Summary generation apparatus, summary generation method, and program

Abstract

To properly generate a summary of a resume.SOLUTION: A summary generation apparatus includes: a summary operation receiving unit which receives, from a user, operation for generating a summary of information used for recruitment; an input information acquisition unit which acquires primary information which is used for generating a summary and has been already input by the user; an extraction unit which extracts an appeal point included in the primary information; a preprocessing unit which generates information in a predetermined data format, on the basis of the primary information acquired by the input information acquisition unit and the appeal point extracted by the extraction unit; and a summary generating unit which generates a summary through inference using a pre-trained machine learning model which uses the information generated by the preprocessing unit as input.SELECTED DRAWING: Figure 14

Description

Technical Field

[0001] The present invention relates to a summary generation device, a summary generation method, and a program.

Background Art

[0002] Conventionally, when an employer or the like newly hires an employee or the like, first, they may make a preliminary judgment on the employability by looking at documents such as the resume and work history of the applicant. Therefore, the resume is useful for the employer or the like to judge the employability. On the other hand, for the applicant or the like, creating a resume takes time and is troublesome. Therefore, as a technology for assisting in creating a resume, a technology for creating an electronic resume with a photo on the network using a multimedia kiosk (MMK) or the like provided in a convenience store or the like is known (for example, see Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] According to the technology as described above, the creation of a resume is assisted by the applicant or the like (applicant) inputting information into a pre-prepared resume format. Here, for the employer who conducts the recruitment work, it is difficult to check every corner of the resumes sent in large quantities, so a "summary" column is often provided in the resume format. However, it is difficult to accurately summarize the appealing points in the limited summary column. Therefore, there has been a demand for the applicant or the like to automatically create a summary in which the appealing points are appropriately reflected in the resume, work history, etc.

[0005] The present invention has been made in view of such circumstances, and an object thereof is to provide a summary generation device, a summary generation method, and a program capable of suitably generating a summary of a resume or a work history.

Means for Solving the Problems

[0006] (1) One aspect of the present invention is a summary generation device including: a summary operation reception unit that receives an operation from a user for generating a summary of information used for human resource recruitment; an input information acquisition unit that acquires primary information which is information used for generating the summary and has already been input by the user; an extraction unit that extracts an appeal point included in the primary information; a preprocessing unit that generates information in a predetermined data format based on the primary information acquired by the input information acquisition unit and the appeal point extracted by the extraction unit; and a summary generation unit that generates a summary by performing inference using a machine learning model that has been pre-trained with the information generated by the preprocessing unit as input.

[0007] (2) Further, one aspect of the present invention is the summary generation device according to (1) described above, wherein the primary information includes a plurality of items and includes information as to whether or not each item has been selected as an appeal point by the user. The extraction unit refers to the primary information and extracts, as an appeal point, an item selected as an appeal point by the user. The number of items that can be selected as an appeal point by the primary information has an upper limit.

[0008] (3) Further, one aspect of the present invention is the summary generation device according to (1) or (2) described above, wherein the primary information includes a selection item in which an arbitrary option is selected from a plurality of options preset for each category, and the appeal point is further selected from the options selected in the selection item.

[0009] (4) Further, in one aspect of the present invention, in the summary generation device described in (3) above, the machine learning model is learned for each of the categories.

[0010] (5) Also, one aspect of the present invention includes a summary operation reception step of receiving an operation from a user for generating a summary of information used for personnel recruitment, an input information acquisition step of acquiring primary information which is information used for generating the summary and has already been input by the user, an extraction step of extracting an appeal point included in the primary information, a preprocessing step of generating information in a predetermined data format based on the primary information acquired in the input information acquisition step and the appeal point extracted in the extraction step, and a summary generation step of generating a summary by performing inference using a machine learning model that has been pre-learned with the information generated in the preprocessing step as input.

[0011] (6) Also, one aspect of the present invention is a program for causing a computer to execute a summary operation reception step of receiving an operation from a user for generating a summary of information used for personnel recruitment, an input information acquisition step of acquiring primary information which is information used for generating the summary and has already been input by the user, an extraction step of extracting an appeal point included in the primary information, a preprocessing step of generating information in a predetermined data format based on the primary information acquired in the input information acquisition step and the appeal point extracted in the extraction step, and a summary generation step of generating a summary by performing inference using a machine learning model that has been pre-learned with the information generated in the preprocessing step as input.

Advantages of the Invention

[0012] According to the present invention, it is possible to provide a summary generation device, a summary generation method, and a program that can suitably generate a summary of a resume.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Mode for Carrying Out the Invention

[0014] [Embodiment] Hereinafter, a preferred embodiment of a summary generation device, a summary generation method, and a program according to an aspect of the present invention will be described in detail with reference to the accompanying drawings. Note that the aspects of the present invention are not limited to these embodiments, and also include those with various modifications or improvements added. That is, the components described below include those that can be easily assumed by those skilled in the art and substantially the same ones, and the components described below can be combined as appropriate. Also, various omissions, substitutions, or changes of the components can be made without departing from the gist of the present invention. Also, in the following drawings, in order to make each configuration easier to understand, the scale, number, etc. of each structure may be different from the scale, number, etc. in the actual structure. Also, in the embodiment described below, the invention will be described by taking the summary of a resume as an example. However, this embodiment is not limited to this example and can be replaced with other application documents such as a curriculum vitae.

[0015] [First Embodiment] First, the first embodiment will be described with reference to FIGS. 1 to 14.

[0016] FIG. 1 is a diagram showing an example of the configuration of the system according to the first embodiment. With reference to this figure, an example of the configuration of the system 1 according to the present embodiment will be described. The system 1 provides a job hunting support service to job seekers who are considering employment (including re-employment and job transfer, etc.). The job hunting support service includes a job transfer support service. An employer of a company considering the employment of employees may request a job applicant who is considering a job transfer to the company to submit a resume (also referred to as a resume. The same applies in the following description). The system 1 according to the present embodiment supports the job transfer of job seekers by automatically generating a resume with a relatively simple operation by the job seekers.

[0017] The system 1 includes a summary generation device 10 and a plurality of user terminal devices 30. In the example shown in the figure, as an example of the plurality of user terminal devices 30, user terminal devices 30-1, 30-2, …, 30-n (n is a natural number of 1 or more) are shown. The plurality of user terminal devices 30 are each connected to the summary generation device 10 via a predetermined communication network NW. The communication network NW includes the Internet, WAN (Wide Area Network), LAN (Local Area Network), public line, provider device, dedicated line, wireless base station, etc.

[0018] The user terminal device 30 is operated by a user. The user terminal device 30 may be, for example, a smartphone, a tablet terminal, or a notebook computer. The user is a job seeker who is looking for an employment destination (including a re-employment destination and a job transfer destination, etc.) using the system 1. The user searches for an employment destination by operating the user terminal device 30. Note that the search for an employment destination performed by the user includes inputting the user's resume, etc., and having the employer search for the user.

[0019] The summary generation device 10 operates by executing a summary generation program. The summary generation device 10 generates a summary of the work history based on the information input by the user terminal device 30. Note that the summary generation program may be implemented as a web application (web app), or may be distributed as an application to the user terminal device 30, installed on each user terminal device 30, and operate on each user terminal device 30.

[0020] FIG. 2 is a diagram showing an example of the screen configuration of the primary information input screen according to the first embodiment. The summary generation program causes, for example, various display screens to be displayed on a display device unit (not shown) provided in the user terminal device 30, and acquires the data selected or input by the job seeker on the screen. Examples of the various display screens include, for example, an employment history input screen, a skill selection screen, and an appeal point selection screen. Each of the various screens will be described in order below.

[0021] FIG. 2(A) is an example of an employment history input screen, and is an example of a display screen for inputting the most recent employment history among the job seeker's history. The employment history input screen shown in the figure includes, as input fields for selecting or inputting data, illustratively, an occupation input field 4a, an affiliated company input field 4b, an employment form input field 4c, and an employment period input field 4d.

[0022] The most recent occupation of the job seeker is input into the occupation input field 4a. In the example shown in the figure, it is configured to be possible to select the most recent occupation of the job seeker from among predetermined occupations. As options for the occupation, for example, sales, general affairs, planning, clerical work, accounting, personnel, sales, customer service, development, and technology can be provided. However, this embodiment is not limited to this example, and necessary occupations can be provided as appropriate according to the actual operation.

[0023] In the affiliated company input field 4b, the applicant inputs the name of the company where the applicant was or is currently affiliated. Since the number of company names is enormous and it is not easy to use a selection input like in the job type input field 4a, it is preferable for the applicant to input the company name as natural language. However, this embodiment is not limited to this example and may be a selection input according to the actual operation. Alternatively, by providing a database for managing company information within the system 1 or connecting to a similar external database via the network NW, it is possible to search and determine the official company name by only inputting a part of the company name in the affiliated company input field 4b.

[0024] In the employment type input field 4c, the employment type at the company entered in the affiliated company input field 4b is input. In the illustrated example, it is configured to be possible to select an arbitrary employment type from a predetermined list of employment types. As options for the employment type, for example, full-time employee, contract employee, dispatched employee, part-time job, etc. can be provided. However, this embodiment is not limited to this example and appropriate employment types can be provided according to the actual operation.

[0025] In the employment period input field 4d, a checkmark indicating whether the applicant is currently employed at the company entered in the affiliated company input field 4b, and the start and end months and years of the employment period are input. Note that the end month and year can be input only when not currently employed, and may be made non-inputtable when currently employed.

[0026] The input items provided on the work history input screen are not limited to the illustrated example. For example, items related to work history such as annual income at the affiliated company may be appropriately input. In addition to the work history input screen, for example, a screen for inputting the applicant's resume including an educational background input screen for inputting the applicant's educational background may be appropriately displayed. Furthermore, when there is no work experience, a screen for inputting information about the job the applicant desires may be displayed.

[0027] FIG. 2(B) is an example of a skill selection screen and an example of a screen for inputting a job seeker's skills. Questions and options related to skills are displayed on the skill selection screen. On the skill selection screen shown in this figure, as exemplary questions related to skills, Question 5a, Question 5c, Question 5e, and Question 5g are displayed. Also, as options for each question, Option 5b, Option 5d, and Option 5f are displayed.

[0028] The questions and options related to skills are created in advance for each job type. Therefore, for example, based on the job type entered in the job type input field 4a of the work history input screen shown in FIG. 2(A), it is preferable to display on the skill selection screen the questions and options related to the skills stored in association with the job type.

[0029] The skill selection screen allows the job seeker to input skills by having the job seeker select one or more applicable options from the options displayed on the screen. At this time, the selected options should be displayed in a form that is easy to visually recognize as being selected, such as making the background color different from the unselected options (it is desirable to make it highlighted). By adopting such a design, not only can more options be listed in a limited space compared to the case of displaying options with a pull-down or check box, but also the selection status (whether it is being selected or not) can be displayed in a more visually recognizable state. Note that the skills that the skill selection screen allows the job seeker to select can include various experiences such as experience fields in industries, operations, or departments in addition to technical capabilities, as well as those related to knowledge and education such as qualifications or degrees held. The job seeker can easily clarify their skills by selecting one or more applicable options from the options according to the questions displayed on the skill selection screen.

[0030] Specifically, Question 5a states, "What tools do you have development / operation experience with in the security system?" The corresponding options in 5b are "AWVS", "Nessus", "SqlMap", "Nmap", "Appscan", "Burp Suite", "X-Way", "AXIOM", and "Others".

[0031] Specifically, Question 5c states, "What regulations / frameworks do you have development / operation experience with in security system development?" The corresponding options in 5d are "ISO27001", "PCI-DSS", "CCPA", "Revised Personal Information Protection Law", and "Others".

[0032] Specifically, Question 5e states, "What target operating systems do you have development / operation experience with?" The corresponding options in 5f are "Windows", "MacOS", "Linux (registered trademark)", and "Other Unix".

[0033] Specifically, Question 5g states, "What languages do you have development experience with?" The corresponding options may be "HTML·CSS", "PHP", "Ruby", "Visual Basic", "Java Script", "Python", "C#", etc. However, since the display area of the screen is limited, they are not shown in the figure. Job seekers can view the presented options by scrolling the screen.

[0034] In addition, for the options displayed on the skill selection screen, further sub-concept options may be provided so that answers to the questions can be selected from among the sub-concept options. In this case, by operating the upper-level option, the sub-options corresponding to the upper-level option may be configured to be viewable.

[0035] FIG. 2(C) is an example of an appeal point selection screen, which is an example of a screen for inputting an applicant's appeal points. On the appeal point selection screen, a plurality of selectable candidates for appeal points are displayed. The applicant inputs the appeal points by selecting one or more appropriate candidates from among the plurality of candidates displayed on the screen.

[0036] As candidates for the appeal points displayed on the appeal point selection screen, for example, the skills selected on the skill selection screen described with reference to FIG. 2(B) can be used. As a result, the applicant can appropriately convey to the employer the skills that the applicant particularly wants to appeal or is confident in among the skills the applicant has selected by selecting one or more skills to appeal from among the skills the applicant has selected.

[0037] On the appeal point selection screen shown in the figure, as selectable candidates for the appeal points, illustratively, candidate 6a regarding the applicant's specialized skills, candidate 6b regarding the qualifications and degrees the applicant holds, candidate 6c regarding the industries and departments the applicant has experience in, and candidate 6d regarding the applicant's specific work experience are displayed.

[0038] Note that the candidates for the appeal points are not limited to the skills selected on the skill selection screen. For example, information the applicant has input in the past, such as information regarding the applicant's work history, may be included as candidates for the appeal points. Also, skills at a lower concept level belonging to the skills selected on the skill selection screen may be added as candidates for the appeal points.

[0039] Specifically, as candidate 6a regarding the applicant's specialized skills, "Docker", "Docker", "Go", "GCS", "Agile / Scrum", "Git", "GCE", and "GKE" are listed as candidates.

[0040] Also, as candidate 6b regarding the qualifications and degrees the applicant holds, "AWS Certified Security" is listed as a candidate.

[0041] In addition, as candidate 6c regarding the industries or departments with which the job seeker has experience, "Service" is listed as a candidate.

[0042] In addition, as candidate 6d regarding the specific work experience of the job seeker, "IAM", "Requirement Definition", and "Design" are listed as candidates.

[0043] The job seeker can select 5 appealing points from among these candidates. Note that the number of selectable appealing points is not limited to an example of 5, but it is preferably finite. If the number of appealing points is too large, the points to be appealed will become blurred, and it will not be easy to create a summary of the resume. In particular, when there is a character limit (including the case of a guideline) in the summary of the resume, it is particularly preferable to limit the number of appealing points.

[0044] FIG. 3 is a flowchart showing an example of the operation flow of the summary generation device according to the first embodiment. While referring to this figure, a series of flows of the processes performed by the summary generation device 10 will be described. In addition, the configuration of the display screen displayed on the user terminal device 30 in each step will be described as appropriate with reference to FIGS. 4 to 10. Note that the display screen is an example when the user terminal device 30 is a smartphone. The display screen may perform a display suitable for the screen size when the user terminal device 30 is a tablet terminal or a notebook personal computer.

[0045] (Step S110) First, the summary generation device 10 displays a job summary input screen on the user terminal device 30. The user can directly input a summary of the resume (hereinafter, may be simply referred to as a job summary) using the touch panel, voice input device, etc. of the summary generation device 10 through this screen.

[0046] FIG. 4 is a diagram showing an example of the screen configuration of the job summary input screen according to the first embodiment. In step S110, a display screen D1 as shown in this figure is displayed. On the display screen D1, the user edits the job summary. The display screen D1 has, as its screen configuration, reference numerals D101, D102, and D103. Reference numeral D101 is a text box in which the user can directly input the job summary. The user enters in the text box indicated as reference numeral D101 his / her previous work experience, things he / she wants to appeal, etc. Input of 400 characters is recommended as a goal for this text box.

[0047] Reference numeral D102 is an AI summary generation execution button. The user starts the AI summary by operating this button. The outline of the AI summary process will be described later.

[0048] Reference numeral D103 is a temporary save button. The user temporarily saves the job summary directly input by the user, the job summary generated by the AI summary, or the job summary generated by the AI summary that the user has modified, by operating this button.

[0049] (Step S120) Returning to FIG. 3, the summary generation device 10 receives an AI summary generation execution operation from the user. The AI summary generation execution operation is received, for example, when the user presses the AI summary generation execution button of reference numeral D102 described above. Here, when the summary generation device 10 receives an AI summary generation execution operation from the user, it may present the precautions for use and seek the user's consent.

[0050] FIG. 5 is a diagram showing an example of the screen configuration of the AI summary generation execution screen according to the first embodiment. As shown in the figure, when the summary generation device 10 receives an AI summary generation execution operation from the user, it displays a display screen D2 presenting the precautions for use and seeks the user's consent.

[0051] In reference numeral D201, "Precautions for Use" is described. Specifically, Please note that the specifications of the AI summary function may be changed or discontinued without prior notice to customers. This function may contain inappropriate language or display information that is not included in your work experience, and may not create a job summary that meets your expectations. This function can be used up to 10 times per day (resets at midnight). The following cautions are stated.

[0052] If the user agrees with the precautions described in reference numeral D201, the user presses the start use button reference numeral D202 to proceed with the process.

[0053] (Step S130) Returning to Figure 3, when an operation to execute AI summary generation is performed, summary generation device 10 determines whether or not to execute AI summary generation. AI summary generation is performed only when sufficient information as described with reference to Figure 2 has been input in advance. In other words, summary generation device 10 refers to the input status of information as described with reference to Figure 2 and determines whether sufficient information has been input to generate a summary of the resume.

[0054] (Step S140) Specifically, summary generation device 10 refers to the input status of items as described with reference to Figure 2, and if it determines that sufficient information has been input to generate a summary of the resume (i.e., step S140; no deficiency), it determines that it is executable and proceeds to step S190. Also, if it determines that sufficient information has not been input to generate a summary of the resume (i.e., step S140; deficiency), it determines that it is not executable and proceeds to step S150.

[0055] (Step S150) If sufficient information has not been entered to generate a resume summary, the summary generation device 10 displays a "generation not possible screen" to the user. The display screen also displays buttons for enhancing the resume, and the user can proceed to the resume input screen by operating the buttons.

[0056] FIG. 6 is a diagram showing an example of the screen configuration of the ungeneratable screen according to the first embodiment. As shown in the figure, when insufficient information for generating a summary of the work history is input, the summary generation device 10 displays a display screen D3, which is an ungeneratable screen, to the user. The display screen D3 includes, as a presentation unit for notifying the user that it is impossible to generate a summary due to insufficient information, a symbol D301, and a symbol D302, which is a button for supplementing the resume, as components of the screen. When the summary generation device 10 detects that the symbol D302 has been operated, it proceeds to the next process.

[0057] (Step S160) Returning to FIG. 3, the summary generation device 10 displays a "Resume Input Promotion Screen" to the user. Specifically, when generating a summary, the user is prompted to input by displaying a question screen for the items that the user has not input among the important input items that are preferably input. The user expands the resume by inputting information according to the instructions on the display screen. By adopting such a form, the user does not need to wonder which item should be expanded among the input items. Also, for the system side, it is possible to prevent the step of "when it is determined that insufficient information for generating a summary of the work history has been input (that is, step S140; insufficient)" in FIG. 2 from being looped by the user expanding input items that are not very important when creating a summary.

[0058] FIG. 7 is a diagram showing an example of the screen configuration of the input promotion screen according to the first embodiment. The figure shows a display screen D4, which is an example of the resume input promotion screen. The display screen D4 includes symbols D401, D402, and D403 as components of the screen.

[0059] Symbol D401 indicates the progress of resume input. In the illustrated example, it is shown to have eight stages of input information. Also, currently, a screen for inputting work experience and academic background, which is the first stage, is shown. Although it is preferable that as much information as possible be input by the user, the summary generation device 10 only needs to be able to obtain sufficient information to generate a summary.

[0060] Symbol D402 indicates a part of the input items for work experience and academic background, which is the first stage. In a specifically illustrated example, a screen for inputting the current annual income is shown. The user selects a range indicating the current annual income from the selected ranges. In the illustrated example, since "<2 million yen" is highlighted, the state is such that "<2 million yen" is selected.

[0061] Symbol D403 is the button to proceed to the next step. The user expands the resume by repeating the input to the displayed items and the operation of the button to proceed to the next step.

[0062] FIG. 8 is a diagram showing an example of the screen configuration of the AI summary generation execution screen after the information according to the first embodiment is enriched. When the expansion of the resume is completed, a display screen D5 as shown in the figure is displayed. The display screen D5 includes symbols D501, D502, D503, and D504 as components of the screen. The display screen D5 is similar to the display screen D1 but is different from the display screen D1 in that it further includes the symbol D504. The phrase "The resume has been enriched" is displayed on the symbol D504. That is, the user can recognize that the resume has been enriched by visually recognizing the display of the symbol D504.

[0063] (Step S170) Returning to FIG. 3, the summary generation device 10 receives an AI summary generation execution operation from the user. The AI summary generation execution operation is received, for example, when the user presses the AI summary generation execution button of the above-described reference sign D502. When the summary generation device 10 receives the AI summary generation execution operation from the user, it may present the precautions for use such as the above-described display screen D2 and seek the user's consent again.

[0064] (Step S180) When the AI summary generation execution operation is received, the resume information is input into the learning model. The learning model may be an AI based on a learning model such as a language model such as a Transformer including GPT (Generative Pretrained Transformer, including GPT-1, GPT-2, GPT-3). Further, the learning model may be an AI based on a multimodal model (for example, including GPT-4, etc.) that can simultaneously understand and process multiple types of data such as text, images, and sounds in addition to the above-described language model. The learning model takes the resume information as input information, makes inferences, and outputs a summary of the resume.

[0065] (Step S190) When the summary of the resume is output as a result of the inference by the learning model, the user confirms the generated product.

[0066] FIG. 9 is a diagram showing an example of the screen configuration of the product confirmation screen according to the first embodiment. When the summary of the resume is output as a result of the inference by the learning model, the summary generation device 10 causes the user terminal device 30 to display a display screen D6 as shown in the figure, thereby allowing the user to confirm the generated product. The display screen D6 has a reference sign D601, a reference sign D602, and a reference sign D603 as screen components.

[0067] The symbol D601 contains a summary of the work history. Specifically, the symbol D601 states, "While working as a full-time employee in the Human Resources / Recruitment department at ABCD Co., Ltd. since November 2018. The recruitment targets include sales, planning / marketing, sales, etc. Also in charge of recruitment interviews such as new graduate interviews and mid-career interviews, and also in charge of human resources system planning for teams of 1 to 20 people. As human resources work, have experience in tasks such as formulating personnel plans, calculating salaries, handling labor issues, and planning human resources systems. Also, as a manager, led and oversaw a team of 5 to 9 subordinates."

[0068] The symbol D602 is a copy button. By operating this copy button, the user can copy the summary of the work history described in the symbol D601 as text data. The copied text data can be used as appropriate according to the user's convenience.

[0069] The symbol D603 is an insert button. By operating this insert button, the user can insert the text data displayed in the symbol D601 into the symbol D501, which is a component of the display screen D5. When the insert button is operated, a display screen in a state where the summary of the work history is inserted is displayed for the symbol D501.

[0070] (Step S200) The user checks the summary of the work history generated by the AI summary generation process, makes corrections if necessary, and saves the data.

[0071] Figure 10 is a diagram showing an example of the screen configuration of the product storage screen according to the first embodiment. The display screen D7 shown in the figure is an example of a display screen in a state where the summary of the work history is inserted for the symbol D501. The display screen D7 has the symbol D701, the symbol D702, and the symbol D703 as screen components. The display screen D7 is similar to the display screen D5, but is different from the display screen D5 in that the summary of the work history generated by the AI summary generation process is displayed in the symbol D701.

[0072] The user edits the text displayed at symbol D701, and after the editing is done, saves the summary of the work history document by operating symbol D703. Note that it is not essential for the summary generation device 10 to have the editing process, and it may be omitted as appropriate, or the edited content may be automatically saved even if the user does not perform the save operation.

[0073] FIG. 11 is a functional configuration diagram showing an example of the functional configuration of the summary generation device according to the first embodiment. With reference to this figure, an example of the functional configuration of the summary generation device 10 for realizing the functions as described above will be described. The summary generation device 10 operates by executing a summary generation program. The summary generation device 10 includes, as functional configurations, a summary operation reception unit 11, an input information acquisition unit 12, a determination unit 13, an information input request unit 14, and a summary generation unit 15.

[0074] The summary operation reception unit 11 receives an operation from the user to generate a summary of information (for example, work history information) used for personnel recruitment. The summary operation reception unit 11 receives, for example, an operation to generate a summary when the user operates symbol D102, which is a button included in the above-described display screen D1. When the summary operation reception unit 11 receives an operation to generate a summary, it instructs the determination unit 13 to start the summary.

[0075] The input information acquisition unit 12 acquires primary information. The primary information is information used for generating the summary and is information that has already been input by the user. Specifically, the primary information is the items as described with reference to FIG. 2. Here, the primary information includes essential input items and optional input items. The essential input items are input items for which it is essential that information is input to generate the summary. The optional input items are input items for which it is not essential that information is input to generate the summary. The essential input items include at least input items related to education or work history.

[0076] Furthermore, the primary information includes selection items and natural language input items. A selection item is an item in which an arbitrary option is selected by the user from a plurality of preset options. The selection items are, for example, the items such as option 5b, option 5d, and option 5f described with reference to FIG. 2(B). A natural language input item is an item in which natural language is input by the user. The natural language input items are, for example, the items such as the affiliated company input field 4b described with reference to FIG. 2(A).

[0077] Based on the primary information acquired by the input information acquisition unit 12, the determination unit 13 determines whether sufficient information for generating a summary has already been input. Specifically, the determination unit 13 may refer to the primary information acquired by the input information acquisition unit 12 and determine that sufficient information for generating a summary has not been input if information has not been input to the essential input items. Note that the determination method of the determination unit 13 is not limited to this example, and other algorithms (for example, machine learning algorithms, etc.) may be used to determine whether sufficient information for generating a summary has already been input.

[0078] When the determination unit 13 determines that sufficient information for generating a summary has not been input, the information input request unit 14 requests the user to input information. Note that the information input request unit 14 may request the user to input all of the primary information, or may request the user to input only the above-described essential input items (or input items including the essential input items). Based on the request made by the information input request unit 14, the user inputs information in addition to the primary information. In some cases, the information generated by the additional input is described as secondary information.

[0079] The summary generation unit 15 generates a summary based on the secondary information generated by the user's additional input. The summary generation unit 15 may generate a summary by machine learning using a language model such as a transformer including GPT. Note that it is preferable for the summary generation unit 15 to generate a summary based on the selection items among the selection items and natural language input items included in the secondary information (or the primary information before addition). For example, when the summary generation unit 15 generates a summary by machine learning, there may be a case where an unintended result is output when making an inference based on the natural language freely input by the user. Therefore, the summary generation unit 15 can generate a highly accurate summary by making an inference based on the selection input items.

[0080] FIG. 12 is a block diagram showing an example of the internal configuration of the summary generation device according to the first embodiment. At least some functions of the summary generation device 10 can be realized using a computer. As shown in the figure, the computer includes a central processing unit 901, a RAM 902, an input / output port 903, input / output devices 904 and 905, etc., and a bus 906. The computer itself can be realized using existing technology. The central processing unit 901 executes instructions included in a program read from the RAM 902 or the like. The central processing unit 901 writes data to the RAM 902, reads data from the RAM 902, and performs arithmetic operations and logical operations according to each instruction. The RAM 902 stores data and programs. Each element included in the RAM 902 has an address and can be accessed using the address. Note that RAM is an abbreviation for "Random Access Memory". The input / output port 903 is a port for the central processing unit 901 to exchange data with external input / output devices and the like. The input / output devices 904 and 905 are input / output devices. The input / output devices 904 and 905 exchange data with the central processing unit 901 via the input / output port 903. The bus 906 is a common communication path used inside the computer. For example, the central processing unit 901 reads and writes data in the RAM 902 via the bus 906. Also, for example, the central processing unit 901 accesses the input / output port via the bus 906. Also, all or part of each functional unit provided in the summary generation device 10 may be realized using hardware such as an ASIC, a PLD, or an FPGA. Also, all or part of each functional unit may be realized by a combination of software and hardware.

[0081] [Summary of the First Embodiment] According to the above-described embodiment, the summary generation device 10 includes a summary operation reception unit 11 to receive from a user an operation for generating a summary of information (e.g., a resume) used for personnel recruitment, and includes an input information acquisition unit 12 to acquire primary information which is information used for generating a summary and has already been input by the user. By including a determination unit 13, it is determined whether sufficient information for generating a summary has already been input based on the primary information acquired by the input information acquisition unit 12. By including an information input request unit 14, when the determination unit 13 determines that sufficient information for generating a summary has not been input, the user is requested to input information. By including a summary generation unit 15, a summary is generated based on secondary information generated by the user adding and inputting information to the primary information based on the request made by the information input request unit 14.

[0082] Here, according to the prior art, the creation of a resume has been assisted by having applicants and the like input information into a pre-prepared resume format. However, the amount of information input into the resume format may vary depending on the applicant and the like, and it is preferable to use resumes with approximately the same amount of information for users and the like to make a fair primary determination of employment eligibility. According to this embodiment, a summary of the resume can be suitably generated based on the primary information that has already been input. Also, the amount of the summary generated by the summary generation device 10 is approximately the same regardless of the type of primary information. Therefore, according to this embodiment, users and the like who check the summary can make a fair primary determination of employment eligibility.

[0083] Also, according to the above-described embodiment, the primary information includes mandatory input items and optional input items. The determination unit 13 refers to the primary information acquired by the input information acquisition unit 12 and determines that sufficient information for generating a summary has not been input when information has not been input into the mandatory input items. Therefore, the summary generation device 10 can easily determine whether the amount of primary information is sufficient.

[0084] Further, according to the above-described embodiment, the mandatory input items at least include input items related to educational background or work experience. That is, the summary generation device 10 generates a summary using educational background or work experience as mandatory input items. Therefore, according to the present embodiment, a summary including at least information about educational background or work experience can be generated. Educational background and work experience are important for the judgment of employment possibility. Thus, according to the present embodiment, a user or the like who checks the summary can fairly make a primary judgment on employment possibility.

[0085] Further, according to the above-described embodiment, the primary information includes a selection item for selecting an arbitrary option from a plurality of preset options and a natural language input item for inputting natural language. The summary generation unit 15 generates a summary based on the selection item. That is, according to the present embodiment, although there is a field for arbitrarily inputting natural language, it is not used for generating the summary, and the summary is generated based on the selection item. Therefore, according to the present embodiment, the risk of generating an unintended summary can be avoided.

[0086] [Second Embodiment] Next, with reference to FIGS. 13 and 14, the second embodiment will be described. The second embodiment is different from the first embodiment in that the data input to the learning model until the summary is generated is shaped to generate a more accurate summary. In the description of the second embodiment, the description of matters already described in the first embodiment may be omitted.

[0087] FIG. 13 is a flowchart showing an example of the operation flow of the summary generation device according to the second embodiment. With reference to this figure, a series of processes performed by the summary generation device 10A according to the second embodiment will be described. In each step described with reference to this figure, the configuration of the display screen displayed on the user terminal device 30 may be the same as that already described as the first embodiment.

[0088] (Step S310) First, the summary generation device 10A receives an AI summary generation execution operation from the user. Step S310 corresponds to Step S120 in the first embodiment. When the summary generation device 10A receives the AI summary generation execution operation from the user, the process proceeds to Step S320.

[0089] (Step S320) Next, the summary generation device 10A checks the resume information (primary information) previously input by the user, and extracts information used for generating the summary, such as the appeal points, work experience, educational background, and expectations. It is preferable that what information to extract is determined in advance. The summary generation device 10A extracts information about predetermined items (for example, the user's appeal points).

[0090] (Step S330) Next, the primary information is formatted into information for inputting to the learning model. The formatting process may be performed, for example, by text processing. In the formatting process, for example, the user's appeal points may be moved to the front and emphasized by attaching specific codes.

[0091] (Step S340) Next, the resume information after formatting is input into the learning model. The learning model makes inferences using the resume information after formatting as input information and outputs a summary of the work history resume.

[0092] (Step S350) When the summary of the work history resume is output as a result of the inference by the learning model, the user checks the generated product.

[0093] (Step S360) The user checks the summary of the work history resume generated by the AI summary generation process, makes corrections if necessary, and saves the data.

[0094] FIG. 14 is a functional configuration diagram showing an example of the functional configuration of the summary generation device according to the second embodiment. With reference to this figure, an example of the functional configuration of the summary generation device 10A for realizing the functions described with reference to FIG. 13 will be described. The summary generation device 10A includes a summary operation reception unit 11, an input information acquisition unit 12, an extraction unit 161, a preprocessing unit 162, and a summary generation unit 15. That is, the summary generation device 10A is different from the summary generation device 10 in that it includes an extraction unit 161 and a preprocessing unit 162 instead of the information input request unit 14. Further, the summary generation device 10A is different from the summary generation device 10 in that it includes a summary generation unit 15A instead of the summary generation unit 15. In the description of the summary generation device 10A, the same components as those of the summary generation device 10 may be denoted by the same reference numerals and the description may be omitted.

[0095] The extraction unit 161 extracts the appeal points included in the primary information pre-input by the user. Here, it is preferable that the primary information includes a plurality of items, and information on whether or not the appeal point has been selected by the user for each item is included. Whether or not it is an appeal point is selected by the user using a display screen as shown in FIG. 2(C). The extraction unit 161 refers to the primary information pre-input by the user and extracts the items selected as appeal points by the user as appeal points.

[0096] Here, when many appeal points are selected by the user, it is not easy to generate a summary that emphasizes the appeal points. Therefore, it is preferable to set an upper limit on the number of items that can be selected as appeal points based on the primary information. As the upper limit number, about 5 is preferably set. Further, as a numerical limit of the appeal points, a lower limit number or a standard number may be determined.

[0097] Also, in this embodiment, the appeal point may be further selected from among the items selected as skills by the user. In other words, it can also be said that in this embodiment, the method of selecting primary information has a hierarchical structure. As the first layer, the skills possessed by the user are selected, and only the skills selected in the second layer are displayed, and the structure is such that the appeal point is selected from among the displayed options. Also, it can be said that the appeal point is further selected from among the options selected in the selection items.

[0098] Based on the primary information acquired by the input information acquisition unit 12 and the appeal point extracted by the extraction unit 161, the preprocessing unit 162 generates information in a predetermined data format. The predetermined data format may be, for example, a text data format, or may be a format in which the appeal point is described in the front, for example.

[0099] The summary generation unit 15A takes the information generated by the preprocessing unit 162 as input and generates a summary by performing inference using a machine learning model that has been pre-trained. The summary generation unit 15A may generate a summary by machine learning using a language model such as a transformer including GPT. Here, it is preferable that the machine learning model used by the summary generation unit 15A for generating the summary is pre-trained for each category. The category may be a classification used when the user selects items suitable for himself or herself, such as educational background, work experience, skills, etc.

[0100] [Summary of the Second Embodiment] According to the above-described embodiment, the summary generation device 10A includes a summary operation reception unit 11 to receive an operation from a user regarding generating a summary of information used for personnel recruitment, and includes an input information acquisition unit 12 to acquire primary information which is information used for generating a summary and has already been input by the user. By including an extraction unit 161, it extracts appeal points included in the primary information. By including a preprocessing unit 162, it generates information in a predetermined data format based on the primary information acquired by the input information acquisition unit 12 and the appeal points extracted by the extraction unit 161. By including a summary generation unit 15A, it takes the information generated by the preprocessing unit 162 as input, and generates a summary by performing inference using a machine learning model that has been pre-trained. By adopting such a configuration, it is possible to generate an accurate summary of a resume based on the already input primary information. According to this embodiment, since an accurate summary of a resume can be generated, the user can shorten the time required for correcting the summary. Also, the user can generate a summary of the resume that accurately reflects the points the user wants to appeal.

[0101] Also, according to the above-described embodiment, the primary information includes a plurality of items and includes information on whether or not each item has been selected as an appeal point by the user. Also, the extraction unit 161 refers to the primary information and extracts items selected as appeal points by the user as appeal points, and there is an upper limit to the number of items that can be selected as appeal points by the primary information. Therefore, according to this embodiment, by setting many appeal points, it is possible to prevent the creation of a summary in which the points to be appealed become ambiguous.

[0102] Also, according to the above-described embodiment, the primary information includes a selection item in which an arbitrary option is selected from a plurality of preset options for each category, and the appeal point is further selected from the options selected in the selection item. In other words, it can also be said that the appeal point is selected in two stages. In the first stage, the user's skills are selected, and in the second stage, the appeal point is selected. According to this embodiment adopting such a configuration, the user can easily generate a summary of the work history document that reflects their own appeal points.

[0103] Also, according to the above-described embodiment, the machine learning model used by the summary generation unit 15A for inference is learned for each category. Therefore, according to this embodiment, more accurate inference can be performed, and a more accurate summary of the work history document can be generated.

[0104] [Third Embodiment] Next, the third embodiment will be described with reference to FIGS. 15 to 17. The third embodiment is different from the first and second embodiments in that, as a result of generating a summary, the summary is generated (regenerated) again according to the user's feedback. In the description of the third embodiment, the description of matters already described in the first or second embodiment may be omitted.

[0105] FIG. 15 is a flowchart showing an example of the operation flow of the summary generation device according to the third embodiment. With reference to this figure, a series of processes performed by the summary generation device 10B according to the third embodiment will be described. In each step described with reference to this figure, the configuration of the display screen displayed on the user terminal device 30 may include the same ones as already described in the first or second embodiment. For the configuration example of the display screen specific to the third embodiment, it will be described separately with reference to the drawings.

[0106] (Step S510) First, the summary generation device 10B receives an AI summary generation execution operation from the user. Step S510 corresponds to Step S120 in the first embodiment. When the summary generation device 10B receives an AI summary generation execution operation from the user, the process proceeds to Step S520.

[0107] (Step S520) When an AI summary generation execution operation is received, the resume information is input into the learning model. The learning model may be an AI equipped with a learning model such as a language model like a Transformer including GPT. The learning model takes the resume information as input information, makes inferences, and outputs a summary of the work history resume.

[0108] (Step S530) When a summary of the work history resume is output as a result of the inference by the learning model, confirmation of the product by the user is performed.

[0109] (Step S540) Here, in the third embodiment, it is different from the first and second embodiments in that there is a step of providing feedback to the product (sometimes simply referred to as FB). In the step of providing feedback to the product, the summary generation device 10B obtains feedback from the user.

[0110] FIG. 16 is a diagram showing an example of the screen configuration of the feedback screen according to the third embodiment. The display screen D8 shown in the figure is the screen displayed to the user in the step of providing feedback to the product. The display screen D8 includes symbol D801, symbol D802, and symbol D803 as components of the screen.

[0111] Symbols D801 and D802 are feedback information acquisition buttons. Specifically, symbol D801 is the positive feedback information acquisition button, and symbol D802 is the negative feedback information acquisition button. By selecting either symbol D801 or D802, the user provides feedback on whether the summary of the displayed resume is good (positive) or not good (negative). After the user selects either symbol D801 or D802, the user submits feedback information to the summary generation device 10B by operating the submission button indicated by symbol D803.

[0112] Note that the feedback information is not limited to information on whether the summary of the displayed resume is good or not, and may include other detailed information. Other detailed information may be, for example, "The text structure is difficult to understand", "There are problems with expression or grammar", "The summary is too long", "The summary is too short", "Skills and achievements are not well presented", "There is incorrect information", "There are parts where information is lacking", "Others", etc., or may be indicated by scoring like "5 - level display" using numbers or the number of "stars", etc. Also, the summary generation device 10B may present these options to the user as detailed information and obtain feedback information based on the user's selection.

[0113] (Step S550) Returning to FIG. 15, the summary generation device 10B determines whether the feedback information is positive. If the feedback information is positive (i.e., Step S550; YES), the summary generation device 10B proceeds to step S580. If the feedback information is negative (i.e., Step S550; NO), the summary generation device 10B proceeds to step S560.

[0114] (Step S560) If the feedback information is negative, the summary generation device 10B performs inference using the learning model again. In the second inference process, the inference may be performed based on the "other information" obtained as the feedback information. Note that in the second inference process, the inference may be performed using a different learning model based on the "other information" obtained as the feedback information.

[0115] (Step S570) Next, the summary generation device 10B presents the summary obtained through the second inference process to the user.

[0116] (Step S580) When the summary of the work history document is output as a result of the inference by the learning model in the second inference process, the user confirms the generated product.

[0117] (Step S590) The user checks the summary of the work history document generated by the AI summary generation process, makes corrections if necessary, and saves the data.

[0118] FIG. 17 is a functional configuration diagram showing an example of the functional configuration of the summary generation device according to the third embodiment. An example of the functional configuration of the summary generation device 10B for realizing the functions as described with reference to FIGS. 15 and 116 will be described while referring to this figure. The summary generation device 10B includes a summary operation reception unit 11, an input information acquisition unit 12, a feedback information acquisition unit 18, a first summary generation unit 151, a second summary generation unit 152, and a presentation unit 17. The summary generation device 10B is different from the summary generation device 10 in that it includes a feedback information acquisition unit 18, a first summary generation unit 151, a second summary generation unit 152, and a presentation unit 17. In the description of the summary generation device 10B, the same components as those of the summary generation device 10 may be denoted by the same reference numerals and the description may be omitted.

[0119] The first summary generation unit 151 acquires information indicating the start of summarization from the summarization operation reception unit 11, and acquires the input information from the input information acquisition unit 12. The first summary generation unit 151 uses the primary information acquired by the input information acquisition unit 12 as input, triggered by the acquisition of information indicating the start of summarization, and performs inference using a machine learning model that has been pre-trained. The first summary generation unit 151 generates a summary by performing inference using the machine learning model. The first summary generation unit 151 provides the result of the generated summary to the presentation unit 17 as first summary information.

[0120] The presentation unit 17 presents the summary generated by the first summary generation unit 151 to the user.

[0121] The feedback information acquisition unit 18 acquires feedback information about the summary presented by the presentation unit 17 from the user. The feedback information acquired by the feedback information acquisition unit 18 at least includes information on whether the user is satisfied with the summary presented by the presentation unit 17. Further, when the user is not satisfied, information about the reason for dissatisfaction may also be included. The information about the reason for dissatisfaction of the user may be information selected from options or text information input by the user.

[0122] The second summary generation unit 152 acquires the input information from the input information acquisition unit 12 and acquires the feedback information from the feedback information acquisition unit 18. The second summary generation unit 152 uses, as inputs, the primary information acquired by the input information acquisition unit 12 and the feedback information acquired by the feedback information acquisition unit 18, and generates a summary by performing inference using a machine learning model that has been pre-trained. The second summary generation unit 152 generates a summary by performing inference using a machine learning model. The second summary generation unit 152 provides the result of the generated summary to the presentation unit 17 as second summary information. Note that the machine learning model used by the first summary generation unit 151 for inference and the machine learning model used by the second summary generation unit 152 for inference may be the same as each other or may be different from each other.

[0123] Note that when the feedback information acquisition unit 18 acquires information about the reason why the user was not satisfied, the second summary generation unit 152 may use, as inputs, the primary information acquired by the input information acquisition unit 12 and the reason why the user was not satisfied included in the feedback information acquired by the feedback information acquisition unit 18, and generate a summary by performing inference using a machine learning model that has been pre-trained. In this case, the machine learning model used by the second summary generation unit 152 for inference may be pre-trained according to the reason why the user was not satisfied.

[0124] For example, when feedback such as "the text structure was difficult to understand" is obtained, inference may be performed based on a machine learning model that is good at creating an easy-to-understand text structure. Also, in advance, a machine learning model characterized by outputting a long summary and a machine learning model characterized by outputting a short summary are prepared, and depending on whether the feedback from the user is "the summary was too long" or "the summary was too short", an appropriate machine learning model may be selected and inference may be performed. In this case, the second summary generation unit 152 may also generate a summary by selecting machine learning models that have been pre-trained based on different teacher data respectively, based on the reason why the user was not satisfied included in the feedback information acquired by the feedback information acquisition unit 18, and performing inference using the selected machine learning models.

[0125] Also, when feedback such as "skills and achievements were not well presented", "there was incorrect information", or "there were parts where information was lacking" is obtained, the user may supplement the information again, and after the second supplementation, summary generation and presentation may be performed for the third and subsequent times.

[0126] [Summary of the Third Embodiment] According to the above-described embodiment, the summary generation device 10B includes a summary operation reception unit 11 to receive an operation from the user regarding generating a summary of information used for personnel recruitment, includes an input information acquisition unit 12 to acquire primary information which is information used for generating a summary and has already been input by the user, includes a first summary generation unit 151 to generate a summary by performing inference using a machine learning model that has been pre-trained with the primary information acquired by the input information acquisition unit 12 as input, includes a presentation unit 17 to present the summary generated by the first summary generation unit 151, includes a feedback information acquisition unit 18 to acquire feedback information regarding the summary presented by the presentation unit 17 from the user, and includes a second summary generation unit 152 to generate a summary by performing inference using a machine learning model that has been pre-trained with the primary information acquired by the input information acquisition unit 12 and the feedback information acquired by the feedback information acquisition unit 18 as input. By adopting such a configuration, it is possible to generate a highly accurate summary of a resume based on feedback from the user. According to this embodiment, since it is possible to generate a highly accurate summary of a resume, the user can shorten the time required for correcting the summary. Further, the user can generate a summary of the resume that reflects the points the user wants to emphasize.

[0127] Also, according to the above-described embodiment, the feedback information at least includes information on whether the user is satisfied with the summary presented by the presentation unit 17, and when the user is not satisfied, it further includes information on the reason for the dissatisfaction. Here, services such as GPT may have inference results that change depending on the timing of inference due to additional learning. Therefore, when the summary generation device 10B performs inference using a service such as GPT, it is assumed that the inference result of the summary generation device 10B will change depending on the timing of inference. Therefore, it is crucial to quickly detect the change. According to this embodiment, in order to obtain feedback from the user, it is possible to detect a change in the learning model based on the feedback.

[0128] Also, according to the above-described embodiment, the second summary generation unit 152 inputs the primary information acquired by the input information acquisition unit 12 and the reason for the user's dissatisfaction included in the feedback information acquired by the feedback information acquisition unit 18, and generates a summary by performing inference using a machine learning model that has been pre-trained. That is, the summary generation device 10B performs inference based on the details of the feedback from the user. Therefore, according to this embodiment, it is possible to generate a more accurate summary of the resume.

[0129] Also, according to the above-described embodiment, the second summary generation unit 152 selects machine learning models that have been pre-trained based on different teacher data respectively based on the reason for the user's dissatisfaction included in the feedback information acquired by the feedback information acquisition unit 18, and generates a summary by performing inference using the selected machine learning model. According to this embodiment, since the summary generation device 10B performs inference using a learning model corresponding to the type of feedback, it is possible to generate a more accurate summary of the resume.

[0130] Note that, for all or some of the functions of each part of the summarization device 10, the summarization device 10A, and the summarization device 10B in the above-described embodiments, a program for realizing these functions may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be read into a computer system and executed. Here, the "computer system" is assumed to include hardware such as an OS and peripheral devices.

[0131] Also, the "computer-readable recording medium" refers to a portable medium such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, etc., and a storage unit such as a hard disk built into a computer system. Further, the "computer-readable recording medium" refers to something that dynamically holds a program for a short time, like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line, and also includes something that holds a program for a certain time, like a volatile memory inside a computer system that serves as a server or a client in that case. Also, the above program may be for realizing a part of the aforementioned functions, and may also be capable of being realized in combination with a program already recorded in the computer system for realizing the aforementioned functions.

[0132] As described above, the embodiments of the present invention have been explained. However, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention.

Explanation of Reference Numerals

[0133] 1... System, 10... Summarization device, 30... User terminal device, 11... Summarization operation reception unit, 12... Acquired information input unit, 13... Determination unit, 14... Information input request unit, 15... Summarization generation unit, 161... Extraction unit, 162... Preprocessing unit, 17... Presentation unit, 18... Feedback information acquisition unit, 151... First summarization generation unit, 152... Second summarization generation unit

Claims

1. A summary operation reception unit that receives an operation for generating a summary of information used for human resource recruitment from a user; An input information acquisition unit that acquires primary information, which is information used for generating the summary and has already been input by the user; An extraction unit that extracts appeal points included in the primary information; A preprocessing unit that generates information in a predetermined data format based on the primary information acquired by the input information acquisition unit and the appeal points extracted by the extraction unit; A summary generation unit that generates a summary by performing inference using a machine learning model that has been pre-trained, with the information generated by the preprocessing unit as input; A summary generation device comprising the above.

2. The primary information includes a plurality of items and includes information on whether or not each item has been selected as an appeal point by the user. The extraction unit refers to the primary information and extracts, as appeal points, the items selected as appeal points by the user. There is an upper limit to the number of items that can be selected as appeal points based on the primary information. The summary generation device according to Claim 1.

3. The primary information includes selection items in which an arbitrary option is selected from a plurality of preset options for each category, and the appeal points are further selected from the options selected in the selection items. The summary generation device according to Claim 1 or Claim 2.

4. The machine learning model is learned for each category. The summary generation device according to Claim 3.

5. A summary generation method including: a summary operation reception step of receiving an operation for generating a summary of information used for human resource recruitment from a user; An input information acquisition step of acquiring primary information, which is information used for generating the summary and has already been input by the user; An extraction step of extracting appeal points included in the primary information; A preprocessing step of generating information in a predetermined data format based on the primary information acquired in the input information acquisition step and the appeal points extracted in the extraction step; A summary generation step of generating a summary by performing inference using a machine learning model that has been pre-trained, with the information generated in the preprocessing step as input. A summary generation method having the above steps.

6. On a computer A summary operation reception step for receiving an operation from a user to generate a summary of information used for personnel recruitment; An input information acquisition step for acquiring primary information which is information used for generating the summary and which has already been input by the user; An extraction step for extracting appeal points included in the primary information; A preprocessing step for generating information in a predetermined data format based on the primary information acquired in the input information acquisition step and the appeal points extracted in the extraction step; A summary generation step for generating a summary by performing inference using a machine learning model that has been pre-trained, with the information generated in the preprocessing step as input; A program for executing the above.

Images