Intelligent personnel selection method and system based on large language model and semantic matching

Abstract

The invention belongs to the field of artificial intelligence, and particularly relates to an intelligent personnel selection method and system based on a large language model and semantic matching, and the method comprises the steps: inputting the data of selected personnel into a large language model through an intelligent matching degree scoring method and an intelligent combination PK method, carrying out the scoring or comparison based on a semantic matching mechanism, and outputting a list of to-be-selected personnel, and decision makers are assisted to determine final winning persons. According to the personnel selection method and system provided by the invention, the problems of low efficiency, high subjectivity and the like of traditional selection can be effectively solved, and the scientificity, accuracy and efficiency of personnel selection are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically to an intelligent personnel selection method and system based on a large language model and semantic matching. Background Art

[0002] In today's rapidly evolving digital and intelligent world, traditional recruitment methods face numerous challenges. Manual processes like screening resumes and conducting interviews are not only time-consuming and labor-intensive, but also susceptible to subjective influences, making it difficult to guarantee the fairness and accuracy of selection results. As companies expand and competition for talent intensifies, how to quickly and accurately select the right talent for a position has become a key challenge in human resources management.

[0003] Large language models, with their powerful natural language processing and knowledge understanding capabilities, as well as their semantic matching mechanisms for precise information analysis, offer a new technical approach for personnel selection. This paper proposes an intelligent personnel selection method and system based on large language models and semantic matching, aiming to overcome the limitations of traditional selection models, improve the efficiency and quality of personnel selection, and provide a more scientific and intelligent solution for talent selection in enterprises and organizations. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent personnel selection method and system based on a large language model and semantic matching. Through intelligent matching scoring and intelligent combined PK methods, candidate information is input into the large language model, scored or compared based on a semantic matching mechanism, and a list of candidates is output to assist decision-makers in determining the final winner. This system and method effectively address the low efficiency and high subjectivity of traditional selection processes, significantly improving the scientific nature, accuracy, and efficiency of personnel selection.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] In the first aspect, the present invention provides an intelligent personnel selection method based on a large language model and semantic matching, including an intelligent matching degree scoring method and an intelligent combined PK method.

[0007] The intelligent matching scoring method includes the following steps:

[0008] S1. Data preprocessing:

[0009] Collect comprehensive data on all candidates, including basic attribute information, professional resume information, and comprehensive evaluation information. Use data cleaning technology to eliminate invalid data, duplicate data, and noise data. Use standardized data processing procedures and unify data format specifications.

[0010] S2. Build a scoring model:

[0011] Based on the responsibilities, skill requirements, and professional competency standards of specific positions, we construct refined semantic matching rules and a quantitative scoring indicator system. Through deep knowledge embedding and model training optimization, we deeply integrate these rules and standards into the large language model, enabling it to accurately understand the semantic connotations and logical relationships of the selection requirements. Based on this, we conduct in-depth semantic analysis and feature extraction of candidate information.

[0012] S3. Matching score:

[0013] The candidate data pre-processed in step S1 is input into the large language model one by one according to the established process. Based on the construction of a complete semantic matching mechanism, the large language model conducts a comprehensive and in-depth quantitative analysis and accurate scoring of the degree of match between each candidate and the selection requirements from multiple key dimensions such as professional skills adaptability, work experience relevance, professional quality fit, and corporate culture integration.

[0014] S4. Generate a candidate list:

[0015] After all candidates have been scored, the system systematically sorts them from highest to lowest match scores using algorithms like intelligent bubble sorting, automatically generating a structured candidate list. Based on this list and the detailed scoring criteria provided by the large language model, decision-makers conduct in-depth evaluations and secondary screening of the candidates, fully leveraging their professional judgment to independently determine the final winner.

[0016] The intelligent combination PK method includes the following steps:

[0017] S1. Data preprocessing:

[0018] Collect comprehensive data on all candidates, including basic attribute information and comprehensive evaluation information. Use data cleaning technology to eliminate invalid, duplicate, and noisy data. Use standardized data processing procedures and unify data format specifications.

[0019] S2. Generate duel matrix:

[0020] After obtaining complete information on all candidates, the system uses combinatorial mathematics algorithms to automatically generate comprehensive and complete two-person PK pairs, ensuring that each candidate can form a comparative pair with all other candidates. Through comprehensive two-on-two duels, the system fully demonstrates the strengths and weaknesses of each candidate in different comparison scenarios.

[0021] S3. PK evaluation:

[0022] The complete basic information and resumes of the two candidates in each PK pair are accurately input into the large language model. Based on the semantic matching mechanism of the selection requirements, the large language model conducts in-depth comparative analysis and intelligent judgment on the strengths and weaknesses of the two candidates in terms of job compatibility from multiple key dimensions.

[0023] S4, statistics and bubble sort:

[0024] After completing the evaluation of all PK combinations, data statistics and analysis algorithms are used to automatically count the number of wins for each candidate. Based on the number of wins, bubble sort and other rules are used to systematically sort the candidates from most to least, generating a list of candidates that intuitively reflects the relative advantages of the candidates. Decision-makers conduct a comprehensive and integrated consideration of the candidates based on the ranking of the number of wins and the detailed winning basis provided by the large language model, and independently decide on the final winner based on their professional experience and judgment.

[0025] Furthermore, the basic attribute information includes but is not limited to age, education level, and major; the professional resume information includes but is not limited to job history, work performance, project experience, and skill certificates; the comprehensive evaluation information includes but is not limited to leadership evaluation, colleague evaluation, and social evaluation.

[0026] In a second aspect, the present invention provides an intelligent personnel selection system based on a large language model and semantic matching, comprising:

[0027] Data management module: supports batch import, editing, and deletion of candidate information and job requirement information, provides data retrieval and screening functions; and also has a data backup and recovery mechanism;

[0028] Model training and configuration module: supports users to adjust parameters and configure semantic matching rules for large language models based on selection scenarios and job requirements; supports model training, optimization, and updating;

[0029] Intelligent Scoring Module: Implements intelligent matching scoring method, automatically completes data preprocessing, scoring model construction, matching scoring, and candidate list generation; also allows users to view scoring progress and results in real time and obtain detailed scoring basis;

[0030] Intelligent PK module: Executes intelligent combination PK method, automatically generates candidate combinations, conducts PK evaluation and statistical ranking, outputs a list of candidates and the basis for winning; supports users to re-evaluate and compare specific combinations;

[0031] Result display and export module: displays candidate information, scoring results and winning criteria in the form of visual charts and lists; at the same time, supports the export of selection results.

[0032] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned intelligent personnel selection method when executing the computer program.

[0033] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned intelligent personnel selection method is implemented.

[0034] The descriptions of the second to fourth aspects of the present invention can refer to the detailed description of the first aspect; and the beneficial effects of the descriptions of the second to fourth aspects can refer to the analysis of the beneficial effects of the first aspect, which will not be repeated here.

[0035] The beneficial effects of the present invention are as follows:

[0036] (1) Deeply integrate the large language model with the semantic matching mechanism and apply it to the field of personnel selection, breaking the traditional model of relying on manual experience and simple screening tools. By utilizing the powerful learning and reasoning capabilities of the large language model and combining it with the semantic matching mechanism to accurately analyze job requirements and candidate information, the intelligent upgrade of personnel selection can be achieved;

[0037] (2) Two innovative selection models are proposed: the intelligent matching scoring method and the intelligent combination PK method. The intelligent matching scoring method can independently evaluate each candidate and provide a quantitative matching score and detailed basis; the intelligent combination PK method compares candidates pairwise to intuitively show the differences in advantages between candidates. The two methods complement each other, providing decision makers with a multi-dimensional selection perspective and effectively improving the reliability and rationality of the selection results;

[0038] (3) The list of candidates, scoring, and winning criteria generated by the system provide objective and comprehensive reference information for decision makers. While reducing human decision-making bias, it retains the decision makers' independent decision-making power, realizes the organic combination of technical assistance and human decision-making, and improves the scientific nature and flexibility of personnel selection decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flowchart of an intelligent personnel selection method based on a large language model and semantic matching provided by an embodiment of the present disclosure;

[0040] Figure 2 A schematic diagram of the flow of the intelligent matching scoring method provided in Example 1 of the present disclosure;

[0041] Figure 3 This is a flow chart of the intelligent combined PK method provided in Example 2 of the present disclosure;

[0042] Figure 4 Schematic diagram of an electronic device provided in Example 4 of the present disclosure. DETAILED DESCRIPTION

[0043] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.

[0044] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The embodiments cannot be described one by one here, but the embodiments of the present invention are not limited to the following embodiments.

[0045] Example 1: An intelligent personnel selection method based on a large language model and semantic matching. This example adopts an intelligent matching degree scoring method. Figure 1 、 2 As shown, the following steps are included:

[0046] S1. Data preprocessing:

[0047] Through multiple channels, we collect full-dimensional information of all candidates, covering basic attribute information, including but not limited to age, education, major, etc.; professional resume information, such as job history, work performance, project experience, skill certificates, etc.; and comprehensive evaluation information, such as leadership evaluation, colleague evaluation, social evaluation, etc. Use data cleaning technology to eliminate invalid data, duplicate data and noise data, and adopt standardized data processing procedures and unified data format specifications to ensure that the data input into the large language model has high accuracy, completeness and consistency, laying a solid data foundation for subsequent selection analysis. The specific steps are as follows: Using a systematic data organization methodology, conduct in-depth structured processing of the basic data of all candidates, build a data model that meets industry standards, and form the following data system with standardized evaluation dimensions, as shown in the following table:

[0048] Table 1 Examples of candidate evaluation dimensions

[0049]

[0050]

[0051] S2. Build a scoring model:

[0052] Based on the specific job responsibilities, skill requirements, and professional quality standards, we build refined semantic matching rules and quantitative scoring indicator systems. Through deep knowledge embedding and model training optimization, we deeply integrate these rules and standards into the large language model, enabling it to accurately understand the semantic connotations and logical relationships of the selection requirements, and based on this, conduct in-depth semantic analysis and feature extraction of candidate information. The specific steps are as follows:

[0053] Pre-develop and design a match scoring agent. This agent possesses intelligent interaction and in-depth analysis capabilities. It receives comprehensive information about the candidate to be scored, along with personalized semantic requirements. After intelligent analysis, it outputs an accurate match score for the candidate and detailed, multi-dimensional scoring criteria. Furthermore, professional match scoring prompts must be carefully designed in advance based on the basic evaluation dimensions established in step S1. Through multiple rounds of training and debugging, the agent's scoring accuracy and reliability must be optimized.

[0054] S3. Matching score:

[0055] The pre-processed candidate information is input into the big language model one by one according to the established process. The big language model is based on the construction of a complete semantic matching mechanism, and conducts a comprehensive and in-depth quantitative analysis and precise scoring of the degree of match between each candidate and the selection requirements from multiple key dimensions such as professional skills adaptability, work experience relevance, professional quality fit, and corporate culture integration. During the scoring process, the big language model carefully considers and evaluates each dimension, and outputs detailed scoring details, and finally integrates and generates a scientific and reasonable comprehensive matching score. At the same time, the big language model will also output a detailed scoring basis report, which systematically explains the specific performance, advantages and disadvantages of each candidate in each indicator, and the complete reasoning process of the scoring, providing sufficient evidence support for the selection decision. The specific steps are as follows:

[0056] The information and personalized semantic requirements of the five candidates A, B, C, D, and E compiled and refined in S1 are sequentially input into the match scoring agent. Relying on the powerful natural language processing and intelligent analysis capabilities of the Large Language Model (LLM), the agent outputs an accurate match score and detailed scoring basis, as shown in the following table:

[0057] Table 2 Example of matching score agent scoring results

[0058] Candidates Match score Rating basis A 85 …… B 90 …… C 72 …… D 68 …… E 53 ……

[0059] S4. Generate a candidate list:

[0060] After all candidates have been scored, the system systematically ranks them from highest to lowest match score using algorithms like intelligent bubble sorting, automatically generating a structured candidate list. Decision-makers can use this list, combined with the detailed scoring criteria provided by the large language model, to conduct in-depth candidate evaluations and secondary screening, fully leveraging their professional judgment to independently determine the winning candidate. This approach, through a comprehensive and independent assessment of each candidate, provides a quantitative and intuitive representation of the candidate's fit with the position, providing a clear, objective, and scientific basis for selection decisions.

[0061] According to the scoring results generated in Table 2, professional bubble sorting and other algorithms are used to rank them, as shown in the following table:

[0062] Table 3 Example of candidate matching score ranking results

[0063] Candidates Match score Ranking B 90 1 A 85 2 C 72 3 D 68 4 E 53 5

[0064] Example 2: An intelligent personnel selection method based on a large language model and semantic matching. This example adopts an intelligent combined PK method, referring to Figure 1 、 3 As shown, the following steps are included:

[0065] S1, data preprocessing: the same as step S1 in Example 1, which will not be repeated here;

[0066] S2. Generate duel matrix:

[0067] After obtaining complete information on all candidates, the system uses combinatorial mathematics algorithms to automatically generate comprehensive and complete two-person PK combinations, ensuring that each candidate can form a comparative combination with all other candidates. Through comprehensive two-on-two duels, the system fully demonstrates the strengths and weaknesses of each candidate in different comparison scenarios. The specific steps are as follows:

[0068] Input the detailed digital information data set of the five candidates A, B, C, D, and E into the system. The system relies on the self-developed intelligent combination generation engine and the advanced combination optimization algorithm to automatically generate all the PK combinations of two people (through the combination of mathematical formulas). Calculate, n is the number of candidates, here Group), forming a structured PK combination matrix, as shown in the following table:

[0069] Table 4 Examples of all PK combinations

[0070] A B C D E A — A vs B A vs C A vs D A vs E B — — B vs C B vs D B vs E C — — — C vs D C vs E D — — — — D vs E E — — — — —

[0071] S3. PK evaluation:

[0072] The full basic information and complete resume information of the two people in each PK combination are accurately input into the big language model. Based on a carefully constructed semantic matching mechanism for selection requirements, the big language model conducts in-depth comparative analysis and intelligent judgment on the pros and cons of the two people in terms of job matching from multiple key dimensions such as professional skill level, project experience richness, problem-solving ability, and job requirement fit. During the evaluation process, the big language model carefully compares the performance differences between the two people in various key indicators, deeply analyzes their fit with job requirements, and accurately determines the person with a better match. At the same time, the big language model will also generate a detailed report on the basis of victory, comprehensively explaining the logical reasoning process and key factors for making judgments. The specific steps are as follows:

[0073] (1) Pre-develop and design a PK evaluation agent. This agent has efficient information processing and intelligent decision-making capabilities. It can receive detailed information and personalized semantic requirements of the two PK players. After in-depth analysis, it outputs the winning and losing results of the PK between the two players and detailed evaluation basis. At the same time, it is necessary to carry out scientific design and rigorous training in advance for the following basic dimensions to achieve accurate comparative evaluation, as shown in the following table:

[0074] Table 5 Examples of evaluation dimensions for PK evaluation agents

[0075]

[0076]

[0077] (2) All combinations and personalized semantic requirements generated in step (1) are input into the PK evaluation agent in sequence. The PK evaluation agent relies on the powerful intelligent analysis capabilities of the large language model (LLM) to evaluate the winners of each group and output detailed winning criteria, as shown in the following table:

[0078] Table 6 Example of PK evaluation agent evaluation results

[0079] PK combination Winner Winning basis A vs B A …… A vs C A …… A vs D A …… A vs E A …… B vs C B …… B vs D B …… B vs E B …… C vs D C …… C vs E C …… D vs E D ……

[0080] S4, statistics and bubble sort:

[0081] After completing the evaluation of all PK combinations, the system uses efficient data statistics and analysis algorithms to automatically count the number of wins for each candidate. Based on the number of wins, scientific bubble sorting and other rules are used to systematically sort the candidates from most to least, generating a list of candidates that intuitively reflects the relative advantages of the candidates. Decision-makers can make a comprehensive and integrated consideration of the candidates based on the ranking of the number of wins and the detailed winning basis provided by the large language model, and independently decide on the final winner based on their professional experience and judgment. The intelligent combination PK method can directly compare the candidates to directly and clearly highlight the advantages and characteristics of the candidates, providing a highly targeted and valuable reference basis for selection decisions. The specific operation steps are as follows:

[0082] According to the PK results generated in Table 6, professional data statistics and bubble sorting algorithms are used to perform statistics and ranking, as shown in the following table:

[0083] Table 7 Example of statistical ranking of candidate PK results

[0084] Candidates Number of battles Number of wins Win rate Ranking A 4 4 100％ 1 B 4 3 75％ 2 C 4 2 50％ 3 D 4 1 25％ 4 E 4 0 0％ 5

[0085] Example 3 This example provides an intelligent personnel selection system based on a large language model and semantic matching, including:

[0086] (1) System architecture design

[0087] The intelligent personnel selection system based on a large language model and semantic matching mechanism adopts a layered architecture design, which is divided into data layer, algorithm layer, function layer and application layer from bottom to top. Each layer is both independent and collaborative to ensure the efficient operation and stable expansion of the system.

[0088] The data layer, serving as the foundation of the system, is responsible for collecting, storing, and managing candidate data and job requirements. This data includes structured data such as basic information, work history, and skill certificates, as well as unstructured data such as project reports and personal statements. Job requirements data includes job responsibilities, job requirements, and compatibility with the corporate culture. The data layer also provides data cleaning, standardization, and update maintenance to ensure the accuracy and timeliness of input data.

[0089] Algorithm layer: This core layer integrates a large language model and a semantic matching algorithm. The large language model has been pre-trained and fine-tuned for personnel selection scenarios, enabling accurate understanding and processing of natural language. The semantic matching algorithm constructs a semantic vector space based on job requirements and candidate information, and uses methods such as cosine similarity and semantic distance calculation to achieve precise semantic matching analysis. Furthermore, the algorithm layer includes a scoring model and a PK evaluation model, providing technical support for the selection process.

[0090] Functional layer: Implements two core functions: intelligent matching scoring and intelligent combination PK. It also provides auxiliary functions such as data management, model configuration, and result export, facilitating user-friendly system settings and data processing.

[0091] Application layer: This user-friendly interface provides a simple and easy-to-use operating platform for human resources managers and decision makers. Users can upload candidate information, set selection criteria, and view the candidate list and scoring criteria through this interface, achieving efficient selection and decision-making through human-computer interaction.

[0092] (2) System Function Module

[0093] 1. Data Management Module: This module supports batch importing, editing, and deleting candidate information and job requirements. It also provides data retrieval and filtering capabilities, allowing users to quickly locate required data. It also includes data backup and recovery mechanisms to ensure data security.

[0094] 2. Model Training and Configuration Module: This module allows users to adjust parameters and configure semantic matching rules for large language models based on different selection scenarios and job requirements. This module supports model training, optimization, and updates, ensuring the system remains adaptable to evolving selection requirements.

[0095] 3. Intelligent Scoring Module: Implements intelligent match scoring, automatically completing processes such as data preprocessing, scoring model construction, match scoring, and candidate list generation. Users can view scoring progress and results in real time and obtain detailed scoring criteria.

[0096] 4. Intelligent PK Module: Executes intelligent combination PK methods, automatically generates candidate combinations, conducts PK evaluation and statistical ranking, and outputs a list of candidates and the basis for winning. It supports users to re-evaluate and compare specific combinations.

[0097] 5. Results Display and Export Module: This module displays candidate information, scoring results, and winning criteria in visual charts and lists, allowing users to intuitively understand the candidate profile. Results can also be exported to Excel, PDF, and other formats for easy archiving and sharing.

[0098] (3) System operation process

[0099] 1. Users log in to the system through the application layer, upload candidate information and job requirements in the data management module, and set selection rules and scoring criteria in the model training and configuration module.

[0100] 2. Based on the selection requirements, the user selects the intelligent matching scoring method or the intelligent combination PK method, and the system calls the corresponding functional modules to execute the selection process.

[0101] 3. The intelligent scoring module independently evaluates the candidates, generates a matching score and a candidate list; the intelligent PK module compares and evaluates the candidate combinations, counts the number of wins, and sorts them to generate a candidate list.

[0102] 4. The result display and export module presents the selection results to users in a visual manner. Users can make final decisions based on the scoring and winning criteria, combined with their own experience, to determine the winners.

[0103] Example 4;

[0104] like Figure 4 As shown, this embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned intelligent personnel selection method is implemented.

[0105] Example 5;

[0106] This embodiment provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the professional skills assessment method of the above embodiment is implemented.

[0107] In the above embodiments provided in this application, it should be understood that the disclosed methods, systems, devices, and media can be implemented in other ways. The above-described methods, systems, devices, and media embodiments are merely illustrative. For example, the division of modules or units is merely a logical functional division, and other division methods may be used in actual implementation. Each functional unit may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit.

[0108] The units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0109] The computer program includes computer program code, which may be in source code form, object code form, executable file, or some intermediate form. Computer-readable media may include any entity or device capable of carrying computer program code, recording media, USB flash drives, removable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory, electrical carrier signals, telecommunications signals, and software distribution media.

[0110] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. An intelligent personnel selection method based on a large language model and semantic matching, characterized by: Including intelligent matching scoring method and intelligent combination PK method, The intelligent matching scoring method includes the following steps: S1. Data preprocessing: Collect comprehensive data on all candidates, including basic attribute information, professional resume information, and comprehensive evaluation information. Use data cleaning technology to eliminate invalid data, duplicate data, and noise data. Use standardized data processing procedures and unify data format specifications. S2. Build a scoring model: Based on the responsibilities, skill requirements, and professional competency standards of specific positions, we construct refined semantic matching rules and a quantitative scoring indicator system. Through deep knowledge embedding and model training optimization, we deeply integrate these rules and standards into the large language model, enabling it to accurately understand the semantic connotations and logical relationships of the selection requirements. Based on this, we conduct in-depth semantic analysis and feature extraction of candidate information. S3. Matching score: The candidate data pre-processed in step S1 is input into the large language model one by one according to the established process. Based on the construction of a complete semantic matching mechanism, the large language model conducts a comprehensive and in-depth quantitative analysis and accurate scoring of the degree of match between each candidate and the selection requirements from multiple key dimensions such as professional skills adaptability, work experience relevance, professional quality fit, and corporate culture integration. S4. Generate a candidate list: After all candidates have been scored, the system systematically sorts them from highest to lowest match scores using algorithms like intelligent bubble sorting, automatically generating a structured candidate list. Based on this list and the detailed scoring criteria provided by the large language model, decision-makers conduct in-depth evaluations and secondary screening of the candidates, fully leveraging their professional judgment to independently determine the final winner. The intelligent combination PK method includes the following steps: S1. Data preprocessing: Collect comprehensive data on all candidates, including basic attribute information and comprehensive evaluation information. Use data cleaning technology to eliminate invalid, duplicate, and noisy data. Use standardized data processing procedures and unify data format specifications. S2. Generate duel matrix: After obtaining complete information on all candidates, the system uses combinatorial mathematics algorithms to automatically generate comprehensive and complete two-person PK pairs, ensuring that each candidate can form a comparative pair with all other candidates. Through comprehensive two-on-two duels, the system fully demonstrates the strengths and weaknesses of each candidate in different comparison scenarios. S3. PK evaluation: The complete basic information and resumes of the two candidates in each PK pair are accurately input into the large language model. Based on the semantic matching mechanism of the selection requirements, the large language model conducts in-depth comparative analysis and intelligent judgment on the strengths and weaknesses of the two candidates in terms of job compatibility from multiple key dimensions. S4, statistics and bubble sort: After completing the evaluation of all PK combinations, data statistics and analysis algorithms are used to automatically count the number of wins for each candidate. Based on the number of wins, bubble sort and other rules are used to systematically sort the candidates from most to least, generating a list of candidates that intuitively reflects the relative advantages of the candidates. Decision-makers conduct a comprehensive and integrated consideration of the candidates based on the ranking of the number of wins and the detailed winning basis provided by the large language model, and independently decide on the final winner based on their professional experience and judgment.

2. The intelligent personnel selection method according to claim 1, characterized in that: The basic attribute information includes but is not limited to age, education background, and major; the professional resume information includes but is not limited to job history, work performance, project experience, and skill certificates; the comprehensive evaluation information includes but is not limited to leadership evaluation, colleague evaluation, and social evaluation.

3. An intelligent personnel selection system based on a large language model and semantic matching, characterized by: include: Data management module: supports batch import, editing, and deletion of candidate information and job requirement information, provides data retrieval and screening functions; and also has a data backup and recovery mechanism; Model training and configuration module: supports users to adjust parameters and configure semantic matching rules for large language models based on selection scenarios and job requirements; supports model training, optimization, and updating; Intelligent Scoring Module: Implements intelligent matching scoring method, automatically completes data preprocessing, scoring model construction, matching scoring, and candidate list generation; also allows users to view scoring progress and results in real time and obtain detailed scoring basis; Intelligent PK module: Executes intelligent combination PK method, automatically generates candidate combinations, conducts PK evaluation and statistical ranking, outputs a list of candidates and the basis for winning; supports users to re-evaluate and compare specific combinations; Result display and export module: displays candidate information, scoring results and winning criteria in the form of visual charts and lists; at the same time, supports the export of selection results.

4. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the intelligent personnel selection method according to any one of claims 1 to 2 is implemented.

5. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the intelligent personnel selection method according to any one of claims 1 to 2 is implemented.

Images