Employment matching method and system based on intelligent data analysis

Abstract

The invention relates to an employment matching method and system based on intelligent data analysis. According to the method, multi-modal recessive skill data of candidates in an occupational cooperation platform, professional field contribution and occupational evaluation are collected, recessive ability vectors of communication cooperation, problem solving and the like are extracted through structured processing and an emotion analysis algorithm, and cooperation compactness and skill complementarity among members are modeled in combination with a team skill relation graph. And based on a game theory multi-objective optimization algorithm, enterprise employment cost, team efficiency and individual development demands are dynamically balanced, employment matching of accurate quantification of recessive skills, team structure suitability optimization and multi-party benefit balance is finally realized, and the precision of employment matching and the sustainability of long-term occupational development are significantly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to an employment matching method and system based on intelligent data analysis. Background Art

[0002] In traditional employment matching systems, it usually relies on text keyword matching of explicit skills (such as education background, certificates, working years) in job seekers' resumes and screening of hard requirements in enterprise job descriptions. Although such methods can quickly complete the initial screening, they have the following significant defects: On the one hand, the implicit abilities of candidates (such as soft skills like communication and coordination, teamwork, problem-solving, etc.) are difficult to accurately quantify through resume texts or standardized tests, resulting in a large number of high-potential talents being overlooked due to insufficient data representation.

[0003] On the other hand, existing technologies usually only focus on the single-point adaptation of individuals to positions, ignoring the skill complementarity between candidates and target team members, the fit with team culture, and the multi-party balance of enterprise employment costs and talent growth needs. For example, traditional models may lead to a large deviation between the recommended results and the actual team needs due to ignoring the team interaction characteristics in cross-platform collaboration records; at the same time, static matching rules cannot dynamically respond to changes in industry skill requirements or enterprise strategic adjustments, making the matching results lack sustainability.

[0004] Therefore, how to achieve employment matching with implicit skill quantification assessment, team structure optimization, and multi-party interest balance through intelligent data analysis has become a technical problem to be urgently solved. Summary of the Invention

[0005] Based on this, it is necessary to provide an employment matching method and system based on intelligent data analysis for the above technical problems.

[0006] In the first aspect, the present application provides an employment matching method based on intelligent data analysis, including: S1: Collect multi-modal implicit skill data of candidates, and perform structured processing on the multi-modal implicit skill data to generate a structured feature matrix; the multi-modal implicit skill data includes collaboration platform interaction records, professional field contribution data, and career assessment interaction data; S2: Based on the structured feature matrix, generate an implicit skill vector through sentiment analysis algorithms and contribution quantification models; S3: Calculate the compatibility score between the candidate and the target team based on the implicit skill vector and the historical collaboration data of the members of the target team; S4: Generate an equilibrium matching plan through a multi-objective game optimization algorithm based on the enterprise utility function, team utility function, individual utility function, and compatibility score; S5: Obtain the feedback results of the candidate on the balanced matching solution, and dynamically iterate the balanced matching solution according to the feedback results.

[0007] In a second aspect, the present application also provides an employment matching system based on intelligent data analysis, including: A data collection and processing module, configured to collect multi-modal implicit skill data of candidates, and perform structured processing on the multi-modal implicit skill data to generate a structured feature matrix; the multi-modal implicit skill data includes collaboration platform interaction records, professional field contribution data, and career assessment interaction data; A feature vector generation module, configured to generate an implicit skill vector based on the structured feature matrix through a sentiment analysis algorithm and a contribution quantification model; A compatibility scoring module, configured to calculate the compatibility score of the candidate with the target team based on the implicit skill vector and the historical collaboration data of the members of the target team; A matching solution generation module, configured to generate a balanced matching solution through a multi-objective game optimization algorithm based on the enterprise utility function, the team utility function, the individual utility function, and the compatibility score; A feedback and iteration module, configured to obtain the feedback results of the candidate on the balanced matching solution, and dynamically iterate the balanced matching solution according to the feedback results.

[0008] In a third aspect, the present application also provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, it implements an employment matching method based on intelligent data analysis as described in the first aspect.

[0009] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements an employment matching method based on intelligent data analysis as described in the first aspect.

[0010] The above employment matching method and system based on intelligent data analysis collect multi-modal implicit skill data of candidates in the professional collaboration platform, professional field contribution, and career assessment, extract implicit ability vectors such as communication and collaboration, problem-solving, etc. through structured processing and sentiment analysis algorithms, model the collaboration tightness and skill complementarity among members through a team skill relationship diagram, and dynamically balance the enterprise's employment cost, team efficiency, and individual development needs based on the game theory multi-objective optimization algorithm, ultimately realizing accurate quantification of implicit skills, optimization of team structure adaptability, and balanced employment matching of multiple parties' interests, significantly improving the accuracy of person-job matching and the sustainability of long-term career development. Description of the Drawings

[0011] To more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the related art. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0012] Figure 1 It is a schematic flowchart of an employment matching method based on intelligent data analysis provided by the present invention; Figure 2 It is a schematic structural diagram of an employment matching system based on intelligent data analysis provided by the present invention. Detailed implementation manners

[0013] In order to make the purpose, technical solutions and advantages of the present application clearer, the following further details the present application in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] Refer to Figure 1 , which shows a schematic flowchart of an employment matching method based on intelligent data analysis provided by the present application. The method includes the following steps: S1: Collect the multi-modal implicit skill data of candidates, and perform structured processing on the multi-modal implicit skill data to generate a structured feature matrix; the multi-modal implicit skill data includes collaborative platform interaction records, professional field contribution data, and career assessment interaction data.

[0015] Specifically, when collecting collaborative platform interaction records, data docking can be performed through the API interfaces of various online collaborative platforms (such as Slack, Microsoft Teams, etc.) to obtain the interaction behavior data of candidates on the above platforms in real time, including but not limited to the discussion topics participated in, the speech frequency, the speech content, the interaction with other members (such as likes, replies, @ mentions, etc.), the performance of problem-solving ability, and the collaborative role in team projects. The above data can reflect the implicit skills of candidates such as communication ability, coordination ability, teamwork spirit, and problem-solving ability in the actual team collaboration environment.

[0016] When collecting data on contributions in a professional field, data scraping techniques can be used to collect data on candidates' contributions in their professional fields from professional academic databases (such as IEEE Xplore, ScienceDirect, etc.), technology communities (such as Stack Overflow, GitHub, etc.), and industry forums. For example, obtain information such as the number of papers published by candidates, paper citation situations, and scientific research projects participated in from academic databases; collect data such as code solutions proposed by candidates, the number of technical questions answered, and open-source projects with code contributions in technology communities; count professional insights shared by candidates and the popularity of discussions participated in from industry forums. The above data helps to evaluate candidates' implicit skills such as knowledge depth, innovation ability, and industry influence in the professional field.

[0017] When collecting data on occupational assessment interactions, a dedicated occupational assessment interaction platform can be developed, and a series of assessment questions and interactive scenarios for implicit skills can be designed to guide candidates to conduct online assessments. The assessment content covers multiple dimensions such as communication ability tests (such as evaluating expression and listening abilities in simulated team meeting scenarios), teamwork ability tests (such as analyzing the degree of cooperation through the completion of group tasks), and problem-solving ability tests (such as presenting complex business problems and asking candidates to propose solutions and evaluating their logic and effectiveness). At the same time, record interaction data such as candidates' operation behaviors, thinking paths, and time allocation during the assessment process to more comprehensively understand the actual application of their implicit skills.

[0018] Clean, preprocess, and extract features from the collected multi-modal implicit skill data. For text data (such as speech content on collaboration platforms, posts on professional field forums, etc.), use natural language processing techniques for lexical analysis, syntactic analysis, and semantic understanding to extract features such as keywords, topic tendencies, and sentiment tendencies; for behavioral data (such as collaboration platform interaction behaviors, occupational assessment operation behaviors, etc.), conduct behavior pattern recognition and quantitative statistics to construct behavioral feature vectors; for structured data (such as the score results of occupational assessments, etc.), directly extract relevant fields as feature values. Finally, integrate features from different modalities into a unified structured feature matrix, and the dimension of the feature matrix for each candidate is adaptively adjusted according to the richness of the data and the diversity of the features to ensure a comprehensive and accurate representation of the candidates' implicit skills.

[0019] S2: Based on the structured feature matrix, generate implicit skill vectors through sentiment analysis algorithms and contribution quantification models.

[0020] Specifically, sentiment analysis algorithms based on deep learning, such as pre-trained models of long short-term memory networks (LSTM) or Transformer architectures, are used to analyze the sentiment tendencies of text data in the structured feature matrix. By constructing a sentiment dictionary and a semantic rule base, positive, negative, or neutral sentiments expressed in the text are identified, and the emotional expression ability and emotional management ability of candidates during communication, teamwork, and problem-solving are quantified. For example, in the discussion records of a collaboration platform, if a candidate can put forward constructive opinions with a positive sentiment tendency and effectively coordinate the emotions of team members, it indicates that they have good emotional intelligence and team leadership potential, and the sentiment analysis algorithm will assign a higher emotional skills score to such candidates.

[0021] For the data on professional field contributions, a contribution quantification model is established. This model comprehensively considers factors such as the quantity, quality, and influence of a candidate's contributions in academic research, technological development, industry exchanges, etc. For example, for academic paper contributions, weighted calculations are performed based on indicators such as the impact factor of the journal where the paper is published, the number of citations, and the authority of the field where the paper is located; for code contributions to a technical community, quantitative evaluations are carried out according to dimensions such as the number of times the code is adopted, the code quality evaluation, and the difficulty of the problems solved; for the sharing of professional insights in industry forums, the reading volume, number of likes, and comment interaction of the posts are referred to measure their contribution value to the industry. Through multi-dimensional quantitative indicators, a comprehensive scoring system that can comprehensively reflect the contribution degree of candidates in the professional field is constructed, and then the corresponding contribution degree feature vector is generated.

[0022] The emotional skills score obtained from the sentiment analysis algorithm and the contribution degree feature vector generated by the contribution quantification model are fused according to a certain weight assignment strategy. At the same time, combined with other implicit skill-related features in the structured feature matrix (such as the ability dimension scores in career assessments, etc.), feature fusion and compression are carried out through dimensionality reduction techniques such as principal component analysis (PCA) or autoencoder, and finally a low-dimensional and highly expressive implicit skill vector is generated. This vector can comprehensively and accurately represent the implicit skill characteristics of candidates and provide the core data basis for subsequent matching calculations.

[0023] S3: Calculate the compatibility score between the candidate and the target team based on the implicit skill vector and the historical collaboration data of the members of the target team.

[0024] Specifically, collect the collaboration data of target team members in past projects, including multi-dimensional information such as communication records between team members, task assignment and completion, conflict resolution cases, and team performance evaluation results. Use social network analysis techniques to construct a collaboration relationship map of team members, and analyze issues such as the collaboration mode within the team, the influence of core members, the communication efficiency of the team, and potential collaboration bottlenecks. At the same time, mine the historical collaboration data through machine learning algorithms to extract the collaboration feature patterns of the team in different project types and business scenarios, such as the rapid response collaboration mode in emergency projects and the refined division of labor collaboration mode in complex projects, to provide a team collaboration benchmark model for subsequent compatibility evaluation.

[0025] Perform multi-dimensional matching and analysis on the latent skill vectors of candidates and the historical collaboration data of the target team. On the one hand, evaluate the degree of fit between the candidate's latent skills and team members' latent skills by calculating the similarity between the candidate's latent skill vector and the set of team members' latent skill vectors (such as using metrics like cosine similarity and Euclidean distance), and judge whether the candidate can integrate into the team's collaboration atmosphere. On the other hand, combined with the collaboration mode characteristics of the target team, analyze whether the candidate's latent skills can make up for the current collaboration shortcoming of the team. For example, if the team lacks innovation thinking and the candidate has strong innovation ability and idea stimulation ability, then it is considered that the candidate has a high compatibility with the team in this dimension. In addition, also consider the impact of factors such as team culture and values on compatibility, and through text analysis and semantic matching techniques, compare the value orientation reflected by the candidate in the career assessment with the team culture description to further optimize the calculation result of the compatibility score.

[0026] S4: Generate an equilibrium matching plan through a multi-objective game optimization algorithm based on the enterprise utility function, team utility function, individual utility function, and compatibility score.

[0027] Specifically, a multi-objective optimization model including an enterprise utility function, a team utility function, and an individual utility function is constructed. The enterprise utility function can consider factors such as the enterprise recruitment cost, the matching degree between the candidate and the position (including explicit skills and implicit skills), and the expected performance contribution after the candidate joins the enterprise. By establishing a mathematical model, these factors are transformed into quantifiable indicators and comprehensively weighted and calculated; the team utility function focuses on enhancing the overall collaboration efficiency of the team, taking key indicators such as the change in the compatibility score of the team after the candidate joins, the improvement of team member satisfaction, and the enhancement of the team's project delivery ability to construct a utility function reflecting the interests of the team; the individual utility function starts from the perspective of the candidate's career development, considering aspects such as the growth opportunities provided by the position, salary and benefits, working environment, and the degree of fit with the individual career plan, and transforms them into quantitative indicators to measure the candidate's personal satisfaction with the position. These three utility functions represent the interest demands of the enterprise, the team, and the individual in the employment matching process respectively, forming a multi-objective optimization problem.

[0028] A multi-objective game optimization algorithm is used to solve the above multi-objective optimization model. This algorithm draws on the Nash equilibrium idea in game theory to seek a balanced solution among multiple objectives. Specifically, the enterprise, the team, and the individual are regarded as game participants, and each participant attempts to maximize its own utility function while being constrained by the utility functions of other participants. The algorithm simulates the game process among the participants and continuously adjusts the matching strategies of the candidates, so that under certain constraint conditions (such as the limit on the number of positions to be recruited, the limit on the job search intentions of the candidates, etc.), the enterprise utility, the team utility, and the individual utility reach a relatively balanced state. In the process of algorithm implementation, intelligent optimization algorithms such as genetic algorithms and particle swarm optimization algorithms are used as solution tools. Through operations such as population initialization, fitness function calculation, selection, crossover, and mutation, the Pareto optimal solution set of the multi-objective optimization problem is gradually searched, and finally an equilibrium matching solution that comprehensively considers the interests of all parties and conforms to the actual recruitment scenario is selected from this solution set.

[0029] S5: Obtain the feedback results of the candidates on the equilibrium matching solution, and dynamically iterate the equilibrium matching solution according to the feedback results.

[0030] Specifically, a feedback collection mechanism is established. Through various methods such as online questionnaires, telephone follow-up, and face-to-face interviews, the feedback results of candidates on the balanced matching plan are obtained. The feedback content includes aspects such as the satisfaction of candidates with the recommended positions, the rationality evaluation of the matching results, the recognition of the system's recommended reasons, and the accuracy judgment of the evaluation of their own implicit skills. At the same time, collect the evaluation feedback of enterprises and teams on the actual performance of the candidates who have been hired, such as the performance of the candidates during their tenure, the integration with the team, and the contribution to team collaboration and enterprise development. Quantify and statistically analyze these feedback data, extract valuable information, such as common dissatisfaction factors, accurate skill evaluation dimensions, effective matching strategies, etc., to provide data support for subsequent system optimization.

[0031] According to the analysis of the feedback results, targeted dynamic iterative optimization is carried out for each link of the employment matching system. At the data collection level, if it is found that the influence of certain data sources on the matching results is biased or insufficient, adjust the data collection strategy in a timely manner, expand or replace the corresponding data collection channels; in terms of the implicit skill evaluation model, if the feedback shows that the evaluation of certain implicit skills is not accurate enough, adjust the parameters or reconstruct the model of the sentiment analysis algorithm, contribution quantification model, etc., to improve the quality of the generation of implicit skill vectors; in terms of the matching algorithm, optimize and adjust the parameter settings, target weight allocation, etc. of the multi-objective game optimization algorithm based on the feedback information to make it more in line with the interest balance requirements in the actual recruitment scenario. In addition, a closed-loop evaluation mechanism for the feedback results and the system optimization effect can be established to regularly evaluate the performance and verify the effect of the optimized system, ensuring that the system can continuously adapt to the changing employment market environment and user needs, and continuously improve the accuracy and satisfaction of employment matching.

[0032] The above-mentioned employment matching method based on intelligent data analysis collects multi-modal implicit skill data of candidates in the professional collaboration platform, professional field contributions, and career assessments, extracts implicit ability vectors such as communication and collaboration, problem-solving, etc. through structured processing and sentiment analysis algorithms, models the collaboration tightness and skill complementarity among members by combining the team skill relationship diagram, and dynamically balances the enterprise's employment costs, team effectiveness, and individual development needs based on the game theory multi-objective optimization algorithm, ultimately realizing accurate quantification of implicit skills, optimization of team structure adaptability, and balanced multi-party interests in employment matching, significantly improving the accuracy of person-job matching and the sustainability of long-term career development.

[0033] In an optional embodiment, S1 includes the following steps: S11: Obtain the desensitized message records of candidates on the collaboration platform through the API interface, and extract the message content and response speed of the desensitized message records as the collaboration platform interaction records.

[0034] Specifically, data docking can be achieved through the API interfaces of various online collaboration platforms (such as Slack, Microsoft Teams, DingTalk, Feishu, etc.). During the data acquisition process, data privacy and security regulations are strictly adhered to, and the message records of candidates are desensitized by removing personal sensitive information such as names and contact information, and only the content related to collaboration behaviors is retained. The extracted message content includes the speech text of candidates in team discussions, the types of topics participated in, and the interaction situations with other members (such as the number of replies, the number of likes, the frequency of @ mentions, etc.). These data can reflect the communication ability, team participation, and collaboration enthusiasm of candidates. At the same time, record the response speed of candidates, that is, the time interval from task assignment or question raising to the candidate's first reply. By analyzing the distribution of response speed, evaluate the timeliness and initiative of candidates in team collaboration.

[0035] S12: Analyze the work records of candidates in their professional fields to obtain the analysis results. Based on the analysis results, extract at least one of the task completion rate of the project management platform, the review passing rate of the design portfolio, or the customer feedback score as the professional field contribution data.

[0036] Specifically, for the work records of candidates in their professional fields, such as task completion situations on project management platforms (such as Jira, Trello, etc.), review results of design portfolios, and customer feedback data, in-depth analysis is carried out. Using data mining and text analysis techniques, extract the task completion rate from the project management platform, that is, the ratio of the number of tasks successfully completed by the candidate to the total number of tasks. This indicator reflects the reliability and efficiency of the candidate in project execution. For the review passing rate of the design portfolio, analyze the ratio of the number of works passed in the professional review process to the total number of submissions to measure the candidate's ability and level in professional creation. The customer feedback score is obtained by collecting the evaluation scores of customers on the candidate's work performance, including but not limited to aspects such as work quality, service attitude, and problem-solving ability, to comprehensively evaluate the candidate's ability to meet customer needs in actual business scenarios.

[0037] S13: Collect the interaction path data of candidates in the gamified tasks of career assessment as the career assessment interaction data; among them, the interaction path data includes the task completion time, decision options, and the number of error corrections.

[0038] Specifically, develop a dedicated professional assessment gamification task platform and design a series of interesting and interactive tasks to comprehensively evaluate the implicit skills of candidates. In task design, focus on simulating real work scenarios, such as teamwork tasks, problem-solving challenges, communication and coordination games, etc., so that candidates can naturally demonstrate their abilities in a relaxed atmosphere. Collect the interaction path data of candidates in these gamification tasks, including the task completion time, that is, the duration from the start to the end of the task, which reflects the decision-making speed and execution efficiency of candidates; decision options, record the choices made by candidates at key task nodes, and analyze their decision-making tendencies and thinking patterns; the number of error corrections, count the number of times candidates correct errors during the task process, and evaluate their learning ability, adaptability, and resilience in problem-solving.

[0039] S14: Normalize the interaction records of the collaboration platform, the contribution data in the professional field, and the interaction path data to generate a structured feature matrix.

[0040] Specifically, normalize the collected interaction records of the collaboration platform, the contribution data in the professional field, and the professional assessment interaction data to eliminate the dimensional and magnitude differences between different data dimensions and ensure the comparability and consistency of the data. Use methods such as Min-Max normalization or Z-score normalization to map each data index to a unified numerical range, such as [0, 1] or [-1, 1]. For the fusion of multi-modal data, construct a structured feature matrix, where the rows represent different feature dimensions and the columns represent different candidates. Fill the normalized data into the matrix according to the predefined feature order, and the feature vector of each candidate consists of its normalized values on each feature dimension.

[0041] In an alternative embodiment, S2 includes the following steps: S21: Perform sentiment analysis on the normalized interaction records of the collaboration platform in the structured feature matrix to generate a communication ability index; communication ability index The calculation formula is: ; Among them, is the sentiment polarity value obtained by performing sentiment analysis on the content of the th message in the interaction records of the collaboration platform through the BERT model; is the response speed of the content of the th message in the interaction records of the collaboration platform; is the content of the th message in the interaction records of the collaboration platform; and are preset weight coefficients, is the total amount of the message content in the interaction records of the collaboration platform.

[0042] Specifically, for the normalized collaborative platform interaction records in the structured feature matrix, natural language processing technology - the BERT model is used for sentiment analysis. The BERT model can accurately understand the semantic information and sentiment tendency in the text through a deep bidirectional Transformer architecture. For each message content , the sentiment polarity value output by the BERT model reflects the degree of positive, negative or neutral sentiment contained in the message, and the value range is usually between [-1, 1], where -1 indicates extremely negative and 1 indicates extremely positive. At the same time, considering the impact of response speed on communication ability, a response speed term is introduced, and its calculation method is the reciprocal of the time interval from the message release time to the relevant task response time. The faster the response, the larger this value. Through preset weight coefficients and , the sentiment polarity value and the response speed are weighted and summed, and the average value of all messages is taken to obtain the communication ability index . This index comprehensively reflects the communication ability and enthusiasm of the candidate in team collaboration.

[0043] S22: Based on the contribution quantification model, calculate the technical leadership score according to the normalized professional field contribution data in the structured feature matrix.

[0044] Specifically, based on the professional field contribution data, a contribution quantification model is constructed to calculate the technical leadership score. Suppose the professional field contribution of a candidate in the software development field is evaluated. Then the contribution quantification model can comprehensively consider multiple contribution indicators of the candidate in the professional field, such as the number of paper citations, code contributions to the technical community, influence in industry forums, etc. For the number of paper citations, the weighted summation method can be used, and different weights are assigned according to the impact factor and field relevance of the cited journals; for the code contributions to the technical community, quantitative evaluation can be carried out based on dimensions such as the number of times the code is adopted, code quality evaluation, and the difficulty of the problems solved; for the influence in industry forums, the contribution value to the industry can be measured by referring to the reading volume, number of likes, comment interaction, etc. of the posts.

[0045] Specifically, the relevant calculation steps for the number of paper citations can be as follows: Refer to the classification and influence of different academic databases on paper types (such as journal papers, conference papers, dissertation papers, etc.), and assign corresponding points to different types of papers. For example, well-known international journal papers have higher points, and ordinary conference papers have relatively lower points.

[0046] It is comprehensively determined based on factors such as the author's ranking in the paper and the degree of contribution. The first author and the corresponding author have a relatively high degree of contribution, and the contribution of other participating authors is proportionally distributed according to factors such as the number of authors. For example, if a paper has 5 authors, the first author can be allocated 50% of the contribution, the corresponding author is allocated 40%, and the remaining authors are evenly allocated 10%.

[0047] Divide the intervals according to the number of times the paper is cited, and different intervals are given different points. For example, the number of citations between 0 - 10 times has a relatively low point, and between 100 - 500 times has a relatively high point, and so on.

[0048] Therefore, the score of each paper = (author contribution degree × journal point + author contribution degree × citation frequency point) × literature type point of the paper. Accumulate the scores of all the papers of the candidate in the professional field to obtain the total score in terms of papers.

[0049] Specifically, the relevant calculation steps for code contributions in the technical community can be as follows: In the open - source community, the evaluation mechanism of the community for code contributions can be referred to. For example, indicators such as the number of likes obtained in code reviews and the proportion of code adopted by core community members are used to measure the quality of contributions. For example, a certain score can be obtained for each like, and the higher the proportion of code adopted by core members, the higher the corresponding score.

[0050] Developers are encouraged to participate in contributions for a long time, and a decay coefficient and a cumulative incentive mechanism are set. For example, developers who have continuously contributed for more than a certain period of time are given additional score rewards; if the contribution is interrupted, the score gradually decays at a certain proportion (such as 85%), but it will not be completely cleared, so that developers can quickly resume their previous influence when they return to the community at any time.

[0051] Comprehensively considering factors such as the quality, quantity, and long - term nature of contributions, a code contribution score model is constructed. For example, code contribution score = (number of contributed projects × basic score per project + number of likes × like score + proportion of code adopted × adoption score) × long - term incentive coefficient.

[0052] Specifically, the relevant calculation steps for the influence in the industry forum can be as follows: Analyze indicators such as the reading volume, number of likes, and comment interaction of the posts published by the candidate in the industry forum to measure their influence. For example, a certain score can be obtained for each like, a corresponding score can also be obtained for each increase in a certain amount of reading volume, and the higher the heat of comment interaction, the more points.

[0053] Through the evaluation and feedback of forum users, such as the number of times the post is recommended as an excellent post and the number of times it is cited by other users, to measure the innovation and practicality of the content. Corresponding score rewards can be set, such as a relatively high score can be obtained for being recommended as an excellent post once, and a certain score can be obtained for being cited once.

[0054] Construct an influence score model for industry forums by considering factors such as comprehensive reading volume, number of likes, comment interactions, featured recommendations, and number of citations. For example, the influence score of an industry forum = (reading volume score + number of likes score + comment interaction score + featured recommendation score + number of citations score).

[0055] Finally, based on the preset weight coefficients, the total scores in the above-mentioned aspects of the paper, the code contribution scores, and the influence scores of industry forums can be weighted and integrated to obtain the corresponding technical leadership ratings.

[0056] S23: Vectorially integrate the communication ability index, the technical leadership rating, and the normalized career assessment interaction data in the structured feature matrix to generate a latent skill vector.

[0057] Specifically, vectorially integrate the communication ability index, the technical leadership rating, and the career assessment interaction data. After the career assessment interaction data is normalized, it is used together with the former two as components of the feature vector. Using dimensionality reduction techniques such as principal component analysis (PCA) or autoencoder, while retaining the key information, map the high-dimensional feature space to a low-dimensional space to generate a compact and highly expressive latent skill vector. This vector can comprehensively and accurately represent the latent skill characteristics of the candidate.

[0058] In an alternative embodiment, S3 includes the following steps: S31: Construct a skill relationship graph based on historical collaboration data, and the edge weights of the skill relationship graph are calculated by the following formula: ; where is the number of collaborative projects between member and member in the target team, is the total number of projects of member ; the edge weight represents the degree of collaboration tightness between member and member in the target team.

[0059] Specifically, based on the historical collaboration data of the target team, construct a skill relationship graph that can intuitively display the collaboration relationships and skill associations among team members. The nodes in the graph represent team members, and the edges represent the collaboration relationships between members. The edge weight is calculated by the formula where represents the number of projects jointly participated by member i and member j, is the total number of projects participated by member i. This weight reflects the degree of closeness of collaboration among members. The higher the weight, the more frequently the two have collaborated in past projects, and the stronger the potential for complementary skills. By constructing a skill relationship graph, the collaboration structure and skill distribution within the target team can be clearly presented, providing a basis for subsequent compatibility assessment.

[0060] S32: Map the implicit skill vector to candidate nodes of the skill relationship graph, and calculate the cosine similarity between the candidate nodes and the team nodes; among them, the team nodes are constructed from historical collaboration data, and each team node corresponds to a skill feature vector of a team member; the cosine similarity represents the skill similarity between the candidate and the members in the target team.

[0061] Specifically, map the candidate's implicit skill vector into the skill relationship graph as candidate nodes. Use the cosine similarity formula , and calculate the skill similarity between the candidate nodes and each member node in the team. Among them, is the candidate's implicit skill vector, is the skill feature vector of the j-th member in the team. The value range of the cosine similarity is between [-1, 1]. The closer the value is to 1, the more similar the two skills are; the closer the value is to -1, the greater the skill difference. Through this method, the degree of fit between the candidate and the members of the target team at the skill level can be quantitatively evaluated.

[0062] S33: Calculate the compatibility score based on the cosine similarity and edge weights; the calculation formula for the compatibility score is: ; Among them, is the cosine similarity, representing the skill similarity between the candidate and the -th member in the target team; is the number of members in the target team.

[0063] Specifically, based on the cosine similarity and the edge weights of the skill relationship graph, calculate the compatibility score of the candidate and the target team. Through the formula , multiply the skill similarity between each team member and the candidate by its corresponding edge weight and sum them up to obtain the final compatibility score. Among them, M is the number of members in the target team. This score comprehensively considers the matching degree between the candidate's skills and the team members and the closeness of the internal collaboration relationship within the team, and can comprehensively reflect the potential adaptability of the candidate to integrate into the target team.

[0064] In an optional embodiment, S4 includes the following steps: S41: Define the enterprise utility function, the team utility function, and the individual utility function based on the compatibility score.

[0065] Specifically, based on the compatibility score, define the enterprise utility function , the team utility function and the individual utility function . The enterprise utility function mainly considers factors such as the enterprise recruitment cost, the matching degree between the candidate and the position (including explicit skills and implicit skills), and the expected performance contribution after the candidate joins the company. By establishing a mathematical model, these factors are transformed into quantifiable indicators and comprehensively weighted. The team utility function focuses on improving the overall collaboration efficiency of the team, taking the changes in the compatibility score of the team after the candidate joins, the improvement of team member satisfaction, and the enhancement of the team's project delivery ability as key indicators. The individual utility function starts from the perspective of the candidate's career development, considering aspects such as the growth opportunities provided by the position, salary and benefits, working environment, and the fit with the individual career plan, and transforms them into quantitative indicators to measure the candidate's personal satisfaction with the position.

[0066] S42: Based on the enterprise utility function, the team utility function, and the individual utility function, construct a multi-objective optimization problem; the expression of the multi-objective optimization problem is: ; where is the enterprise utility function, is the team utility function, is the individual utility function, , and are dynamic weights obtained by training with historical HR decision data.

[0067] Specifically, based on the above three utility functions, construct a multi-objective optimization problem, and its expression is: . Where , and are dynamic weights obtained by training with historical HR decision data, which are used to balance the relative importance of the interests of the enterprise, the team, and the individual in the optimization problem. These weights reflect the attention degree and priority setting of enterprise decision-makers to different stakeholders in the actual recruitment scenario, and can be dynamically adjusted according to different recruitment needs and market environments to ensure that the optimization results are more in line with the actual situation.

[0068] S43: For the multi-objective optimization problem and the preset constraint conditions, solve the Pareto optimal solution set through the NSGA-II algorithm.

[0069] Specifically, the NSGA-II algorithm is used to solve the constructed multi-objective optimization problem and the preset constraints (such as the number of job vacancies, the job-seeking intention of candidates, etc.). NSGA-II (fast non-dominated sorting genetic algorithm) is an efficient multi-objective optimization algorithm that continuously evolves and searches in the population by simulating natural selection and genetic mechanisms to find the Pareto optimal solution set that can optimize multiple objective functions at the same time. During the execution of the algorithm, a random candidate population is first initialized, and then iteratively evolves step by step through operations such as fitness evaluation, selection, crossover, and mutation to generate a series of non-dominated solutions on the Pareto frontier. These solutions represent the optimal matching solutions under different trade-offs.

[0070] S44: Select the solution with the highest comprehensive score from the Pareto optimal solution set as the equilibrium matching solution.

[0071] Specifically, the final balanced matching solution is selected from the Pareto optimal solution set based on the principle of the highest comprehensive score. The comprehensive score can be calculated by weighted summing the enterprise utility, team utility and personal utility corresponding to each solution according to the preset business rules or decision maker's preferences, or by evaluating through other multi-criteria decision-making methods. The selected balanced matching solution can achieve a relatively balanced state between the enterprise's recruitment needs, team collaboration optimization and personal career development, which not only meets the enterprise's employment requirements, but also takes into account the team's collaborative efficiency and the candidate's personal wishes, and realizes the coordination and unification of the interests of multiple parties.

[0072] In an optional embodiment, S5 includes the following steps: S51: Output a list of recommended positions and an interpretable report for a balanced matching solution. The interpretable report includes matching degree, team compatibility score, and salary range.

[0073] Specifically, the generated balanced matching scheme is converted into an intuitive and easy-to-understand list of recommended positions and presented to the candidates. The list contains multiple matching positions, each of which describes in detail the basic information such as the position name, company, and work location. At the same time, an explainable report is generated for each recommended position, which covers key information such as matching degree, team compatibility score, and salary range. The matching degree is based on the comprehensive evaluation results of the enterprise utility function, team utility function, and personal utility function, and is presented in the form of a percentage, reflecting the overall fit between the candidate and the position. The team compatibility score details the match between the candidate and the target team in terms of skills, personality, work style, etc., and explains the role and contribution he may play in the team. The salary range clearly shows the salary range provided by the position to ensure that the candidate has a clear understanding of the salary package.

[0074] S52: If the candidate refuses the recommendation, a rejection reason label is collected, and the dynamic weight is updated according to the rejection reason label.

[0075] Specifically, if a candidate refuses a recommended position, the rejection reason label is collected in a timely manner. The rejection reason label is obtained through a designed questionnaire survey or interview feedback, covering multiple dimensions, such as job mismatch, salary dissatisfaction, inconvenience of work location, etc. The collected rejection reasons are classified and quantified to build a rejection reason feedback data set. Based on historical HR decision data and new rejection reason feedback data, dynamic weights are retrained , and . Using machine learning algorithms, such as linear regression and neural networks, the rejection reasons are used as feature inputs, and the weights are adjusted to minimize the prediction error, so that the new dynamic weights can better reflect the actual needs and preferences of the candidates.

[0076] S53: Regenerate a balanced matching solution according to the updated dynamic weight.

[0077] Specifically, the multi-objective optimization process is re-executed using the updated dynamic weights. The new weight values ​​are substituted into the expression of the multi-objective optimization problem, and the Pareto optimal solution set is solved again through the NSGA-II algorithm in combination with the enterprise utility function, team utility function, and personal utility function. During the operation of the algorithm, the priority of the objective function is adjusted according to the new weights to ensure that the optimization results are more in line with the expectations of the candidates. Finally, the solution with the highest comprehensive score is selected from the new Pareto optimal solution set as the regenerated equilibrium matching solution, and the recommended job list and explainability report are output again until the candidate accepts the recommendation or the preset number of iterations is reached.

[0078] The above-mentioned employment matching method based on intelligent data analysis integrates multimodal data such as interaction records of career collaboration platforms, contribution data in professional fields, and career assessment interactions, constructs a structured feature matrix and extracts implicit skill vectors such as communication skills and problem-solving skills, and combines the team skill relationship diagram to quantify the collaboration closeness and skill complementarity between candidates and target teams. It dynamically balances the company's employment costs, team effectiveness, and personal development demands based on the game theory multi-objective optimization model, and realizes adaptive optimization of matching solutions through a weight iteration mechanism driven by user feedback. Ultimately, it achieves a comprehensive technical effect of accurate quantification of implicit skills, coordinated optimization of multi-party interests, enhanced adaptability of team structure, and improved interpretability of job matching decisions, thereby solving the systematic defects of traditional methods in implicit ability assessment, dynamic team adaptation, and long-term career development support.

[0079] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are sequentially shown as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0080] Based on the same inventive concept, an embodiment of the present application also provides a system for implementing the employment matching method based on intelligent data analysis described above. The implementation solution provided by this system to solve problems is similar to the implementation solution described in the above method. Therefore, the specific limitations in one or more embodiments of the employment matching system based on intelligent data analysis provided below can refer to the limitations on an employment matching method based on intelligent data analysis in the above text, and will not be repeated here.

[0081] In an exemplary embodiment, as Figure 2 shown, an employment matching system 20 based on intelligent data analysis is provided, including: A data collection and processing module 21, configured to collect multi-modal implicit skill data of candidates, and perform structured processing on the multi-modal implicit skill data to generate a structured feature matrix; the multi-modal implicit skill data includes collaboration platform interaction records, professional field contribution data, and career assessment interaction data.

[0082] A feature vector generation module 22, configured to generate an implicit skill vector based on the structured feature matrix through a sentiment analysis algorithm and a contribution quantification model.

[0083] A compatibility scoring module 23, configured to calculate the compatibility score between a candidate and a target team based on the implicit skill vector and the historical collaboration data of the members of the target team.

[0084] A matching scheme generation module 24, configured to generate an equilibrium matching scheme through a multi-objective game optimization algorithm based on an enterprise utility function, a team utility function, a personal utility function, and a compatibility score.

[0085] A feedback and iteration module 25, configured to obtain the feedback result of a candidate on the equilibrium matching scheme, and perform dynamic iteration on the equilibrium matching scheme according to the feedback result.

[0086] Optionally, the data collection and processing module 21 includes: The collaborative data acquisition and extraction unit 211 is used to obtain the anonymized message records of candidates on the collaborative platform through the API interface, and extract the message content and response speed of the anonymized message records as the collaborative platform interaction records.

[0087] The professional data parsing and extraction unit 212 is used to parse the work records in the professional fields of candidates to obtain parsing results, and based on the parsing results, extract at least one of the task completion rate on the project management platform, the review passing rate of the design portfolio, or the customer feedback score as the professional field contribution data.

[0088] The career assessment data collection unit 213 is used to collect the interaction path data of candidates in the career assessment gamification tasks as the career assessment interaction data; among them, the interaction path data includes the task completion time, decision options, and the number of error corrections.

[0089] The data integration and processing unit 214 is used to perform normalization processing on the collaborative platform interaction records, professional field contribution data, and interaction path data to generate a structured feature matrix.

[0090] Optionally, the feature vector generation module 22 includes: The communication assessment unit 221 is used to perform sentiment analysis on the normalized collaborative platform interaction records in the structured feature matrix to generate a communication ability index; the communication ability index The calculation formula is: ; Where is the sentiment polarity value obtained by performing sentiment analysis on the th message content in the collaborative platform interaction records through the BERT model; is the response speed of the th message content in the collaborative platform interaction records; and are preset weight coefficients, and is the total amount of message content in the collaborative platform interaction records. is the total amount of message content in the collaborative platform interaction records.

[0091] The technology assessment unit 222 is used to calculate the technology leadership score based on the contribution quantification model according to the normalized professional field contribution data in the structured feature matrix.

[0092] The skill integration unit 223 is used to perform vector fusion on the communication ability index, technology leadership score, and the normalized career assessment interaction data in the structured feature matrix to generate a latent skill vector.

[0093] Optionally, the compatibility scoring module 23 includes: A collaboration relationship modeling unit 231, configured to construct a skill relationship graph according to historical collaboration data, and the edge weights of the skill relationship graph are calculated by the following formula: ; where is the number of collaboration projects between member and member in the target team, is the total number of projects of member ; the edge weight represents the collaboration closeness between member and member in the target team.

[0094] A skill similarity calculation unit 232, configured to map the implicit skill vector to candidate nodes of the skill relationship graph, and calculate the cosine similarity between the candidate nodes and the team nodes; wherein, the team nodes are constructed from historical collaboration data, and each team node corresponds to a skill feature vector of a team member; the cosine similarity represents the skill similarity between the candidate and the members in the target team.

[0095] A compatibility score calculation unit 233, configured to calculate the compatibility score based on the cosine similarity and the edge weight; the calculation formula of the compatibility score is: ; where is the cosine similarity, representing the skill similarity between the candidate and the th member in the target team; is the number of members in the target team.

[0096] Optionally, the matching scheme generation module 24 includes: A utility function definition unit 241, configured to define an enterprise utility function, a team utility function, and a personal utility function based on the compatibility score.

[0097] An optimization problem construction unit 242, configured to construct a multi-objective optimization problem based on the enterprise utility function, the team utility function, and the personal utility function; the expression of the multi-objective optimization problem is: ; where is the enterprise utility function, is the team utility function, is the personal utility function, , and are dynamic weights obtained by training with historical HR decision data.

[0098] The optimal solution solving unit 243 is used to solve the Pareto optimal solution set for the multi-objective optimization problem and the preset constraint conditions through the NSGA-II algorithm.

[0099] The solution selection unit 244 is used to select the solution with the highest comprehensive score from the Pareto optimal solution set as the equilibrium matching solution.

[0100] Optionally, the feedback and iteration module 25 includes: The result output unit 251 is used to output the recommended position list and the interpretability report of the equilibrium matching solution. The interpretability report includes the matching degree, the team compatibility score, and the salary range.

[0101] The feedback collection and analysis unit 252 is used to collect the rejection reason tags if the candidate rejects the recommendation, and update the dynamic weights according to the rejection reason tags.

[0102] The solution update unit 253 is used to regenerate the equilibrium matching solution according to the updated dynamic weights.

[0103] An embodiment of the present application also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps in the foregoing method embodiments are implemented.

[0104] An embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.

[0105] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiments described above are only illustrative. The components described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present disclosure solution. Those of ordinary skill in the art can understand and implement it without creative work.

[0106] The above embodiments only express several implementation manners of the embodiments of the present application. The description is relatively specific and detailed, but it cannot be understood as a limitation on the patent scope of the application embodiments. It should be noted that for those of ordinary skill in the art, without departing from the concept of the embodiments of the present application, several deformations and improvements can still be made, and these all belong to the protection scope of the embodiments of the present application.

Claims

1. An employment matching method based on intelligent data analysis, characterized in that: The method comprises: S1: Collecting the candidate's multimodal implicit skill data, and performing structured processing on the multimodal implicit skill data to generate a structured feature matrix; the multimodal implicit skill data includes collaborative platform interaction records, professional field contribution data, and career assessment interaction data; S2: Based on the structured feature matrix, an implicit skill vector is generated through a sentiment analysis algorithm and a contribution quantification model; S3: Calculating a compatibility score between the candidate and the target team based on the implicit skill vector and historical collaboration data of members of the target team; S4: Based on the enterprise utility function, the team utility function, the individual utility function and the compatibility score, an equilibrium matching solution is generated through a multi-objective game optimization algorithm; S5: Obtain the candidate's feedback result on the balanced matching solution, and dynamically iterate the balanced matching solution according to the feedback result.

2. The method according to claim 1, characterized in that The S1 includes: S11: Obtaining the candidate's desensitized message record on the collaboration platform through an API interface, extracting the message content and response speed of the desensitized message record as the collaboration platform interaction record; S12: parsing the candidate's professional field work record to obtain parsing results, and based on the parsing results, extracting at least one of the task completion rate of the project management platform, the review pass rate of the design portfolio, or the customer feedback score as the professional field contribution data; S13: Collecting the candidate's interaction path data in the career assessment gamification task as the career assessment interaction data; wherein the interaction path data includes task completion time, decision options, and number of error corrections; S14: normalizing the collaboration platform interaction records, the professional field contribution data, and the interaction path data to generate the structured feature matrix.

3. The method according to claim 2, characterized in that The S2 includes: S21: Perform sentiment analysis on the normalized interaction records of the collaboration platform in the structured feature matrix to generate a communication ability index; the communication ability index The calculation formula is: ； in, In order to use the BERT model to analyze the interaction records of the collaboration platform Message content Sentiment polarity value obtained by sentiment analysis; The first Message content Response speed; and is the preset weight coefficient, The total amount of message content in the interaction record of the collaboration platform; S22: Based on the contribution quantification model, calculate the technical leadership score according to the normalized professional field contribution data in the structured feature matrix; S23: Performing vector fusion on the communication ability index, the technical leadership score, and the normalized career assessment interaction data in the structured feature matrix to generate the implicit skill vector.

4. The method according to claim 1, characterized in that The S3 includes: S31: construct a skill relationship graph based on the historical collaboration data, and the edge weights of the skill relationship graph Calculated by the following formula: ； in, Members of the target team With members The number of collaborative projects For Members The total number of items; the edge weight Indicates the members of the target team With members The closeness of collaboration between them; S32: Mapping the implicit skill vector to a candidate node of the skill relationship graph, and calculating the cosine similarity between the candidate node and the team node; wherein the team node is constructed by the historical collaboration data, and each team node corresponds to a skill feature vector of a team member; and the cosine similarity represents the skill similarity between the candidate and the members of the target team; S33: Calculate the compatibility score based on the cosine similarity and the edge weight; the compatibility score The calculation formula is: ； in, is the cosine similarity, representing the difference between the candidate and the target team. The skill similarity of each member; is the number of members in the target team.

5. The method according to any one of claims 1 to 4, characterized in that The S4 includes: S41: Based on the compatibility score, define the enterprise utility function, the team utility function and the personal utility function; S42: Based on the enterprise utility function, the team utility function and the personal utility function, a multi-objective optimization problem is constructed; the expression of the multi-objective optimization problem is: ； in, is the enterprise utility function, is the team utility function, is the personal utility function, , and The dynamic weights are obtained through training of historical HR decision data; S43: solving the Pareto optimal solution set by using the NSGA-II algorithm for the multi-objective optimization problem and the preset constraints; S44: Selecting the solution with the highest comprehensive score from the Pareto optimal solution set as the balanced matching solution.

6. The method according to claim 5, characterized in that The S5 includes: S51: Outputting a recommended job list and an interpretability report of the balanced matching solution, wherein the interpretability report includes a matching degree, a team compatibility score, and a salary range; S52: If the candidate refuses the recommendation, collect the rejection reason label, and update the dynamic weight according to the rejection reason label; S53: Regenerate the balanced matching solution according to the updated dynamic weight.

7. An employment matching system based on intelligent data analysis, characterized in that: The system comprises: A data collection and processing module is used to collect the multimodal implicit skill data of candidates and perform structured processing on the multimodal implicit skill data to generate a structured feature matrix; the multimodal implicit skill data includes collaborative platform interaction records, professional field contribution data, and career assessment interaction data; A feature vector generation module, used to generate an implicit skill vector based on the structured feature matrix through a sentiment analysis algorithm and a contribution quantification model; a compatibility scoring module, configured to calculate a compatibility score between the candidate and the target team based on the implicit skill vector and historical collaboration data of members of the target team; A matching scheme generation module, for generating an equilibrium matching scheme through a multi-objective game optimization algorithm based on the enterprise utility function, the team utility function, the individual utility function and the compatibility score; The feedback and iteration module is used to obtain the candidate's feedback results on the balanced matching solution, and dynamically iterate the balanced matching solution according to the feedback results.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images