People and post matching method, device and equipment, medium and product

Abstract

The invention discloses a person and post matching method and device, equipment, a medium and a product, and relates to the technical field of information processing, and the method comprises the steps: receiving recruitment post data and resume data; and performing data matching on the recruitment post data and the resume data through a pre-constructed person-post matching model to obtain a matching result, the person-post matching model being constructed based on a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model. Therefore, data matching is carried out on the recruitment post data and the resume data through the human-post matching model constructed by the recruitment vector model, the information extraction model, the weight evaluation model and the matching degree scoring model to obtain the matching result, and the problem that post requirements and resume advantages are not aligned in the prior art is solved. Therefore, the recruitment quality of the enterprise is not high and the talents are lost can be solved, and the accuracy of man-post matching can be improved.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a person-job matching method, device, equipment, medium and product. Background Art

[0002] When a company is recruiting, it will receive resume information from multiple applicants. In order to find the most suitable applicant based on these resume information, it is necessary to extract information and match people with jobs. Information extraction is an important part of the task of matching people with jobs. The quality of key data extraction directly affects the matching results. Matching people with jobs is because the requirements of different positions and the personal advantages of resume information have different emphases.

[0003] However, the information extraction methods designed for existing job-person matching solutions have not fully integrated domain knowledge, the extraction performance is limited, and the potential of the matching algorithm cannot be fully utilized. Secondly, the extraction dimensions designed in the existing job-person matching solutions are relatively small, which will make it impossible to evaluate the degree of matching between jobs and talent resumes from multiple perspectives. In the matching part, different jobs have different emphases. Although the corresponding weights can be evaluated by designing job requirement emphasis dimensions, this solution is only applicable to the scenario of matching resumes based on jobs, and does not take into account the advantages reflected in resumes when matching jobs based on talents in job-person matching tasks.

[0004] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of this application is to provide a person-job matching method, device, equipment, medium and product, aiming to solve the technical problem in the prior art that job requirements are not aligned with resume advantages, resulting in low quality of corporate recruitment and talent loss.

[0006] To achieve the above purpose, the present application proposes a person-job matching method, which includes: Receive recruitment position data and resume data; The recruitment position data and resume data are matched with each other through a pre-built person-job matching model to obtain a matching result. The person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model.

[0007] In one embodiment, before the step of performing data matching on the recruitment position data and the resume data using a pre-built person-position matching model to obtain a matching result, the method further includes: Obtain historical job information and historical resume information; Performing cluster analysis on the historical job information to obtain person-job matching dimensions; Based on the person-job matching dimension, extract the historical job information and historical resume information through a first deep learning and natural language processing model to obtain a first extraction result; Constructing a vector model optimization data set according to the first extraction result, and training an initial vector model based on the vector model optimization data set to obtain a recruitment vector model; Performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an information extraction model; Performing a dimension weight evaluation on the person-job matching dimension and historical resume information through a second deep learning and natural language processing model to obtain an evaluation result; Performing knowledge distillation on the second deep learning and natural language processing model according to the evaluation result to obtain a weight evaluation model; Inputting the first extraction result and the evaluation result into a third deep learning and natural language processing model for matching scoring to obtain a first scoring result; Performing knowledge distillation on the third deep learning and natural language processing model according to the first scoring result to obtain a matching scoring model; A person-job matching model is constructed based on the recruitment vector model, information extraction model, matching degree scoring model and weight evaluation model.

[0008] In one embodiment, the step of performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an information extraction model includes: Based on the person-job matching dimension, the first extraction result is dimensionally classified by the first deep learning and natural language processing model to obtain a first classification result; Based on the person-job matching dimension, the first extraction result is dimensionally classified by a classification model to obtain a second classification result; Cross-validating the first classification result and the second classification result to obtain a dimension classification data set; Performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an initial information extraction model; According to the dimensionally classified data set, the sub-network of the initial information extraction model is activated through a gating network, and the initial information extraction model is nonlinearly transformed through an activation function to obtain an information extraction model.

[0009] In one embodiment, the step of performing knowledge distillation on the third deep learning and natural language processing model according to the first scoring result to obtain a matching scoring model includes: Performing knowledge distillation on the third deep learning and natural language processing model to obtain an initial matching degree scoring model; Inputting the first extraction result and the evaluation result into the initial matching degree scoring model to obtain a second scoring result; Performing difference calculation based on the first scoring result and the second scoring result to obtain a cross entropy loss; Performing a difference calculation based on the probability distribution of the first scoring result and the probability distribution of the second scoring result to obtain a divergence loss; The initial matching score model is optimized according to the cross entropy loss and the divergence loss to obtain a matching score model.

[0010] In one embodiment, the step of performing data matching on the recruitment position data and the resume data by using a pre-built person-position matching model to obtain a matching result includes: Extracting information from the recruitment position data and resume data using the information extraction model to obtain a second extraction result; The weight evaluation model is used to evaluate the recruitment position data and resume data to obtain the position dimension weight and the resume dimension weight; vectorizing the second extraction result by using the recruitment vector model to obtain a vector representation of the second extraction result; Based on the weight of the position dimension and the weight of the resume dimension, the second extraction results are sorted by similarity to obtain information of several candidates; The plurality of candidate information are matched using the matching degree scoring model to obtain a matching result.

[0011] In one embodiment, after the step of performing data matching on the recruitment position data and the resume data using a pre-built person-position matching model to obtain a matching result, the method further includes: Receive job application results; According to the job application results, query the resume data to obtain the applicant information; Sending an initial employee evaluation form to an employee management terminal according to the applicant information, and the employee management terminal filling in the initial employee evaluation form to obtain a final employee evaluation form; The final employee evaluation form is received, and the person-job matching model is optimized according to the final employee evaluation form to obtain a model optimization result.

[0012] In addition, to achieve the above purpose, the present application also proposes a person-job matching device, the person-job matching device comprising: Receiving module, used to receive recruitment position data and resume data; The matching module is used to match the recruitment position data and resume data through a pre-built person-job matching model to obtain a matching result. The person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model, and a matching degree scoring model.

[0013] In addition, to achieve the above-mentioned purpose, the present application also proposes a person-job matching device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the person-job matching method described above.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the person-job matching method described above are implemented.

[0015] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the person-job matching method described above.

[0016] One or more technical solutions proposed in this application have at least the following technical effects: The embodiment of the present application proposes a person-job matching method, device, equipment, medium and product, which receives job recruitment data and resume data; performs data matching on the job recruitment data and resume data through a pre-constructed person-job matching model to obtain a matching result, and the person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model. Thus, the job recruitment data and resume data are matched through a person-job matching model constructed by a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model to obtain a matching result, which solves the problem in the prior art that the job requirements are not aligned with the resume advantages, resulting in low quality of corporate recruitment and talent loss, and improves the accuracy of person-job matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A flowchart of the first embodiment of the applicant-job matching method is provided; Figure 2 This is a flow chart of the process of building a person-job matching model for the applicant-job matching method; Figure 3 A flow chart of the second embodiment of the applicant-job matching method is provided; Figure 4 A schematic diagram of the matching process of the applicant-job matching method involving the person-job matching model; Figure 5 A brief flowchart of the person-job matching method provided in Example 2 of the present application; Figure 6 This is a schematic diagram of the module structure of the person-job matching device according to an embodiment of the present application; Figure 7 Schematic diagram of the device structure of the hardware operating environment involved in the person-job matching method in the embodiment of the present application.

[0020] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0022] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solution of the embodiment of the present application is: obtaining historical job information and historical resume information; performing cluster analysis on the historical job information to obtain a person-job matching dimension; based on the person-job matching dimension, extracting information from the historical job information and historical resume information through a first deep learning and natural language processing model to obtain a first extraction result; constructing a vector model optimization data set according to the first extraction result, and training an initial vector model based on the vector model optimization data set to obtain a recruitment vector model; performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an information extraction model; and extracting information from the historical job information and historical resume information through a second deep learning and natural language processing model to obtain a first extraction result. The deep learning and natural language processing model performs dimension weight evaluation on the person-job matching dimension and historical resume information to obtain an evaluation result; the second deep learning and natural language processing model performs knowledge distillation according to the evaluation result to obtain a weight evaluation model; the first extraction result and the evaluation result are input into the third deep learning and natural language processing model for matching scoring to obtain a first scoring result; the third deep learning and natural language processing model performs knowledge distillation according to the first scoring result to obtain a matching scoring model; the person-job matching model is constructed according to the recruitment vector model, the information extraction model, the matching scoring model and the weight evaluation model. Based on the person-job matching dimension, the first extraction result is dimensionally classified through the first deep learning and natural language processing model to obtain a first classification result; based on the person-job matching dimension, the first extraction result is dimensionally classified through the classification model to obtain a second classification result; the first classification result and the second classification result are cross-validated to obtain a dimensionally classified data set; according to the first extraction result, the first deep learning and natural language processing model is knowledge distilled to obtain an initial information extraction model; according to the dimensionally classified data set, the sub-network of the initial information extraction model is activated through the gating network, and the initial information extraction model is nonlinearly transformed through the activation function to obtain an information extraction model. Perform knowledge distillation on the third deep learning and natural language processing model to obtain an initial matching scoring model; input the first extraction result and the evaluation result into the initial matching scoring model to obtain a second scoring result; perform difference calculation based on the first scoring result and the second scoring result to obtain a cross entropy loss; perform difference calculation based on the probability distribution of the first scoring result and the probability distribution of the second scoring result to obtain a divergence loss; optimize the initial matching scoring model according to the cross entropy loss and the divergence loss to obtain a matching scoring model.The information extraction model is used to extract information from the recruitment position data and resume data to obtain a second extraction result; the weight evaluation model is used to evaluate the weight of the recruitment position data and resume data to obtain the position dimension weight and the resume dimension weight; the second extraction result is vectorized by the recruitment vector model to obtain a vector representation of the second extraction result; the second extraction result is sorted by similarity based on the position dimension weight and the resume dimension weight to obtain a number of candidate information; the several candidate information is matched by the matching scoring model to obtain a matching result. The job application result is received; the resume data is queried according to the job application result to obtain the applicant information; the initial employee evaluation form is sent to the employee management end according to the applicant information, and the employee management end fills the initial employee evaluation form to obtain the final employee evaluation form; the final employee evaluation form is received, and the person-job matching model is optimized by the final employee evaluation form to obtain a model optimization result. Thus, the problem of not aligning the job requirements with the resume advantages in the prior art, resulting in low recruitment quality and talent loss in the enterprise, is solved, the job matching is achieved, and the accuracy of person-job matching is improved. Based on the scheme of the present invention, a person-job matching method is designed based on the problem that the number of large model parameters is multiplied in reality, and the matching time and accuracy will be affected under limited computing power conditions, and the length of the content generated by model prediction will increase the time consumption of person-job matching, thereby reducing the accuracy. The effectiveness of the person-job matching method of the present invention is verified when matching jobs, and finally the accuracy of person-job matching performed by the method of the present invention is significantly improved.

[0024] In this embodiment, for the convenience of description, the following description is made with the person-job matching device as the execution subject.

[0025] Since the information extraction in the existing technology during the job matching is not fully integrated with the domain knowledge, the extraction performance is limited and the potential of the matching algorithm cannot be fully utilized. Secondly, the extraction dimensions designed in the existing job matching scheme are relatively small, and the extraction dimensions of the job and the resume are not aligned. The misalignment of the dimensions will cause more information interference when the feature extraction is performed in the vector space later, and the small number of extraction dimensions will not be able to evaluate the matching degree between the job and the talent resume from multiple perspectives. In addition, different jobs have different focuses during job matching. For this reason, most schemes are designed to evaluate the corresponding weights according to the focus dimensions of job requirements. However, this scheme is only applicable to the scenario of matching resumes based on job matching, and does not consider the advantages reflected in the resume when matching jobs based on talents in the job matching task. In addition, the matching sorting part is different from the HR evaluation method in the real scenario, resulting in the need to improve the interpretability of the quantitative matching results, which will also reduce the accuracy of job matching in the existing technology.

[0026] This application provides a solution, in which a person-job matching model is constructed by a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model, and the person-job matching model is used to match the received recruitment position data and resume data, thereby improving the accuracy and intelligence of person-job matching and providing users with better services.

[0027] It can be seen from the above embodiments that the present application matches the recruitment position data and resume data through a person-job matching model constructed by a recruitment vector model, an information extraction model, a weight evaluation model and a matching score model to obtain a matching result. This solves the problem in the prior art that the job requirements are not aligned with the resume advantages, resulting in low quality of corporate recruitment and talent loss, and improves the accuracy of person-job matching.

[0028] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device that can realize the above functions, a person-job matching device, etc. The following takes the person-job matching device as an example to illustrate this embodiment and the following embodiments.

[0029] Based on this, the present application embodiment provides a person-job matching method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the applicant-job matching method.

[0030] In this embodiment, the person-job matching method includes steps S01-S02: Step S01, receiving recruitment position data and resume data; Before explaining the scheme of this embodiment, it should be clear that in the job recruitment of an enterprise, the resume information of multiple applicants will be received. In order to find the most matching applicant based on these resume information, information extraction and person-job matching are required. Information extraction is an important part of the person-job matching task. The quality of key data extraction directly affects the matching results. Person-job matching is because different positions have different focuses. Therefore, it is necessary to evaluate the corresponding weights according to the focus dimensions of job requirements.

[0031] However, the information extraction methods designed for existing job matching solutions have not fully integrated domain knowledge, and the extraction performance is limited. The main problems are concentrated on: ① The information extraction link lacks domain knowledge integration, resulting in incomplete identification of key data; ② The evaluation dimension design is single, which makes it difficult to cover the diverse characteristics of job requirements; ③ The matching algorithm only focuses on the weight of job requirements, ignoring the core advantages of talents reflected in the resume (such as special skills, project results, etc.), resulting in mismatching of high-potential candidates and hidden talent loss.

[0032] Therefore, in this embodiment, the recruitment position data and resume data are received first. The recruitment position data includes the basic attributes of the position (such as position name, job grade system, salary range and work location, etc.), ability requirements (such as technology stack weight, framework requirements, certificate hard conditions and project experience years, etc.) and implicit data (such as technical depth indicators, team collaboration characteristics and industry-specific data, etc.). These recruitment position data are formulated by the employer and can be modified according to their specific needs. For example, for a software development position, the basic attributes can be: (1) Job titles: software engineer, front-end development engineer, back-end development engineer, full-stack engineer, etc.

[0033] (2) Job level system: such as junior, intermediate, senior, or more detailed job level systems such as intern, junior engineer, senior engineer, technical manager, technical director, etc.; (3) Salary range: for example, annual salary range, monthly salary range, or other salary structures (such as fixed salary + bonus, stock options, etc.); (4) Workplace: including company headquarters, branches, or remote office (Hybrid / Remote). It may also be broken down into specific cities or regions; (5) Nature of company: for example, start-up company, large Internet company, foreign-invested company, traditional industry, etc.

[0034] Competency requirements include: (1) Technology stack weight: the programming languages ​​(such as Java, Python, JavaScript, etc.), tools (such as Docker, Kubernetes, etc.), databases (such as MySQL, PostgreSQL, etc.), and related platforms (such as cloud platforms, DevOps tools, etc.) that candidates need to master; (2) Framework requirements: For front-end development positions, React, Vue.js or Angular may be required, and back-end development may require Spring Boot, Django, etc.; (3) Certification requirements: Specific technical certification or certificates are required, such as AWS certification, Google Cloud certification, PMP, etc. (4) Years of project experience: more than 3 years of project experience or years of experience in a specific technology stack is required; (5) Educational background: Bachelor degree or above is required, or whether there are major requirements in specific fields (such as computer science, software engineering, etc.); (6) Language requirements: Fluency in English is required, especially for foreign companies or multinational corporations.

[0035] The implicit data includes: (1) Technical depth indicators: Applicants are required to have certain architectural design capabilities, in-depth understanding of algorithms and data structures, or experience in designing large-scale distributed systems; (2) Team collaboration characteristics: The relevant positions require good teamwork skills, cross-departmental communication skills, project management skills, etc.; (3) Industry-specific data: requires knowledge of a specific industry, such as specific development experience in the financial industry or the medical industry; (4) Soft skills requirements: communication skills, problem-solving skills, ability to cope with stress, etc.; (5) Work model / culture: Emphasis on agile development and DevOps culture, or requiring support for rapid iteration and innovative thinking, etc.

[0036] In the process of talent recruitment, in addition to recruitment position data (i.e. company requirements), resume data (i.e. applicants' relevant information) is also required, including applicants' basic personal information (such as name, contact information, job search intentions, and personal profile), educational background (such as school name, degree, major, and graduation time), work experience (such as company name and position, working hours, job responsibilities and achievements, and use of technology or tools), skills and certificates (such as technical skills, soft skills, and certificates). However, when applicants submit their resumes, they may apply for them at many places, that is, send their resume information to multiple recruitment platforms, companies, or positions. This will cause resumes that do not meet the requirements to be sent to the company's recruitment department, which will cause the company to waste resources for screening when recruiting talents. In addition, some applicants' resumes are relatively simple or fail to grasp the key points of the company's requirements. The above two situations may lead to the failure to consider the advantages of the resume when matching job talents, resulting in talent loss.

[0037] Step S02, matching the recruitment position data and resume data with a pre-built person-job matching model to obtain a matching result, wherein the person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model, and a matching degree scoring model.

[0038] After receiving the recruitment position data and resume data, it can be assumed that the talent matching task of the enterprise has been received. At this time, the recruitment position data and the person-position matching data are matched by the person-position matching model constructed by the recruitment vector model, the information extraction model, the weight evaluation model and the matching score model to obtain the applicant resume (i.e., the matching structure) that best matches the above-mentioned recruitment position data. In order to solve the problem of talent loss, the recruitment vector model in this embodiment converts the description of the recruitment position and the content in the resume into a vector form. The text information of each position and resume will be converted into a numerical vector to facilitate the calculation of the similarity between them. The information extraction model automatically extracts key information from the position recruitment data and resume data, such as position name, skill requirements, work experience, educational background, etc. The weight evaluation model assigns different weights to different parts of the recruitment position and resume. For example, the "work experience" part of the position may have a greater weight than the "interests and hobbies" part, and the "professional skills" part of the resume has a higher weight. The weights of different information can be adjusted according to the important requirements of the position, and the matching score model calculates the matching score between the resume and the position according to the similarity between the position and the resume, as well as the weight of each part of the information, and outputs a matching score.

[0039] In this embodiment, a person-job matching model constructed by a recruitment vector model, an information extraction model, a weight evaluation model, and a matching score model is used to match recruitment job data and resume data to obtain the most suitable resume of the applicant, thereby solving the problem in the prior art of failing to align job requirements with resume advantages, resulting in low quality of corporate recruitment and talent loss.

[0040] Specifically, before using the person-job matching model to match the recruitment position data and the resume data, it is necessary to complete the construction of the person-job matching model in advance. Therefore, in the above step S02, before the step of matching the recruitment position data and the resume data by using the pre-constructed person-job matching model to obtain the matching result, the method further includes: Step S0201, obtaining historical job information and historical resume information; Step S0202, performing cluster analysis on the historical job information to obtain person-job matching dimensions; Step S0203, based on the person-job matching dimension, extract the historical job information and historical resume information through a first deep learning and natural language processing model to obtain a first extraction result; Step S0204, constructing a vector model optimization data set according to the first extraction result, and training the initial vector model based on the vector model optimization data set to obtain a recruitment vector model; Step S0205, performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an information extraction model; Step S0206, performing dimension weight evaluation on the person-job matching dimension and historical resume information through a second deep learning and natural language processing model to obtain an evaluation result; Step S0207, performing knowledge distillation on the second deep learning and natural language processing model according to the evaluation result to obtain a weight evaluation model; Step S0208, inputting the first extraction result and the evaluation result into a third deep learning and natural language processing model for matching scoring to obtain a first scoring result; Step S0209, performing knowledge distillation on the third deep learning and natural language processing model according to the first scoring result to obtain a matching scoring model; Step S02010, constructing a person-job matching model based on the recruitment vector model, information extraction model, matching degree scoring model and weight evaluation model.

[0041] The scheme of model construction in this embodiment is as follows Figure 2 As shown, the deep learning and natural language processing model used is DeepSeek R1, and it can also be other deep learning and natural language processing models, such as the GPT series (GPT-4o), Gemini, Kimi, and ERNIE Bot. After receiving historical job information and historical resume information, a cluster analysis is performed on the requirements of a large number of jobs, and the dimensions that need to be matched can be divided into six parts (i.e., person-job matching dimensions), namely: basic information, educational experience, industry background, project experience, professional skills, and general capabilities.

[0042] The vector matching and matching degree evaluation in the subsequent links need to be based on the extracted information. The quality of the extraction will directly affect the retrieval effect. Therefore, information extraction is not only the identification of relevant content, but also the analysis, classification, and summary of the content. To this end, this embodiment constructs an information extraction MoE model through a series of processes, that is, information is extracted based on historical job information and historical resume information through a first deep learning and natural language processing model. After obtaining the first extraction result, the first deep learning and natural language processing model is subjected to knowledge distillation to obtain an information extraction model.

[0043] This embodiment also constructs a recruitment field vector model fine-tuning dataset based on the first extraction result, and trains the vector model based on the dataset, so that the structure of the vector space is more consistent with the semantic relationship in the field, enhancing the understanding of professional vocabulary, and based on the fine-tuning vector model, the distinction between different positions and technical fields (such as front-end and back-end development) can be improved in the semantic space, thereby improving the targeted matching of people and positions.

[0044] In addition, the extracted information will be used for matching evaluation and Top n retrieval tasks. Using the extracted information in the matching evaluation task can effectively shorten the context of the model input, make the key information densely distributed, and reduce the impact of interference information on the matching results. Similarly, vectorizing the extracted text and performing matching calculations in various dimensions based on the vector can effectively improve the recall rate of this link. Therefore, in this embodiment, the first extraction result is also used to construct a vector model optimization data set, and then the initial vector model is trained with the vector model optimization data set to obtain a recruitment vector model, thereby improving the vector model's ability to vectorize specific information.

[0045] In addition, while large-scale parameter models improve the comprehensive capabilities of the model, they also increase the demand for computing resources. The tasks in the recruitment field are relatively clear. Therefore, this embodiment distills the capabilities of DeepSeek R1 in the recruitment field into a small model and strengthens it in a targeted manner, which can achieve better results with limited resources. In particular, in the matching evaluation task, the use of the distilled small model can achieve fast scoring, reordering and recommendation within limited computing resources. For this reason, this embodiment selects DeepSeek R1 as the teacher model, and subsequently completes the dimension weight evaluation and matching scoring to form a knowledge distillation data set corresponding to each task. Finally, knowledge distillation is performed on the knowledge distillation data set to obtain a matching scoring model and a weight evaluation model.

[0046] It can be seen from this that in this embodiment, deep learning and natural language processing models are used to extract information, evaluate dimension weights, and score matching in the recruitment field. Subsequently, in order to enhance the accuracy of the model and reduce the computing power required for the model, knowledge distillation is performed on the deep learning and natural language processing models that perform information extraction, dimension weight evaluation, and matching scoring, to obtain an information extraction model, a matching scoring model, and a weight evaluation model. Finally, the construction of the person-job matching model is completed, so that the model can more accurately extract information, evaluate weights, and score matching when dealing with recruitment position data and resume data, thereby solving the problem that the person-job matching task does not take into account the advantageous content in the resume, leading to talent loss.

[0047] More specifically, after the step S02 of performing data matching on the recruitment position data and the resume data by using a pre-built person-position matching model to obtain a matching result, the method further includes: Step S03, receiving job application results; Step S04, querying the resume data according to the job application result to obtain the applicant information; Step S05, sending an initial employee evaluation form to the employee management terminal according to the applicant information, and the employee management terminal fills in the initial employee evaluation form to obtain a final employee evaluation form; Step S06, receiving the final employee evaluation form, optimizing the person-job matching model according to the final employee evaluation form, and obtaining a model optimization result.

[0048] The system receives job application results from job search platforms or recruitment systems, which usually include applicants' resumes, job positions, application dates, and other information. It then stores the received application results in a database and identifies each application record with its corresponding job information.

[0049] Then, query the resume data based on the job application results to obtain the applicant information, and query the corresponding resume data based on the application results to obtain the applicant's detailed information (such as name, contact information, education, work experience, skills, etc.) to provide data support for subsequent analysis.

[0050] A corresponding initial employee evaluation form is generated according to the applicant information, namely the employee's personal information and work completion status, and the initial employee evaluation form is sent to the employee management end, wherein the employee management end in this embodiment includes at least the HR system and the employee's corresponding superior leader.

[0051] The employee evaluation form contains some basic questions, such as work experience, skill level, job suitability, etc., for HR personnel and superiors to make preliminary evaluations of applicants. However, the above content is only a basic evaluation of employees and cannot optimize the model specifically. Therefore, multiple scoring indicators can be added, such as: skill matching, experience matching, cultural adaptability, etc., to finally form a detailed employee evaluation form. The detailed employee evaluation form will also include the corresponding content filled in the resume information of previous employees, which can further deepen the efficiency and accuracy of information processing during the model training process.

[0052] After the final employee evaluation form is completed on the employee management side, the person-job matching model is optimized based on the final employee evaluation form using the collected feedback data. This process can be broken down into the following steps: (1) Recruitment vector model optimization: The recruitment vector model is optimized using the actual matching conditions in the employee evaluation form (such as employee adaptability and job performance). For example, the keyword weights or semantic vectors in the recruitment positions and resumes are adjusted so that the model can generate vector representations of positions and resumes that are more in line with actual job requirements and candidate performance. (2) Information extraction model optimization: Based on the actual feedback from the final employee evaluation form, the rules and strategies of the information extraction model are adjusted so that the model can more accurately extract key information from resumes and job descriptions, such as the importance of "work experience" or "professional skills". If certain skills or experience are highlighted as important in an employee's performance, the model will update the weights and pay more attention to the extraction and identification of these features; (3) Weight evaluation model optimization: Based on the scores in the final evaluation form, the weights of various features in the position and resume are adjusted. For example, if HR or superiors give particularly high evaluations to certain skills, the model will automatically increase the weight of the skill in the matching score, ensuring that more attention is paid to the skills and experience that the recruiter values ​​most during the matching process. (4) Optimization of the matching scoring model: The matching scoring model is optimized using the data from the final employee evaluation form to make the scoring more consistent with the actual job requirements and employee performance. Through regression analysis or classification algorithms, the model will be gradually adjusted so that the matching score can better reflect the actual recruitment results. If the employee evaluation form shows that the matching score of certain positions is low, the model will be adjusted through reinforcement learning and other technologies to retrain the scoring system in order to better predict and match new applicants.

[0053] Thus, the model optimization result can be obtained. After the optimization of the above steps, the updated person-job matching model is finally obtained. The model will more accurately evaluate the matching degree between applicants and positions, and improve the efficiency and accuracy of recruitment.

[0054] In this embodiment, the person-job matching model is finely adjusted by continuously receiving feedback (final employee evaluation form), and each sub-model (recruitment vector model, information extraction model, weight evaluation model, and matching scoring model) is optimized, thereby improving recruitment efficiency and ensuring a more accurate matching between positions and candidates.

[0055] This embodiment uses the above scheme to specifically receive job data and resume data; and matches the job data and resume data through a pre-built job matching model to obtain a matching result. The job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model, and a matching degree scoring model. Thus, the job data and resume data are matched through a job matching model constructed by a recruitment vector model, an information extraction model, a weight evaluation model, and a matching degree scoring model to obtain a matching result, which solves the problem in the prior art that the job requirements are not aligned with the resume advantages, resulting in low quality of corporate recruitment and talent loss, and improves the accuracy of job matching.

[0056] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the first embodiment can refer to the above introduction, and will not be repeated in the following. Figure 3 In step S0205, in the step of performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an information extraction model, the information extraction method further includes steps S02051 to S02055: Step S02051, based on the person-job matching dimension, classify the first extraction result by dimension through the first deep learning and natural language processing model to obtain a first classification result; Step S02052, based on the person-job matching dimension, classify the first extraction result by dimension through a classification model to obtain a second classification result; Step S02053, cross-validating the first classification result and the second classification result to obtain a dimension classification data set; Step S02054, performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an initial information extraction model; Step S02055, according to the dimension classification data set, activate the sub-network of the initial information extraction model through the gating network, and perform nonlinear transformation on the initial information extraction model through the activation function to obtain the information extraction model.

[0057] In this embodiment, the purpose is to reasonably distribute text fragments to the expert models corresponding to each dimension to adapt to the situation where discontinuous information is extracted from the file. For example, the text content extracted from the resume file in PDF format by OCR technology contains many text fragments. In order to better train the gating network focusing on content distribution, this embodiment constructs a dimension classification data set, wherein the specific method of constructing the dimension classification data set includes: With the help of DeepSeek R1, the collected resume and job text fragments are classified into corresponding dimensions. There is a situation where one fragment corresponds to multiple dimensions. At the same time, the same classification is performed based on the classification model (i.e., the bean bag model). Subsequently, the classification results of the two models are cross-validated, and the inconsistent parts in the cross-validation are manually processed to complete the production of the dimensional classification data set.

[0058] Then, knowledge distillation is performed on the first deep learning and natural language processing model according to the first extraction result obtained previously to obtain an initial information extraction model.

[0059] Then, based on the dimension classification data set, this embodiment targets the gated network Whether to activate the i-th expert model Set dimensional vector space , where 0 means inactive and 1 means active. is the input, is the gating network parameter, which can be expressed as follows:

[0060] In this embodiment, whether each expert model is activated is regarded as a binary classification task, and the Sigmoid activation function is selected to express the output of the gated network as:

[0061] in, is the weight parameter, is the bias parameter.

[0062] At the same time, based on the enhanced knowledge distillation dataset, the objective function is established as follows:

[0063] in, is the input sample size, is the number of experts, Indicates that the gating network is Expert in input The output, i.e. the dimension The extraction result.

[0064] After completing the activation and nonlinear transformation of the distilled initial information extraction model through the above scheme, a more accurate information extraction model is obtained. When facing the recruitment field, more accurate job information and resume information can be matched, thereby solving the problem of inaccurate information extraction, resulting in errors in subsequent weight allocation and score matching, and thus talent loss.

[0065] More specifically, in the above embodiment, the step S0209 of performing knowledge distillation on the third deep learning and natural language processing model according to the first scoring result to obtain a matching scoring model includes: Step S02091, performing knowledge distillation on the third deep learning and natural language processing model to obtain an initial matching degree scoring model; Step S02092, inputting the first extraction result and the evaluation result into the initial matching degree scoring model to obtain a second scoring result; Step S02093, performing difference calculation based on the first scoring result and the second scoring result to obtain a cross entropy loss; Step S02094, performing a difference calculation based on the probability distribution of the first scoring result and the probability distribution of the second scoring result to obtain a divergence loss; Step S02095, optimizing the initial matching score model according to the cross entropy loss and the divergence loss to obtain a matching score model.

[0066] The main task of this step is to transfer DeepSeek R1's capabilities in six-dimensional information extraction, weight allocation, and matching evaluation to the student model through distillation. Although the tasks are different, the overall methods are similar. Taking the matching scoring model as an example, the specific process is as follows: (1) Perform knowledge distillation on the DeepSeek R1 model that performs matching evaluation to distill out the initial matching scoring model; (2) The DeepSeek R1 model that performs matching evaluation is used as the teacher model, and the initial matching scoring model is used as the student model. The probability distribution of the teacher model and the probability distribution of the student model are calculated using the following formula:

[0067] in, represents the DeepSeek R1 model, represents the student model, Indicates the task input samples, Represents the model for the input The resulting output probability distribution.

[0068] During the knowledge distillation process performed in this embodiment, the student model needs to learn the probability distribution of the teacher model output while learning the target task. To this end, the cross entropy loss and KL divergence loss are combined in the distillation process to comprehensively evaluate the training situation. The cross entropy loss is used to evaluate the difference between the student model output and the true label, while the KL divergence loss is used to measure the difference between the output distribution of the student model and the teacher model.

[0069] (3) Through this dual loss mechanism (i.e., cross entropy loss and KL divergence loss), the student model can better imitate the behavior of the teacher model, thereby improving its performance. The specific loss calculation is as follows: The calculation formula of the cross entropy loss is:

[0070] The calculation formula of KL divergence loss is:

[0071] The total distillation loss is calculated as:

[0072] in, is the cross entropy loss, is the divergence loss, is the total number of samples, is the total distillation loss, is the final output of the teacher model, is the cross entropy loss coefficient.

[0073] (4) After obtaining the total loss of the student model and the teacher model, the initial matching score model can be optimized based on the total loss to obtain the matching score model.

[0074] In this embodiment, it should be noted that for different tasks, Corresponding adjustments need to be made. Taking the information extraction model as an example, in the distillation process of the expert models of each dimension of information extraction, this embodiment has been strengthened accordingly based on the output of the teacher model according to the characteristics of the extraction dimension, so that the model has stronger domain information mining capabilities. This embodiment will collect relevant public information from universities, enterprises, social organizations, etc. to build a corresponding knowledge base, and provide each expert model with more timely information in a RAG manner. Specifically, the information extraction tasks focused on by the expert models of each dimension are as follows: The basic information extraction expert model extracts information including mobile phone number, email, age, work location, etc. By adding certain inference data, the model has a certain inference ability and can further complete the information, such as calculating age based on date of birth and completing province based on city.

[0075] The educational experience extraction expert model extracts information including academic qualifications, whether full-time, whether enrolled in the national college entrance examination, school name and national ranking, major studied, etc. Combined with the information provided by the knowledge base, the model can identify and complete the missing information.

[0076] The industry background extraction expert model focuses on mining the industry attributes involved in resumes and job description information. In essence, the model completes a multi-classification task. In actual scenarios, a description may involve multiple industry attributes. For example, the project experience description of the wind power and power conversion efficiency detection system mentioned in the resume involves two industries: new energy and software information services. For jobs, this dimension expert model can obtain more information about job providers based on the knowledge base, thereby extracting industry background.

[0077] The project experience extraction expert model mainly focuses on information such as the project scale, content, complexity, and the role of the position in the project. Many resumes do not directly mention information such as project scale when describing project experience. Therefore, this dimension expert model analyzes other mentioned content and mines potential information.

[0078] The expert model for professional ability extraction focuses on skill nouns and proficiency for extraction. The focus of this dimension includes two parts: skill extraction and cleaning. In order to improve the targeted matching, the parts irrelevant to the expected position are ignored in the skill noun extraction. For example, the irrelevant software operation capabilities such as office and PS are ignored in the skill nouns mastered by Java development engineers. Professional skills are one of the important dimensions of person-job matching. Improving the relevance of the extracted skills and positions can effectively improve the recall rate in the vector retrieval link.

[0079] The general competence extraction expert model focuses on analyzing the comprehensive qualities of innovation, communication and collaboration, and stress resistance from the self-description and project description in the resume. For positions, this dimension expert model can combine the company introduction and position description information in the knowledge base to analyze the general competence requirements for talents in this position.

[0080] That is, by distilling the model in the above manner, we can obtain optimized matching scoring model, information extraction model and weight evaluation model, which improves the overall accuracy of the person-job matching model and solves the problem that the person-job matching task does not consider the advantages of the resume, leading to talent loss.

[0081] Furthermore, in the above step S02, the step of performing data matching on the recruitment position data and the resume data by using a pre-built person-position matching model to obtain a matching result includes: Step S021, extracting information from the recruitment position data and resume data through the information extraction model to obtain a second extraction result; Step S022, weight evaluation is performed on the recruitment position data and resume data through the weight evaluation model to obtain position dimension weight and resume dimension weight; Step S023, vectorizing the second extraction result by using the recruitment vector model to obtain a vector representation of the second extraction result; Step S024, sorting the second extraction results by similarity based on the position dimension weight and the resume dimension weight to obtain information on several candidates; Step S025, matching the plurality of candidate information by using the matching degree scoring model to obtain a matching result.

[0082] The overall solution of this embodiment is as follows Figure 4 As shown, after receiving new resumes and job data, weight allocation and information extraction of each dimension are completed in parallel in an asynchronous manner. The extracted texts of the six dimensions are vectorized and stored separately through the recruitment vector model. For positions with relatively simple descriptions, the requirements for candidates in terms of project experience and other aspects may not be mentioned. This embodiment designs a preset value filling method to guide the vector model and the vector database for encoding and retrieval. Taking the Java development position as an example, when the project experience is not mentioned, the preset value filled is: "It is required to have participated in Java development related projects, and those who have participated in the optimization design of complex scenarios such as high concurrency are preferred."

[0083] Then, people can be matched directly to positions. The specific principles are as follows: Let the vector of each dimension be , the vectors of each dimension of the position are , each position and resume has a corresponding weight, and the weight of each dimension of the position is , the weight of each dimension of resume is , taking the process of matching resumes according to job information as an example, the calculation method of the initial screening recommendation score can be expressed as:

[0084] After sorting based on similarity, the top n candidates can be initially screened out. The information of various dimensions of the candidates who have passed the initial screening can be obtained from the vector database. The matching evaluation model is used to simulate the real HR evaluation principles for scoring and sorting, so as to obtain the final result of matching resumes according to positions.

[0085] Similarly, based on , a similar method can be used to match positions based on talents.

[0086] This embodiment adopts the above scheme, specifically by dimensionally classifying the first extraction result based on the person-job matching dimension through the first deep learning and natural language processing model to obtain a first classification result; based on the person-job matching dimension, dimensionally classifying the first extraction result through the classification model to obtain a second classification result; cross-validating the first classification result and the second classification result to obtain a dimension classification data set; performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an initial information extraction model; according to the dimension classification data set, activating the sub-network of the initial information extraction model through the gated network, and performing nonlinear transformation on the initial information extraction model through the activation function to obtain an information extraction model. Thus, the person-job matching model constructed by the recruitment vector model, the information extraction model, the weight evaluation model and the matching score model is used to match the recruitment position data and the resume data to obtain a matching result, which solves the problem in the prior art that the job requirements are not aligned with the resume advantages, resulting in low recruitment quality and talent loss in the enterprise, and improves the accuracy of person-job matching.

[0087] For example, to help understand the implementation process of the person-job matching method obtained by combining this embodiment with the above-mentioned embodiment 1, please refer to Figure 5 , Figure 5 A brief flowchart of a person-job matching method is provided, specifically: Step S1 is the construction of the model. Prioritize the acquisition of historical job information and historical resume information, and then clean and desensitize them to form a person-job matching data set. Finally, DeepSeek R1 is selected as the teacher model and the small parameter model is selected as the student model. Distillation is performed on the six dimensions of information extraction, weight allocation, and matching evaluation to form a six-dimensional information extraction MoE model, a weight allocation model, and a matching scoring model.

[0088] Step S is the preparation before matching, that is, after receiving new job data and resume data, the information fragments of different parts are handed over to the MoE model to complete the information extraction, and the weight distribution model is used to complete the weight distribution of each resume and demand. Then the extracted information is vectorized by dimension and stored in the vector database together with the weight information of each dimension.

[0089] Step S3 is the matching stage, which is to calculate based on semantic similarity, screen out TOP n based on weights, obtain preliminary screening results, and finally construct contextual information of resumes and positions with the extracted information, and submit it to the matching evaluation model for scoring, and make the final refinement to give the matching results.

[0090] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the job matching method of the present applicant. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0091] This application also provides a person-job matching device, please refer to Figure 6 , the person-job matching device comprises: Receiving module 10, used to receive recruitment position data and resume data; The matching module 20 is used to match the recruitment position data and resume data through a pre-built person-job matching model to obtain a matching result. The person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model.

[0092] The person-job matching device provided by the present application adopts the person-job matching method in the above embodiment, which can solve the technical problem that the job requirements and resume advantages are not aligned in the prior art, resulting in low recruitment quality and talent loss in enterprises. Compared with the prior art, the beneficial effects of the person-job matching device provided by the present application are the same as the beneficial effects of the person-job matching method provided by the above embodiment, and the other technical features in the person-job matching device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0093] The present application provides a person-job matching device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the person-job matching method in the above-mentioned embodiment 1.

[0094] Reference below Figure 7 , which shows a schematic diagram of the structure of a person-job matching device suitable for implementing the embodiment of the present application. The person-job matching device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The person-job matching device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0095] like Figure 7As shown, the person-job matching device may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory 1002 or the program loaded from the storage device 1003 to the random access memory 1004. In the random access memory 1004, various programs and data required for the operation of the xxx device are also stored. The processing device 1001, the read-only memory 1002 and the random access memory 1004 are connected to each other through a bus 1005. The input / output interface 1006 is also connected to the bus. Generally, the following systems can be connected to the input / output interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the person-job matching device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a person-job matching device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided alternatively.

[0096] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0097] The person-job matching device provided by the present application adopts the person-job matching method in the above embodiment, which can solve the technical problem that the job requirements and resume advantages are not aligned in the prior art, resulting in low recruitment quality and talent loss in enterprises. Compared with the prior art, the beneficial effects of the person-job matching device provided by the present application are the same as the beneficial effects of the person-job matching method provided by the above embodiment, and the other technical features in the person-job matching device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0098] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0099] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0100] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the person-job matching method in the above-mentioned embodiment.

[0101] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0102] The computer-readable storage medium may be included in the person-job matching device; or may exist independently without being assembled into the person-job matching device.

[0103] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the person-job matching device, the person-job matching device: receives recruitment position data and resume data; performs data matching on the recruitment position data and resume data through a pre-built person-job matching model to obtain a matching result. The person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model.

[0104] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0105] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0106] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0107] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned person-job matching method, and can solve the technical problem in the prior art that the job requirements are not aligned with the resume advantages, resulting in low quality of corporate recruitment and talent loss. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the person-job matching method provided by the above-mentioned embodiment, and will not be repeated here.

[0108] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned person-job matching method when executed by a processor.

[0109] The computer program product provided by this application can solve the technical problem in the prior art that job requirements are not aligned with resume advantages, resulting in low recruitment quality and talent loss in enterprises. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as those of the person-job matching method provided in the above embodiment, and will not be elaborated here.

[0110] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A person-job matching method, characterized in that: The person-job matching method comprises: Receive recruitment position data and resume data; The recruitment position data and resume data are matched with each other through a pre-built person-job matching model to obtain a matching result. The person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model and a matching degree scoring model.

2. The person-job matching method according to claim 1, characterized in that: Before the step of performing data matching on the recruitment position data and the resume data by using the pre-built person-position matching model to obtain a matching result, the method further includes: Obtain historical job information and historical resume information; Performing cluster analysis on the historical job information to obtain person-job matching dimensions; Based on the person-job matching dimension, extract the historical job information and historical resume information through a first deep learning and natural language processing model to obtain a first extraction result; Constructing a vector model optimization data set according to the first extraction result, and training an initial vector model based on the vector model optimization data set to obtain a recruitment vector model; Performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an information extraction model; Performing a dimension weight evaluation on the person-job matching dimension and historical resume information through a second deep learning and natural language processing model to obtain an evaluation result; Performing knowledge distillation on the second deep learning and natural language processing model according to the evaluation result to obtain a weight evaluation model; Inputting the first extraction result and the evaluation result into a third deep learning and natural language processing model for matching scoring to obtain a first scoring result; Performing knowledge distillation on the third deep learning and natural language processing model according to the first scoring result to obtain a matching scoring model; A person-job matching model is constructed based on the recruitment vector model, information extraction model, matching degree scoring model and weight evaluation model.

3. The method according to claim 2, characterized in that The step of performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an information extraction model includes: Based on the person-job matching dimension, the first extraction result is dimensionally classified by the first deep learning and natural language processing model to obtain a first classification result; Based on the person-job matching dimension, the first extraction result is dimensionally classified by a classification model to obtain a second classification result; Cross-validating the first classification result and the second classification result to obtain a dimension classification data set; Performing knowledge distillation on the first deep learning and natural language processing model according to the first extraction result to obtain an initial information extraction model; According to the dimensionally classified data set, the sub-network of the initial information extraction model is activated through a gating network, and the initial information extraction model is nonlinearly transformed through an activation function to obtain an information extraction model.

4. The method according to claim 2, characterized in that The step of performing knowledge distillation on the third deep learning and natural language processing model according to the first scoring result to obtain a matching scoring model comprises: Performing knowledge distillation on the third deep learning and natural language processing model to obtain an initial matching degree scoring model; Inputting the first extraction result and the evaluation result into the initial matching degree scoring model to obtain a second scoring result; Performing difference calculation based on the first scoring result and the second scoring result to obtain a cross entropy loss; Performing a difference calculation based on the probability distribution of the first scoring result and the probability distribution of the second scoring result to obtain a divergence loss; The initial matching score model is optimized according to the cross entropy loss and the divergence loss to obtain a matching score model.

5. The method according to claim 1, characterized in that The step of performing data matching on the recruitment position data and the resume data by using the pre-built person-position matching model to obtain a matching result comprises: Extracting information from the recruitment position data and resume data using the information extraction model to obtain a second extraction result; The weight evaluation model is used to evaluate the recruitment position data and resume data to obtain the position dimension weight and the resume dimension weight; vectorizing the second extraction result by using the recruitment vector model to obtain a vector representation of the second extraction result; Based on the weight of the position dimension and the weight of the resume dimension, the second extraction results are sorted by similarity to obtain information of several candidates; The plurality of candidate information are matched using the matching degree scoring model to obtain a matching result.

6. The method according to claim 1, characterized in that After the step of performing data matching on the recruitment position data and the resume data by using the pre-built person-position matching model to obtain a matching result, the method further includes: Receive job application results; According to the job application results, query the resume data to obtain the applicant information; Sending an initial employee evaluation form to an employee management terminal according to the applicant information, and the employee management terminal filling in the initial employee evaluation form to obtain a final employee evaluation form; The final employee evaluation form is received, and the person-job matching model is optimized according to the final employee evaluation form to obtain a model optimization result.

7. A person-job matching device, characterized in that: The person-job matching device comprises: Receiving module, used to receive recruitment position data and resume data; The matching module is used to match the recruitment position data and resume data through a pre-built person-job matching model to obtain a matching result. The person-job matching model is constructed based on a recruitment vector model, an information extraction model, a weight evaluation model, and a matching degree scoring model.

8. A person-job matching device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the person-job matching method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the person-job matching method as described in any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the person-job matching method according to any one of claims 1 to 6 are implemented.

Images