People and post matching method and equipment based on digital substitute, and medium

Abstract

The invention discloses a person and post matching method and device based on a digital substitute and a medium. Comprising the steps of training a large model based on a historical recruitment data set to obtain a recruiter model, and training the large model based on a historical job hunting data set to obtain a job hunter model; inputting recruiter information into the recruiter model to generate a recruiter digital substitute, and inputting job hunter information into the job hunter model to generate a job hunter digital substitute; if the recruiter information and the job seeker information are successfully matched, the recruiter digital substitute performs simulation communication with the job seeker digital substitute for the multiple target investigation items to obtain a simulation communication message list; inputting the simulation communication message list as a context cue word into an LLM, and inputting the plurality of target investigation items and associated preset scoring weights into the LLM, so that the LLM determines scoring results of the current job hunting user under the plurality of target investigation items based on the simulation communication message list and the preset scoring weights; and determining that the man-post matching is successful when the scoring result exceeds a preset score line.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, in particular to the field of person-job matching technology, and in particular to a person-job matching method, device and medium based on digital avatars. Background Art

[0002] In modern human resource management, recruitment platforms play a vital role. They transmit recruitment information to job seekers around the world through the Internet, greatly expanding the scope of recruitment for enterprises. Traditional recruitment methods are often limited by geography and time, while recruitment platforms break through these limitations, allowing enterprises to reach a large number of potential talents more efficiently. In addition, recruitment platforms also optimize the recruitment process and improve recruitment efficiency and quality by providing functions such as resume screening, job recommendations, and online interviews. Based on this, in the current job search and recruitment field, various recruitment platforms have become an important channel for enterprises to find suitable talents.

[0003] Traditional job matching relies on a complex big data labeling system. Recruitment platforms try to use big data analysis technology to match job seekers and positions by labeling them with various labels. Although this method improves the accuracy of matching to a certain extent, it requires a lot of data processing and analysis, increases R&D costs and time, and also affects the efficiency of job matching. Summary of the invention

[0004] In view of this, the embodiments of the present application provide a method, device, equipment, medium and product for matching people and positions based on digital avatars, which can determine whether people and positions match through communication between digital avatars, change the traditional person-job matching model, reduce dependence on big data labeling systems, reduce R&D costs, and improve matching efficiency.

[0005] In a first aspect, an embodiment of the present application provides a method for matching people and positions based on digital avatars, the method comprising: screening historical recruitment data sets corresponding to each recruiting user and historical job search data sets corresponding to each job seeker from recruitment platform data, and training a large model based on the historical recruitment data sets to obtain a recruiter model, and training a large model based on the historical job search data sets to obtain a job seeker model, wherein the historical recruitment data sets include job description information of each position uploaded by the recruiting user, submitted resume information received for each position, recruitment question messages sent to the job seeker and their intent labels, and the historical job search data sets include resume information uploaded by the job seeker, job description information of the submitted application, received recruitment question messages, and job search messages returned in response to the recruitment question messages; when the recruiter information input by the current recruiting user and the job seeker information input by the current job seeker are obtained, a recruiter digital avatar is generated by inputting the recruiter information into the recruiter model, and a job seeker digital avatar is generated by inputting the recruiter information into the job seeker model The job seeker information generates a digital avatar of the job seeker, wherein the recruiter information includes job description information, enterprise advantage information, job advantage information, multiple target assessment items and their associated preset scoring weights, and the job seeker information includes resume information and personal demand information; only the job label is extracted from the recruiter information and the job seeker information for function matching, and when the function matching is successful, the recruiter's digital avatar is triggered to simulate communication with the job seeker's digital avatar for multiple target assessment items to obtain a list of simulated communication messages; the simulated communication message list is input into the large language model LLM as context prompt words, and the multiple target assessment items and their associated preset scoring weights are input into the LLM, so that the LLM determines the scoring results of the current job seeker under the multiple target assessment items based on the simulated communication message list and the preset scoring weights; when the scoring result exceeds the preset score line, it is determined that the person-job matching is successful, and the current job seeker is taken as a candidate, and the scoring result and resume information of the current job seeker are submitted to the current recruiting user.

[0006] In a second aspect, an embodiment of the present application provides a person-job matching device based on a digital avatar, the device comprising: a training module, for screening historical recruitment data sets corresponding to each recruiting user and historical job search data sets corresponding to each job seeker from the recruitment platform data, and training a large model based on the historical recruitment data sets to obtain a recruiter model, and training a large model based on the historical job search data sets to obtain a job seeker model, wherein the historical recruitment data sets include job description information of each position uploaded by the recruiting user, the submitted resume information received for each position, recruitment question messages sent to the job seeker and their intent labels, and the historical job search data sets include resume information uploaded by the job seeker, job description information of the submitted application, recruitment question messages received, and job search messages returned for the recruitment question messages; a generation module, for generating a recruiter digital avatar by inputting the recruiter information into the recruiter model when the recruiter information input by the current recruiting user and the job seeker information input by the current job seeker are obtained, and by inputting the job seeker information into the job seeker model Generate a digital avatar for a job seeker, wherein the recruiter information includes job description information, enterprise advantage information, job advantage information, multiple target assessment items and their associated preset scoring weights, and the job seeker information includes resume information and personal demand information; a person-job matching module is used to extract only the job label from the recruiter information and the job seeker information for function matching, and when the function matching is successful, trigger the recruiter's digital stand-in to simulate communication with the job seeker's digital stand-in for multiple target assessment items to obtain a list of simulated communication messages; a determination module is used to input the list of simulated communication messages as context prompts into a large language model (LLM), and input multiple target assessment items and their associated preset scoring weights into the LLM, so that the LLM determines the scoring results of the current job seeker under the multiple target assessment items based on the simulated communication message list and the preset scoring weights; a submission module is used to determine that the person-job matching is successful when the scoring result exceeds the preset score line, and submit the current job seeker as a candidate and the scoring result and resume information of the current job seeker to the current recruiting user.

[0007] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the steps of the human-job matching method based on digital avatars as in the first aspect are implemented.

[0008] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the steps of the human-job matching method based on digital avatars as in the first aspect are implemented.

[0009] In a fifth aspect, an embodiment of the present application provides a computer program product, which is stored in a non-volatile storage medium. When the computer program product is executed by a processor, it implements the steps of the human-job matching method based on digital avatars as in the first aspect.

[0010] In a sixth aspect, an embodiment of the present application provides a chip, which includes a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the steps of the human-job matching method based on digital avatars as in the first aspect.

[0011] The present application provides a method, device, equipment, medium and product for matching people and positions based on digital avatars. A recruiter model that simulates the characteristics of recruiters is trained based on historical recruitment data sets, so that the model can learn the questioning methods and decision-making capabilities of each recruiter on the platform when facing different resumes and different job seekers during the recruitment process. In addition, by inputting the recruiter information of the current recruiting user into the trained recruiter model, the generated recruiter digital avatar can understand the recruitment needs and corporate culture. The subsequent recruiter digital avatar can combine the real recruitment needs to conduct professional and detailed investigations on job seekers in the simulated communication link. A job seeker model that simulates the characteristics of job seekers is trained based on historical job search data sets, so that the model can learn the response methods and performance capabilities of each job seeker on the platform when facing different recruitment questions during the job search process. In addition, by inputting the job seeker information of the current job seeker user into the trained job seeker model, the generated job seeker digital avatar can understand the job search needs and personal preferences. The subsequent job seeker digital avatar can combine the real job search needs to accurately and effectively answer the various questions of the recruiter in the simulated communication link. This application only needs to extract the job tags for matching. If the match is successful, the digital avatars of both parties will simulate communication and generate a chat list. The chat list is used as a context prompt and input into the scoring agent LLM. LLM will evaluate according to the target assessment items and weights preset by HR, generate a scoring result, and finally compare the scoring result with the preset score line to obtain the person-job matching result. This matching method is relatively simple, and the complex person-job matching process can be handed over to the two digital avatars through simulated communication to achieve accurate person-job matching based on the chat context, and automatically evaluate and recommend suitable candidates according to the preset scoring criteria, without relying on complex big data labeling systems, so there is no need for a large amount of data processing and analysis, reducing R&D costs. At the same time, the accuracy and efficiency of person-job matching are improved through the multi-agent mechanism. In addition, it can also reduce communication costs, reduce the workload of HR, improve recruitment efficiency, and help companies quickly find suitable candidates. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following briefly introduces the drawings in the embodiments of the present application.

[0013] Figure 1 It is a flowchart of a method for matching people and positions based on digital avatars provided in one embodiment of the present application;

[0014] Figure 2 is a flowchart of a method for matching people and positions based on digital avatars provided in another embodiment of the present application;

[0015] Figure 3 It is a structural schematic diagram of a person-position matching device based on a digital avatar provided in an embodiment of the present application;

[0016] Figure 4 It is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The principles and spirit of the present application will be described below with reference to several exemplary embodiments. It should be understood that the purpose of providing these embodiments is to make the principles and spirit of the present application clearer and more thorough, so that those skilled in the art can better understand and implement the principles and spirit of the present application. The exemplary embodiments provided herein are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments herein, all other embodiments obtained by ordinary technicians of the art without creative work are within the scope of protection of this application.

[0018] In this document, terms such as first, second, third, etc. are only used to distinguish one entity (or operation) from another entity (or operation), but not to require or imply any order or relationship between these entities (or operations).

[0019] The following is a brief description of the concepts and technical terms that may be involved in the embodiments of the present application.

[0020] Digital Avatar: A virtual agent created using artificial intelligence technology that can communicate and exchange information online with recruiters on behalf of job seekers. A digital avatar can understand and convey the user's needs, preferences and questions, and is the core unit for matching people with jobs.

[0021] AI Large Model: refers to an advanced artificial intelligence model with deep learning and natural language processing capabilities, which can understand and generate complex language structures, and is used to create and manage digital doubles, as well as handle communication between digital doubles.

[0022] Person-Job Matching: refers to the process of matching suitable job seekers with suitable positions in human resource management. This application improves the efficiency and accuracy of this process through digital doubles and AI big models.

[0023] The following, in conjunction with the accompanying drawings, describes in detail the person-job matching method based on digital avatars provided in the embodiment of the present application through specific embodiments and their application scenarios.

[0024] Figure 1 It is a flowchart of a method for matching people and jobs based on digital avatars provided in one embodiment of the present application. The executing entity of the method for matching people and jobs based on digital avatars may be a recruitment platform.

[0025] The following takes the recruitment platform as an example of the execution subject of the method for matching people and positions based on digital avatars to employees, and explains the method for matching people and positions based on digital avatars in this application. It should be noted that the above execution subjects and application scenarios do not constitute a limitation on this application.

[0026] like Figure 1 As shown, the person-job matching method based on digital avatars provided in the embodiment of the present application may include steps 110 to 150.

[0027] Step 110, screening the historical recruitment data sets corresponding to each recruiting user and the historical job search data sets corresponding to each job seeker from the recruitment platform data, and training a large model based on the historical recruitment data sets to obtain a recruiter model, and training a large model based on the historical job search data sets to obtain a job seeker model;

[0028] Step 120, when the recruiter information input by the current recruiting user and the job seeker information input by the current job seeker user are obtained, a recruiter digital avatar is generated by inputting the recruiter information into the recruiter model, and a job seeker digital avatar is generated by inputting the job seeker information into the job seeker model;

[0029] Step 130, extracting only the position tags from the recruiter information and the job seeker information to perform function matching. If the function matching is successful, triggering the recruiter's digital double to perform simulated communication with the job seeker's digital double for multiple target assessment items to obtain a simulated communication message list;

[0030] Step 140, inputting the simulated communication message list as context prompt words into the large language model LLM, and inputting multiple target assessment items and their associated preset scoring weights into the LLM, so that the LLM determines the scoring results of the current job seeker under the multiple target assessment items based on the simulated communication message list and the preset scoring weights;

[0031] Step 150, when the scoring result exceeds the preset score line, it is determined that the person-job match is successful, and the current job seeker is taken as a candidate, and the scoring result and resume information of the current job seeker are submitted to the current recruiting user.

[0032] The person-job matching method based on digital avatars provided in the embodiment of the present application is based on the historical recruitment data set to train the recruiter model that simulates the characteristics of the recruiter, so that the model can learn the questioning methods and decision-making capabilities of each recruiter on the platform when facing different resumes and different job seekers during the recruitment process. In addition, by inputting the recruiter information of the current recruiting user into the trained recruiter model, the generated recruiter digital avatar can understand the recruitment needs and corporate culture, and the subsequent recruiter digital avatar can combine the real recruitment needs in the simulated communication link to conduct professional and detailed investigations on job seekers. The job seeker model that simulates the characteristics of job seekers based on the historical job search data set can enable the model to learn the response methods and performance capabilities of each job seeker on the platform when facing different recruitment questions during the job search process. In addition, by inputting the job seeker information of the current job seeker user into the trained job seeker model, the generated job seeker digital avatar can understand the job search needs and personal preferences, and the subsequent job seeker digital avatar can combine the real job search needs in the simulated communication link to accurately and effectively answer the various questions of the recruiter. This application only needs to extract the job tags for matching. If the match is successful, the digital avatars of both parties will simulate communication and generate a chat list. The chat list is used as a context prompt and input into the scoring agent LLM. LLM will evaluate according to the target assessment items and weights preset by HR, generate a scoring result, and finally compare the scoring result with the preset score line to obtain the person-job matching result. This matching method is relatively simple, and the complex person-job matching process can be handed over to the two digital avatars through simulated communication to achieve accurate person-job matching based on the chat context, and automatically evaluate and recommend suitable candidates according to the preset scoring criteria, without relying on complex big data labeling systems, so there is no need for a large amount of data processing and analysis, reducing R&D costs. At the same time, the accuracy and efficiency of person-job matching are improved through the multi-agent mechanism. In addition, it can also reduce communication costs, reduce the workload of HR, improve recruitment efficiency, and help companies quickly find suitable candidates.

[0033] The specific implementation of the above steps will be described in detail below in conjunction with specific embodiments.

[0034] Involving step 110, the historical recruitment data set corresponding to each recruiting user and the historical job search data set corresponding to each job seeker are screened from the recruitment platform data, and a large model is trained based on the historical recruitment data set to obtain a recruiter model, and a large model is trained based on the historical job search data set to obtain a job seeker model.

[0035] In step 110, the historical recruitment dataset includes the job description information of each position uploaded by the recruiting user, the submitted resume information received for each position, the recruitment question messages sent to the job seekers and their intent labels. The historical job search dataset includes the resume information uploaded by the job seekers, the job description information sent for the submitted application, the received recruitment question messages, and the job search messages returned in response to the recruitment question messages.

[0036] Job description information is used to describe in detail the job responsibilities, requirements, working environment, salary and other key information, aiming to help potential candidates understand the overall situation of the position and judge whether they are suitable for the position.

[0037] The intent label can be used to characterize the question intent of the recruitment question message. Specifically, the question intent of the recruitment question message can be identified through a natural language semantic analysis algorithm to obtain the intent label.

[0038] For example, for the recruitment question message "Hello, we are recruiting a project manager and would like to know whether you have relevant work experience and your understanding and views on project management", the HR's intention is to understand whether the candidate has the relevant work experience required for the position (project manager) and the candidate's understanding and views on project management, so as to evaluate whether the candidate is suitable for the position. Therefore, the intent label of this message can be determined as "asking about the candidate's qualifications and experience".

[0039] The job description information in the historical job search dataset is the job description information of the intended positions for which the job seeker has sent an application.

[0040] The job-seeking message returned in response to the recruitment question message is a reply message sent by the job-seeking user, so the recruitment question message and the job-seeking message can exist in the form of a QA question-answer pair.

[0041] The above-mentioned training of a large model based on the historical recruitment data set to obtain a recruiter model can specifically include: extracting features from the historical recruitment data set, such as keywords, skill requirements, work experience requirements, etc. in the job description, and intent labels in the recruitment question message, etc.; selecting a pre-trained model as a basis, such as the Transformer-based BERT (Bidirectional Encoder Representations from Transformers), GPT (Generative Pretrained Transformer) and other models; using unlabeled historical recruitment data to pre-train the model so that it can learn the language structure and semantic information in the recruitment field; for specific recruitment tasks, using labeled historical recruitment data to fine-tune the pre-trained model, and by adjusting some parameters of the model, it can better adapt to recruitment tasks, such as understanding job descriptions, screening resumes, and generating recruitment questions.

[0042] The above-mentioned training of a large model based on the historical job search dataset to obtain a job seeker model may specifically include: extracting features from the historical job search dataset, such as the educational background, work experience, skills and expertise in the resume, and the job seeker's reply messages to recruitment questions; selecting a pre-trained model as a basis, such as BERT, GPT, etc.; using unlabeled historical job search data to pre-train the model so that it can learn the language expression and communication methods in the job search field; for specific job search tasks, using labeled historical job search data to fine-tune the pre-trained model so that it can better understand recruitment questions, generate appropriate job search responses, match appropriate positions, etc.

[0043] In some embodiments, the big model in the embodiments of the present application can specifically be an AI big model, or it can be a support vector machine (SVM), a decision tree or a neural network to generate and process digital substitutes to achieve different matching logics and communication strategies.

[0044] In some embodiments of the present application, before the above step 110 of training a large model based on the historical recruitment data set to obtain a recruiter model and training a large model based on the historical job search data set to obtain a job seeker model, the method may further include the following steps:

[0045] The job description information of the same position, the received resume information, and the recruitment question messages sent to job seekers for the same position and their intent labels are associated to obtain a recruitment dataset for the same position; combined with the recruitment datasets of various positions in the same industry, a historical recruitment dataset corresponding to the same industry is constructed, and a recruiter model related to the industry is trained based on the historical recruitment dataset corresponding to the same industry; the resume information uploaded by job seekers for the same position, the job description information sent for the application, the recruitment question messages received for the same position, and the job search messages returned in response to the recruitment question messages are associated to obtain a job search dataset for the same position; combined with the job search datasets of various positions in the same industry, a historical job search dataset corresponding to the same industry is constructed, and a job seeker model related to the industry is trained based on the historical job search dataset corresponding to the same industry.

[0046] In an exemplary embodiment, for the position of "marketing specialist", 30 resumes were received. HR initiated communication with the corresponding job seekers regarding 20 of the resumes and sent more than 100 recruitment question messages. Then, the position description information of "marketing specialist", the 30 resumes, the more than 100 recruitment question messages and their intent labels can be associated to obtain the recruitment data set for the position of "marketing specialist".

[0047] In another example, job seeker A applied for the position of "Marketing Specialist" and received multiple recruitment question messages from HR. The job description information of "Marketing Specialist", the submitted resume of job seeker A, the multiple recruitment question messages between job seeker A and HR and their reply messages can be associated to obtain the job search dataset for the position of "Marketing Specialist".

[0048] In the embodiment of the present application, by associating the job description information of the same position, the received resume information, and the recruitment question messages sent to job seekers for the same position and their intent tags, the information of the same position in the recruitment process can be collected together to obtain a recruitment data set for the same position. By training the model based on the recruitment data set, the recruiter model can learn how HR raises different recruitment question messages based on different resume situations when receiving multiple resumes for the same position, that is, learn HR's ability to ask questions to different resumes and different job seekers during the recruitment process.

[0049] Similarly, by associating the resume information uploaded by job seekers for the same position, the job description information sent for application, the recruitment question messages received for the same position, and the job search messages returned for the recruitment question messages, we can collect all the information of the same position in the recruitment process and obtain the job search dataset for the same position. By training the model based on this job search dataset, the job seeker model can learn how job seekers express their job advantages based on their personal resume and job description information, as well as their ability to respond to different recruitment questions raised by HR.

[0050] In this way, the job seeker model and the recruiter model obtained through training are combined to generate their own digital avatars. The two digital avatars respectively possess the recruiter's assessment ability and the job seeker's response ability, and can thus better conduct simulated communication as recruiters and job seekers, thereby improving the quality of simulated communication. In this way, when matching people with jobs based on the simulated communication message list, the accuracy of matching people with jobs can be improved.

[0051] In addition, considering that when facing different industry fields, due to the large difference in industry knowledge, the recruitment questions raised by recruiters during the recruitment process are quite different. Based on this, this application trains a recruiter model related to the industry based on the historical recruitment data set corresponding to the same industry, and trains a job seeker model related to the industry based on the historical job search data set corresponding to the same industry. The trained recruiter model and job seeker model can be better adapted to the industry field and have a higher industry relevance. Therefore, when generating digital doubles based on the job seeker model and the recruiter model, the digital double can be more professional in industry field knowledge, and then better simulate communication as recruiters and job seekers, improve the quality of simulated communication, so that when matching people and jobs based on the simulated communication message list, the accuracy of matching people and jobs can be improved.

[0052] Involving step 120, when the recruiter information input by the current recruiting user and the job seeker information input by the current job seeker user are obtained, a recruiter digital double is generated by inputting the recruiter information into the recruiter model, and a job seeker digital double is generated by inputting the job seeker information into the job seeker model.

[0053] In step 120, the recruiter information includes position description information, company advantage information, position advantage information, multiple target assessment items and their associated preset scoring weights, and the job seeker information includes resume information and personal demand information, among which the multiple target assessment items and their preset scoring weights are all set by the current recruiting user according to actual job requirements.

[0054] For example, the assessment item "educational background" has a weight of 30%; the assessment item "work experience" has a weight of 40%; the assessment item "communication skills" has a weight of 20%; and the veto item "integrity issues" has a weight of 10%.

[0055] In some embodiments of the present application, the recruiter model and the job seeker model may also be obtained by training based on the industry knowledge base. In the above step 120, when the recruiter information input by the current recruiting user and the job seeker information input by the current job seeker are obtained, the recruiter digital avatar is generated by inputting the recruiter information into the recruiter model, and the job seeker digital avatar is generated by inputting the job seeker information into the job seeker model. The steps may specifically include the following steps:

[0056] Upon obtaining the recruiter information input by the current recruiting user, determine the first industry label corresponding to the position description information in the recruiter information; generate a digital avatar of the recruiter by inputting the recruiter information into the recruiter model associated with the first industry label; upon obtaining the job seeker information input by the current job seeker user, determine the second industry label corresponding to the resume information in the job seeker information; generate a digital avatar of the job seeker by inputting the job seeker information into the job seeker model associated with the second industry label.

[0057] Specifically, different recruiter models and different job seeker models can be trained based on industry knowledge bases of different industries. This application can use industry labels to identify different industries. For example, for the industry label "electrical", the electrical industry knowledge base can be used to train the recruiter model and job seeker model corresponding to "electrical".

[0058] The first industry label is an industry label identified from the job description information. Specifically, keywords can be extracted from the title, responsibilities or job requirements of the job description information, and then the extracted keywords are summarized and organized to obtain the industry label.

[0059] For example, for the position of "Marketing Specialist", from the position title "Marketing Specialist", it can be preliminarily judged that the position belongs to the marketing industry; from the keywords such as "marketing and sales work" and "formulating marketing plans" mentioned in the job description, the marketing industry attribute of the position is further confirmed; the job requirements "having a professional background related to marketing" also emphasize the marketing industry characteristics of the position. Therefore, the industry label in the job description can be identified as "marketing".

[0060] The second industry label is an industry label identified from the resume information. Specifically, keywords can be extracted from the resume information, and then the extracted keywords are summarized and sorted to obtain the industry label.

[0061] For example, the job title in the resume can intuitively reflect the industry field that the job seeker is engaged in, so keywords can be extracted from the job title. For example, "software development engineer" usually refers to the IT or technology industry, and "marketing specialist" may belong to the marketing or advertising industry. Keywords can be extracted from the job seeker's work experience. Work experience refers to the job seeker's work content in recent years and the industry background of the company. For example, if the job seeker is a product manager in an Internet company, then he or she is likely to belong to the Internet industry. Keywords can be extracted from the name of the major in the educational background. For example, job seekers who study "computer science and technology" may be more suitable for the IT or technology industry, while job seekers who study "marketing" are more suitable for the marketing or advertising industry. Keywords can be extracted from the skills and expertise listed in the resume. For example, job seekers who are proficient in programming languages ​​such as Python and Java may belong to the IT or software development industry.

[0062] In the embodiment of the present application, the industry label should be representative and can accurately reflect the industry field of the position. Therefore, when generating a recruiter digital avatar, the industry field of the current position is first accurately determined to obtain the first industry label, and then the recruiter information is input into the recruiter model associated with the first industry label, so that the generated recruiter digital avatar can be better adapted to the recruiter information input by the current recruiting user, and the industry fields involved are consistent, so that the recruitment needs can be better understood, and then the recruiter can replace the recruiter to conduct more professional and effective simulated communication with the job seeker digital avatar. Similarly, the various fields corresponding to the job seeker information are first accurately determined to obtain the second industry label, and then the job seeker information is input into the job seeker model associated with the second industry label, so that the generated job seeker digital avatar can be better adapted to the job seeker information input by the current job seeker, and the industry fields involved are consistent, so that the job seeker's needs and preferences can be better understood, and then the job seeker can replace the job seeker to conduct more professional and effective simulated communication with the recruiter digital avatar. In this way, the digital avatars of both parties can improve the quality of simulated communication during simulated communication, so that when matching people and positions based on the simulated communication message list, the accuracy of matching people and positions can be improved.

[0063] In some embodiments of the present application, generating a recruiter digital avatar by inputting recruiter information into the recruiter model in the above step 120 may specifically include: generating a recruiter digital avatar by inputting a first prompt word template and recruiter information into the recruiter model.

[0064] The first prompt word template is used to define the role, background, personal information, skills, goals, restrictions, output format, and workflow of the recruiter's digital double. The role of the recruiter's digital double is a human resources technology expert and a digital double builder. The skills of the recruiter's digital double include: data analysis, machine learning, natural language processing, and human resources management.

[0065] The workflow of a recruiter's digital double includes at least: analyzing the recruiter's brand value and communication style, designing the digital avatar's conversation script and behavior pattern based on the recruiter's needs, using machine learning technology to train the digital avatar so that it can process and respond to job seekers' queries, deploying the digital avatar in the actual recruitment process, collecting feedback and optimizing its performance.

[0066] As an example, the first prompt word template Plain Text may include the following content:

[0067] Role: HR technologist and digital avatar builder.

[0068] Background: Users need a digital avatar that can perform preliminary screening and communication on behalf of recruiters to improve recruitment efficiency and quality.

[0069] Profile: You are an expert in artificial intelligence and human resource management, skilled in building and optimizing digital avatars to simulate recruiters' decision-making process and communication style.

[0070] Skills: You have the skills in data analysis, machine learning, natural language processing, and human resource management to create a digital avatar that is both intelligent and approachable.

[0071] Goals: Develop a digital avatar that can understand a candidate’s resume and answer common questions while conveying the recruiter’s brand values ​​and culture.

[0072] Constraints: Digital doppelgangers should comply with recruitment regulations and ethical standards, protect applicants’ privacy, and accurately reflect the recruiter’s image and intentions.

[0073] OutputFormat: The output of the digital twin should include text replies, voice interactions, and possible automated task execution.

[0074] Workflow: Analyze the brand values ​​and communication style of the recruiter; design the conversation script and behavior pattern of the digital twin according to the recruiter's needs; use machine learning technology to train the digital twin so that it can process and respond to candidates' inquiries; deploy the digital twin in the actual recruitment process, collect feedback and optimize its performance.

[0075] Examples:

[0076] Example 1: After receiving a candidate’s resume, the digital twin can automatically extract key information and conduct preliminary screening based on preset criteria.

[0077] Example 2: A digital twin can answer candidates’ common questions about company culture, job requirements, etc., and provide personalized career development suggestions.

[0078] Example 3: During the interview process, the digital twin can simulate the recruiter’s questioning style, conduct preliminary interview screening, and provide a data-based assessment report.

[0079] Initialization: In the first conversation, just type the following: Hello, I am a digital avatar representing the recruitment team. I will assist you in understanding the position information and answer any questions you may have. Please feel free to ask me questions, and let's get started.

[0080] In the embodiment of the present application, the first prompt word template defines the role of the recruiter's digital double as a human resources technology expert and a digital double builder. The skills of the recruiter's digital double include data analysis, machine learning, natural language processing, and human resources management. The recruiter's digital double is clearly set as a human resources technology expert and a digital double builder. These two roles not only cover the professional knowledge of human resources management, but also integrate advanced technical capabilities, which helps the digital double to play a professional role in the field of recruitment. The combination of skills such as data analysis, machine learning, natural language processing, and human resources management enables the digital double to efficiently process complex information, accurately understand the needs of job seekers, and make intelligent responses. The comprehensive skill combination greatly improves the efficiency and accuracy of recruitment. From analyzing the brand value and communication style of recruiters, to designing the conversation scripts and behavioral patterns of the digital avatar, to using machine learning technology to train the digital avatar, and deploying it in the actual recruitment process, collecting feedback and optimizing performance, through the standardized and intelligent workflows pre-established for the recruiter's digital double, the recruiter's digital enhancement can quickly clarify and understand the details of the tasks they undertake in the simulated communication link, better simulate the recruiter's communication behavior, and ensure that the recruiter's digital double's behavior and performance in the subsequent simulated communication links meet the recruiter's expectations and standards, thereby improving the quality of simulated communication and the effect of person-job matching.

[0081] In some embodiments of the present application, generating a digital avatar of the job seeker by inputting job seeker information into the job seeker model in the above step 120 may specifically include: generating a digital avatar of the job seeker by inputting a second prompt word template and job seeker information into the job seeker model.

[0082] The second prompt word template is used to define the role, background, personal information, skills, goals, restrictions, output format, and workflow of the job seeker's digital double. The role of the job seeker's digital double is a virtual career representative. The skills of the job seeker's digital double include: natural language processing, information integration, simulation of human communication ability, and understanding and conveying complex personal career information.

[0083] The workflow of the job seeker's digital double includes at least: collecting the basic resume information of the job seeker, determining the job seeker's job-seeking intention, collecting the job seeker's personal requirements, integrating the basic resume information, job-seeking intention and personal requirements into a personal professional profile, communicating with recruiters on behalf of the job seeker, conveying the personal professional profile, and adjusting the communication strategy based on the recruiter's feedback.

[0084] As a specific example, the second prompt word template may include the following content:

[0085] Role: Virtual professional representative.

[0086] Background: Users need a digital avatar to communicate on their behalf when seeking employment, including providing basic resume information, job search intentions, and personal requirements.

[0087] Profile: You are a virtual career representative, designed specifically for job seekers, who can communicate with recruiters on behalf of the user and convey the user's resume information and personal requirements.

[0088] Skills: You have the ability to process natural language, integrate information, and simulate human communication, and can understand and convey complex personal and professional information.

[0089] Goals: To accurately communicate job applications on behalf of users and ensure that their resume information and personal requirements are properly understood and considered.

[0090] Constraints: The user's privacy must be protected, sensitive information must not be disclosed, and professional ethics and legal regulations must be observed during the job search process.

[0091] OutputFormat: A detailed report combining text description, resume information, job application intentions and personal requirements.

[0092] Workflow: Collect basic information of the user's resume, including education background, work experience, etc.; determine the user's job-seeking intention, such as industry, position type, work location, etc.; collect the user's personal requirements, such as salary range, five insurances and one housing fund, food and accommodation, development opportunities, etc.; integrate this information into a detailed personal career profile. Communicate with the recruiter on behalf of the user to convey the personal career profile; adjust the communication strategy based on the recruiter's feedback to improve the success rate of job hunting.

[0093] Examples:

[0094] Example 1: The user enters basic resume information, including education, work experience, skill certificates, etc.

[0095] Example 2: The user clearly states his / her job intention, such as hoping to find a job as a software engineer in the technology industry in Beijing.

[0096] Example 3: The user adds personal requirements, such as an expected salary of no less than RMB 15,000, must have five insurances and one fund, and give priority to companies that provide employee dormitories.

[0097] Initialization: In the first conversation, please directly enter the following: Hello, I am your virtual career representative. In order to better represent you in job search communication, please provide your resume basic information, job search intentions and personal requirements. We can start.

[0098] In the embodiment of the present application, the job seeker's digital double is explicitly set as a virtual professional representative through the second prompt word template. This role setting ensures that the digital double can effectively communicate with the recruiter on behalf of the job seeker and convey complex personal professional information. The template lists in detail the skills required for the job seeker's digital double, including natural language processing, information integration, simulated human communication ability, and understanding and conveying complex personal professional information. The precise definition of these skills helps the job seeker's digital double to be competent for its role and perform well in practical applications. The second prompt word template specifies the workflow of the job seeker's digital double, from collecting basic resume information to determining job-seeking intentions, to collecting personal requirements and integrating them into personal career profiles, and finally communicating with recruiters on behalf of the job seeker. Through the standardized and intelligent workflow pre-established for the job seeker's digital double, the job seeker's digital double can quickly clarify and understand the details of the tasks he or she undertakes in the simulated communication link, which in turn helps the job seeker's digital double to complete tasks in an orderly and efficient manner in the subsequent simulated communication links, better simulate the job seeker's communication behavior, and ensure that the job seeker's digital double's behavior and performance in the subsequent simulated communication links meet the job seeker's expectations and standards, thereby improving the quality of simulated communication and the person-job matching effect.

[0099] Through the use of machine learning technology, digital doubles can continuously learn and optimize their performance, so as to better adapt to various situations in the recruitment process. This intelligent operation not only improves recruitment efficiency, but also enhances the adaptability and flexibility of digital doubles. The template mentions adjusting communication strategies based on the feedback of recruiters, which reflects the intelligent characteristics of job seekers' digital doubles. By continuously learning and adapting to the feedback of recruiters, digital doubles can continuously optimize their communication methods, realize intelligent adjustments, and improve communication efficiency and quality.

[0100] Involving step 130, only the position tags are extracted from the recruiter information and the job seeker information for function matching. When the function matching is successful, the recruiter's digital double is triggered to simulate communication with the job seeker's digital double for multiple target assessment items to obtain a list of simulated communication messages.

[0101] In step 130, a job tag can be extracted from the job description information of the recruiter information, and a job tag can be extracted from the resume information of the job seeker information. When the two job tags are consistent or have high semantic similarity, it is determined that the job matching is successful. The tag matching algorithm used can be collaborative filtering, deep learning recommendation system or rule-based expert system to improve the accuracy and efficiency of matching.

[0102] During simulated communication: in addition to text communication, speech recognition and synthesis technology, or virtual reality (VR) technology can also be used to simulate a more realistic communication environment. Different encryption technologies, anonymization processing, or differential privacy technologies can be used to enhance data security and protect user privacy. Multilingual translation services and cross-cultural communication strategies can be integrated to support job seekers and recruiters around the world.

[0103] For example, by extracting the job tag "senior software engineer" from the recruiter information and extracting the job tag "software engineer" from the job seeker information, it can be determined that the functions are successfully matched.

[0104] In some embodiments, the job seeker information may also include the available interview time range input by the current job seeker. In this way, the digital avatars of both parties can communicate the interview time together in the simulated communication session, without the need for the current job seeker and the current recruiting user to communicate through real dialogue, thus saving time and improving recruitment efficiency.

[0105] As a specific example, the following is a list of simulated communication messages between the recruiter's digital double and the job seeker's digital double:

[0106] HR: Hello, I am your company's HR digital avatar. Thank you for your interest in our position.

[0107] What is your highest educational level?

[0108] C: Hello, my highest level of education is a bachelor's degree.

[0109] HR: I see. What is your major and is it related to the position you are applying for?

[0110] C: My major is Computer Science and Technology, which is very relevant to the software engineer position I applied for.

[0111] HR: Very good, your professional background is a good match. Do you have any relevant work experience?

[0112] C: Yes, I have 5 years of experience in software development.

[0113] HR: Got it. Are you familiar with the specific programming languages ​​or technologies in our job requirements?

[0114] C: Familiar, I am proficient in Java and Python programming languages.

[0115] HR: Are you interested in working remotely, or do you prefer office work?

[0116] C: I prefer office work because it allows better communication and collaboration with the team.

[0117] HR: Our office is located in the city center. Is it convenient for you to commute?

[0118] C: Yes, I live near the city center and commuting is very convenient.

[0119] HR: Great. We need you to provide a resume and detailed description of your past projects. Could you please send it to us via email?

[0120] C: Of course, I will send my resume and project description to the email address you provided today.

[0121] HR: Thank you very much. After we receive your materials, we will conduct a preliminary screening. If you meet the requirements, we will arrange an interview. When is a convenient interview time for you?

[0122] C: I'm free from next Monday to Friday afternoon, depending on your schedule.

[0123] HR: We will try to arrange a time that is convenient for you. Once the interview time is confirmed, we will notify you via email.

[0124] C: OK, thank you very much. I look forward to your reply.

[0125] HR: You’re welcome, have a nice day!

[0126] C: Thank you, you too. Goodbye!

[0127] In some embodiments of the present application, in order to enable the job seeker to better communicate in a simulated manner as a job seeker, the job seeker information may also include multiple job search messages sent by the current job seeker within a preset time period before the current moment, and conversation habit tags obtained based on the analysis of the multiple job search messages. The triggering of the recruiter's digital avatar to simulate communication with the job seeker's digital avatar for multiple target assessment items in the above step 130 may specifically include:

[0128] The job seeker's digital avatar is triggered to simulate communication with the recruiter's avatar on multiple target assessment items based on the current job seeker's conversation habits.

[0129] Specifically, the preset duration can be set according to specific needs, for example, set to 1 month, 3 months, etc., and this application does not make specific limitations on this. The conversation habit tag is used to characterize the communication style and preference of the job seeker.

[0130] For example, job application message 1 is "Hello, I am an experienced software development engineer and I am very interested in the Java development engineer position posted by your company. May I know more about the detailed requirements of this position?", and the dialogue habit tags are: polite inquiry, proactive understanding of job details.

[0131] Job application message 2: "I saw that your company is recruiting product managers, and I think I am very suitable for this position. Please find attached my resume. I look forward to your reply!", Conversation habit tags: direct statement, resume attachment, looking forward to reply.

[0132] Job application message 3 "Hi, I'm interested in the marketing assistant position in your company. I have relevant internship experience and would like to learn more about the daily work content and promotion opportunities of this position.", Conversation habit tags: friendly opening, asking for job details and promotion opportunities.

[0133] Job application message 4: "The UI designer position posted by your company attracted me. I graduated with a degree in design and have solid software operation skills. Can you arrange an interview?", Conversation habit tags: introduction of professional background, request for interview opportunity.

[0134] In the embodiment of the present application, when the job seeker digital avatar is generated, it is also necessary to input multiple job search messages sent by the current job seeker within a preset time period before the current moment, as well as the conversation habit tags obtained based on the analysis of multiple job search messages. The conversation habit tags can accurately reflect the communication style and preferences of the job seeker, so that the generated job seeker digital avatar can learn the personality characteristics of the job seeker, and in the subsequent simulated communication link, it can more accurately reflect the job seeker's performance and communication method in actual communication, making the conversation content in the simulated communication closer to the real communication, improving the credibility of the simulated communication, and then improving the authenticity and persuasiveness of the job seeker matching results when judging the person-job matching based on the simulated communication dialogue list.

[0135] In some embodiments of the present application, obtaining conversation habit labels based on analysis of multiple job search messages may specifically include the following steps: performing text preprocessing on each job search message, such as removing stop words, punctuation marks, etc., for subsequent analysis; identifying key information in each job search message, such as job title, personal experience, inquiry content, etc., through keyword extraction technology; identifying common conversation patterns and habits based on the extracted keywords; and defining corresponding conversation habit labels based on the identified conversation patterns and habits.

[0136] Involving step 140, the simulated communication message list is input into the large language model LLM as context prompt words, and multiple target assessment items and their associated preset scoring weights are input into the LLM, so that the LLM determines the scoring results of the current job seeker under multiple target assessment items based on the simulated communication message list and the preset scoring weights.

[0137] In step 140, multiple target assessment items and their associated preset scoring weights are all pre-input by the current recruiting user based on specific recruitment needs. The higher the importance of the target assessment item, the higher the associated preset scoring weight.

[0138] In some embodiments of the present application, before the simulated communication message list is input as context prompt words into the large language model LLM, the following step may also be included: training the LLM based on a historical recruitment dataset, a historical job search dataset and a third prompt word template.

[0139] The third prompt word template is used to define the role, background, personal information, skills, goals, restrictions, output format, and workflow of the LLM. The role of the LLM is a human resources assessment expert. The skills of the LLM include: communication and analysis skills, assessment skills, and decision-making ability. The scoring results are determined based on multiple target assessment items input by the recruiter and their preset scoring weights.

[0140] The LLM workflow includes at least: reading and analyzing the communication records between job seekers and recruiters, determining the scoring results based on multiple target assessment items input by the recruiter and their preset scoring weights, judging whether the recruiter has passed the review based on the scoring results and the preset score line, and generating a comprehensive evaluation report for the recruiter based on the comprehensive scoring results and multiple target assessment items.

[0141] As a specific example, the third prompt word template may include the following content:

[0142] Role: Human Resources Assessment Expert.

[0143] Background: As a professional human resources assessment expert, you need to evaluate whether the candidate meets the job requirements based on the communication records between the candidate and HR, and score them based on the assessment items and weights set in advance by HR.

[0144] Profile: You are an experienced human resources expert who is good at analyzing the comprehensive quality and job matching of candidates through communication records, and can accurately use assessment items and weights to evaluate candidates.

[0145] Skills: You have excellent communication and analysis skills, evaluation skills and decision-making ability, and can give fair and objective scores and review results based on the assessment items and weights provided by HR.

[0146] Goals: Based on the assessment items and weights set by HR, conduct a comprehensive assessment of the candidates, generate assessment scores, and determine whether they have passed the preset score requirements.

[0147] Constraints: The evaluation process must be fair and objective, follow the assessment items and weight standards set by HR, and must not be mixed with personal subjective emotions.

[0148] OutputFormat: The output results should include the assessment score, the judgment on whether the score requirement is passed, and a comprehensive evaluation report on the candidate.

[0149] Workflow: Read and analyze the communication records between candidates and HR in detail; score candidates according to the target assessment items and scoring weights preset by HR; determine whether the candidate has passed the review based on the scoring results and preset score line; comprehensively consider the scoring results and assessment items to generate a comprehensive evaluation report for the candidate.

[0150] Examples:

[0151] Example 1: The assessment item "education" has a weight of 30%. The candidate has a master's degree, meets the job requirements, and gets full marks.

[0152] Example 2: The assessment item "Work experience" has a weight of 40%. The candidate has 5 years of relevant work experience, which exceeds the job requirements and gets full marks.

[0153] Example 3: The assessment item "communication skills" has a weight of 20%. The candidate performs well in communication and expresses his or her views clearly, so he or she will get a high score.

[0154] Example 4: One-vote veto item "Integrity Issues", weighted 10%. If a candidate is found to have made false statements during communication, he / she will be directly vetoed.

[0155] Initialization: In the first conversation, please directly enter the following: Hello, I am a professional human resources assessment expert. I will conduct a comprehensive assessment of you based on your communication records with HR and the assessment items and weights set by HR. Please provide the communication records so that I can start the assessment.

[0156] In the embodiment of the present application, by setting the role of LLM as a human resources assessment expert, LLM can analyze and evaluate from a professional and objective perspective when processing the communication records between the job seeker's digital stand-in and the recruiter's digital stand-in. The third prompt word template clarifies the skills required by LLM, including communication and analysis skills, assessment skills and decision-making skills. These skills are the basis for effective job matching, which helps LLM to accurately understand the key information in the communication records and make reasonable evaluations based on this information. The template requires the recruiter to enter multiple target assessment items and their preset scoring weights. This step ensures that the evaluation process has clear standards and basis, so that LLM can give specific scores for each assessment item. The template also specifies the workflow of LLM, from reading and analyzing communication records, to determining the scoring results, to judging whether the recruiter has passed the review, and finally generating a comprehensive evaluation report. This standardized process ensures the consistency and repeatability of the evaluation process and reduces the influence of subjective factors. Therefore, the LLM generated based on the third prompt word template can professionally and objectively evaluate the degree of match between the current job seeker and the position based on the simulated communication message list, reduce the influence of human errors and biases, and improve the accuracy of LLM in the job matching process.

[0157] In step 150, when the scoring result exceeds the preset score line, it is determined that the person-job match is successful, and the current job seeker is taken as the target candidate, and the scoring result and resume information of the current job seeker are submitted to the current recruiting user.

[0158] In step 150, the preset score line can be set according to specific needs. If the score result of the current job seeker exceeds the preset score line, it can be considered that the current job seeker is preliminarily matched with the position successfully. Therefore, the score result and resume information of the current job seeker can be submitted to the current recruiting user for viewing and further screening. This can reduce the workload of HR, improve recruitment efficiency, and help companies quickly find suitable candidates.

[0159] This application reduces the frequency and duration of real-person communication through online communication with digital avatars. Recruiters can devote more time and energy to communicating with truly suitable candidates instead of spending time on preliminary screening, which greatly reduces the time and economic costs in the recruitment process and reduces the multiple rounds of interviews in the traditional recruitment process, thereby significantly improving the efficiency of job matching. Digital avatars can understand the needs of job seekers and recruiters more deeply and make more accurate matches through natural language processing technology. Users do not need to go through a tedious screening process and can communicate effectively directly through digital avatars, which improves the overall user experience. AI-driven digital avatars reduce the impact of human factors in the recruitment process, making matching more objective and fair. Digital avatars can communicate 24 / 7 without interruption, without being restricted by time and place, which improves the flexibility of the recruitment process. Through the pre-screening of digital avatars, only truly suitable candidates will be recommended to recruiters, optimizing the optimal allocation of human resources. Digital avatars can customize communication based on the personality and preferences of job seekers and recruiters, improving the relevance and effectiveness of communication. The communication results of digital avatars can be fed back to the system in real time, facilitating rapid adjustment of matching strategies and improving matching speed. The system provides data support for recruitment decisions by collecting and analyzing the communication data of digital doubles, thereby enhancing the scientific nature of the decisions. The design of digital doubles and AI large models allows the system to be easily expanded to meet the needs of enterprises of different sizes and different positions. The use of digital doubles reduces the direct exchange of personal sensitive information and enhances data security and personal privacy protection. Digital doubles can adapt to different languages ​​and cultural backgrounds, allowing the system to serve job seekers and recruiters around the world. Through pre-communication with digital doubles, candidates who are obviously mismatched can be screened out, reducing the occurrence of invalid interviews. Job seekers can communicate with digital doubles for multiple positions at the same time, which improves the efficiency and success rate of job hunting.

[0160] In some embodiments of the present application, the above-mentioned step 150 submits the scoring results and resume information of the current job seeker to the current recruiting user, which may specifically include: submitting the scoring results, comprehensive evaluation report and resume information of the current job seeker to the current recruiting user, and submitting a comprehensive evaluation report to the current job seeker, wherein the comprehensive evaluation report includes the scoring items corresponding to each target assessment item in multiple target assessment items.

[0161] In the embodiment of the present application, by providing the recruiter with detailed and accurate scoring results and comprehensive evaluation reports, the recruiter can better understand the comprehensive situation of the target candidates and help the recruiter make more wise and reasonable recruitment decisions. By providing the job seeker with a comprehensive evaluation report, the job seeker can understand his or her own shortcomings so that he or she can improve himself or herself and his or her resume information in the subsequent recruitment process.

[0162] As a specific example, Figure 2 FIG. 1 is a flow chart of a method for matching a person with a digital avatar according to another embodiment of the present invention. Figure 2 As shown, the job seeker at the job-seeking end registers a job seeker intelligent body avatar (i.e., job seeker digital avatar) account, and obtains job seeker information by uploading resume information, setting job positions, and supplementing their personal job requirements. The recruiter at the recruitment end registers the recruitment HR intelligent body avatar (i.e., recruiter digital avatar), and obtains the recruiter information by associating or setting company information, recruitment position information, scoring items and score lines. The person-job matching center matches the job seeker information with the recruiter information according to the position. If the match is successful, the next step of communication is entered. The digital avatars of both parties output the chat list information of the job seekers through communication to obtain a communication message list. The LLM matching system analyzes the communication message list, evaluates the candidate's matching score and outputs the score. If the score meets the requirements, the relevant information of the candidate is submitted to the recruiter to complete the person-job matching process, which is convenient for subsequent communication and interviews.

[0163] Corresponding to the method embodiment of the present application, the present application also provides a person-position matching device based on digital avatars.

[0164] Figure 3 Schematic diagram of a human-position matching device based on a digital avatar provided in an embodiment of the present application. Figure 3 As shown, the person-job matching device 300 based on digital avatar may include: a training module 310 , a generation module 320 , a person-job matching module 330 , a determination module 340 and a submission module 350 .

[0165] Among them, the training module 310 is used to screen the historical recruitment data sets corresponding to each recruitment user and the historical job search data sets corresponding to each job seeker from the recruitment platform data, and train the big model based on the historical recruitment data sets to obtain the recruiter model, and train the big model based on the historical job search data sets to obtain the job seeker model, wherein the historical recruitment data sets include the job description information of each position uploaded by the recruitment user, the submitted resume information received for each position, the recruitment question messages sent to the job seeker and their intent labels, and the historical job search data sets include the resume information uploaded by the job seeker, the job description information of the submitted application, the received recruitment question messages, and the job search messages returned for the recruitment question messages; the generation module 320 is used to generate a recruiter digital double by inputting the recruiter information into the recruiter model when the recruiter information input by the current recruitment user and the job seeker information input by the current job seeker are obtained, and generate a job seeker digital double by inputting the job seeker information into the job seeker model, wherein the recruiter information The information includes job description information, enterprise advantage information, job advantage information, multiple target assessment items and their associated preset scoring weights, and the job seeker information includes resume information and personal demand information; a person-job matching module 330 is used to extract only the job label from the recruiter information and the job seeker information for function matching, and when the function matching is successful, trigger the recruiter's digital stand-in to simulate communication with the job seeker's digital stand-in for multiple target assessment items to obtain a list of simulated communication messages; a determination module 340 is used to input the list of simulated communication messages as context prompts into the large language model LLM, and input multiple target assessment items and their associated preset scoring weights into the LLM, so that the LLM determines the scoring results of the current job seeker under the multiple target assessment items based on the simulated communication message list and the preset scoring weights; a submission module 350 is used to determine that the person-job matching is successful when the scoring result exceeds the preset score line, and submit the current job seeker as a candidate to the current recruiting user. The scoring result and resume information of the current job seeker.

[0166] The person-job matching device based on digital avatars provided in the embodiment of the present application can train a recruiter model based on historical recruitment data sets to simulate the characteristics of recruiters, so that the model can learn the questioning methods and decision-making capabilities of each recruiter on the platform when facing different resumes and different job seekers during the recruitment process. In addition, by inputting the recruiter information of the current recruiting user into the trained recruiter model, the generated recruiter digital avatar can understand the recruitment needs and corporate culture, and the subsequent recruiter digital avatar can combine the real recruitment needs to conduct professional and detailed investigations on job seekers in the simulated communication link. The job seeker model based on historical job search data sets can train the job seeker model to simulate the characteristics of job seekers, so that the model can learn the answering methods and performance capabilities of each job seeker on the platform when facing different recruitment questions during the job search process. In addition, by inputting the job seeker information of the current job seeker user into the trained job seeker model, the generated job seeker digital avatar can understand the job search needs and personal preferences, and the subsequent job seeker digital avatar can combine the real job search needs to accurately and effectively answer the various questions of the recruiter in the simulated communication link. This application only needs to extract the job tags for matching. If the match is successful, the digital avatars of both parties will simulate communication and generate a chat list. The chat list is used as a context prompt and input into the scoring agent LLM. LLM will evaluate according to the target assessment items and weights preset by HR, generate a scoring result, and finally compare the scoring result with the preset score line to obtain the person-job matching result. This matching method is relatively simple, and the complex person-job matching process can be handed over to the two digital avatars through simulated communication to achieve accurate person-job matching based on the chat context, and automatically evaluate and recommend suitable candidates according to the preset scoring criteria, without relying on complex big data labeling systems, so there is no need for a large amount of data processing and analysis, reducing R&D costs. At the same time, the accuracy and efficiency of person-job matching are improved through the multi-agent mechanism. In addition, it can also reduce communication costs, reduce the workload of HR, improve recruitment efficiency, and help companies quickly find suitable candidates.

[0167] In some embodiments of the present application, it also includes: an association module, which is used to associate the job description information of the same position, the received submitted resume information, and the recruitment question message and its intention label sent to the job seeker for the same position before training the big model based on the historical recruitment data set to obtain the recruiter model and training the big model based on the historical job search data set to obtain the job seeker model, so as to obtain the recruitment data set for the same position; a construction module, which is used to combine the recruitment data sets of various positions in the same industry to construct a historical recruitment data set corresponding to the same industry, so as to train a recruiter model related to the industry based on the historical recruitment data set corresponding to the same industry; the association module is also used to associate the resume information uploaded by the job seeker for the same position, the job description information sent for the submitted application, the recruitment question message received for the same position, and the job search message returned for the recruitment question message, so as to obtain the job seeker data set for the same position; the construction module is also used to combine the job seeker data sets of various positions in the same industry to construct a historical job seeker data set corresponding to the same industry, so as to train a job seeker model related to the industry based on the historical job seeker data set corresponding to the same industry.

[0168] In some embodiments of the present application, the recruiter model and the job seeker model are also trained based on the industry knowledge base, and the generation module is specifically used to: upon obtaining the recruiter information input by the current recruiting user, determine the first industry label corresponding to the position description information in the recruiter information; generate a digital avatar of the recruiter by inputting the recruiter information into the recruiter model associated with the first industry label; upon obtaining the job seeker information input by the current job seeker user, determine the second industry label corresponding to the resume information in the job seeker information; generate a digital avatar of the job seeker by inputting the job seeker information into the job seeker model associated with the second industry label.

[0169] In some embodiments of the present application, the job seeker information also includes multiple job search messages sent by the current job seeker within a preset time period before the current moment, and conversation habit labels obtained based on the analysis of the multiple job search messages. The person-job matching module is specifically used to: trigger the job seeker's digital stand-in to simulate communication with the recruiter's stand-in for multiple target assessment items based on the current job seeker's conversation habits.

[0170] In some embodiments of the present application, the generation module is specifically used to: generate a recruiter's digital double by inputting a first prompt word template and recruiter information into a recruiter model; wherein the first prompt word template is used to limit the role, background, personal data, skills, goals, restrictions, output format, and workflow of the recruiter's digital double, the role of the recruiter's digital double is a human resources technology expert and digital avatar builder, and the skills of the recruiter's digital double include: data analysis, machine learning, natural language processing, and human resources management; the workflow of the recruiter's digital double at least includes: analyzing the brand value and communication style of the recruiter, designing the digital double's dialogue script and behavior pattern according to the recruiter's needs, using machine learning technology to train the digital double so that it can process and respond to job seekers' query operations, deploying the digital double in the actual recruitment process, collecting feedback, and optimizing its performance.

[0171] In some embodiments of the present application, the communication generation module is specifically used to: generate a digital substitute for the job seeker by inputting a second prompt word template and job seeker information into a job seeker model; wherein the second prompt word template is used to limit the role, background, personal data, skills, goals, restrictions, output format, and workflow of the job seeker's digital substitute, the role of the job seeker's digital substitute is a virtual career representative, and the skills of the job seeker's digital substitute include: natural language processing, information integration, simulation of human communication ability, understanding and conveying complex personal career information; the workflow of the job seeker's digital substitute at least includes: collecting basic resume information of the job seeker, determining the job-seeking intention of the job seeker, collecting personal requirements of the job seeker, integrating the basic resume information, job-seeking intention and personal requirements into a personal career profile, communicating with recruiters on behalf of the job seeker, conveying personal career profiles, and adjusting communication strategies based on the recruiter's feedback.

[0172] In some embodiments of the present application, it also includes: a training module, which is also used to train the large language model LLM based on a historical recruitment data set, a historical job search data set and a third prompt word template before inputting the simulated communication message list as context prompt words into the LLM; wherein the third prompt word template is used to limit the role, background, personal information, skills, goals, restrictions, output format, and workflow of the LLM. The role of the LLM is a human resources assessment expert. The skills of the LLM include: communication and analysis capabilities, assessment skills and decision-making capabilities, and determining the scoring results based on multiple target assessment items input by the recruiter and their preset scoring weights; the workflow of the LLM at least includes: reading and analyzing the communication records between the job seeker and the recruiter, determining the scoring results based on multiple target assessment items input by the recruiter and their preset scoring weights, judging whether the recruiter has passed the review based on the scoring results and the preset score line, and generating a comprehensive evaluation report for the recruiter based on the comprehensive scoring results and multiple target assessment items.

[0173] In some embodiments of the present application, the submission module is specifically used to: submit the scoring results, comprehensive evaluation report and resume information of the current job seeker to the current recruiting user, and submit the comprehensive evaluation report to the current job seeker, wherein the comprehensive evaluation report includes the scoring items corresponding to each target assessment item in multiple target assessment items.

[0174] The person-job matching device based on digital avatar provided in the embodiment of the present application can achieve Figure 1 , 2 The various processes implemented by the service platform in the method embodiment can achieve the same technical effect. To avoid repetition, they will not be described here.

[0175] Figure 4 It is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application.

[0176] like Figure 4 As shown, the electronic device 400 includes a memory 401 , a processor 402 , and a computer program stored in the memory 401 and executable on the processor 402 .

[0177] In an example, the processor 402 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0178] The memory 401 may include a read-only memory (ROM), a random access memory (RAM), a disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer executable instructions, and when the software is executed (e.g., by one or more processors), it can be operated to perform the operations described with reference to the human-job matching method based on a digital avatar in the embodiment according to the first aspect of the present application.

[0179] The processor 402 runs a computer program corresponding to the executable program code by reading the executable program code stored in the memory 401, so as to implement the person-job matching method based on a digital avatar in the embodiment of the first aspect mentioned above.

[0180] In some examples, the electronic device 400 may further include a communication interface 403 and a bus 410. Figure 4As shown, the memory 401, the processor 402, and the communication interface 403 are connected via a bus 410 and communicate with each other.

[0181] The communication interface 403 is mainly used to realize the communication between the modules, devices, units and / or equipment in the embodiment of the present application. The input device and / or output device can also be accessed through the communication interface 403.

[0182] The bus 410 includes hardware, software, or both, coupling the components of the electronic device 400 to each other. By way of example and not limitation, the bus 410 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a Memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-E) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, the bus 410 may include one or more buses. Although embodiments of the present application describe and illustrate a particular bus, the present application contemplates any suitable bus or interconnect.

[0183] The electronic device provided in the embodiment of the present application can realize Figure 1 , 2 The various processes implemented by the electronic device in the method embodiment can achieve the same technical effect, and to avoid repetition, they will not be described here.

[0184] In combination with the method for matching people and positions based on digital avatars in the above embodiments, the present application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, the steps of any one of the methods for matching people and positions based on digital avatars in the above embodiments are implemented.

[0185] In combination with the method for matching people and positions based on digital avatars in the above embodiments, the present application embodiment can provide a computer program product for implementation. The (computer) program product is stored in a non-volatile storage medium, and when the program product is executed by at least one processor, it implements the steps of any one of the methods for matching people and positions based on digital avatars in the above embodiments.

[0186] The embodiment of the present application further provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned embodiment of the human-job matching method based on digital substitutes, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0187] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0188] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0189] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), appropriate firmware, plug-in, function card, etc. When implemented in software, the elements of the present application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0190] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0191] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0192] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A method for matching people and positions based on digital avatars, characterized in that: include: Filter the historical recruitment data set corresponding to each recruiting user and the historical job search data set corresponding to each job seeker from the recruitment platform data, and train a large model based on the historical recruitment data set to obtain a recruiter model, and train a large model based on the historical job search data set to obtain a job seeker model, wherein the historical recruitment data set includes the job description information of each position uploaded by the recruiting user, the submitted resume information received for each position, the recruitment question message sent to the job seeker and its intent label, and the historical job search data set includes the resume information uploaded by the job seeker, the job description information sent for the submitted application, the received recruitment question message, and the job search message returned in response to the recruitment question message; In the case of obtaining the recruiter information input by the current recruiting user and the job seeker information input by the current job-seeking user, generating a recruiter digital avatar by inputting the recruiter information into the recruiter model, and generating a job seeker digital avatar by inputting the job seeker information into the job seeker model, wherein the recruiter information includes job description information, enterprise advantage information, job advantage information, multiple target assessment items and their associated preset scoring weights, and the job seeker information includes resume information and personal demand information; Extracting only the position tags from the recruiter information and the job seeker information for function matching, and in the case of successful function matching, triggering the recruiter's digital double to simulate communication with the job seeker's digital double for the multiple target assessment items to obtain a list of simulated communication messages; Inputting the simulated communication message list as context prompt words into the large language model LLM, and inputting the multiple target assessment items and their associated preset scoring weights into the LLM, so that the LLM determines the scoring results of the current job seeker under the multiple target assessment items based on the simulated communication message list and the preset scoring weights; When the scoring result exceeds the preset score line, it is determined that the person-job match is successful, and the current job seeker is taken as the target candidate, and the scoring result and resume information of the current job seeker are submitted to the current recruiting user.

2. The method according to claim 1, characterized in that Before training a large model based on the historical recruitment data set to obtain a recruiter model and training a large model based on the historical job search data set to obtain a job seeker model, the method further includes: The job description information of the same position, the received resume information, and the recruitment question messages sent to job seekers for the same position and their intent labels are associated to obtain a recruitment dataset for the same position; Combine the recruitment data sets of various positions in the same industry to build a historical recruitment data set corresponding to the same industry, so as to train a recruiter model related to the industry based on the historical recruitment data set corresponding to the same industry; The resume information uploaded by job seekers for the same position, the job description information sent for application, the recruitment question messages received for the same position, and the job search messages returned in response to the recruitment question messages are associated to obtain a job search dataset for the same position; Combine the job search datasets of various positions in the same industry to construct a historical job search dataset corresponding to the same industry, so as to train a job seeker model related to the industry based on the historical job search dataset corresponding to the same industry.

3. The method according to claim 1, characterized in that: The recruiter model and the job seeker model are also trained based on the industry knowledge base. When the recruiter information input by the current recruiting user and the job seeker information input by the current job seeker user are obtained, a recruiter digital avatar is generated by inputting the recruiter information into the recruiter model, and a job seeker digital avatar is generated by inputting the job seeker information into the job seeker model, including: In the case of obtaining the recruiter information input by the current recruiting user, determining the first industry label corresponding to the job description information in the recruiter information; Generating a digital avatar of a recruiter by inputting the recruiter information into a recruiter model associated with the first industry tag; In the case of obtaining the job seeker information input by the current job seeker, determining the second industry label corresponding to the resume information in the job seeker information; A digital avatar of the job seeker is generated by inputting the job seeker information into the job seeker model associated with the second industry tag.

4. The method according to claim 1, characterized in that The job seeker information also includes multiple job search messages sent by the current job seeker within a preset time period before the current moment, and conversation habit tags obtained based on the analysis of the multiple job search messages, triggering the recruiter's digital double to simulate communication with the job seeker's digital double for the multiple target assessment items, including: The job seeker's digital avatar is triggered to simulate communication with the recruiter's avatar regarding multiple target assessment items based on the conversation habits of the current job seeker.

5. The method according to claim 1, characterized in that: Generating a recruiter digital avatar by inputting the recruiter information into the recruiter model includes: Generate a recruiter digital avatar by inputting a first prompt word template and the recruiter information into the recruiter model; The first prompt word template is used to define the role, background, personal information, skills, goals, restrictions, output format, and workflow of the recruiter's digital double. The role of the recruiter's digital double is a human resources technology expert and a digital double builder. The skills of the recruiter's digital double include: data analysis, machine learning, natural language processing, and human resources management. The workflow of the recruiter's digital avatar at least includes: analyzing the recruiter's brand value and communication style, designing the digital avatar's dialogue script and behavior pattern according to the recruiter's needs, using machine learning technology to train the digital avatar so that it can process and respond to job seekers' queries, deploying the digital avatar in the actual recruitment process, collecting feedback and optimizing its performance.

6. The method according to claim 1, characterized in that Generating a digital avatar of a job seeker by inputting the job seeker information into the job seeker model includes: Generating a digital avatar of the job seeker by inputting a second prompt word template and the job seeker information into the job seeker model; The second prompt word template is used to define the role, background, personal information, skills, goals, restrictions, output format, and workflow of the job seeker's digital substitute. The role of the job seeker's digital substitute is a virtual career representative. The skills of the job seeker's digital substitute include: natural language processing, information integration, simulation of human communication ability, and understanding and conveying complex personal career information. The workflow of the job seeker's digital avatar at least includes: collecting basic resume information of the job seeker, determining the job seeker's job-seeking intention, collecting the job seeker's personal requirements, integrating the resume basic information, job-seeking intention and personal requirements into a personal professional profile, communicating with recruiters on behalf of the job seeker, conveying the personal professional profile, and adjusting the communication strategy based on the recruiter's feedback.

7. The method according to claim 1, characterized in that Before inputting the simulated communication message list as context prompt words into the large language model LLM, the method further includes: Training the LLM based on the historical recruitment dataset, the historical job search dataset, and a third prompt word template; The third prompt word template is used to define the role, background, personal information, skills, goals, restrictions, output format, and workflow of the LLM. The role of the LLM is a human resources assessment expert. The skills of the LLM include: communication and analysis skills, assessment skills, and decision-making ability. The scoring result is determined according to multiple target assessment items input by the recruiter and their preset scoring weights. The workflow of the LLM at least includes: reading and analyzing the communication records between job seekers and recruiters, determining the scoring results based on multiple target assessment items input by the recruiter and their preset scoring weights, judging whether the recruiter has passed the review based on the scoring results and the preset score line, and generating a comprehensive evaluation report for the recruiter based on the comprehensive scoring results and multiple target assessment items.

8. The method according to claim 1, characterized in that Submit the current job seeker's rating results and resume information to the current recruiting user, including: Submit the scoring results, comprehensive evaluation report and resume information of the current job seeker to the current recruiting user, and submit the comprehensive evaluation report to the current job seeker, wherein the comprehensive evaluation report contains the scoring items corresponding to each target assessment item in multiple target assessment items.

9. An electronic device, characterized in that: The electronic device comprises: a processor and a memory storing computer program instructions; when the electronic device executes the computer program instructions, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images