evaluation for each overall evaluation model. In Figure 8, the vertical axis shows the name or identification information of each applicant. For example, the identification information may be the applicant number, etc. On the other hand, the horizontal axis shows the setting name of the talent requirement model. For example, the setting name of each talent requirement model may be the setting name assigned by the recruiter when setting each talent requirement model. Also, when one of the talent requirement models shown on the horizontal axis is selected, the applicant data may be sorted and redisplayed based on the overall evaluation by the selected talent requirement model. This allows the recruiter to switch between talent requirement models and determine which applicants to extract for the next step in the recruitment process. If each talent requirements model has a threshold used to determine whether or not to proceed to the next step in the recruitment process, it may be indicated whether or not the overall evaluation exceeds that threshold.For example, overall evaluations that exceed a certain threshold may be highlighted in a different color from other overall evaluations. Furthermore, the thresholds may differ for each personnel requirement. Additionally, the highest, lowest, and average values for each overall evaluation may be displayed for each test-taker.
[0039] In this way, the evaluation server 101 generates a comprehensive evaluation model based on the talent requirements model set by the recruiter, and applies the generated comprehensive evaluation model to the applicant dataset to perform a comprehensive evaluation that aligns with the requirements of the target company. This makes it possible to efficiently identify applicants with high basic abilities that meet the requirements of the target company. Note that the flowchart in Figure 4 is for illustrative purposes only, and the comprehensive evaluation performed by the evaluation server 101 is not limited to this. For example, the order of processing may be different. As an example, the evaluation server 101 may acquire applicant data (S403) before determining the comprehensive evaluation model (S402). This makes it possible to use one or more applicant datasets created based on interviews conducted in advance, apply multiple different talent requirements models, and make decisions on which applicants should proceed to the next step in the recruitment process.
[0040] (Second example of processing) In this example, the evaluation server 101 generates a talent requirements model using the information stored in the candidate information database 121, and then uses that talent requirements model to perform an overall evaluation of the applicants. Figure 9 shows an example of the processing flow executed by the evaluation server 101 in this example. In Figure 9, the same reference numbers are used for processes similar to those in Figure 4, and their explanations are omitted. The evaluation server 101 acquires candidate data (S901). For example, the evaluation server 101 acquires a candidate dataset of successful candidates from the target company's past recruitment processes stored in the candidate information database 121. Note that the candidate dataset does not necessarily have to include information that identifies each applicant. Also, the candidate information database 121 may contain information other than the data of successful candidates. For example, the candidate information database 121 may contain data of those who did not pass (unsuccessful candidates). The set of scores for predetermined evaluation items assigned to each unsuccessful candidate will be called the unsuccessful candidate dataset. For example, the unsuccessful applicants dataset may include information that identifies each applicant (name, applicant number, etc.) and at least a portion of the scores assigned to each of the predetermined evaluation items. The unsuccessful applicants dataset does not have to include information that identifies each applicant. The evaluation server 101 may also retrieve the unsuccessful applicants dataset of the target company stored in the successful applicants database 121. Figure 10 shows an example of a successful applicants dataset and an unsuccessful applicants dataset. In Figure 10, the successful applicants dataset and the unsuccessful applicants dataset are similarly structured, but the values of the acceptance / rejection information are different. The acceptance / rejection information may be set to a value of 1 for successful applicants and a value of 0 for unsuccessful applicants.
[0041] The evaluation server 101 generates a talent requirements model using one or more datasets of successful candidates and one or more datasets of unsuccessful candidates (S902). For example, the evaluation server 101 may use one or more datasets of successful candidates to calculate statistical values such as the mean, median, maximum, and minimum values for each of the predetermined evaluation items. The evaluation server 101 may use any of the calculated statistical values, or a combination thereof, as weights for each evaluation item to generate a talent requirements model.
[0042] The evaluation server 101 may use one or more successful candidate datasets and one or more unsuccessful candidate datasets to calculate the correlation value between each evaluation item and acceptance / rejection, and use that correlation value to generate a talent requirements model. For example, the evaluation server 101 may determine that a particular evaluation item has a high correlation with acceptance / rejection if the score in the successful candidate dataset (e.g., mean or median) is high, the score in the unsuccessful candidate dataset is low, and the difference between them is greater than a threshold. The evaluation server 101 may also determine that a particular evaluation item has a low correlation with acceptance / rejection if the difference between the score in the successful candidate dataset and the score in the unsuccessful candidate dataset is less than a threshold. Furthermore, the evaluation server 101 may determine that a particular evaluation item has a high correlation with acceptance / rejection if the score in the successful candidate dataset is low, the score in the unsuccessful candidate dataset is high, and the difference between them is greater than a threshold. Furthermore, the evaluation server 101 may determine that a particular evaluation item has a high correlation to acceptance or rejection based solely on the pass / fail dataset, if the mean value of that evaluation item is higher than the threshold and the variance is lower than the threshold. The method by which the evaluation server 101 evaluates the correlation between a particular evaluation item and acceptance or rejection is not limited to these. The evaluation server 101 may generate a talent requirements model such that the higher the correlation, the higher the weight of that evaluation item, and the lower the correlation, the lower the weight of that evaluation item.
[0043] The evaluation server 101 determines an overall evaluation model using the generated talent requirements model (S402). The evaluation server 101 then applies the determined overall evaluation model to the applicant data to perform an overall evaluation (S403-S405). In this way, by generating a talent requirements model using the target company's pass / fail data set and rejection data set, the need for recruiters to create a talent requirements model and set it in the evaluation server 101 is eliminated, thus streamlining the recruitment process. Furthermore, because the talent requirements model is generated based on data generated from the recruitment process, an overall evaluation can be performed using objective talent requirements that do not include the subjectivity of recruiters.
[0044] The evaluation server 101 may generate a trained model for performing an overall evaluation by machine learning using one or more successful candidate datasets and one or more unsuccessful candidate datasets. In this case, the successful candidate datasets and unsuccessful candidate datasets used in machine learning may include scores and acceptance / rejection information assigned to each of the predetermined evaluation items. Each of the data included in the successful candidate dataset and unsuccessful candidate dataset may be used as training data for generating the trained model. The generated trained model may be used as an overall evaluation model (S402). By inputting the candidate dataset to be evaluated into the generated trained model, the evaluation server 101 may output a value in the range of 0 to 1 indicating whether the candidate is closer to acceptance or rejection as an overall evaluation (S404). A value closer to 1 may be output as the candidate is closer to acceptance. In this way, the evaluation server 101 can perform an overall evaluation of candidates using the trained model.
[0045] (Third processing example) In this example, the evaluation server 101 generates multiple personnel requirement models using information stored in the successful candidate information database 121, and performs an overall evaluation of the examinee using each of the generated personnel requirement models. Figure 11 shows an example of the processing flow executed by the evaluation server 101 in this example. In Figure 11, the same reference numbers are used for processes similar to those in Figures 4 and 9, and explanations are omitted. To generate multiple personnel requirement models, the evaluation server 101 obtains one or more successful candidate datasets and one or more unsuccessful candidate datasets (S901), and uses them to create multiple groups (S1101). For example, the evaluation server 101 may create groups of successful candidate datasets using the departments within the target company of each successful candidate included in the successful candidate dataset. For example, departments may be General Affairs Department, Product Development Department, Sales Department, etc. Departments may also be more specific departments or sections within these, or broader units including multiples of these. In this case, the successful candidate dataset may include the score assigned to each of the predetermined evaluation items and information indicating the current or past departments of those successful candidates. The evaluation server 101 may generate groups by grouping passer datasets belonging to the same organization. A single passer dataset may be included in multiple groups, and there may be passer datasets that are not included in any group. The method by which the evaluation server 101 generates groups using passer datasets is not limited to this, and any method for generating multiple groups may be applied. For example, the evaluation server 101 may generate groups based on when the passers were hired.
[0046] The evaluation server 101 generates a talent requirements model for each group of the generated pass-through datasets (S1102). For example, the evaluation server 101 may generate a talent requirements model for each group using the method described in S902. If the talent requirements model uses the correlation between a specific evaluation item and acceptance / rejection, the evaluation server 101 may determine the correlation using the pass-through dataset of a specific group and the pass-through datasets of other groups. For example, if the score of a specific evaluation item is high in the pass-through dataset of a specific group and low in the pass-through datasets of other groups, it may be determined that there is a high correlation between that specific evaluation item and acceptance / rejection. The evaluation server 101 may generate a talent requirements model such that the weight of that specific evaluation item is higher for that specific group.
[0047] Furthermore, the evaluation server 101 may perform both the creation of groups using the passer datasets and the generation of talent requirement models (S1101, S1102). For example, the evaluation server 101 performs clustering on one or more acquired passer datasets. Since the passer datasets can be viewed as N-dimensional vectors, where N is the number of predetermined evaluation items, the evaluation server 101 can generate multiple groups of passer datasets by performing clustering in the N-dimensional space. For example, the k-means method may be used for clustering. The evaluation server 101 can select a predetermined number of groups in descending order of the number of passer datasets contained in each generated group, and create a talent requirement model corresponding to each group. For example, the evaluation server 101 may determine the passer dataset closest to the center of the selected cluster as the talent requirement model for that group. The evaluation server 101 may also calculate statistical values such as the mean and median for each evaluation item in the passer datasets contained in each group, and use these statistical values to generate the respective talent requirement models.
[0048] Figure 12 shows an example of clustering in this processing example. Figure 12 shows a mapping of the values of each item X and item Y in the passer dataset onto a two-dimensional plane with item X on the horizontal axis and item Y on the vertical axis. That is, each point in Figure 12 represents the values of item X and item Y in each passer dataset. Item X and item Y are any two selected from a predetermined N evaluation items. In Figure 12, regions 1201, 1202, and 1203 are circles of the same radius and contain 11, 14, and 8 data points, respectively. In this way, if regions of the same size contain more than a threshold (e.g., 8 points), it can be determined that these data form a single region (cluster). In actual clustering processing, any combination of 1 to N items can be arbitrarily selected. For example, when clustering is performed based on three items, spheres of the same radius can be used to determine the data that form a cluster. Note that the regions used for clustering determination may not be circles or spheres, but other shapes. Furthermore, the method for determining data that form a cluster is not limited to this; for example, any region containing more than a predetermined number of data points may be determined as a single cluster. In Figure 12, data 1211, data 1212, and data 1213 are examples of data close to the centers of regions 1201, 1202, and 1203, respectively. A dataset of passersby containing these data points can be determined as a talent requirements model for each region. Alternatively, the data closest to the center of each region may be selected as the talent requirements model, or one or more data points within a predetermined distance from the center of each region may be selected as the talent requirements model based on random numbers or the like.
[0049] The evaluation server 101 may present the recruiter with multiple generated talent requirement models and accept their selection of the talent requirement model to be used. For example, if clustering is performed, the evaluation server 101 may display the pass-through dataset closest to the center of each cluster as a sample of the group formed by the pass-through datasets included in that cluster on the recruiter's terminal 111, etc. For example, the pass-through dataset may be displayed in the form of a radar chart or bar graph. Information about the pass-through dataset associated with that dataset may also be presented. As an example, a video of the pass-through interview may be displayed. By presenting information about the pass-through dataset along with information indicating the pass-through dataset, the recruiter can decide whether or not to use the talent requirement model for that group based on diverse information. The talent requirement model selected by the company representative may be input to the evaluation server 101 via an operation screen or the like provided by the evaluation server 101.
[0050] Figure 13 shows an example of how the evaluation server 101 presents multiple generated talent requirement models to recruiters and accepts their selection. Figure 13 shows an example where, for example, talent requirement models 1301 to 1303 corresponding to areas 1201 to 1203 in Figure 12 are presented. For example, talent requirement models 1301 to 1303 show the values of each evaluation item included in the passer dataset corresponding to each of the data 1211 to 1213 using radar charts. Talent requirement models can be shown in a format other than radar charts. For example, a talent requirement model can be shown with multiple bar graphs corresponding to each evaluation item. In Figure 13, the number of talent requirement models displayed may correspond to the number of talent requirement models generated by the clustering process. In addition to the displayed talent requirement models, recruiters can refer to information on passers corresponding to each talent requirement model. For example, the evaluation server 101 may display information on the corresponding passers based on the selection of test results 1311 to 1313 by the recruiter. For example, the evaluation server 101 may display video footage of the interview of the corresponding successful candidate. The recruiter may select a talent requirements model to be used for the overall evaluation. For example, the evaluation server 101 may store the corresponding talent requirements model as the talent requirements model to be used for the overall evaluation, based on the selection of talent requirements additions 1321 to 1323 corresponding to each talent requirements model.
[0051] The evaluation server 101 determines an overall evaluation model using the determined personnel requirements model (S903). For example, if multiple personnel requirements models are selected, the evaluation server 101 may generate multiple overall evaluation models using each of the personnel requirements models. For example, if each evaluation item is weighted as a personnel requirement model, the evaluation server 101 may generate an overall evaluation model like the one shown in Equation 1 above. Alternatively, the evaluation server 101 may generate a trained model for each group as an overall evaluation model using the passer dataset included in each group. In this case, the passer dataset included in that group and the passer dataset not included in that group or the unsuccessful applicant dataset may be used as training data. The passer dataset included in that group may contain information with scores assigned to each of the predetermined evaluation items and a value of 1 indicating that it is included in the group. On the other hand, the passer dataset not included in that group or the unsuccessful applicant dataset may contain information with scores assigned to each of the predetermined evaluation items and a value of 0 indicating that it is not included in the group. The evaluation server 101 can output a value in the range of 0 to 1 indicating whether or not a candidate belongs to a group as an overall evaluation by inputting the candidate dataset to be evaluated into the generated trained model. A value closer to 1 is likely to be output as the candidate is more likely to belong to a group. The evaluation server 101 applies each generated overall evaluation model to the candidate dataset to perform an overall evaluation of each candidate (S1104) and outputs the overall evaluation for each overall evaluation model (S1105). For example, the overall evaluations using each talent requirement model (overall evaluation model) may be output together in the format shown in Figure 8.
[0052] (Fourth example of processing) In this example, the evaluation server 101 updates the information stored in the passer information database 121 using the information of the successful candidates, and performs an overall evaluation of the candidates using the updated information. Figure 14 shows an example of the processing flow executed by the evaluation server 101 in this example. In Figure 14, the same reference numbers are used for processes similar to those in Figure 9, and explanations are omitted. Note that this example can be executed in combination with the respective processing examples shown in Figures 9 and 11. In Figure 14, for example, the evaluation server 101 outputs an overall evaluation for each candidate (S405). Based on the overall evaluation output by the evaluation server 101, candidates who pass may be determined (S1401). For example, candidates whose overall evaluation output by the evaluation server 101 exceeds a threshold may be determined as successful candidates. Furthermore, subsequent steps of the recruitment process are performed based on the overall evaluation output by the evaluation server 101, and the final successful candidates may be determined.
[0053] Information on successful candidates may be stored in the successful candidate information database 121 shown in Figure 1. For example, the candidate dataset of successful candidates may be stored in the successful candidate information database 121 as a successful candidate dataset. Information on unsuccessful candidates may also be stored in the successful candidate information database 121. In this case, the candidate dataset of unsuccessful candidates may be stored in the successful candidate information database 121 as an unsuccessful candidate dataset. If the evaluation server 101 updates the information on successful candidates (YES in S1402), it stores the information on successful and unsuccessful candidates in the successful candidate information database 121 (S1403). If the evaluation server 101 does not update the information on successful candidates (NO in S1402), it terminates the process.
[0054] The added information on successful and unsuccessful candidates can be used to create a talent requirements model (S902). For example, the evaluation server 101 can use the information after the addition of successful and unsuccessful candidates to perform a comprehensive evaluation of subsequent test takers. This makes it possible to generate a talent requirements model that responds to changes in the talent required by the target company. The evaluation server 101 may create multiple groups using a new set of datasets that includes the added information on successful candidates (S1101). In this case, the addition of a new dataset of successful candidates may generate new groups formed by clusters of different datasets. In addition, different successful candidates may be selected as successful candidates closer to the center of the cluster. Therefore, a new talent requirements model may be created. In this case, each time new information on successful candidates is added, the evaluation server 101 may add a new talent requirements model while retaining the previously generated talent requirements model, or it may replace the previous talent requirements model with the new one.
[0055] (modified version) The above explanation primarily used examples where the evaluation items included in the examinee's test results were the same as those included in the talent requirements model. However, the evaluation items included in the examinee's test results may be different from those included in the talent requirements model. Furthermore, the number of evaluation items included in the examinee's test results may be different from the number of evaluation items included in the talent requirements model. For example, the evaluation items included in the examinee's test results may be "proactiveness," "initiative," and "execution ability" in Figure 6, while the evaluation item included in the talent requirements model may be "the ability to take initiative." Similarly, the evaluation items included in the examinee's test results may be "problem-finding ability," "planning ability," and "creativity" in Figure 6, while the evaluation item included in the talent requirements model may be "the ability to think things through." In this way, when the evaluation items included in the examinee's test results are sub-concepts of the evaluation items in the talent requirements model, the evaluation server 101 can perform a conversion process between evaluation items, calculating the score of one evaluation item using the score of the other evaluation item. For example, the evaluation server 101 can create a talent requirements model for use in overall evaluation that includes high-level concepts such as "the ability to take initiative," "the ability to think critically," and "the ability to work in a team." On the other hand, the evaluation server 101 can acquire a test-taker dataset that includes low-level concepts such as "proactiveness," "the ability to take initiative," and "the ability to execute." Furthermore, the evaluation server 101 can perform processes such as summing, averaging, and weighting the scores of "proactiveness," "the ability to take initiative," and "the ability to execute" included in the test-taker dataset in order to calculate the score for "the ability to take initiative" based on the test-taker dataset. In this way, the evaluation server 101 can calculate (convert) the scores of each evaluation item included in the talent requirements model from the scores of each evaluation item included in the test-taker dataset. The evaluation server 101 can perform an overall evaluation for each test-taker using the talent requirements model and the converted test-taker dataset. In this way, when the number of evaluation items included in the talent requirements model is less than the number of evaluation items included in the test-taker's test results, it becomes easier for recruiters to set up and select a talent requirements model.
[0056] The evaluation items included in the applicant's test results and the evaluation items included in the talent requirements model may be completely different. For example, the company that provides the interview service for the applicant may be different from the company that provides the overall evaluation of the applicant. In this case, the evaluation server 101 that performs the overall evaluation of the applicant will use the test results that are in line with the evaluation items used by the other company that provides the interview service to perform the overall evaluation of the applicant. For this reason, the evaluation server 101 may have a conversion model created based on the correlation between the evaluation items included in the applicant's test results and the evaluation items included in the talent requirements model. For example, if there is a high correlation between the scores of some of the evaluation items included in the applicant's test results and the scores of specific evaluation items included in the talent requirements model, the evaluation server 101 may use a conversion model that utilizes this correlation to convert the scores of the evaluation items included in the applicant's test results to the scores of the evaluation items included in the talent requirements model. By performing such a conversion process, it becomes possible to perform the overall evaluation of the applicant even if the company that conducts the interview and the company that performs the overall evaluation are independent of each other.
[0057] As described above, according to this embodiment, the evaluation server 101 identifies the weights for each of the predetermined evaluation items for evaluating the applicant and generates an evaluation model that includes these weights. The evaluation server 101 also obtains a set of scores assigned to the applicant based on the employment interview conducted with the applicant. The evaluation server 101 then applies the evaluation model to the set of scores assigned to the applicant to perform an overall evaluation of the applicant and outputs the overall evaluation. With this configuration, it becomes possible to evaluate applicants by assigning weights to the predetermined evaluation items that match the requirements of the company. As a result, it becomes possible to efficiently extract applicants who meet the requirements of the company and have high basic abilities, thereby streamlining the company's recruitment process.
[0058] The invention is not limited to the embodiments described above, and various modifications and changes are possible within the scope of the gist of the invention. [Explanation of Symbols]
[0059] 101: Evaluation server, 102: Interview server, 111: Recruitment staff terminal, 112: Test taker terminal, 121: Successful candidate information database, 122: Test result database, 131: Network
Claims
1. An information processing device for evaluating applicants in job interviews, A means for identifying the weight of each of a predetermined set of evaluation items for evaluating the examinee, A generation means for generating an evaluation model that includes weights for each of the predetermined plurality of evaluation items, A first acquisition means for acquiring a set of scores for at least a portion of the predetermined evaluation items, which are assigned to the applicant based on the employment interview conducted with the applicant; The system includes an output means that outputs an overall evaluation of the examinee, obtained by applying the evaluation model to the set of points assigned to the examinee. An information processing device characterized by the following:
2. The system further includes a second acquisition means for acquiring the set of scores for each of the predetermined multiple evaluation items, which are assigned to candidates who have passed the recruitment process of the target company in the aforementioned recruitment interview. The identification means uses a first set including the set of points assigned to each of the multiple qualifiers to determine the weight for each of the predetermined multiple evaluation items. The information processing apparatus according to feature 1.
3. The identifying means generates a plurality of second sets using at least a portion of the first set, and identifies the weight for each of the predetermined plurality of evaluation items for each of the second sets. The generation means generates the evaluation model for each of the second sets. The information processing apparatus according to feature 2.
4. Each of the aforementioned second sets represents the result of a classification based on the similarity of the sets of scores for each of the predetermined evaluation items with respect to the first set. The information processing apparatus according to claim 3.
5. The system further includes a receiving means for receiving the selection of the evaluation model to be used for the overall evaluation from among the evaluation models generated for each of the second sets. The information processing apparatus according to claim 3.
6. The system further includes an updating means for updating the first set using the set of scores for each of the predetermined multiple evaluation items, which are assigned to the applicants who have passed the recruitment process of the target company. The identification means uses the updated first set to identify the weights for each of the predetermined plurality of evaluation items. The information processing apparatus according to feature 2.
7. The system further includes a receiving means for receiving the setting of a set of weights for each of the predetermined multiple evaluation items, The identifying means identifies the weight set for each of the predetermined plurality of evaluation items as the weight for each of the predetermined plurality of evaluation items. The information processing apparatus according to feature 1.
8. The receiving means receives the settings for each of the multiple sets of weights, each of which has at least some of the weights for each of the predetermined multiple evaluation items different. The generation means generates a plurality of evaluation models corresponding to each of the sets of weights. The information processing apparatus according to feature 7.
9. The identifying means identifies a plurality of sets of weights in which at least some of the weights for each of the predetermined plurality of evaluation items are different, The generation means generates a plurality of evaluation models corresponding to each of the plurality of sets of weights, The output means outputs the overall evaluation of the examinee, obtained by applying each of the multiple evaluation models to the set of scores assigned to the examinee. The information processing apparatus according to feature 1.
10. The aforementioned employment interview is characterized by being an interview of the applicant conducted by artificial intelligence (AI). The information processing apparatus according to claim 1.
11. An evaluation method performed by an information processing device that evaluates applicants for employment interviews, To identify the weights for each of the predetermined evaluation items used to evaluate the examinee, To generate an evaluation model that includes weights for each of the predetermined multiple evaluation items, Obtaining a set of scores for at least some of the predetermined evaluation items, which are assigned to the applicant based on the employment interview conducted with the applicant; This includes outputting an overall evaluation of the examinee, obtained by applying the evaluation model to the set of points assigned to the examinee. An evaluation method characterized by the following.
12. A program for causing a computer to function as one of the means of the information processing device described in claim 1.