JP7507525B1 - Job application document recommendation system, job application document recommendation method and program - Google Patents

Abstract

[Problem] To efficiently screen documents. [Solution] An employment application document recommendation system that recommends employment application documents learns an annotated learning dataset for training employment application documents, generates a trained model based on the learning results, acquires new employment application documents, and uses the generated trained model to extract points to be checked in document screening for the acquired new employment application documents, and recommends the extracted points to be checked. [Selected Figure] Figure 1

Description

This invention relates to technology that is effective in recruiting personnel.

In recent years, technology related to document screening in personnel recruitment has been attracting attention.
For example, Patent Document 1 discloses a technology that uses AI (Artificial Intelligence) technology to provide a certain evaluation of job applicants based on objective data, thereby reducing the workload of hiring managers.
Furthermore, Patent Document 2 provides a technique that enables efficient management of interview results of multiple types of interviews.

JP 2018-116710 A JP 2018-181257 A

In recruiting, companies and organizations collect application documents from job candidates and screen them. In such document screening, the recruiter manually checks the contents of the application documents and screens the documents.
However, there was a problem in that it was inefficient for recruiters to manually screen applications.
Therefore, there is a demand for technology to efficiently screen documents, in particular technology to automatically extract points in job application documents that hiring managers should check and make recommendations to them.
However, the techniques described in Patent Documents 1 and 2 are not capable of automatically extracting points in job application documents that a hiring manager should check and making recommendations to the hiring manager.

The present invention aims to provide an employment application document recommendation system, employment application document recommendation method, and program that can automatically extract areas of employment application documents that hiring managers should check and recommend them to hiring managers, thereby enabling efficient document selection.

The present invention provides a job application document recommendation system that recommends job application documents,
A learning unit that annotates any part of a training application document by adding tags and metadata , and learns a training dataset that uses the annotated parts and contents as a dataset ;
a model generation unit that generates a trained model based on the training result;
A document acquisition department that acquires new employment application documents;
A document screening extraction unit that compares the annotated parts and contents in the generated trained model with the acquired new employment application documents and extracts parts in the new employment application documents that match or are similar to the annotated parts and contents as confirmation parts to be confirmed in document screening;
a recommendation unit that recommends the extracted confirmation portion;
The present invention provides a job application document recommendation system comprising:

According to the present invention, a trained model that has been trained in advance using training job application documents is used to extract points to be checked in new job application documents and recommend the extracted points to be checked.
As a result, the system automatically extracts the points that hiring managers should check and makes recommendations to them, making it possible to efficiently screen applications.

This invention is in the category of a system, but the same effects and advantages can be achieved with methods and programs.

The present invention makes it possible to efficiently screen documents.

FIG. 1 is a diagram illustrating an overview of a job application document recommendation system 1. A diagram showing the functional configuration of a job application document recommendation system 1. FIG. 2 is a flowchart showing a model generation process executed by a computer 10. FIG. 13 is a flowchart showing a modified example of the model generation process executed by the computer 10. FIG. 11 is a flowchart showing a recommendation process executed by a computer 10. 10 is a diagram illustrating an example of a UI displayed on the person in charge terminal 3. FIG. FIG. 13 is a flowchart showing a modified example of a recommendation process executed by the computer 10. FIG. 11 is a flowchart showing a first model update process executed by the computer 10. FIG. 11 is a flowchart showing a second model update process executed by the computer 10. FIG. 11 is a flowchart showing a change process executed by the computer 10. FIG. 11 is a flowchart showing a reflection process executed by the computer 10. FIG. 11 is a flowchart showing a selection content output process executed by the computer 10. FIG. 13 is a flowchart showing a fictitious selection process executed by the computer 10.

Below, a detailed description of the mode for carrying out the present invention (hereinafter, the embodiment) will be given with reference to the attached drawings. In the following drawings, the same elements are given the same numbers or symbols throughout the description of the embodiment.

[Overview of Job Application Document Recommendation System 1]
Fig. 1 is a schematic diagram for explaining an overview of a job application document recommendation system 1. Components of the job application document recommendation system 1 will be explained based on Fig. 1.
The employment application document recommendation system 1 is a system for recommending employment application documents, which is composed of a computer 10 having at least a server function. In this embodiment, the employment application document recommendation system 1 includes, in addition to the computer 10, a person in charge terminal 3 managed by a recruiter 2 who is in charge of recruiting personnel for a company, organization, etc.
The computer 10 has a server function and may be realized, for example, by a single computer, or may be realized by multiple computers such as a cloud computer.
In this specification, a cloud computer may refer to either a computer that uses any computer in a scalable manner to perform a specific function, or a computer that includes multiple functional modules to realize a system, the functions of which can be freely combined.
The staff member terminal 3 is a terminal device managed by the recruiter 2, and is, for example, a tablet terminal, a personal computer, a laptop computer, etc. The number of staff member terminals 3 may be a number corresponding to the number of recruiters 2, is not particularly limited, and can be designed as appropriate.

This section provides an overview of the processing steps that the job application document recommendation system 1 follows when recommending job application documents.

The computer 10 learns the annotated learning data set from the learning job application documents (step S1).
The computer 10 acquires a training application document from an external system, etc. The computer 10 outputs the acquired training application document to a worker terminal (not shown) managed by a worker who performs annotation, and accepts annotations by the worker. The computer 10 generates a training data set based on the accepted annotations, and learns from this training data set.

The computer 10 generates a trained model based on the learning results (step S2).
The computer 10 generates a trained model using a predetermined algorithm.

The computer 10 acquires new employment application documents (step S3).
The computer 10 acquires new employment application documents from a candidate terminal (not shown) managed by the employment candidate or an agent terminal (not shown) managed by the hiring agent.

The computer 10 uses the generated trained model to extract points that should be checked during document screening for the acquired new employment application documents (step S4).
The computer 10 compares the new employment application documents with the annotated parts in the trained model and the content of the annotations for these parts, and extracts the parts to be confirmed.

The computer 10 recommends the extracted check points (step S5).
The computer 10 recommends the extracted check points to the recruiter 2 via a specified UI (User Interface).

The above is an overview of the processing steps of the job application document recommendation system 1.
According to this employment application document recommendation system 1, it is possible to efficiently screen documents.

[Device configuration]
Fig. 2 is a block diagram showing the configuration of the job application document recommendation system 1. The device configuration of the job application document recommendation system 1 will be described with reference to Fig. 2.
The employment application document recommendation system 1 is a system that recommends employment application documents, and is composed of at least a computer 10. In this embodiment, the employment application document recommendation system 1 is composed of, in addition to the computer 10, a recruiter terminal 3 managed by a recruiter 2 who is in charge of recruiting personnel for a company, organization, etc.
The job application document recommendation system 1 is a system in which a computer 10 is connected to a person in charge terminal 3 so as to be capable of data communication via a network 8 such as a public line network.
In addition to the computer 10 and staff terminal 3 described above, the employment application document recommendation system 1 may also include other terminals and devices such as a worker terminal (tablet terminal, personal computer, laptop computer, etc.) managed by a worker who annotates employment application documents, a candidate terminal (tablet terminal, personal computer, laptop computer, etc.) managed by a candidate for employment, an agent terminal (tablet terminal, personal computer, laptop computer, etc.) managed by a hiring agent, an external system with a server function for storing employment application documents for study, and an external system with a server function for storing job data and document screening result information from companies, organizations, etc., and the number, type, and functions thereof are not particularly limited and can be designed as appropriate.

The computer 10 has a server function and may be realized, for example, by a single computer, or may be realized by multiple computers such as a cloud computer.
The computer 10 includes, as a control unit, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), etc., and, as a communication unit, includes a device for enabling communication with other terminals and devices, a document acquisition unit 11 for acquiring new employment application documents, and a recommendation unit 12 for recommending extracted confirmation points.
The computer 10 includes a storage unit such as a hard disk, a semiconductor memory, a recording medium, a memory card, or the like for storing data.
The computer 10 includes, as processing units, various devices for executing various processes, a learning unit 13 for learning an annotated learning data set from training employment application documents, a model generation unit 14 for generating a trained model based on the training results, and a document screening extraction unit 15 for using the generated trained model to extract points to be checked during document screening from the obtained new employment application documents.

In computer 10, the control unit loads a specified program and works in cooperation with the communication unit to realize a learning document acquisition module, a learning document output module, a learning annotation reception module, a new document acquisition module, a job data acquisition module, a recommendation module, an evaluation acquisition module, a comment acquisition module, a conditions reception module, a past result acquisition module, a selection content output module, a fictitious document output module, a fictitious annotation reception module, and an evaluation result output module.
In addition, in the computer 10, the control unit reads a predetermined program and cooperates with the storage unit to realize a storage module.
In addition, in computer 10, the control unit reads a specified program and works in cooperation with the processing unit to realize a learning dataset generation module, a learning module, a model generation module, a document selection extraction module, a manner change module, a reason generation module, an interview selection extraction module, a confirmation point change module, a reflection module, a reflection result output module, a selection content generation module, a fictitious document generation module, a fictitious selection extraction module, a comparison module, an evaluation module, and an improvement comment generation module.

The person in charge terminal 3 is a terminal device managed by a recruiter 2 who is in charge of recruiting personnel for a company, organization, etc., and is, for example, the terminal device described above.
The staff terminal 3 is equipped with a CPU, GPU, RAM, ROM, etc. as a terminal control unit, a device for enabling communication with other terminals and devices, etc. as a communication unit, and various devices for performing input and output of specified data, etc. as an input/output unit.

Below, each process executed by the job application document recommendation system 1 will be described together with the processes executed by each of the modules described above.
In this specification, each module may execute its processing contents as its own function, or may execute its processing contents via a predetermined application.

[Model Generation Process Executed by Computer 10]
The model generation process executed by the computer 10 will be described with reference to Fig. 3. The figure is a flowchart showing the model generation process executed by the computer 10. This model generation process is a detailed description of a learning process (step S1) for learning an annotated learning data set from the above-mentioned learning employment application documents, and a model generation process (step S2) for generating a trained model based on the learning results.

The study document acquisition module acquires study job application documents (step S10).
The training application documents are documents used in document screening and interview screening in personnel recruitment, such as resumes, CVs, skill sheets, and reference sheets, and are intended to be used for generating a trained model, which will be described later. The training application documents are data in a predetermined data format (PDF (Portable Document Format), document file, spreadsheet file, etc.).
The study document acquisition module acquires study application documents from an external system (not shown) having a server function for storing study application documents.
The external system transmits the study employment application documents to the computer 10 based on a request from the computer 10 or a specified input from an administrator of the external system or the like.
The study document acquisition module receives the study job application document and acquires the study job application document.

The study document output module outputs the acquired study job application documents (step S11).
The learning document output module transmits the learning application document to a worker terminal (not shown) managed by the worker who performs the annotation. The worker terminal receives the learning application document and displays it on its display unit.
The study document output module outputs the acquired study job application document by displaying the acquired study job application document on the worker terminal.

The learning annotation receiving module receives annotations for the acquired learning job application documents (step S12).
Annotation is the addition of tags and metadata to a learning employment application document. Annotations include symbols such as ◯, ×, and △, scores, ranks, and the number of stars. Specific examples of annotations include adding ◯ to positive parts, × to negative parts, and △ to parts that are neither positive nor negative, adding scores to any part according to its importance (higher scores for more important parts, lower scores for less important parts), adding ranks according to its importance (higher ranks for more important parts, lower ranks for less important parts), and adding a number of stars according to its importance (more stars for more important parts, fewer stars for less important parts, according to a preset number of stars (maximum 5, minimum 1, etc.)).
The worker terminal accepts annotation input for the study employment application document from the worker. For example, the worker terminal accepts input of each of ◯, ×, and △ from the worker in any part of the study employment application document (part that the recruiter 2 should check in document screening or interview selection (educational history, work history, license, qualification, job content, skills, personality, strengths, weaknesses, achievements, participating projects, etc.)). The worker taps any part (part that the recruiter 2 should check, etc.) in the study employment application document displayed by the worker terminal, and inputs either ◯, ×, or △ in this part by direct input using a virtual keyboard or by selection input from options. The worker terminal may directly annotate the study employment application document, or may annotate the study employment application document read by a specified application and displayed on a specified UI.
The worker terminal transmits the location where the annotation was accepted and the content of the annotation to the computer 10 as learning annotation data.
The learning annotation receiving module receives this learning annotation data and receives annotations for the acquired learning job application documents.

It is also possible for the computer 10 itself to be configured to annotate the study-use job application documents. In this case, it is possible to annotate the study-use job application documents based on a predetermined character string that has been set in advance and the contents of the annotation that has been set in advance for this character string.

The learning dataset generation module generates an annotated learning dataset for the learning job application documents (step S13).
The learning dataset generation module generates a learning dataset based on the received annotations by treating the parts of the learning job application documents where annotations were received and the contents of the annotations as a dataset.
The computer 10 repeats the above-described processes of steps S11 to S13 until a training data set is generated in an amount necessary for training the training data set, which will be described later.

The learning module learns the annotated learning data set against the training job application documents (step S14).
The learning method executed by the learning module may be machine learning using supervised learning, unsupervised learning, reinforcement learning, etc., or deep learning using a convolutional neural network, a recurrent neural network, long/short-term memory, etc. In this embodiment, machine learning using supervised learning will be described as an example.
The learning module performs supervised machine learning using the generated learning data set, and learns the annotated learning data set from the training employment application documents. The learning module performs learning using the parts where annotations have been accepted and the contents of the annotations as teachers, and learns the parts of the employment application documents that should be checked by recruiter 2 and the contents of those annotations.

The model generation module generates a trained model based on the learning results (step S15).
The model generation module generates a trained model using a predetermined algorithm (linear regression, decision tree, logistic regression, k-nearest neighbor method, support vector machine, etc.) based on the learning results.
The algorithm used in the trained model can be designed appropriately depending on the learning method, etc., and is not limited to the example described above.

The memory module stores the generated trained model (step S16).

This completes the model generation process.
The job application document recommendation system 1 performs the processing described below using the trained model generated by the model generation processing.

[Modification of the model generation process executed by the computer 10]
A modified example of the model generation process executed by the computer 10 will be described with reference to Fig. 4. The figure is a flowchart showing a modified example of the model generation process executed by the computer 10. This modified example is a process that is performed when requested by the recruiter 2. This modified example is a process that is performed after execution of the generation process of a trained model by at least the model generation process. Detailed explanations of processes similar to those described above will be omitted.

The study document acquisition module outputs study job application documents (step S20).
The study document acquisition module transmits the study job application documents acquired by the process of step S10 described above to the person in charge terminal 3 based on a request from the person in charge terminal 3 or the like.
The person in charge terminal 3 receives the study-use employment application documents and displays them on its own display unit.
The study document output module outputs the study job application document by displaying the acquired study job application document on the person in charge terminal 3.

The learning annotation receiving module receives annotations for the learning job application documents (step S21).
The process of step S21 is a process executed by the person in charge terminal 3, which is the same as the process executed by the worker terminal in the process of step S12 described above.

The learning dataset generation module generates an annotated learning dataset for the learning job application documents (step S22).
The process of step S22 is similar to the process of step S13 described above.
In this modified example, the computer 10 may repeat the processing of steps S20 to S22 described above until a learning data set of a quantity necessary for learning the learning data set described below is generated, or may repeat the processing of steps S20 to S22 described above until a predetermined number is satisfied, or may perform the processing of steps S20 to S22 described above only once without repeating the processing.

The learning module adds the learning data set based on the annotations received for the training job application documents to the learning data set annotated for the training job application documents, and performs learning (step S23).
The learning module adds the learning data set generated by the processing of step S22 to the learning data set generated by the processing of step S13 described above, and learns the annotated learning data set against the training job application documents.
The learning method executed by the learning module is similar to the process in step S14 described above.

The model generation module generates a trained model based on the learning results (step S24).
The process of step S24 is similar to the process of step S15 described above.

The memory module stores the generated trained model (step S25).

The above is a modified example of the model generation process.
By performing annotations, the hiring manager 2 can improve the accuracy of the selection process during the document screening and interview screening described below.

[Recommendation process executed by computer 10]
The recommendation process executed by the computer 10 will be described with reference to Fig. 5. The figure is a diagram showing a flowchart of the recommendation process executed by the computer 10. This recommendation process is a detailed description of the document acquisition process (step S3) for acquiring the above-mentioned new employment application documents, the document screening extraction process (step S4) for extracting confirmation points to be confirmed in document screening from the acquired new employment application documents using the generated trained model, and the recommendation process (step S5) for recommending the extracted confirmation points.

The new employment document acquisition module acquires new employment application documents of the employment candidate (step S30).
New employment application documents are documents used for document screening and interview selection in personnel recruitment, such as resumes, skill sheets, and reference sheets. These new employment application documents are data in a specified data format (PDF, document file, spreadsheet file, etc.). Note that learning employment application documents and new employment application documents are different.
The new document acquisition module acquires the new employment application documents from a candidate terminal (not shown) managed by the employment candidate or an agent terminal (not shown) managed by the hiring agent.
The candidate terminal or agent terminal transmits the new employment application documents to the computer 10 based on predetermined input from the candidate or the hiring agent.
The new employment document acquisition module receives the new employment application documents and acquires the new employment application documents of the employment candidate.

The job offer data acquisition module acquires job offer data (step S31).
The employment data is data on employment opportunities at companies, organizations, etc., and includes information on job content, application qualifications, work location, salary, working hours, working days of the week, job type, number of positions available, annual holidays, treatment, employee benefits, etc.
The job data acquisition module acquires job data from an external system (not shown) that manages job data of companies, organizations, etc.
The external system transmits job vacancy data to the computer 10 based on a request from the computer 10 .
The job data acquisition module receives the job data and acquires the job data.
The computer 10 may omit the process of step S31. In this case, the computer 10 may execute the process described below with a configuration that excludes the job offer data.

The document screening extraction module uses the generated trained model to extract points that should be checked during document screening from the acquired new employment application documents (step S32).
The document screening extraction module extracts check points from the acquired new employment application documents using the generated trained model and the acquired job data. The trained model in step S32 is based on the learning results of the check points to be checked in the document screening.
Things to check include, for example, whether the candidate holds excellent qualifications, has ample management experience, whether the candidate has a strong tendency to be a job hopper, which is risky, whether the candidate makes many typos and is therefore easily distracted, etc.
The document screening extraction module extracts any part in the new employment application documents that matches the job data, compares the extracted part with the part to be extracted in the trained model and the annotation of this part, and extracts the part that matches or is similar as the confirmation part. When extracting the confirmation part from the new employment application documents, the document screening extraction module extracts not only the location of the confirmation part in the new employment application documents, but also the content of the annotation in the trained model that corresponds to this part.
The document screening extraction module may extract the check points using only the generated trained model. In this case, the document screening extraction module compares the new employment application documents with the points to be extracted in the trained model and the annotations of these points, and extracts those that match or are similar as check points.

The display mode change module changes the display mode of the extracted check portion based on a predetermined condition (step S33).
The predetermined conditions are, for example, positive, negative, and importance. Positive refers to desirable aspects of the employment candidate (qualifications, skills, work history, projects participated in, etc.). Negative refers to undesirable aspects of the employment candidate (rewards and punishments, reasons for leaving, typos, etc.). Importance refers to the aspects that the hiring manager 2 places importance on (whether or not the candidate has qualifications, whether or not the candidate has work history, etc.), and may be weighted and ranked for each aspect, or may be equal without weighting. The predetermined conditions may be set in advance by the hiring manager 2, may be set based on job recruitment data, or may be set by other methods.
The display mode is, for example, character color (highlight, colored, etc.), character size (large, medium, small, etc.), and character thickness (bold, thin, medium, etc.).
The mode change module changes the display mode of the extracted confirmation points to a positive one. For example, the mode change module changes at least one of the character color, character size, and character thickness to a more noticeable one (for example, the character color is highlighted with red or a bright color, for example, the character size is made larger, and for example, the character thickness is made thicker). The mode change module changes the display mode of the extracted confirmation points to a negative one. For example, the mode change module changes at least one of the character color, character size, and character thickness to a less noticeable one (for example, the character color is highlighted with gray or a dark color, for example, the character size is made smaller, and for example, the character thickness is made thinner). The mode change module changes the display mode of the extracted confirmation points to a display mode according to the importance. For example, the mode change module changes at least one of the character color, character size, and character thickness to a more noticeable one for a higher importance and a less noticeable one for a lower importance.
The display mode and the predetermined conditions are not limited to the above-mentioned examples, and can be designed as appropriate. The method of changing the display mode performed by the mode change module is also not limited to the above-mentioned examples, and can be designed as appropriate.

The reason generation module generates a check reason for checking the extracted check point (step S34).
The reason for checking is the reason why the recruiter 2 is checking the check items, for example, whether it matches the recruitment data, whether the person possesses skills and qualifications, etc.
The reason generation module generates a verification reason for the extracted items that correspond to the predetermined required contents of the employment candidate (qualifications, skills, work history, projects participated, etc.).

The recommendation module recommends the extracted check points (step S35).
The recommendation module recommends the extracted check locations, together with the generated check reasons for the check locations, to the recruiter 2 via a predetermined UI. The recommendation module associates the check reasons with the check locations.
The recommendation module transmits the extracted check points and the check reasons linked to the check points to the person in charge terminal 3.
The person in charge terminal 3 receives the check points and check reasons and displays them on its own display unit using a specific UI (see FIG. 6). The hiring person 2 views the check points and check reasons displayed using this specific UI.
The recommendation module recommends the extracted checking points by displaying the checking points and the checking reasons on the person in charge terminal 3.
The recommendation module can also be configured to recommend only the checked points.

The UI displayed by the person in charge terminal 3 will be described with reference to Fig. 6. This figure is a diagram showing a model of the UI displayed by the person in charge terminal 3. In this figure, a UI 20 is shown.
The UI 20 includes a check location field 21 for displaying the extracted check location, and a check reason field 22 for displaying the generated check reason.
The person in charge terminal 3 displays text (sentences, phrases, words, etc.) corresponding to the extracted check point in a check point field 21. When the person in charge terminal 3 receives a predetermined input (such as a tap operation) from the hiring manager 2 into this check point field 21, it displays a check reason field 22 as a pop-up at a predetermined position in the UI 20 (such as near the check point field 21 where the tap operation was received). In addition to the generated check reason, the check reason field 22 also displays the reason for changing the display mode, such as positive, negative, importance, etc. (in the figure, a circle indicating a positive state is shown).
The hiring manager 2 looks at the check points column 21 and the check reason column 22 and understands the check points and check reasons.

That's all about recommendation processing.

[Modification of recommendation process executed by computer 10]
A modified example of the recommendation process executed by the computer 10 will be described with reference to Fig. 7. The figure is a flowchart of a modified example of the recommendation process executed by the computer 10. This modified example is a process for recommending points to be checked during interview selection. Detailed explanations of processes similar to those described above will be omitted.

The interview selection extraction module uses the generated trained model to extract points that should be checked during interview selection from the acquired new employment application documents (step S40).
The trained model in step S40 is based on the learning results of the points to be checked during interview selection.
Things to check include, for example, whether the candidate holds excellent qualifications, has ample management experience, whether the candidate has a strong tendency to be a job hopper, which is risky, whether the candidate makes many typos and is therefore easily distracted, etc.
The interview selection extraction module compares the new employment application documents with the parts to be extracted in the trained model and the annotations for these parts, and extracts those that match or are similar as confirmation parts. When extracting confirmation parts from new employment application documents, the interview selection extraction module extracts not only the positions of the confirmation parts in the new employment application documents, but also the contents of the annotations in the trained model that correspond to these parts.
The interview selection extraction module may extract check points from the acquired new employment application documents using the acquired job data in addition to the generated trained model. In this case, the interview selection extraction module may extract any part in the new employment application documents that matches the job data, compare the extracted part with the part to be extracted in the trained model and the annotation of this part, and extract the parts that match or are similar as check points.

The display mode change module changes the display mode of the extracted check portion based on a predetermined condition (step S41).
The process of step S41 is similar to the process of step S33 described above.

The reason generation module generates a check reason for checking the extracted check point (step S42).
The reason for checking is the reason why the recruiter 2 checks the check items, such as matching the image of the employee, willingness to join the company, matching of the hiring conditions, etc. with the wishes of the candidate, etc.
The reason generation module generates the reason for confirmation based on the intention for the extracted confirmation points that correspond to the contents (personality, motivation to join the company, conditions, etc.) that are asked in advance during the interview selection process.

The recommendation module recommends the extracted check points (step S43).
The process of step S43 is similar to the process of step S35 described above.

The above is a modified example of the recommendation process.
The computer 10 may simultaneously recommend both the points to be checked in the document screening process and the points to be checked in the interview screening process in the recommendation process, or may recommend the points to be checked using only the new employment application documents of candidates who have advanced to the interview screening process.

[First model update process executed by computer 10]
The first model update process executed by the computer 10 will be described with reference to Fig. 8. Fig. 8 is a flowchart showing the first model update process executed by the computer 10.

The evaluation acquisition module acquires an evaluation of the recommended check points (step S50).
The evaluation may be in a text format such as good or bad, in a symbolic format such as O, X, or △, in a ranking format such as the number of stars, or in some other manner.
The evaluation acquisition module acquires this evaluation from the person in charge terminal 3.
The person in charge terminal 3 accepts input of evaluations for the recommended check points (check points in document screening and check points in interview screening). The person in charge terminal 3 accepts input of evaluations from the hiring manager 2 using a predetermined button, icon, input form, UI, etc. The person in charge terminal 3 transmits the accepted evaluations to the computer 10.
The evaluation acquisition module receives the evaluation and acquires an evaluation of the recommended check points.

The learning module adds the recommended check points and the obtained evaluations to a learning data set and performs re-learning (step S51).
The learning module adds the recommended check points and the obtained evaluation to the learning data set generated by the process of step S13 or S22. Here, the recommended check points correspond to the points in the learning data set that have received annotations, and the obtained evaluation corresponds to the content of the annotations in the learning data set.
The learning module re-learns using the learning dataset to which the recommended check points and obtained evaluations have been added.
The re-learning method executed by the learning module may be the same as the learning method in the process of step S14 described above.

The model generation module updates the trained model based on the re-learning results (step S52).
The model generation module updates the trained model based on the re-learning result by using a predetermined algorithm in the processing of step S15 described above. When the model generation module updates the trained model, it may mean changing the content of the existing trained model based on the current re-learning result, or it may mean deleting the existing trained model and generating a new trained model.

The memory module stores the updated trained model (step S53).

This completes the first model update process.

[Second Model Update Process Executed by Computer 10]
The second model update process executed by the computer 10 will be described with reference to Fig. 9. Fig. 9 is a flowchart showing the second model update process executed by the computer 10.

The comment acquisition module acquires comments on the recommended check points (step S60).
Comments can be in text format, such as good, bad, helpful, not helpful, etc.
The comment acquisition module acquires the comment from the person in charge terminal 3.
The person in charge terminal 3 accepts input of comments on the recommended check points (check points in document screening and check points in interview screening). The person in charge terminal 3 accepts input of comments from the hiring manager 2 through a predetermined input form, UI, etc. The person in charge terminal 3 transmits the accepted comments to the computer 10.
The comment acquisition module receives the comments and acquires the comments on the recommended confirmation points.

The learning module adds the recommended check points and the acquired comments to a learning dataset and performs re-learning (step S61).
The learning module adds the recommended check points and the acquired comments to the learning data set generated by the process of step S13 or S22. Here, the recommended check points correspond to the points in the learning data set that have received annotations, and the acquired comments correspond to the contents of the annotations in the learning data set.
The learning module re-learns using the learning dataset to which the recommended check points and acquired comments have been added.
The re-learning method executed by the learning module may be the same as the learning method in the process of step S14 described above.

The model generation module updates the trained model based on the re-learning results (step S62).
The model generation module updates the trained model based on the re-learning result by using a predetermined algorithm in the processing of step S15 described above. When the model generation module updates the trained model, it may mean changing the content of the existing trained model based on the current re-learning result, or it may mean deleting the existing trained model and generating a new trained model.

The memory module stores the updated trained model (step S63).

This completes the second model update process.

The computer 10 may perform a combination of the first model update process and the second model update process described above. In this case, the computer 10 obtains evaluations and comments for the recommended check points, adds them to the learning data set, and performs re-learning.

[Change Processing Executed by Computer 10]
The change processing executed by the computer 10 will be described with reference to Fig. 10. Fig. 10 is a flowchart showing the change processing executed by the computer 10.

The condition receiving module receives input of predetermined conditions for the check points to be extracted (step S70).
The predetermined conditions relate to the amount of check points to be extracted (check points in document screening and check points in interview screening), and are a number of preset modes (harsh mode/normal mode/lenient mode, simple mode/normal mode/detailed mode, etc.). For example, in the simple mode, only important check points such as positive check points and check points with high importance are extracted from the check points, in the normal mode, more check points than in the simple mode are extracted, such as positive check points, negative check points, and check points with a higher importance than a predetermined level are extracted from the check points, and in the detailed mode, more check points than in the normal mode are extracted, such as positive check points, negative check points, and check points of all importance levels are extracted from the check points.
The staff terminal 3 accepts selection input (tap operation, input via a virtual keyboard, etc.) for each mode through options, lists, direct input, etc. displayed on its display unit using a specified input form or UI, etc., and accepts input of the mode desired by the hiring staff 2.
The person in charge terminal 3 transmits the accepted mode to the computer 10 .
The condition receiving module receives this mode and receives input of a predetermined condition for the confirmation portion to be extracted.

The check portion change module changes the check portions to be extracted based on a predetermined condition (step S71).
The check portion change module changes the amount of check portions to be extracted in the processes of steps S32 and S40 described above in accordance with the received mode.
In the processes of steps S32 and S40 described above, the computer 10 extracts the confirmation points according to this changed extraction amount.

That's all for the change process.

[Reflection process executed by computer 10]
The reflection process executed by the computer 10 will be described with reference to Fig. 11. Fig. 11 is a flowchart showing the reflection process executed by the computer 10.

The reflection module reflects the recommendation results in the new employment application documents (step S80).
The reflection module reflects the check points extracted from the new employment application documents (check points in document screening and check points in interview selection) in the new employment application documents. The reflection module reflects the changed display mode of the extracted check points in the new employment application documents. Furthermore, the reflection module also reflects the generated check reasons for these check points. The reflection module changes the display mode of the check points in the new employment application documents, links the check reasons to these check points, and reflects the recommendation results in the new employment application documents.

The reflection result output module outputs the new employment application documents reflecting the recommendation results in a predetermined format (step S81).
The predetermined format is, for example, a PDF, a document file, a spreadsheet file, etc. This predetermined format may be set in advance or may be set by the recruiter 2.
The reflection result output module outputs the new employment application documents in a prescribed format in which the display mode of the confirmation points has been changed and the confirmation reasons are linked to these confirmation points.

This is the reflection process.
The computer 10 can also be configured to output the new employment application documents reflecting the recommendation results to the person in charge terminal 3, etc. The person in charge terminal 3 receives the new employment application documents sent by the computer 10 and displays them on its display unit. The recruiter 2 views the new employment application documents reflecting the recommendation results, and displays the confirmation reason linked to the confirmation section by accepting a predetermined input operation (tapping operation, etc.) for the confirmation section.

[Selection Content Output Process Executed by Computer 10]
The selection details output process executed by the computer 10 will be described with reference to Fig. 12. Fig. 12 is a flowchart showing the selection details output process executed by the computer 10.

The past result acquisition module acquires information on actual past document screening results (step S90).
The document screening result information is information on the results of document screening conducted in the past on employment application documents, such as which employment application documents passed the document screening, the basis for the document screening decision, and the reasons for the document screening decision.
The past result acquisition module acquires document screening result information from an external system that stores document screening result information of companies, organizations, etc.
The external system transmits document screening result information to the computer 10 based on a request from the computer 10 .
The past result acquisition module receives the document screening result information and acquires the past actual document screening result information.

The selection content generation module generates a handover comment of the selection pass rate, the actual selection result, and the selection reason based on the extracted check points, the recommendation result, and the obtained document selection result information (step S91).
The recommendation result in step S91 is the confirmation reason (positive, negative, importance, etc.).
The selection pass rate is the probability (percentage, ratio, etc.) of actually passing the document screening. The actual selection result is the result of the actual document screening of new employment application documents (passed, rejected, etc.). The handover comment is the basis or reason for the decision result of the document screening.
The selection content generation module compares the extracted check points and recommendation results with the document screening result information and calculates the percentage of match or approximation. The selection content generation module generates a selection pass rate based on the calculated percentage. The selection content generation module compares the extracted check points and recommendation results with the document screening result information, compares the new employment application documents with past actual employment application documents and the results of document screening for these employment application documents, and judges new employment application documents that meet the preset criteria (the number of check points is a predetermined number or more, the number of positive recommendation results is a predetermined number or more, the importance is a predetermined number or more, etc.) as passed, and judges new employment application documents that do not meet the criteria to be passed, and generates this judgment result as the actual selection result. The selection content generation module generates a handover comment (mostly positive recommendation results, most of high importance, etc.) based on the generated selection result.
The selection content generation module may determine that new employment application documents with a screening pass rate below a predetermined probability are to be passed over as a result of document screening, and generate the result as a screening result. In this case, the selection content generation module may generate a handover comment indicating that the screening pass rate is below a predetermined probability.

The selection content output module outputs the generated selection pass rate, the actual selection results, and the handover comments on the reasons for the selection (step S92).
The selection content output module transmits the generated selection pass rate, the actual selection results, and the handover comments on the selection reasons to the person in charge terminal 3.
The person in charge terminal 3 receives the selection pass rate, the actual selection result, and the handover comment on the selection reason, and displays them on its display unit using a predetermined UI, etc. The hiring manager 2 views the displayed selection pass rate, the actual selection result, and the handover comment on the selection reason.
The selection content output module outputs the generated selection pass rate, actual selection results, and handover comments on the selection reasons by displaying the selection pass rate, actual selection results, and handover comments on the selection reasons on the person in charge terminal 3.

This completes the selection content output process.

[Fictitious Selection Process Executed by Computer 10]
The hypothetical selection process executed by the computer 10 will be described with reference to Fig. 13. Fig. 13 is a flowchart showing the hypothetical selection process executed by the computer 10.

The fictitious document generation module generates fictitious job application documents (step S100).
The fictitious employment application documents are fictitious documents used in document screening and interview screening in personnel recruitment, such as resumes, skill sheets, reference sheets, etc. The fictitious employment application documents are data in a predetermined data format (PDF, document file, spreadsheet file, etc.).
The fictitious document generation module generates fictitious employment application documents based on application documents of candidates who have been employed in the past, sample application documents generally distributed via public line networks, etc.
The method by which the fictitious document generation module generates fictitious job application documents is not particularly limited and can be designed as appropriate.

The fictitious document output module outputs the generated fictitious job application documents (step S101).
The fictitious document output module transmits the generated fictitious employment application documents to the person in charge terminal 3. The person in charge terminal 3 receives the respective employment application documents and displays them on its own display unit.
The fictitious document output module outputs the generated fictitious employment application document by displaying the generated fictitious employment application document on the person in charge terminal 3.

The fictitious annotation receiving module receives annotations for the output fictitious job application documents (step S103).
The annotation is the same as that in step S12 described above.
The person in charge terminal 3 accepts annotation input for the fictitious employment application documents from the recruiter 2. For example, the person in charge terminal 3 accepts input of ◯, ×, △ from the recruiter 2 in any part of the fictitious employment application documents (parts that the recruiter 2 should check in document screening and interview selection (educational history, work history, license, qualifications, job content, skills, personality, strengths, weaknesses, achievements, participating projects, etc.)). The recruiter 2 taps any part (parts that the recruiter 2 should check, etc.) in the fictitious employment application documents displayed on the person in charge terminal 3, and inputs either ◯, ×, or △ in this part by direct input using a virtual keyboard or by selection input from options. The person in charge terminal 3 may directly annotate the fictitious employment application documents, or may annotate the fictitious employment application documents read in a specified application and displayed in a specified UI.
The person in charge terminal 3 transmits the location where the annotation was accepted and the content of the annotation to the computer 10 as fictitious annotation data.
The fictitious annotation receiving module receives this fictitious annotation data and accepts annotations for the output fictitious employment application documents.

The fictitious selection extraction module uses the generated trained model to extract fictitious confirmation points that should be checked during document screening for the generated fictitious employment application documents (step S103).
The fictitious selection extraction module compares fictitious job application documents with the parts to be extracted in the trained model and the annotations for these parts, and extracts those that match or are similar as fictitious confirmation parts.

The reason generation module generates a verification reason for verifying the extracted fictitious verification location (step S104).
The reason for checking is the reason why the recruiter 2 checks the fictitious check points, for example, a match with job data, the holder of skills and qualifications, etc.
The reason generation module generates a verification reason for the extracted fictitious verification items that correspond to the predetermined required contents of a job candidate (qualifications, skills, work history, projects participated, etc.).

The comparison module compares the received annotation with the extracted fictitious confirmation portion (step S105).
The comparison module compares the annotated parts of the fictitious employment application documents with the extracted fictitious confirmation parts to determine the degree of match or approximation. Furthermore, the comparison module compares the contents of the annotations in the fictitious employment application documents with the reasons for checking the fictitious confirmation parts to determine the match, deviation, difference, etc. of the contents.

The evaluation module evaluates the received annotation based on the comparison result (step S106).
The rating may be, for example, by points, rank, or number of stars.
The evaluation module evaluates the received annotations based on the percentage of match or approximation between the parts of the fictitious job application documents for which annotations were received and the extracted fictitious check parts, and based on the match, discrepancy, or difference between the content of the annotations in the fictitious job application documents and the reasons for and content of the fictitious check parts.
The evaluation module evaluates the annotations based on a score according to the degree of match or approximation of the annotation received portion, and a score according to the degree of match, deviation, difference, etc. of the annotation content. The evaluation module also evaluates the annotations based on ranks such as A, B, C, etc. according to the degree of match or approximation of the annotation received portion, and the degree of match, deviation, difference, etc. of the annotation content. The evaluation module also evaluates the annotations based on the number of stars according to the degree of match or approximation of the annotation received portion, and the degree of match, deviation, difference, etc. of the annotation content.

The improvement comment generating module generates an improvement comment for the evaluation result (step S107).
The improvement comment generation module generates improvement comments that indicate what parts should be annotated to improve the score, rank, number of stars, etc. of the evaluation results (increasing the score, increasing the rank, increasing the number of stars, etc.).

The evaluation result output module outputs the evaluation result and comments for improvement (step S108).
The evaluation result output module transmits the evaluation result and the improvement comments to the person in charge terminal 3. The person in charge terminal 3 receives the evaluation result and the improvement comments and displays them on its own display unit.
The evaluation result output module outputs the generated fictitious employment application documents by displaying the evaluation results and improvement comments on the person in charge terminal 3.

The above is the fictitious selection process.

Although each of the above processes is described as a separate process, the computer 10 can also be configured to execute a combination of some or all of the above processes.

The above-mentioned means and functions are realized by a computer (including a CPU, an information processing device, and various terminals) reading and executing a specified program. The program may be provided, for example, from the computer via a network (Software as a Service: SaaS) or as a cloud service. The program may also be provided in a form recorded on a computer-readable recording medium. In this case, the computer reads the program from the recording medium, transfers it to an internal or external recording device, records it, and executes it. The program may also be recorded in advance on a recording device (recording medium) and provided from the recording device to the computer via a communication line.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above-mentioned embodiments. Furthermore, the effects described in the embodiments of the present invention are merely a list of the most favorable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the embodiments of the present invention.

A first aspect disclosed in this embodiment is a job application document recommendation system that recommends job application documents,
a learning unit that learns an annotated learning dataset from training employment application documents;
a model generation unit that generates a trained model based on the training result;
A document acquisition department that acquires new employment application documents;
A document screening extraction unit that uses the generated trained model to extract confirmation points to be confirmed in document screening for the acquired new employment application documents;
a recommendation unit that recommends the extracted confirmation portion;
The present invention provides a job application document recommendation system comprising:

A second aspect disclosed in this embodiment includes a job data acquisition unit that acquires job data;
Further comprising:
The document screening extraction unit adds the acquired job data and extracts the confirmation points.
The present invention provides a job application recommendation system according to a first aspect.

A third aspect disclosed in the present embodiment includes a reason generating unit that generates a check reason for checking the check point;
Further comprising:
The recommendation unit makes a recommendation together with the generated confirmation reason.
The present invention provides a job application recommendation system according to a first aspect.

A fourth aspect disclosed in the present embodiment includes an evaluation acquisition unit that acquires an evaluation of the recommended confirmation portion;
Further comprising:
The learning unit adds the recommended confirmation points and the obtained evaluation to the learning data set and re-learns.
The present invention provides a job application recommendation system according to a first aspect.

A fifth aspect disclosed in the present embodiment includes a comment acquisition unit that acquires comments on the recommended confirmation portion;
Further comprising:
The learning unit adds the recommended confirmation points and the acquired comments to the learning data set and re-learns.
The present invention provides a job application recommendation system according to a first aspect.

A sixth aspect disclosed in the present embodiment is a display mode changing unit that changes a display mode of the extracted confirmation portion based on a predetermined condition;
The job application document recommendation system according to the first aspect is further provided.

A seventh aspect disclosed in this embodiment is a part change unit that changes the confirmation part to be extracted based on a predetermined condition;
The job application document recommendation system according to the first aspect is further provided.

An eighth aspect disclosed in this embodiment is an interview selection extraction unit that uses the generated trained model to extract confirmation points to be confirmed in an interview selection process from the acquired new employment application documents;
Further comprising:
The recommendation unit recommends points to be confirmed in the extracted interview selection.
The present invention provides a job application recommendation system according to a first aspect.

A ninth aspect disclosed in this embodiment is the recommendation unit making recommendations through a predetermined user interface.
The present invention provides a job application recommendation system according to a first aspect.

A tenth aspect disclosed in this embodiment includes a reflection unit that reflects a recommendation result in the new employment application document;
a reflection output unit that outputs the new employment application documents reflecting the recommendation results in a predetermined format;
The job application document recommendation system according to the first aspect is further provided.

An eleventh aspect disclosed in this embodiment includes a result information acquisition unit that acquires past actual document screening result information;
A selection content generation unit that generates a selection pass rate, an actual selection result, and a handover comment of the selection reason based on the extracted confirmation points, the recommendation result, and the obtained document selection result information;
a selection content output unit that outputs the generated selection pass rate, the actual selection result, and a handover comment on the selection reason;
The job application document recommendation system according to the first aspect is further provided.

A twelfth aspect disclosed in this embodiment is a learning annotation receiving unit that receives annotations for the learning job application document;
Further comprising:
The learning unit adds a learning data set based on the annotations received for the training job application documents to a learning data set annotated for the training job application documents, and performs training.
The present invention provides a job application recommendation system according to a first aspect.

A thirteenth aspect disclosed in this embodiment includes a fictitious document generation unit that generates a fictitious job application document;
a fictitious document output unit that outputs the generated fictitious employment application documents;
a fictitious annotation receiving unit that receives annotations for the output fictitious employment application documents;
A fictitious selection extraction unit that uses the generated trained model to extract fictitious selection confirmation points to be confirmed in document screening for the generated fictitious employment application documents;
A comparison unit that compares the annotations received for the fictitious employment application documents with the extracted fictitious selection confirmation points;
an evaluation unit that evaluates annotations received for the fictitious job application based on a comparison result;
an evaluation result output unit that outputs the evaluation result;
The job application document recommendation system according to the first aspect is further provided.

A fourteenth aspect disclosed in the present embodiment is an improvement comment generating unit that generates an improvement comment for an evaluation result;
Further comprising:
The evaluation result output unit outputs the generated improvement comment.
A job application document recommendation system according to a thirteenth aspect is provided.

REFERENCE SIGNS LIST 1 Employment application document recommendation system 2 Recruiter 3 Personnel terminal 8 Network 10 Computer 11 Document acquisition unit 12 Recommendation unit 13 Learning unit 14 Model generation unit 15 Document selection extraction unit 20 UI
21 Check item column 22 Check reason column

Claims (17)

An employment application document recommendation system that recommends employment application documents,
A learning unit that annotates any part of a training application document by adding tags and metadata, and learns a training dataset that uses the annotated parts and contents as a dataset;
a model generation unit that generates a trained model based on the training result;
A document acquisition department that acquires new employment application documents;
A document screening extraction unit that compares the annotated parts and contents in the generated trained model with the acquired new employment application documents and extracts parts in the new employment application documents that match or are similar to the annotated parts and contents as confirmation parts to be confirmed in document screening;
a recommendation unit that recommends the extracted confirmation portion;
A job application document recommendation system that includes: A job data acquisition unit that acquires job data;
Further comprising:
When extracting the confirmation points, the document selection extraction unit compares the annotated points and contents in the generated trained model with the job data, and extracts the points in the new employment application documents that match the job data as the confirmation points.
The job application recommendation system of claim 1 . a reason generating unit that generates a reason for checking the check points corresponding to the contents required of the recruitment candidate that are preset;
Further comprising:
The recommendation unit makes a recommendation together with the generated confirmation reason.
The job application recommendation system of claim 1 . an evaluation acquisition unit that acquires an evaluation of the recommended confirmation portion;
Further comprising:
The learning unit adds the recommended confirmation points and the obtained evaluation to the learning data set and re-learns.
The job application recommendation system of claim 1 . a comment acquisition unit that acquires comments on the recommended confirmation portion;
Further comprising:
The learning unit adds the recommended confirmation points and the acquired comments to the learning data set and re-learns.
The job application recommendation system of claim 1 . a mode changing unit that changes the display mode of the extracted confirmation portion based on desirable contents of the employment candidate in the new employment application documents, undesirable contents of the employment candidate, or contents that a hiring manager considers important;
Further comprising:
The job application recommendation system of claim 1 . a part changing unit that changes the confirmation part to be extracted based on a preset mode related to an extraction amount of the confirmation part;
Further comprising:
The job application recommendation system of claim 1 . An interview selection extraction unit that compares the annotated parts and contents in the generated trained model with the acquired new employment application documents and extracts parts that match or are similar to the confirmation parts and contents to be confirmed in the interview selection in the new employment application documents as confirmation parts to be confirmed in the interview selection;
Further comprising:
The recommendation unit recommends points to be confirmed in the extracted interview selection.
The job application recommendation system of claim 1 . The recommendation unit makes recommendations through a predetermined user interface.
The job application recommendation system of claim 1 . A reflection unit that reflects the recommendation result in the new employment application document;
a reflection output unit that outputs the new employment application documents reflecting the recommendation results in a predetermined format;
The job application recommendation system of claim 1 further comprising: A result information acquisition unit that acquires information on actual past document screening results;
A selection content generation unit that generates a selection pass rate, an actual selection result, and a handover comment of the selection reason based on the extracted confirmation points, the recommendation result, and the obtained document selection result information;
a selection content output unit that outputs the generated selection pass rate, the actual selection result, and a handover comment on the selection reason;
Further comprising:
The selection content generation unit,
Calculate the rate at which the extracted confirmation points and recommendation results match or approximate the document screening result information as the screening pass rate;
The extracted confirmation points are compared with the recommendation result, and the new employment application documents are compared with the past actual document screening result information, and an actual selection result is generated based on whether or not the documents satisfy a preset criterion.
Based on the generated selection result, a state of the recommendation result is generated as the handover comment.
The job application recommendation system of claim 1 . a learning annotation receiving unit that receives annotations for the learning job application document;
Further comprising:
The learning unit adds a learning data set based on the annotations received for the training job application documents to a learning data set annotated for the training job application documents, and performs training.
The job application recommendation system of claim 1 . a fictitious document generation unit for generating fictitious employment application documents, which are fictitious employment application documents;
a fictitious document output unit that outputs the generated fictitious employment application documents;
a fictitious annotation receiving unit that receives annotations for adding tags or metadata to any portion of the output fictitious employment application documents;
a fictitious selection extraction unit that compares the annotated parts and contents in the generated trained model with the annotated parts and contents received for the generated fictitious employment application documents, and extracts parts that match or are similar to the annotated parts and contents received for the generated fictitious employment application documents as fictitious selection confirmation parts to be confirmed in document screening;
A comparison unit that compares the annotations received for the fictitious employment application documents with the extracted fictitious selection confirmation points;
an evaluation unit that evaluates the annotations received for the fictitious job application documents based on a comparison result, based on a rate of agreement or approximation between the portion of the fictitious job application documents where the annotations have been received and the extracted fictitious selection confirmation portion;
an evaluation result output unit that outputs the evaluation result;
The job application recommendation system of claim 1 further comprising: a fictitious document generation unit for generating fictitious employment application documents, which are fictitious employment application documents;
a fictitious document output unit that outputs the generated fictitious employment application documents;
a fictitious annotation receiving unit that receives annotations for adding tags or metadata to any portion of the output fictitious employment application documents;
a fictitious selection extraction unit that compares the annotated parts and contents in the generated trained model with the annotated parts and contents received for the generated fictitious employment application documents, and extracts parts that match or are similar to the annotated parts and contents received for the generated fictitious employment application documents as fictitious selection confirmation parts to be confirmed in document screening;
a fictitious reason generating unit for generating a reason for checking the fictitious selection confirmation points corresponding to the contents required of a candidate for employment that are preset;
A comparison unit that compares the annotations received for the fictitious employment application documents with the extracted fictitious selection confirmation points;
an evaluation unit that evaluates the received annotation based on at least one of a match, a deviation, and a difference between the content of the annotation in the fictitious employment application document based on a comparison result and the content of the confirmation reason in the fictitious selection confirmation part ;
an evaluation result output unit that outputs the evaluation result;
The job application recommendation system of claim 1 further comprising : an improvement comment generating unit that generates an improvement comment for the evaluation result, the improvement comment indicating a portion to be annotated for improving the evaluation result;
Further comprising:
The evaluation result output unit outputs the generated improvement comment.
15. The job application document recommendation system according to claim 13 or 14. 1. A computer-implemented method for recommending job applications, comprising:
Annotating any part of the training job application document by adding tags and metadata, and learning a training dataset that uses the annotated parts and contents as a dataset;
generating a trained model based on the training results;
Obtaining new employment application documents;
A step of comparing the annotated parts and contents in the generated trained model with the acquired new employment application documents, and extracting parts in the new employment application documents that match or are similar to the annotated parts and contents as confirmation parts to be confirmed in document screening;
A step of recommending the extracted confirmation portion;
The employment application document recommendation method includes: A computer that recommends job applications
Annotating any part of the training employment application document by adding tags and metadata, and learning a training dataset using the annotated parts and contents as a dataset;
generating a trained model based on the training results;
Steps for obtaining new employment application documents,
A step of comparing the annotated parts and contents in the generated trained model with the acquired new employment application documents, and extracting parts in the new employment application documents that match or are similar to the annotated parts and contents as confirmation parts to be confirmed in document screening;
A step of recommending the extracted confirmation portion;
A program for executing.