Recruitment full-process automatic collaboration method and system based on large language model

Abstract

The invention discloses a recruitment full-process automatic cooperation method and system based on a large language model, and relates to the technical field of artificial intelligence and human resource management, and the method specifically comprises the following steps: a recruitment platform receives recruitment position information issued by an enterprise recruiter, and generates a standardized recruitment position information description; performing text semantic analysis on the standardized recruitment post information description, and extracting a job key field of a recruitment post; extracting job seeker resume key fields from the job seeker resume; converting the job key field of the recruitment post into a high-dimensional semantic query vector, constructing a resume semantic vector library based on the resume key field of the job seeker, and performing query matching to generate a candidate resume list set; generating an interview comprehensive evaluation report based on the interview data; returning the interview comprehensive evaluation report to a training library of the large language model, and updating parameters of the large language model; and the enterprise recruiter determines an intention job seeker according to the interview comprehensive evaluation report, and generates an in-job notification and an in-job guide process.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and human resource management technology, specifically a method and system for automating the entire recruitment process based on a large language model. Background Technology

[0002] As enterprises accelerate their digital transformation and competition for talent intensifies, traditional recruitment processes face pain points such as fragmented processes, low efficiency, and insufficient semantic understanding. By building an intelligent, precise, and scenario-based talent management system, organizations can significantly improve the accuracy of talent identification and the scientific nature of management decisions.

[0003] Chinese invention application CN120317847A discloses an AI chatbot control system and method for human resources, comprising the following steps: S1, acquiring and preprocessing recruitment data to form a recruitment dataset; S2, performing semantic analysis using a hierarchical Transformer model to generate job-job seeker feature representations; S3, optimizing the Transformer model through neural architecture search to obtain an optimized model; S4, calculating job matching scores based on the optimized model and generating a candidate ranking list by combining it with a knowledge graph; S5, calling an MX chatbot to conduct intelligent interviews with candidates, dynamically adjusting the difficulty of questions, and acquiring interview data; S6, integrating job seeker interview scores and job matching scores to generate a final candidate recommendation list; S7, generating a recruitment decision support report based on the recommendation list. This invention employs a hierarchical Transformer model, neural architecture search optimization, and an MX chatbot working collaboratively to achieve precise optimization of job matching and automated decision support for job seeker interview evaluation.

[0004] Existing technologies suffer from the following problems: they rely heavily on manual input and repetitive operations in resume screening, job analysis, and candidate communication, leading to low recruitment efficiency; furthermore, delays in candidate communication and interview scheduling result in talent loss; and they fall short in terms of talent matching speed, communication quality, and recruitment cycle control, failing to meet companies' needs for rapid recruitment and precise talent selection. Therefore, there is an urgent need for an automated, collaborative method and system for the entire recruitment process based on a large language model to address the problems of existing technologies. Summary of the Invention

[0005] To address the aforementioned technical problems, the present invention aims to provide an automated collaborative method for the entire recruitment process based on a large language model. This method includes the following steps: Step 1: The recruitment platform receives job postings from employers and generates standardized job descriptions by combining the industry knowledge graph of the job postings. Step 2: Perform text semantic parsing on the standardized job description information to extract key fields related to the job duties; Step 3: Parse the job seeker's resume into a structured text representation based on the multimodal model, and extract the key fields from the job seeker's resume; Step 4: Convert the key fields of the job posting into high-dimensional semantic query vectors, build a resume semantic vector library based on the key fields of the job seeker's resume, and perform query matching to generate a list of candidate resumes. Step 5: Send interview invitations based on the candidate resume list and in conjunction with the company's HRM system, generate interview data, and generate a comprehensive interview evaluation report based on the interview data; Step 6: Send the comprehensive interview evaluation report back to the training library of the large language model to update the parameters of the large language model; Step 7: Based on the comprehensive evaluation of the interview, the company recruiter determines the candidates to be shortlisted and generates an offer letter and onboarding guidance through the company's HRM system.

[0006] In a preferred embodiment, the specific process of generating standardized job posting information descriptions includes: The recruitment platform receives job postings submitted by employers through the front-end interface, parses the job postings, and further combines them with the industry knowledge graph of the job postings to perform semantic reorganization and normalization, ultimately obtaining a standardized description of the job posting information. The specific process of semantic reorganization and normalization includes: intelligently completing missing job information items using the job industry knowledge graph; and standardizing and correcting job information that is not standardized or ambiguous based on the standard expressions in the job industry knowledge graph.

[0007] In a preferred embodiment, the key fields for job postings specifically include: basic job information, job responsibilities, skills and abilities requirements, education and background requirements, work experience requirements, salary and benefits information, and recruitment restrictions.

[0008] In a preferred embodiment, the key fields of the job seeker's resume specifically include: job application information, basic personal information, educational background, work experience, skills and abilities, and multimodal additional information.

[0009] In a preferred embodiment, the specific process of generating a candidate resume list includes: The key fields for job postings are preprocessed and semantically enhanced. These preprocessed and semantically enhanced fields are then transformed into high-dimensional semantic query vectors using a vectorization encoding algorithm. The same preprocessing and semantic enhancement are applied to the key fields of job applicants' resumes. These preprocessed and semantically enhanced fields are then input into a large language model to generate resume semantic vectors. A resume semantic vector library is built based on these vectors. A vector similarity calculation algorithm is used to calculate the similarity between the high-dimensional semantic query vectors and the resume semantic vectors in the library. The top N resumes with similarity values ​​reaching a preset similarity threshold are selected to form a candidate resume list.

[0010] In a preferred embodiment, the specific process of generating a comprehensive interview evaluation report based on interview data includes: The enterprise HRM system receives a list of candidate resumes, obtains personal contact information from the resumes, and automatically sends interview invitations. For those who accept the invitations, the system schedules interviews with interviewers, records the entire interview process, and generates interview data. The system then uses data analysis software to perform comprehensive calculations and multi-dimensional evaluations of the generated interview data, ultimately generating a comprehensive interview evaluation report.

[0011] In another aspect, this invention also discloses an automated collaborative system for the entire recruitment process based on a large language model. The system includes the following modules: information standardization module, key field extraction module one, key field extraction module two, candidate resume collection generation module, evaluation report generation module, feedback and update module, and onboarding management module. The information standardization module: The recruitment platform receives job postings from employers and, in conjunction with an industry knowledge graph of the job postings, generates standardized job descriptions. The first key field extraction module performs text semantic parsing on the standardized job description information to extract key fields for the job position. The second key field extraction module parses the job seeker's resume into a structured text representation based on a multimodal model and extracts the key fields from the resume. The candidate resume collection generation module transforms the key fields of job postings into high-dimensional semantic query vectors, constructs a resume semantic vector library based on the key fields of job seekers' resumes, performs query matching, and generates a candidate resume list collection. The evaluation report generation module sends interview invitations based on the candidate resume list and in conjunction with the enterprise HRM system, generates interview data, and generates a comprehensive interview evaluation report based on the interview data. The feedback and update module feeds back the comprehensive interview evaluation report to the training library of the large language model to update the parameters of the large language model. The onboarding management module allows employers to identify potential job seekers based on comprehensive interview evaluations and generate onboarding offer notifications and onboarding guidance processes through the company's HRM system.

[0012] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention extracts key fields for job positions by performing text semantic parsing on standardized job descriptions; further, it parses job applicant resumes into structured text representations using a multimodal model and extracts key fields from the resumes; it further transforms the key fields of job positions into high-dimensional semantic query vectors, constructs a resume semantic vector library based on these fields, performs query matching, and generates a candidate resume list; it further sends interview invitations based on the candidate resume list and in conjunction with the company's HRM system, generates interview data, and generates a comprehensive interview evaluation report based on the interview data; it further sends the comprehensive interview evaluation report back to the training library of a large language model to update the model's parameters; and finally, the company recruiter determines potential job applicants based on the comprehensive interview evaluation report and generates onboarding offer notifications and onboarding guidance processes through the company's HRM system. Attached Figure Description

[0013] Figure 1 This is a schematic diagram illustrating the steps of a collaborative method for automating the entire recruitment process based on a large language model, as described in an embodiment of this application.

[0014] Figure 2 This is a schematic diagram showing the connections between various modules of a recruitment full-process automated collaborative system based on a large language model, according to an embodiment of this application. Detailed Implementation

[0015] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0016] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in one or more embodiments of the present invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0017] Example 1 Please see Figure 1 As shown in the figure, this application provides a method for automated collaborative recruitment process based on a large language model. The method includes the following steps: Step 1: The recruitment platform receives job postings from employers and generates standardized job descriptions by combining the industry knowledge graph of the job postings. Based on the above embodiments, the specific process of generating standardized job posting information descriptions includes: the recruitment platform receives job posting information submitted and published by enterprise recruiters through the front-end interface, parses the job posting information, further combines the job posting industry knowledge graph, performs semantic reorganization and normalization expression, and finally obtains standardized job posting information descriptions. The specific process of semantic reorganization and normalization includes: intelligently completing missing job information items using the job industry knowledge graph; and standardizing and correcting job information that is not standardized or ambiguous based on the standard expressions in the job industry knowledge graph. It should be noted that the recruitment platforms specifically include: online recruitment websites and social recruitment platforms; online recruitment websites specifically include: for example, BOSS Zhipin, Zhaopin, 51job, Liepin, etc.; social recruitment platforms specifically include: such as LinkedIn, Maimai, etc., which have professional social networking and job posting functions; The aforementioned job industry knowledge graph refers to a pre-constructed knowledge system used to support the semantic understanding of job positions. Specifically, the job industry knowledge graph includes: the hierarchical structure of job classification in various industries, the semantic relationship between job responsibilities and skill requirements, knowledge entities in fields such as skill tags, professional qualifications, and educational requirements, and the synonym and near-synonym mapping relationship of job names in different companies. Step 2: Perform text semantic parsing on the standardized job description information to extract key fields related to the job duties; Based on the above embodiments, the key fields for job positions specifically include: basic job information, job responsibilities, skills and abilities requirements, education and background requirements, work experience requirements, salary and benefits information, and recruitment restrictions. It should be noted that the basic job information specifically includes: job title, job category, department, work location, etc.; the job responsibilities information specifically includes, but is not limited to: core work tasks, daily work content, teamwork requirements, etc.; the skills and abilities requirements specifically include, but are not limited to: essential skills (hard skills, such as programming languages, accounting standards), preferred skills (soft skills, such as communication and coordination, innovation ability), etc.; the education and background requirements specifically include, but are not limited to: minimum education requirements, professional background, qualification certificates, industry certifications, etc.; the work experience requirements specifically include, but are not limited to: years of work experience, experience in specific fields, project experience, etc.; the salary and benefits information specifically includes, but is not limited to: salary range, benefits, promotion conditions, etc.; the recruitment restrictions specifically include, but are not limited to: age limit, language requirements, frequency of business trips, etc. Step 3: Parse the job seeker's resume into a structured text representation based on the multimodal model, and extract the key fields from the job seeker's resume; Based on the above embodiments, the key fields of the job seeker's resume specifically include: job application information, basic personal information, educational background, work experience, skills and abilities, and multimodal additional information; It should be noted that job application information specifically includes, but is not limited to: job intention, expected salary, and available start date; basic personal information specifically includes, but is not limited to: name, gender, age, contact information, and location; educational background specifically includes, but is not limited to: degree level, university, major, start and end dates, and degree certificates; work experience specifically includes, but is not limited to: company name, job title, length of service, job responsibilities, and project experience; skills and abilities specifically include, but are not limited to: professional skills (such as programming languages ​​and accounting software), foreign language skills, software skills, and teamwork skills; multimodal supplementary information specifically includes, but is not limited to: portfolio links, professional qualification certificates, professional licenses, and awards; and multimodal models specifically include, but are not limited to: text semantic models and image recognition models. Step 4: Convert the key fields of the job posting into high-dimensional semantic query vectors, build a resume semantic vector library based on the key fields of the job seeker's resume, and perform query matching to generate a list of candidate resumes. Based on the above embodiments, the specific process of generating a candidate resume list includes: preprocessing and semantically enhancing the key fields of the job positions, and converting the preprocessed and semantically enhanced key fields of the job positions into high-dimensional semantic query vectors using a vectorized encoding algorithm. For key fields in job seekers' resumes, the same preprocessing and semantic enhancement processing as for key fields in job positions is applied. The preprocessed and semantically enhanced key fields of job seekers' resumes are then input into a large language model to generate resume semantic vectors. A resume semantic vector library is then built based on these resume semantic vectors. Using a vector similarity calculation algorithm (such as the cosine similarity algorithm), the high-dimensional semantic query vector is compared with the resume semantic vectors in the resume semantic vector library to obtain the corresponding similarity value. The top N resumes (N is a positive integer, which can be set according to actual recruitment needs) with similarity values ​​reaching a preset similarity threshold are selected to form a candidate resume list. It should be noted that the preset similarity threshold was set by those skilled in the art through experiments based on historical recruitment data. It can be dynamically adjusted according to the urgency of the job posting, the expected number of candidates, and historical recruitment data to ensure that the candidate resume list not only meets the basic requirements of the job but also provides sufficient and high-quality candidates for subsequent screening. Large language models specifically include models such as GPT series, BERT series, ERNIE, LLaMA, and GLM. Step 5: Send interview invitations based on the candidate resume list and in conjunction with the company's HRM system, generate interview data, and generate a comprehensive interview evaluation report based on the interview data; Based on the above embodiments, the specific process of generating a comprehensive interview evaluation report based on interview data includes: the enterprise HRM system receives a list of candidate resumes, obtains personal contact information from the job seeker's basic personal information in the resume, and automatically sends an interview invitation; for job seekers who receive the interview invitation, an interview time and an interviewer are arranged for the interview; the entire interview process is recorded, and interview data is generated; the generated interview data is comprehensively calculated and evaluated and analyzed in multiple dimensions using data analysis software, and finally a comprehensive interview evaluation report is generated. It should be noted that an enterprise HRM system (Human Resource Management System, or HRM for short) is an information platform used by enterprises for human resource management and employee lifecycle management. It supports the entire process from recruitment, onboarding, training, attendance, payroll, performance appraisal, promotion to departure management through software or cloud services. Interview data includes, but is not limited to: real-time interview scores recorded by the interviewer (covering dimensions such as professional skills, communication skills, teamwork, and job suitability); technical assessment results (such as programming question accuracy, code quality score, and completion time); and behavioral performance scores (such as speaking speed, pauses, emotional fluctuations, body language, and facial expressions). The comprehensive interview evaluation report includes, but is not limited to: an interview overview, detailed evaluation results for each dimension, strengths and areas for improvement, comprehensive recommendations, and a summary of interview data. The report supports multiple presentation formats such as text and tables, so that HR and recruitment decision-makers can intuitively obtain effective information. Step 6: Send the comprehensive interview evaluation report back to the training library of the large language model to update the parameters of the large language model; It should be noted that the purpose of sending the comprehensive interview evaluation report back to the training library of the large language model to update the model parameters is to continuously optimize the large language model's understanding and analysis accuracy of various evaluation dimensions in the recruitment scenario, so that it can more accurately capture the matching relationship between candidate traits and job requirements. At the same time, by continuously absorbing effective interview data from actual interview cases, the accuracy of generating resume semantic vectors can be improved, thereby enhancing the selection accuracy of the entire recruitment process in an automated and collaborative manner. The parameters for updating the large language model include: word vector parameters, attention weight parameters, loss function parameters, and feedback learning parameters. Step 7: Based on the comprehensive evaluation of the interview, the company recruiter determines the candidates to be shortlisted and generates an offer letter and onboarding guidance through the company's HRM system. It should be noted that the job offer notification includes core elements such as job title, job responsibilities, salary and benefits, start date, and reporting location; the onboarding process includes a checklist of pre-onboarding preparations (such as material preparation and system account pre-registration guidance), an explanation of the onboarding process (such as check-in, department introduction, and familiarization with the office environment), and an outline of the post-onboarding training plan, ensuring that job seekers clearly understand the entire onboarding process and improving their onboarding experience and efficiency.

[0018] Example 2 Please see Figure 2 As shown, in another embodiment of the present invention, the present invention also discloses an automated collaborative system for the entire recruitment process based on a large language model. The system includes the following modules: information standardization module, key field extraction module one, key field extraction module two, candidate resume collection generation module, evaluation report generation module, feedback and update module, and onboarding management module. The modules described above are connected via wired and / or wireless means to enable data transmission between them. The information standardization module: The recruitment platform receives job postings from employers and, in conjunction with an industry knowledge graph of the job postings, generates standardized job descriptions. The first key field extraction module performs text semantic parsing on the standardized job description information to extract key fields for the job position. The second key field extraction module parses the job seeker's resume into a structured text representation based on a multimodal model and extracts the key fields from the resume. The candidate resume collection generation module transforms the key fields of job postings into high-dimensional semantic query vectors, constructs a resume semantic vector library based on the key fields of job seekers' resumes, performs query matching, and generates a candidate resume list collection. The evaluation report generation module sends interview invitations based on the candidate resume list and in conjunction with the enterprise HRM system, generates interview data, and generates a comprehensive interview evaluation report based on the interview data. The feedback and update module feeds back the comprehensive interview evaluation report to the training library of the large language model to update the parameters of the large language model. The onboarding management module allows employers to identify potential job seekers based on comprehensive interview evaluations and generate onboarding offer notifications and onboarding guidance processes through the company's HRM system.

[0019] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0020] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A collaborative method for automating the entire recruitment process based on a large language model, characterized in that, The method includes the following steps: Step 1: The recruitment platform receives job postings from employers and generates standardized job descriptions by combining the industry knowledge graph of the job postings. Step 2: Perform text semantic parsing on the standardized job description information to extract key fields related to the job duties; Step 3: Parse the job seeker's resume into a structured text representation based on the multimodal model, and extract the key fields from the job seeker's resume; Step 4: Convert the key fields of the job posting into high-dimensional semantic query vectors, build a resume semantic vector library based on the key fields of the job seeker's resume, and perform query matching to generate a list of candidate resumes. Step 5: Send interview invitations based on the candidate resume list and in conjunction with the company's HRM system, generate interview data, and generate a comprehensive interview evaluation report based on the interview data; Step 6: Send the comprehensive interview evaluation report back to the training library of the large language model to update the parameters of the large language model; Step 7: Based on the comprehensive evaluation of the interview, the company recruiter determines the candidates to be shortlisted and generates an offer letter and onboarding guidance through the company's HRM system.

2. The automated collaborative recruitment process based on a large language model as described in claim 1, characterized in that, The specific process of generating standardized job description information includes: The recruitment platform receives job postings submitted by employers through the front-end interface, parses the job postings, and further combines them with the industry knowledge graph of the job postings to perform semantic reorganization and normalization, ultimately obtaining a standardized description of the job posting information. The specific process of semantic reorganization and normalization includes: intelligently completing missing job information items using the job industry knowledge graph; and standardizing and correcting job information that is not standardized or ambiguous based on the standard expressions in the job industry knowledge graph.

3. The automated collaborative recruitment process based on a large language model as described in claim 1, characterized in that, Key fields for job postings include: basic job information, job responsibilities, required skills, educational background, work experience, salary and benefits, and recruitment restrictions.

4. The automated collaborative recruitment process based on a large language model as described in claim 1, characterized in that, The key fields of the job seeker's resume include: job information, basic personal information, educational background, work experience, skills and abilities, and multimodal additional information.

5. The automated collaborative method for the entire recruitment process based on a large language model according to claim 1, characterized in that, The specific process of generating a list of candidate resumes includes: The key fields for job postings are preprocessed and semantically enhanced. These preprocessed and semantically enhanced fields are then transformed into high-dimensional semantic query vectors using a vectorization encoding algorithm. The same preprocessing and semantic enhancement are applied to the key fields of job applicants' resumes. These preprocessed and semantically enhanced fields are then input into a large language model to generate resume semantic vectors. A resume semantic vector library is built based on these vectors. A vector similarity calculation algorithm is used to calculate the similarity between the high-dimensional semantic query vectors and the resume semantic vectors in the library. The top N resumes with similarity values ​​reaching a preset similarity threshold are selected to form a candidate resume list.

6. The automated collaborative method for the entire recruitment process based on a large language model according to claim 1, characterized in that, The specific process of generating a comprehensive interview evaluation report based on interview data includes: The enterprise HRM system receives a list of candidate resumes, obtains personal contact information from the resumes, and automatically sends interview invitations. For those who accept the invitations, the system schedules interviews with interviewers, records the entire interview process, and generates interview data. The system then uses data analysis software to perform comprehensive calculations and multi-dimensional evaluations of the generated interview data, ultimately generating a comprehensive interview evaluation report.

7. A fully automated collaborative recruitment system based on a large language model, employing the fully automated collaborative recruitment method based on a large language model as described in any one of claims 1-6, characterized in that, The system includes the following modules: information standardization module, key field extraction module 1, key field extraction module 2, candidate resume collection generation module, evaluation report generation module, feedback and update module, and onboarding management module; The information standardization module: The recruitment platform receives job postings from employers and, in conjunction with an industry knowledge graph of the job postings, generates standardized job descriptions. The first key field extraction module performs text semantic parsing on the standardized job description information to extract key fields for the job position. The second key field extraction module parses the job seeker's resume into a structured text representation based on a multimodal model and extracts the key fields from the resume. The candidate resume collection generation module transforms the key fields of job postings into high-dimensional semantic query vectors, constructs a resume semantic vector library based on the key fields of job seekers' resumes, performs query matching, and generates a candidate resume list collection. The assessment report generation module; Based on the candidate resume list and in conjunction with the company's HRM system, interview invitations are sent, interview data is generated, and a comprehensive interview evaluation report is generated based on the interview data. The backhaul and update module; The comprehensive interview evaluation report is sent back to the training library of the large language model to update the parameters of the large language model; The onboarding management module; Company recruiters determine potential job seekers based on comprehensive interview evaluations and generate onboarding offer notifications and onboarding guidance processes through the company's HRM system.

Images