Multi-source data driven talent potential analysis and demand matching method and system

Through the multi-source data-driven talent potential analysis and demand matching method, the problem of difficult to analyze the dynamic growth potential and development trends of talents in the existing technology is solved, and a comprehensive assessment and efficient matching of candidates is achieved, and the recruitment success rate is improved.

CN119027077AInactive Publication Date: 2024-11-26ZHEJIANG SUPENG INFORMATION TECHNOLOGY CO LTD

Patent Information

Application Number
CN202411137759.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-16
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing talent recommendation method based on big data analysis is difficult to effectively analyze the dynamic growth potential and development trend of talents, making it difficult to discover high-potential talents.

Method used

Through the multi-source data-driven talent potential analysis and demand matching method, the initial growth potential level of candidates is set, real-time active data is collected, potential level is adjusted, multi-dimensional potential data is obtained, features are extracted, high-order potential features are generated, and growth potential evaluation is constructed. The model combines comprehensive quality scores and development momentum scores to generate a portrait of the candidate's growth potential, and match the job requirements to recommend candidates.

Benefits of technology

A comprehensive and three-dimensional assessment of candidates has been achieved, the accuracy and efficiency of talent recommendations have been improved, the recommended talents have been highly matched with positions, and the recruitment success rate has been significantly improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119027077A_ABST
    Figure CN119027077A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of big data analysis talent recommendation, and discloses a multi-source data driven talent potential analysis and demand matching method and system, and the method comprises the steps: setting the initial growth potential level of a candidate, collecting the real-time active data of the candidate, and adjusting the initial growth potential level according to the real-time data. Multi-dimensional potential data of candidates are obtained from a plurality of data sources, feature extraction and combined association are carried out, comprehensive quality scores and growth potential scores of the candidates are generated, a growth potential evaluation model is constructed, and growth potential portraits of the candidates are generated in combination with the comprehensive quality scores. And when an enterprise puts forward a talent demand, the candidate matched with the post requirement is retrieved from the portrait library, and a recommendation report is generated. Through association and analysis of multi-source heterogeneous data, accurate talent recommendation is provided, and the efficiency and quality of enterprise recruitment are significantly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of big data analysis talent recommendation technology, and more specifically, to a talent potential analysis and demand matching method and system driven by multi-source data. Background Art

[0002] With the continuous development of big data analysis technology, using big data to mine talent value and build talent portraits has become a new trend in human resource management. Existing technologies have begun to apply big data analysis methods to support talent evaluation and recommendation.

[0003] The patent with publication number CN105160498A discloses a method and system for calculating personnel value based on big data. The method extracts personnel data from a large number of resumes, scores the data according to a scoring system, selects key fields by constructing a sparse group structure penalty function with an organizational structure prior, establishes a regression model with the goal of predicting salary, and finally calculates the value score of the personnel. The invention can extract valuable elements from complex resume data and price talents with standardized variable calculations. The patent with publication number CN113656687A discloses a method for constructing a teacher portrait based on teaching and research data. The method obtains corresponding data from a teacher feature model, a subject teaching model, a professional training model, and a teaching and research achievement model to construct teacher portrait data containing multiple dimensions. The invention uses multi-source heterogeneous teaching and research data to accurately portray teachers, effectively improving the richness of teacher portraits.

[0004] At present, the existing talent recommendation methods based on big data analysis are insufficient in analyzing the dynamic growth potential and development trends of talents, making it difficult to discover and recommend high-potential talents. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a talent potential analysis and demand matching method and system driven by multi-source data.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A multi-source data-driven talent potential analysis and demand matching method, including:

[0008] Set the candidate's initial growth potential level, collect the candidate's real-time active data, adjust the initial growth potential level according to the real-time active data, obtain the candidate's growth potential level, and set the collection parameters of different data sources based on the growth potential level;

[0009] Acquire multidimensional potential data of candidates from multiple data sources according to acquisition parameters, perform feature extraction on the multidimensional potential data to form multidimensional potential features of candidates, perform combined correlation extraction on the multidimensional potential features of candidates to generate high-order potential features;

[0010] Generate a comprehensive quality score for the candidate based on multi-dimensional potential characteristics; construct a candidate growth potential assessment model, and obtain the candidate's growth potential score and development momentum score based on the candidate growth potential assessment model; combine the growth potential score, comprehensive quality score, and development momentum score to generate a candidate's growth potential portrait, and store it in the candidate portrait library;

[0011] When an enterprise raises talent needs, candidates matching the job requirements are retrieved from the candidate portrait library; candidates matching the job requirements are screened, and a candidate recommendation report is generated, presenting key recommendation information of the candidates.

[0012] Further, the method comprises:

[0013] Collect the basic attribute data of the candidates, and set the initial growth potential level of the candidates according to the basic attribute data of the candidates;

[0014] Collect real-time active data of candidates;

[0015] Calculate the growth potential score based on the candidate's real-time active data and initial growth potential level; set the numerical range of the growth potential level, and determine the candidate's final growth potential level based on the numerical range of the growth potential score;

[0016] Set the collection parameters for different data sources based on the growth potential level; the collection parameters include data source priority, data collection breadth and data collection depth;

[0017] For candidates with high growth potential, increase the priority of data sources and expand the breadth and depth of data collection; for candidates with low growth potential, reduce the priority of data sources and narrow the breadth and depth of data collection.

[0018] Furthermore, the method for calculating the growth potential score according to the candidate's real-time activity data and initial growth potential level includes:

[0019]

[0020] Among them, L f is the growth potential score, L0 is the initial growth potential level, S i is the score of real-time active data, W i is the basic weight of real-time active data, t i For real-time active data Si The timestamp of , t0 is the current time, α is the adjustment coefficient, n is the number of types of real-time active data, γ is the time decay factor, i is the index of the real-time active data type, and e is the base of the natural logarithm.

[0021] Further, the method comprises:

[0022] By using the common fields between data sources, the original data of the same candidate from different data sources are associated and matched to form a unified data view of the candidate;

[0023] Based on multi-dimensional potential data, we can mine the association rules between different candidates and form an association network between candidates;

[0024] Integrate the unified data view of the same candidate and the association network between different candidates to build an association map of multi-source heterogeneous data from different data sources;

[0025] Taking the candidate as the central node, extract the characteristics of the candidate's neighbor nodes in the associated network to form the candidate's network potential characteristics;

[0026] Combine and correlate the multi-dimensional potential characteristics of candidates, explore the interaction patterns and influence mechanisms between the multi-dimensional potential characteristics, and form combined potential characteristics;

[0027] The network potential characteristics and combined potential characteristics of the candidates are integrated to form high-order potential characteristics.

[0028] Further, the method comprises:

[0029] Multidimensional potential characteristics include educational characteristics, occupational characteristics, and skill characteristics;

[0030] Based on the candidate's educational characteristics, the candidate's educational quality score is formed; based on the candidate's professional characteristics, the candidate's professional quality score is formed; based on the candidate's skill characteristics, the candidate's skill quality score is formed; the candidate's comprehensive quality score is formed by combining the candidate's educational quality score, professional quality score and skill quality score;

[0031] Construct a candidate growth potential assessment model to obtain the candidate's growth potential score and development momentum score based on the candidate's growth potential level, real-time active data, multi-dimensional potential characteristics, high-level potential characteristics and growth potential assessment model;

[0032] The candidate's total growth potential score is formed based on the candidate's growth potential score, comprehensive quality score and development momentum score;

[0033] Extract the candidate's key tags and keywords from the candidate's multi-dimensional potential characteristics and high-order potential characteristics to form a keyword portrait of the candidate.

[0034] Generate a candidate's growth potential profile based on the total growth potential score and keyword profile;

[0035] Furthermore, the method of forming a candidate's educational quality score based on the candidate's educational characteristics includes:

[0036]

[0037] Among them, E is the candidate's professional quality score, D j is the grade or score of the jth educational characteristic, T j is the duration of the jth educational feature, R j is the correlation of the jth educational characteristic, w j is the weight of the jth educational feature, λ is the adjustment coefficient, ε is the time decay factor, Δt j is the time difference from the current time, m is the total number of educational features, and j is the index of the educational feature.

[0038] Further, the method comprises:

[0039] Collect and analyze the talent needs raised by the enterprise, extract key job requirements and qualifications, and form a structured job profile;

[0040] Perform feature representation and vectorization on the job profile and the candidate's growth potential profile to generate a job profile feature vector and a growth potential profile feature vector;

[0041] Based on the job profile feature vector and the growth potential profile feature vector, the matching index between the two is calculated; a matching threshold is set, and if the matching index is greater than the matching threshold, the candidate is determined to be suitable for the job;

[0042] Sort the matching indexes of candidates suitable for the position and select the top N candidates with the highest matching index as the recommendation results; generate a recommendation report for the candidates, presenting key recommendation information of the candidates.

[0043] A multi-source data-driven talent potential analysis and demand matching system, which is used to implement the multi-source data-driven talent potential analysis and demand matching method, and the system includes:

[0044] Growth potential assessment module: used to collect basic attribute data of candidates and set the initial growth potential level; collect real-time active data of candidates and dynamically adjust the growth potential level; set the collection parameters of different data sources based on the growth potential level;

[0045] Multi-dimensional potential data collection module: used to obtain the multi-dimensional potential data of candidates from multiple data sources according to the collection parameters, pre-process and structure the multi-dimensional potential data, and extract the multi-dimensional potential characteristics of the candidates;

[0046] Association graph construction module: used to associate and match records from different data sources of the same candidate to form a unified view, mine association rules between different candidates, form an association network, integrate the unified view and association network, and construct an association graph for multi-source heterogeneous data;

[0047] High-order potential feature extraction module: used to extract the network potential features of candidates based on the association map, explore the combined associations between the multi-dimensional potential features of candidates, form combined potential features, fuse network potential features and combined potential features, and generate high-order potential features;

[0048] Portrait generation module: used to generate comprehensive quality scores of candidates based on multi-dimensional potential characteristics, build a growth potential assessment model, generate growth potential scores and development momentum scores, and combine comprehensive quality scores, growth potential scores and development momentum scores to generate growth potential portraits of candidates and store them in the candidate portrait library;

[0049] Talent recommendation module: used to retrieve candidates matching the job requirements from the candidate portrait library; screen the candidates matching the job requirements, generate a candidate recommendation report, and present key recommendation information of the candidates.

[0050] An electronic device includes a memory, a central processing unit, and a computer program stored in the memory and executable on the central processing unit. When the central processing unit executes the computer program, the above-mentioned multi-source data-driven talent potential analysis and demand matching method is implemented.

[0051] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the above-mentioned multi-source data-driven talent potential analysis and demand matching method.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] Through big data analysis technology, the present invention can obtain multi-dimensional potential data of candidates from multiple data sources, extract features and combine them to generate comprehensive quality scores and growth potential scores. This can ensure that the recommended talents are highly matched with job requirements and significantly improve the success rate of recruitment.

[0054] The present invention collects real-time active data of candidates and dynamically adjusts the growth potential level of candidates based on these data, so that enterprises can obtain the latest status and potential assessment of candidates in a timely manner, thereby making more accurate recruitment decisions and avoiding missing out on outstanding talents due to information lag.

[0055] The present invention utilizes the association and analysis of multi-source heterogeneous data, builds an association map, and deeply explores the association rules and interaction patterns between candidates to form high-level potential characteristics. This method can comprehensively and three-dimensionally evaluate the potential and quality of candidates and provide a richer and more detailed talent portrait.

[0056] The present invention constructs a candidate growth potential assessment model, and generates a growth potential score and a development momentum score based on the candidate's growth potential level, real-time active data, multi-dimensional potential characteristics, and high-order potential characteristics. This personalized assessment model can perform differentiated assessments based on the characteristics of different candidates, ensuring the scientificity and objectivity of the assessment results.

[0057] The present invention performs feature representation and vectorization on the job profile and the candidate's growth potential profile, calculates the matching index between the two, and sets a matching threshold to ensure that the recommended candidates highly match the job requirements. At the same time, the matching index of the candidates suitable for the job is sorted, and the top N candidates with the highest matching index are selected, which greatly improves the efficiency and accuracy of talent matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 A flowchart of a talent potential analysis and demand matching method driven by multi-source data in the present invention;

[0060] Figure 2 A flow chart of a method for establishing a correlation map between different data sources to generate high-order potential features in the talent potential analysis and demand matching method driven by multi-source data of the present invention;

[0061] Figure 3 A method flow chart of the multi-source data-driven talent potential analysis and demand matching method based on the association graph of the present invention;

[0062] Figure 4 A flow chart of a method for generating a candidate's growth potential portrait in the talent potential analysis and demand matching method driven by multi-source data of the present invention;

[0063] Figure 5 This is a functional module diagram of the multi-source data-driven talent potential analysis and demand matching system in the present invention. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] Example 1

[0066] See also Figure 1 As shown, this embodiment provides a talent potential analysis and demand matching method driven by multi-source data, including:

[0067] Step S1000, setting the initial growth potential level of the candidate, collecting the real-time active data of the candidate, adjusting the initial growth potential level according to the real-time active data, obtaining the growth potential level of the candidate, and setting the collection parameters of different data sources based on the growth potential level;

[0068] Furthermore, step S1000 includes:

[0069] Step S1100, collecting basic attribute data of the candidate, and setting the candidate's initial growth potential level according to the candidate's basic attribute data;

[0070] Furthermore, step S1100 includes:

[0071] Step S1110, collecting basic attribute data of the candidate;

[0072] Step S1120, setting the initial growth potential level classification rules according to the basic attribute data of the candidates;

[0073] Step S1130, evaluating the initial growth potential level of the candidate according to the initial growth potential level classification rule, and generating the initial growth potential level of the candidate;

[0074] Specifically, the initial growth potential level classification rules are set according to the basic attribute data of the candidates, such as education, major, years of work, etc. The classification rules can be customized according to the characteristics of the industry and the needs of the enterprise to ensure the rationality and operability of the level classification. The basic attribute data of the candidates are collected and entered into the system through manual reporting, data import, etc. The data collection process requires necessary data cleaning and formatting to ensure the accuracy and consistency of the data. The level assessment process can be automatically executed by the rule engine, or combined with manual review to improve the reliability of the assessment results. The setting of the initial growth potential level provides a benchmark for subsequent dynamic evaluation. By comprehensively considering the basic attributes of the candidates such as education, major, years of work, etc., their growth potential can be preliminarily judged. For example, candidates with high education, relevant majors, and rich work experience usually have higher growth potential and can be classified as higher initial levels. Candidates with lower education, irrelevant majors, and lack of work experience may have lower initial levels. The reasonable setting of the initial level can provide a reference for subsequent level adjustment and potential assessment, and also help enterprises quickly screen out high-potential candidates and improve the efficiency of talent recruitment.

[0075] Step S1200, collecting real-time active data of candidates;

[0076] Specifically, real-time active data includes participating in projects, publishing papers, receiving honors, etc. The real-time active data of candidates reflects their latest work performance and growth, and is an important basis for evaluating their growth potential. For example, candidates who frequently participate in difficult projects, continuously produce high-quality papers, and continuously receive industry honors usually show strong learning ability and development potential, and their growth potential level can be improved. On the other hand, candidates who have no new projects for a long time, stagnant paper output, and slow honor acquisition may have a decline in their growth potential. The collection of real-time active data can help companies keep abreast of the latest status of candidates, dynamically evaluate their growth potential, and discover outstanding talents with outstanding potential or obvious improvement. At the same time, it can also identify talents with stagnant development or declining potential, and provide a reference for talent management decisions.

[0077] Step S1300, calculating the growth potential score according to the candidate's real-time active data and initial growth potential level; setting a numerical range for the growth potential level, and determining the candidate's final growth potential level according to the numerical range in which the growth potential score is located;

[0078] Based on the candidate's real-time activity data and initial growth potential level, the growth potential score is calculated in the following ways:

[0079]

[0080] in:

[0081] L f : Growth potential score;

[0082] L0: initial growth potential level;

[0083] S i : The score of real-time active data can be scored according to the candidate's specific performance in a certain aspect. For example, publishing a high-quality paper may score 10, and participating in a difficult project may score 20;

[0084] W i : The basic weight of real-time active data. The weight is set according to the impact of different types of active data on growth potential. For example, the weight of participating in a difficult project may be greater than the weight of publishing a paper;

[0085] t i :Real-time active dataS i Timestamp

[0086] t0: current time;

[0087] α: Adjustment coefficient, used to adjust the impact of real-time activity data on the growth potential level.

[0088] n: the number of types of real-time active data;

[0089] γ: time decay factor, used to indicate the importance of data freshness. A higher γ value indicates a higher emphasis on recent data.

[0090] i: index of real-time active data types;

[0091] e: base of natural logarithm;

[0092] The adjustment coefficient α is determined by analyzing historical data to determine the optimal adjustment coefficient for the impact of real-time active data on the growth potential level. It can be calibrated using methods such as regression analysis. The basic weight of real-time active data W i It can be set according to the specific needs of the enterprise and the characteristics of the industry, and can be determined by historical data analysis and other methods. Through the time decay factor γ, the formula can reflect the latest status of the candidate in real time, giving priority to recent active data. Introducing the logarithmic function log(S i +1) and multiple real-time active data factors (such as project participation, paper publication, honors received, etc.), this formula can more comprehensively and accurately assess the growth potential of candidates.

[0093] Set the numerical range of the growth potential level. For example, the growth potential level can be divided into levels 1-5, and the corresponding numerical range can be set as:

[0094] Level 1: L f<1.5;

[0095] Level 2: 1.5 ≤ L f <2.5;

[0096] Level 3: 2.5 ≤ L f <3.5;

[0097] Level 4: 3.5 ≤ L f <4.5;

[0098] Level 5: L f ≥4.5;

[0099] The division of numerical ranges can be adjusted according to the actual situation and management needs of the enterprise to better match the actual situation of talent development. Setting a reasonable numerical range can not only reflect the differences in the growth potential of candidates, but also avoid the increase in management complexity caused by overly detailed level division.

[0100] Combining a candidate's real-time active data with their initial growth potential level can more accurately and comprehensively assess the candidate's development potential. The initial growth potential level reflects the candidate's basic qualities and development baseline, while the real-time active data reflects the candidate's immediate performance and growth rate. The organic combination of the two can dynamically present the candidate's development and change trends. For example, if a candidate with a high initial growth potential level continues to perform well in the real-time active data, his or her growth potential level can be further improved; conversely, if the performance continues to be sluggish, the level may be lowered. For candidates with a low initial growth potential level, if they perform well in the real-time active data, their level can be greatly improved, achieving overtaking on the curve. The dynamic level adjustment mechanism makes growth potential assessment more flexible and timely, helping companies to promptly discover fast-growing outstanding talents, or formulate improvement plans for talents with stagnant development, thereby optimizing the construction of talent echelons.

[0101] Step S1400, setting collection parameters of different data sources based on the growth potential level; the collection parameters include data source priority, data collection breadth and data collection depth;

[0102] Further, step S1400 includes:

[0103] Step 1410, setting collection parameters of different data sources according to the candidate's latest growth potential level;

[0104] Step S1420: for candidates with high growth potential, the data source priority is increased, and the breadth and depth of data collection are expanded; for candidates with low growth potential, the data source priority is decreased, and the breadth and depth of data collection are reduced;

[0105] Step S1430, updating the collection parameters of the data source in real time according to the candidate's latest growth potential level;

[0106] Specifically, according to the candidate's latest growth potential level, the collection parameters of different data sources such as education background, career experience, and skill tags are set, and the hierarchical settings are made to configure data collection differently to improve the pertinence and efficiency of data collection. For example, for candidates with high growth potential levels, detailed information such as academic performance, research projects, and academic papers in their education background can be deeply mined to fully understand their knowledge reserves and learning abilities; in terms of career experience, key indicators such as the scale, difficulty, and performance of their leading projects can be captured to evaluate their work ability and performance; in the skill tag dimension, information such as the skill training, skill certification, and skill competitions they have participated in can be obtained to judge their skill level and development potential. For candidates with lower growth potential levels, the dimensions and content of data collection can be relatively simplified, and the focus can be on the core information related to their current positions, saving collection resources. The dynamic adjustment of collection parameters makes data collection more flexible and changeable, which can not only fully mine the rich information of high-potential talents, but also save the collection cost of low-potential talents, and achieve the optimal allocation of human resources.

[0107] Step S2000, acquiring multidimensional potential data of the candidate from multiple data sources according to the acquisition parameters, performing feature extraction on the multidimensional potential data to form multidimensional potential features of the candidate, performing combined correlation extraction on the multidimensional potential features of the candidate to generate high-order potential features;

[0108] Furthermore, step S2000 includes:

[0109] Step S2100, acquiring multi-dimensional potential data of the candidate from multiple data sources according to the acquisition parameters;

[0110] Furthermore, step S2100 includes:

[0111] Step S2110, extracting the candidate's original data from multiple data sources according to the data source priority;

[0112] Step S2120, preprocessing the original data extracted from the multiple data sources to generate preprocessed original data;

[0113] Step S2130, performing structured storage on the pre-processed raw data to form multi-dimensional potential data of the candidate; the multi-dimensional potential data includes educational background data, professional experience data and skill label data;

[0114] Specifically, according to the set data source priority, connect internal and external data sources such as education background, professional experience, skill tags, etc. in turn to extract the candidate's original data. Perform preprocessing operations such as cleaning, deduplication, and formatting on the original data to improve data quality. The preprocessing process can use regular expressions, data mapping and other technologies to achieve automatic data processing and conversion. The preprocessed raw data is structured and stored according to the predefined data model to form the candidate's multi-dimensional potential data. The data model needs to comprehensively consider factors such as the logical relationship of the data, query efficiency, and storage cost, and reasonably design data tables and fields.

[0115] Obtaining multi-dimensional potential data of candidates from multiple data sources can comprehensively characterize the characteristics of candidates in terms of knowledge, experience, ability, etc., and provide a rich information basis for subsequent comprehensive analysis. For example, obtaining information such as the candidate's academic qualifications, major, school ranking, and academic performance from the education background data source reflects their knowledge reserves and learning ability; obtaining information such as the candidate's years of work, company he / she works for, main responsibilities, and project experience from the career experience data source reflects his / her practical experience and work ability; obtaining information such as the candidate's professional skills, tool usage, language ability, and certificate certification from the skill label data source reflects his / her professional level and comprehensive quality. The acquisition and integration of multi-dimensional potential data can break through the limitations of a single data source, reduce information blind spots, and achieve a three-dimensional, multi-angle analysis of candidates, providing a more comprehensive and accurate basis for talent assessment and decision-making.

[0116] Step S2200, extracting features from the multidimensional potential data to form multidimensional potential features of the candidate, the multidimensional potential features including education features, occupation features and skill features;

[0117] Specifically, for educational background data, key features such as the candidate's academic level, graduation school, professional field, academic performance, and awards are extracted to form an education feature vector. Feature extraction can use keyword matching, rule parsing and other technologies to automatically identify and extract key information. For professional experience data, key features such as the candidate's years of work, company, job responsibilities, project experience, and performance are extracted to form a professional feature vector. Feature extraction can be combined with external knowledge sources such as corporate knowledge graphs and job profiles to enhance the semantic understanding and extraction accuracy of features. For skill label data, key features such as the candidate's professional skills, tool use, language ability, and certificate certification are extracted to form a skill feature vector. Feature extraction can use natural language processing technologies such as named entity recognition and keyword extraction to automatically identify and classify skill labels.

[0118] Feature extraction of multidimensional potential data can transform heterogeneous data into structured and standardized feature representations, which is convenient for subsequent data analysis and mining. For example, extracting key features such as the candidate's education level, graduation school, and professional field from the education background data can characterize the candidate's education level and professional background; extracting key features such as the candidate's years of work, company, and job responsibilities from the career experience data can characterize the candidate's work experience and career development trajectory; extracting key features such as the candidate's professional skills, tool usage, and language ability from the skill label data can characterize the candidate's professional ability and comprehensive quality. Through feature extraction, the candidate's multi-dimensional data can be converted into feature vectors that can be understood and processed by computers, laying the foundation for subsequent data association analysis, similarity calculation, and talent portrait construction. At the same time, feature extraction can also filter out redundant and noise information to improve the efficiency and accuracy of data processing.

[0119] Step S2300, establishing a correlation map between different data sources based on multi-dimensional potential data;

[0120] Furthermore, if Figure 2 As shown, step S2300 includes:

[0121] Step S2310, using the common fields between data sources, the original data of the same candidate from different data sources are associated and matched to form a unified data view of the candidate;

[0122] Step S2320, based on the multi-dimensional potential data, mining association rules between different candidates to form an association network between the candidates;

[0123] Step S2330, integrating the unified data view of the same candidate and the association network between different candidates to construct an association graph of multi-source heterogeneous data from different data sources;

[0124] Specifically, by using the common fields between data sources, such as name, ID number, email address, etc., records from different data sources of the same candidate are associated and matched to form a unified data view of the candidate. Association matching can use technologies such as exact matching, fuzzy matching, and approximate matching to improve the accuracy and fault tolerance of matching. Based on information such as candidates' educational background, professional experience, and skill tags, association rules between different candidates are mined, such as graduating from the same college, working in the same company, and having the same skills, to form an association network between candidates. Association rule mining can use data mining algorithms such as frequent item set analysis and correlation calculation to discover implicit connections between candidates. The unified data view of the same candidate and the association network between different candidates are integrated to construct an association map of multi-source heterogeneous data. The association map combines the individual characteristics and group relationships of candidates, and can characterize the association patterns and interaction rules between candidates from multiple angles and levels.

[0125] Establishing associations between different data sources can break through the limitations of data silos and achieve data interconnection and fusion analysis. For example, by associating the educational background, career experience, skill tags and other data of the same candidate, we can fully understand the candidate's growth trajectory and development process, discover the correlation between his educational experience and career development, and reveal the matching degree between his skill level and work performance, so as to portray the candidate's career portrait in a more three-dimensional and dynamic way. For example, by mining the association rules between different candidates, we can discover the group characteristics and behavior patterns of candidates, such as the distribution of educational backgrounds of excellent candidates, the talent preferences of different companies, the role and influence of key skills, etc., to provide valuable group insights for talent selection and training. In addition, association analysis can also reveal the social network and influence between candidates, such as the activity and centrality of candidates in alumni networks, workplace circles, and technical communities, reflecting their personal resources and social capital. Data association analysis expands the evaluation of candidates from static, isolated individual analysis to dynamic, associated network analysis, greatly broadening the breadth and depth of talent insights.

[0126] Step S2400, based on the association graph, performing combined association extraction on the multi-dimensional potential features of the candidate to generate high-order potential features;

[0127] Furthermore, if Figure 3 As shown, step S2400 includes:

[0128] Step S2410, taking the candidate as the central node, extracting the characteristics of the candidate's neighbor nodes in the associated network to form the candidate's network potential characteristics;

[0129] Step S2420, combining and associating the multi-dimensional potential characteristics of the candidate, exploring the interaction patterns and influence mechanisms between the multi-dimensional potential characteristics, and forming a combined potential characteristic;

[0130] Step S2430, fusing the candidate's network potential features and combined potential features to form a high-order potential feature;

[0131] Specifically, taking the candidate as the central node, the characteristics of the candidate's neighbor nodes in the association network are extracted, such as the career development of the alumni of the same school as the candidate, the ability level of the technical experts in the same field as the candidate, etc., to form the candidate's network potential characteristics. The network potential characteristics reflect the role and status of the candidate in the association network, and are an effective supplement to their multidimensional potential characteristics. Using association rules, the multidimensional potential characteristics of the candidate are combined and associated, and the interaction patterns and influence mechanisms between the characteristics are explored to form combined potential characteristics. For example, the correlation between the candidate's academic performance and his career development is analyzed to explore the influence of educational background on career success; for another example, the matching degree between the candidate's skill level and his project experience is analyzed to explore the role of practical experience in promoting ability improvement. The candidate's network potential characteristics and combined potential characteristics are integrated to form high-order potential characteristics, which comprehensively characterize the candidate's comprehensive quality and development potential. High-order potential characteristics integrate the static attributes and dynamic relationships, explicit characteristics and implicit characteristics of the candidate, and can more accurately and comprehensively evaluate the candidate's professional competence and development prospects.

[0132] By using the association graph to extract high-level potential features, we can explore the implicit features and potential capabilities of candidates from multiple dimensions and achieve a deep portrait of the candidates. For example, by analyzing the activity and influence of candidates in the alumni network, we can evaluate their social skills and leadership potential; by analyzing the contribution and recognition of candidates in the technical community, we can evaluate their professional capabilities and innovation potential; by analyzing the key nodes and bridges of candidates in the flow of talents across fields and enterprises, we can evaluate their adaptability and development potential. High-level potential feature extraction breaks through the isolated evaluation of candidates, fully considers the role and influence of candidates in the association network, and explores the hidden advantages and potential value of candidates, making talent evaluation more comprehensive, accurate and in-depth. At the same time, high-level potential feature extraction also provides directional guidance for candidates' career development planning and capacity improvement, such as discovering the shortcomings and weaknesses of candidates in certain areas and formulating targeted improvement plans for them; for example, discovering the cross-border potential and compound advantages of candidates to open up a broader career development path for them.

[0133] Step S3000, based on the multi-dimensional potential characteristics, generate a comprehensive quality score of the candidate; construct a candidate growth potential assessment model, and obtain the candidate's growth potential score and development momentum score according to the candidate growth potential assessment model; combine the growth potential score, comprehensive quality score and development momentum score to generate a candidate's growth potential portrait, and store it in the candidate portrait library;

[0134] Furthermore, step S3000 includes:

[0135] Step S3100: Based on the candidate's education characteristics, an education quality score of the candidate is formed; based on the candidate's occupational characteristics, an occupational quality score of the candidate is formed; based on the candidate's skill characteristics, a skill quality score of the candidate is formed; and the candidate's education quality score, occupational quality score and skill quality score are combined to form a comprehensive quality score of the candidate;

[0136] Specifically, based on educational characteristics, the candidate's knowledge level, learning ability, professional quality, etc. are evaluated to form the candidate's educational quality score. The evaluation process can adopt scoring models such as weighted summation and logistic regression, reasonably set the weights and evaluation criteria of various indicators, and objectively and impartially evaluate the candidate's educational quality. Based on professional characteristics, the candidate's work experience, performance, leadership ability, etc. are evaluated to form the candidate's professional quality score. The evaluation process can introduce external evaluation mechanisms such as peer review and superior evaluation to comprehensively measure the candidate's professional quality and job competency. Based on skill characteristics, the candidate's professional skills, tool mastery, continuous learning ability, etc. are evaluated to form the candidate's skill quality score. The evaluation process can combine standardized methods such as skill certification and ability assessment to objectively and accurately evaluate the candidate's skill level and development potential. Comprehensive evaluation can adopt multidimensional decision-making methods such as hierarchical analysis method and fuzzy comprehensive evaluation to scientifically and reasonably determine the importance and weight of each quality, and comprehensively and objectively evaluate the candidate's comprehensive quality and development potential.

[0137] Comprehensively considering the multi-dimensional characteristics of candidates for quality assessment can comprehensively and systematically measure the professional competence and development potential of candidates. For example, it is necessary to pay attention to the candidate's educational background and academic level, assess their knowledge reserves and learning ability, and also pay attention to their professional experience and work performance, and assess their practical experience and performance contribution; it is necessary to examine the candidate's professional skills and tool mastery, assess their job matching and operational capabilities, and also consider their willingness to continue learning and potential for self-improvement, and assess their development plasticity and growth space. Comprehensive assessment avoids one-sided judgment of a single dimension, and can more comprehensively and objectively judge the comprehensive quality of candidates, identify their strengths and weaknesses, and provide a scientific reference for subsequent talent selection, training, and motivation. At the same time, comprehensive assessment also provides panoramic feedback for candidates' self-cognition and career development, helping them to clarify their strengths and weaknesses, formulate career development goals and improvement plans, and achieve the optimal growth of their personal careers.

[0138] Methods for forming a candidate's educational quality score based on the candidate's educational characteristics include:

[0139]

[0140] in:

[0141] E: Candidate's professional quality score;

[0142] D j : The grade or rating of the jth educational characteristic;

[0143] T j : duration of the jth educational feature;

[0144] R j : The relevance of the jth educational feature, indicating the matching degree between this educational feature and the candidate’s current position;

[0145] w j : The weight of the jth educational feature;

[0146] λ: adjustment factor, used to prevent division by zero errors, usually a small positive value;

[0147] ε: time decay factor, indicating the importance of data freshness;

[0148] Δt j : The time difference from the current time, that is, the time interval from the occurrence of the jth educational feature to the present;

[0149] m: total number of educational characteristics;

[0150] j: index of educational characteristics;

[0151] The weight of the educational feature is set according to the importance of the educational feature, which can be determined through historical data analysis. For example, the weight of core courses may be higher than that of elective courses. Different educational features have different effects on the educational quality score, and this difference is reflected by the weight. The larger the weight, the greater the impact on the final score. The product of the weight and the score determines the contribution of each educational feature to the overall score. The grade or score of an educational feature refers to the score based on the candidate's performance on a certain educational feature. For example, the course grade can be directly used as the score, and the paper score can be determined by the review score. The grade or score of an educational feature directly reflects the candidate's performance on a certain educational feature. The higher the score, the better the performance of the feature. The duration of an educational feature refers to the duration of recording each educational feature. The shorter the duration, the higher the efficiency and the greater the impact on the score. For example, the duration of a course can be expressed by the length of the semester, and the project completion time can be expressed by the length of time from start to end. The time decay factor can be determined through historical data analysis. A higher ε value indicates a higher degree of attention to recent data. The newer the data, the greater the impact on the score. Through the time decay factor, the recent educational features contribute more to the score. Δt jReflects the time interval between occurrences of educational features. The larger the time difference, the older the feature is and the smaller the impact on the score is. This makes more recent educational characteristics contribute more to the score. R represents the degree of match between educational characteristics and the candidate's current position, which can be determined through expert scoring or historical data analysis. The higher the correlation, the more important the characteristic is to the current position and the greater the impact on the score. For example, if the professional course is highly correlated with the job requirements, the correlation score is high. Using the logarithmic function log(1+R j ) Ensure that the impact of correlation on the score is nonlinear, and avoid that highly correlated but mismatched features have too much impact on the score. The calculation formula of the professional quality score achieves a quantitative assessment of educational quality by weighted scoring of multiple educational projects. At the same time, multiple factors such as the performance, duration, correlation and time decay of educational activities are considered to make the score more comprehensive and objective.

[0152] Methods for forming a candidate's skill quality score based on the candidate's skill characteristics include:

[0153]

[0154] in:

[0155] SI: Skill quality score, which reflects the comprehensive skill quality of the candidate;

[0156] J k : The knowledge level of the kth skill, obtained through skill testing or qualification certification;

[0157] μ k : The weight of the kth skill, reflecting the influence of this skill on the overall quality;

[0158] E′ k : The experience value of the kth skill, evaluated by years of work experience and related project experience;

[0159] T′ k : The number of trainings for the kth skill, obtained through the number of training courses or seminars attended;

[0160] A k : The application frequency of the kth skill, evaluated by the frequency of using the skill in actual work;

[0161] D′ k : The difficulty coefficient of the kth skill is set according to the complexity and market scarcity of the skill;

[0162] R′ k : The correlation coefficient of the kth skill, reflecting the matching degree between the skill and the recruitment position;

[0163] F k : The latest coefficient of the k-th skill, reflecting the latest degree and update frequency of the skill;

[0164] M: The total number of skill items, i.e. the number of skill types the candidate possesses;

[0165] k: index of skill feature;

[0166] μ k According to the needs of the enterprise and the job description, it is determined through historical data analysis to ensure that key skills have a greater impact on the total score and reflect the focus of the enterprise's needs. k It directly reflects the candidate's theoretical mastery of the skill. A high knowledge level means that the candidate has a solid theoretical foundation, which accounts for an important proportion of the skill quality score. Skills with high knowledge levels have a more significant effect on improving the overall score. E′ k It reflects the breadth and depth of the candidate's application of the skill in actual work. Rich experience means that the candidate has applied the skill many times in actual work. The richer the experience, the greater the positive impact on the score. Number of training sessions T′ k Reflects the candidate's willingness to continuously learn and improve skills. Frequent training shows that the candidate continues to improve his or her skills, reflecting learning ability and enterprising spirit. Skills with more training times contribute more to the score. Application frequency A k Evaluate the degree to which the candidate relies on the skill in actual work. Skills with high application frequency indicate that the candidate relies on the skill in daily work. The higher the frequency, the greater the impact of the skill on the candidate's work ability, thus increasing the total score. Difficulty coefficient D' k Consider the complexity of the skills and ensure that difficult skills are given due weight in the scoring. Skills with high difficulty coefficients are usually more challenging and scarce in the market. Therefore, in the scoring formula, skills with high difficulty coefficients account for a larger proportion of the total score, reflecting the candidate's ability level. k Based on the job requirements and skill matching assessment, the relevance coefficient ensures that the score is closely aligned with the job requirements. Skills with high relevance are scored higher, ensuring that the candidate has the core competencies required for the job. k Based on the skill update frequency and market change assessment, it reflects the updating frequency and timeliness of the candidate's skills. Skills with high up-to-dateness indicate that the candidate keeps up with industry development trends, which has a greater positive impact on the score.

[0167] By assigning specific values ​​to each skill, the formula can objectively evaluate the skill quality of candidates and reduce subjective bias. The formula takes into account multiple dimensions such as knowledge level, experience, training, application frequency, difficulty, relevance and recency, providing a comprehensive evaluation. Parameters can be dynamically adjusted according to industry changes and corporate needs to ensure the timeliness and accuracy of the evaluation results. Weights and coefficients can be adjusted according to different positions and corporate needs to provide personalized evaluation plans. Through a quantitative scoring system, high-quality candidates can be quickly screened out and recruitment efficiency can be improved. Companies can formulate targeted training and development plans based on the scoring results to improve the overall quality of employees.

[0168] Step S3200, constructing a candidate growth potential assessment model, and obtaining the candidate's growth potential score and development momentum score based on the candidate's growth potential level, real-time active data, multi-dimensional potential characteristics, high-order potential characteristics and growth potential assessment model;

[0169] Further, step S3200 includes:

[0170] Step S3210, collecting the candidate's historical growth data, including past growth potential level changes, real-time active data accumulation values, multi-dimensional potential characteristics and high-order potential characteristics, as training data for the candidate's growth potential assessment model;

[0171] Step S3220, preprocessing and feature engineering the training data to construct a feature vector of the growth potential assessment model; the feature vector includes growth potential level features, real-time active data features, multi-dimensional potential features and high-order potential features, as well as interaction features and statistical features of these features;

[0172] Step S3230, using a machine learning method to construct a growth potential assessment model, setting two neurons in the output layer of the model to output a growth potential score and a development momentum score respectively; using the feature vector as the input of the model, and training the model parameters through supervised learning;

[0173] Step S3240, using a cross-validation method to evaluate the generalization performance of the model, and tuning the model according to the evaluation results, optimizing the model's hyperparameters, and improving the model's prediction accuracy and stability;

[0174] Step S3250, applying the trained growth potential assessment model to the feature data of the current candidate, the model maps the input growth potential level, real-time activity data, multi-dimensional potential features and high-order potential features into a growth potential score and a development momentum score;

[0175] Past growth potential level changes refer to changes in the growth potential level of candidates over the past period of time. These changes can reflect the dynamic trends of candidates' abilities and performance. Real-time active data accumulation refers to the accumulation and summary of various real-time active data of candidates over a certain period of time. Specifically, real-time active data accumulation can be understood as the historical records and accumulation of activities such as candidates' participation in projects, publication of papers, and honors. By accumulating these data, the overall activity and performance of candidates over a period of time can be reflected. Interactive features are new features generated by combining existing features. These combinations can capture the interactions between features. Common interactive feature generation methods include feature product, feature quotient, feature sum, feature difference, and polynomial features. By constructing interactive features, complex relationships between original features can be captured, and the expression ability of the model can be improved. Statistical features are statistics calculated from original data. These statistics can summarize certain properties of data. Common statistical features include mean, variance, standard deviation, maximum value, and minimum value.

[0176] The growth potential score and development momentum score of candidates are predictive judgments on their career development prospects. The growth potential score focuses on evaluating the long-term development potential of candidates, taking into account the basic qualities, knowledge and skills, practical experience, development speed and other factors of candidates, reflecting the possibility of candidates growing into high-end talents in the future. The development momentum score focuses on evaluating the short-term development trend of candidates, focusing on the outstanding performance and progress of candidates in the recent period, reflecting the upward momentum of candidates at the current stage. The two scores judge the development potential of candidates from different time scales, complement each other, and provide a more comprehensive and three-dimensional portrait of talent development. For example, a candidate has a high growth potential score but an average development momentum score, indicating that he has a good long-term development foundation, but his recent performance is mediocre, and he may need to be motivated to release his potential; while the two scores of another candidate are both high, reflecting that he has outstanding development potential in both the long and short term, and is a high-end talent selected from the best. Through a comprehensive analysis of the two scores, it can help companies fully grasp the current status and future trends of candidates' career development, and provide a scientific basis for talent selection and training. At the same time, dynamically updating the candidate's potential score can also continuously track the candidate's growth and changes, adjust the talent development strategy in a timely manner, and maximize the stimulation and utilization of the candidate's potential value.

[0177] Step S3300, combining the growth potential score, comprehensive quality score and development momentum score to generate a growth potential portrait of the candidate, and storing it in a candidate portrait library to form a three-dimensional portrait that combines the static description and dynamic description of the candidate.

[0178] Furthermore, if Figure 4As shown, step S3300 includes:

[0179] Step S3310, forming a total growth potential score for the candidate based on the candidate's growth potential score, comprehensive quality score, and development momentum score;

[0180] Step S3320, extracting the key tags and keywords of the candidate from the candidate's multi-dimensional potential characteristics and high-order potential characteristics to form a keyword profile of the candidate;

[0181] Step S3330, generating a growth potential portrait of the candidate based on the growth potential total score and the keyword portrait;

[0182] According to certain weights, the candidate's growth potential score, comprehensive quality score and development momentum score are weighted and combined to form the candidate's total growth potential score. The weight setting needs to comprehensively consider the importance and reliability of each score, and can be reasonably determined by methods such as data analysis.

[0183] Key tags are used to identify important characteristics of candidates in different dimensions. These tags are usually presented in a brief and summarized form to facilitate quick identification and classification. Key tags can include the following:

[0184] Education background labels: for example, "PhD", "Master", "graduated from a prestigious university", "computer science major", etc.

[0185] Career experience tags: for example, “10 years of work experience”, “project manager”, “technical supervisor”, “work experience in multinational companies”, etc.

[0186] Skill tags: for example, "Python Programming", "Data Analysis", "Machine Learning", "Project Management Certification", etc.

[0187] Keywords are extracted from the candidate's multidimensional potential characteristics and high-order potential characteristics to more finely describe the candidate's specific abilities and experience. These keywords are usually extracted from text data such as the candidate's resume, project description, paper title, etc. Keywords can include the following:

[0188] Specific technologies and tools: for example, "TensorFlow", "SQL", "Docker", "AWS", etc.

[0189] Project experience: for example, "big data processing project", "AI model development", "cloud computing architecture design", etc.

[0190] Achievements and honors: for example, "Best Employee Award", "Industry Paper Publication", "Patent Invention", etc.

[0191] Keyword portraits summarize the core characteristics of candidates in a concise and clear way, which is easy to quickly identify and remember. Growth potential portraits comprehensively integrate various information of candidates, condense the core values ​​and development characteristics of candidates, and lay a solid foundation for subsequent talent matching and recommendation. Growth potential portraits form a comprehensive and three-dimensional talent portrait by integrating and analyzing the multi-dimensional characteristics of candidates. It not only includes static attributes such as relatively stable educational background and work experience of candidates, but also covers dynamic characteristics such as dynamically changing ability level and development momentum of candidates. It not only evaluates the current comprehensive quality of candidates, but also predicts the future growth potential of candidates, realizing the static and dynamic combination and overall coordination of talent evaluation. For example, the growth potential portrait of a candidate may be: high potential talent / doctoral degree / 5 years of work experience / technical expert / rapid growth / strong innovation ability. This portrait uses keyword tags to intuitively present the candidate's educational background, work experience, professional field, development trend, ability characteristics, etc., and outlines the candidate's talent image in three dimensions, which is convenient for enterprises to quickly understand the candidate's core competitiveness. The application of growth potential portraits can not only help companies identify and discover outstanding talents more accurately and efficiently, but also help candidates more clearly understand their own strengths and weaknesses, formulate reasonable career development plans, and achieve continuous growth and value creation of talents themselves.

[0192] Step S4000: When an enterprise puts forward a talent demand, it retrieves candidates matching the job requirements from the candidate portrait library; screens the candidates matching the job requirements, generates a candidate recommendation report, and presents key recommendation information of the candidates.

[0193] Furthermore, step S4000 includes:

[0194] Step S4100: When an enterprise proposes a talent demand, a candidate matching the job requirements is retrieved from the candidate portrait library;

[0195] Furthermore, step S4100 includes:

[0196] Step S4110, collecting and analyzing the talent needs proposed by the enterprise, extracting key job requirements and qualifications, and forming a structured job profile;

[0197] Step S4120, feature representation and vectorization of the job profile and the candidate's growth potential profile to generate a job profile feature vector and a growth potential profile feature vector;

[0198] Step S4130, based on the position portrait feature vector and the growth potential portrait feature vector, calculate the matching index between the two; set a matching threshold, if the matching index is greater than the matching threshold, then determine that the candidate is suitable for the position;

[0199] Job profiles include the main responsibilities, required skills, experience requirements, and qualifications for the position, and are the basis for talent matching. Feature representation and vectorization are performed on job profiles and candidate profiles to convert unstructured text information into structured feature vectors, which facilitates subsequent similarity calculations and matching evaluations. Common feature representation methods include TF-IDF, Word2Vec, etc., which can be selected based on data characteristics and task requirements. Based on the job profile feature vector and the growth potential profile feature vector, the matching index between the two is calculated using cosine similarity, Euclidean distance and other metrics. The setting of the matching threshold needs to balance the matching accuracy and recall rate, and can be dynamically adjusted according to the business scenario and talent supply and demand. If the matching index is greater than the matching threshold, the candidate is judged to be suitable for the position; if the matching index is less than or equal to the matching threshold, the candidate will be judged to be unsuitable for the position.

[0200] By matching the job profile with the candidate's growth potential profile, intelligent matching and accurate recommendation of job information can be achieved. For example, a software company released a job requirement for a big data development engineer. The job profile extracted keywords: Java development, Spark, more than 3 years of experience, bachelor's degree or above, etc. The system automatically searched the candidate portrait library and found that the growth potential profile of a candidate was labeled as: Java development engineer, Spark technical expert, 5 years of work experience, master's degree, which matched the job profile very well, so it was recommended to the company for further screening and evaluation. The matching of the job profile with the candidate's growth potential profile not only considers the semantic similarity of the two at the keyword level, but also considers the degree of fit between the candidate's academic background, skill level, work experience and other aspects of the job requirements, ensuring the high relevance and matching of the recommendation results. Intelligent matching not only improves the efficiency and accuracy of talent recommendation, but also saves HR's time cost, allowing it to devote more energy to in-depth evaluation and communication and coordination of talents. At the same time, accurate talent recommendation also improves the candidate's job search experience, helps them quickly locate suitable positions, shortens the job search cycle, and achieves efficient matching of people and jobs.

[0201] Step S4200, sort the matching indexes of the candidates suitable for the position, select the top N candidates with the highest matching index as the recommendation results; generate a recommendation report for the candidates, presenting key recommendation information of the candidates.

[0202] Key recommendation information includes but is not limited to the candidate's growth potential profile, total growth potential score, matching index with the position, and recommendation reasons. Sorting and screening the recommendation results can help companies quickly lock in the most suitable high-quality talents and improve the efficiency and success rate of recruitment. For example, for a product manager position, the system screens out the top 10 candidates with the highest matching degree from a large number of candidates through matching degree calculation, and generates a recommendation report for HR reference. The report focuses on presenting information highly relevant to the position, such as the candidate's product design works, user growth performance, product thinking characteristics, etc., and provides comprehensive quality evaluation and development potential prediction of the candidate to help HR fully grasp the situation of the candidate. Based on the system's recommendation sorting, HR can focus on evaluating the most matching candidates without having to screen massive resumes one by one, which greatly improves the pertinence and efficiency of recruitment. At the same time, intelligent screening also ensures the quality of the recommendation results, reduces the screening cost and error rate of HR, and matches the best and most suitable talents for the company.

[0203] Example 2

[0204] This embodiment provides a talent potential analysis and demand matching system driven by multi-source data on the basis of embodiment 1, such as Figure 5 As shown, including:

[0205] Growth potential assessment module: used to collect basic attribute data of candidates and set the initial growth potential level; collect real-time active data of candidates and dynamically adjust the growth potential level; set the collection parameters of different data sources based on the growth potential level;

[0206] Multi-dimensional potential data collection module: used to obtain the multi-dimensional potential data of candidates from multiple data sources according to the collection parameters, pre-process and structure the multi-dimensional potential data, and extract the multi-dimensional potential characteristics of the candidates;

[0207] Association graph construction module: used to associate and match records from different data sources of the same candidate to form a unified view, mine association rules between different candidates, form an association network, integrate the unified view and association network, and construct an association graph for multi-source heterogeneous data;

[0208] High-order potential feature extraction module: used to extract the network potential features of candidates based on the association map, explore the combined associations between the multi-dimensional potential features of candidates, form combined potential features, fuse network potential features and combined potential features, and generate high-order potential features;

[0209] Portrait generation module: used to generate comprehensive quality scores of candidates based on multi-dimensional potential characteristics, build a growth potential assessment model, generate growth potential scores and development momentum scores, and combine comprehensive quality scores, growth potential scores and development momentum scores to generate growth potential portraits of candidates and store them in the candidate portrait library;

[0210] Talent recommendation module: used to retrieve candidates matching the job requirements from the candidate portrait library; screen the candidates matching the job requirements, generate a candidate recommendation report, and present key recommendation information of the candidates.

[0211] Example 3

[0212] This embodiment discloses an electronic device, including a memory, a central processing unit, and a computer program stored in the memory and executable on the central processing unit. When the central processing unit executes the computer program, the multi-source data-driven talent potential analysis and demand matching method provided above is implemented.

[0213] Since the electronic device introduced in this embodiment is an electronic device used to implement the talent potential analysis and demand matching method driven by multi-source data in the embodiment of this application, based on the talent potential analysis and demand matching method driven by multi-source data introduced in the embodiment of this application, the technical personnel of this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not described in detail here. As long as the technical personnel of this field implement the electronic device used by the talent potential analysis and demand matching method driven by multi-source data in the embodiment of this application, it belongs to the scope of protection of this application.

[0214] Example 4

[0215] This embodiment discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed, the above-mentioned multi-source data-driven talent potential analysis and demand matching method is implemented.

[0216] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters, weights and thresholds in the formula are set by technicians in this field according to actual conditions.

[0217] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through a wired network or a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD) or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0218] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention 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 to be beyond the scope of the present invention.

[0219] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0220] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0221] The units 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 distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0222] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0223] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0224] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A talent potential analysis and demand matching method driven by multi-source data, characterized by: include: Set the candidate's initial growth potential level, collect the candidate's real-time active data, adjust the initial growth potential level according to the real-time active data, obtain the candidate's growth potential level, and set the collection parameters of different data sources based on the growth potential level; Acquire multidimensional potential data of candidates from multiple data sources according to acquisition parameters, perform feature extraction on the multidimensional potential data to form multidimensional potential features of candidates, perform combined correlation extraction on the multidimensional potential features of candidates to generate high-order potential features; Generate a comprehensive quality score for the candidate based on multi-dimensional potential characteristics; construct a candidate growth potential assessment model, and obtain the candidate's growth potential score and development momentum score based on the candidate growth potential assessment model; combine the growth potential score, comprehensive quality score, and development momentum score to generate a candidate's growth potential portrait, and store it in the candidate portrait library; When an enterprise raises talent needs, it retrieves candidates that match the job requirements from the candidate profile database; Screen candidates that match the job requirements, generate a candidate recommendation report, and present key recommendation information of the candidates.

2. The method for talent potential analysis and demand matching driven by multi-source data according to claim 1 is characterized in that: The method comprises: Collect the basic attribute data of the candidates, and set the initial growth potential level of the candidates according to the basic attribute data of the candidates; Collect real-time active data of candidates; Calculate the growth potential score based on the candidate's real-time active data and initial growth potential level; set the numerical range of the growth potential level, and determine the candidate's final growth potential level based on the numerical range of the growth potential score; Set the collection parameters for different data sources based on the growth potential level; the collection parameters include data source priority, data collection breadth and data collection depth; For candidates with high growth potential, increase the priority of data sources and expand the breadth and depth of data collection; for candidates with low growth potential, reduce the priority of data sources and narrow the breadth and depth of data collection.

3. The method for talent potential analysis and demand matching driven by multi-source data according to claim 2 is characterized in that: The method for calculating the growth potential score according to the candidate's real-time active data and initial growth potential level includes: Among them, L f is the growth potential score, L0 is the initial growth potential level, S i is the score of real-time active data, W i is the basic weight of real-time active data, t i For real-time active data S i The timestamp of , t0 is the current time, α is the adjustment coefficient, n is the number of types of real-time active data, γ is the time decay factor, i is the index of the real-time active data type, and e is the base of the natural logarithm.

4. The method for talent potential analysis and demand matching driven by multi-source data according to claim 1, characterized in that: The method comprises: By using the common fields between data sources, the original data of the same candidate from different data sources are associated and matched to form a unified data view of the candidate; Based on multi-dimensional potential data, we can mine the association rules between different candidates and form an association network between candidates; Integrate the unified data view of the same candidate and the association network between different candidates to build an association map of multi-source heterogeneous data from different data sources; Taking the candidate as the central node, extract the characteristics of the candidate's neighbor nodes in the associated network to form the candidate's network potential characteristics; Combine and correlate the multi-dimensional potential characteristics of candidates, explore the interaction patterns and influence mechanisms between the multi-dimensional potential characteristics, and form combined potential characteristics; The network potential characteristics and combined potential characteristics of the candidates are integrated to form high-order potential characteristics.

5. The method for talent potential analysis and demand matching driven by multi-source data according to claim 1, characterized in that: The method comprises: Multidimensional potential characteristics include educational characteristics, occupational characteristics, and skill characteristics; Based on the candidate's educational characteristics, the candidate's educational quality score is formed; based on the candidate's professional characteristics, the candidate's professional quality score is formed; based on the candidate's skill characteristics, the candidate's skill quality score is formed; the candidate's comprehensive quality score is formed by combining the candidate's educational quality score, professional quality score and skill quality score; Construct a candidate growth potential assessment model to obtain the candidate's growth potential score and development momentum score based on the candidate's growth potential level, real-time active data, multi-dimensional potential characteristics, high-level potential characteristics and growth potential assessment model; The candidate's total growth potential score is formed based on the candidate's growth potential score, comprehensive quality score and development momentum score; Extract key tags and keywords of candidates from their multi-dimensional potential characteristics and high-level potential characteristics to form keyword portraits of candidates; Generate a growth potential profile of the candidate based on the total growth potential score and keyword portrait.

6. The method for talent potential analysis and demand matching driven by multi-source data according to claim 5 is characterized in that: The method for forming a candidate's educational quality score based on the candidate's educational characteristics includes: Among them, E is the candidate's professional quality score, D j is the grade or score of the jth educational characteristic, T j is the duration of the jth educational feature, R j is the correlation of the jth educational characteristic, w j is the weight of the jth educational feature, λ is the adjustment coefficient, ε is the time decay factor, Δt j is the time difference from the current time, m is the total number of educational features, and j is the index of the educational feature.

7. The method for talent potential analysis and demand matching driven by multi-source data according to claim 1, characterized in that: The method comprises: Collect and analyze the talent needs raised by the enterprise, extract key job requirements and qualifications, and form a structured job profile; Perform feature representation and vectorization on the job profile and the candidate's growth potential profile to generate a job profile feature vector and a growth potential profile feature vector; Based on the job profile feature vector and the growth potential profile feature vector, the matching index between the two is calculated; a matching threshold is set, and if the matching index is greater than the matching threshold, the candidate is determined to be suitable for the job; Sort the matching indexes of candidates suitable for the position and select the top N candidates with the highest matching index as the recommendation results; generate a recommendation report for the candidates, presenting key recommendation information of the candidates.

8. A multi-source data-driven talent potential analysis and demand matching system, which is used to implement the multi-source data-driven talent potential analysis and demand matching method according to any one of claims 1 to 7, characterized in that: The system comprises: Growth potential assessment module: used to collect basic attribute data of candidates and set the initial growth potential level; collect real-time active data of candidates and dynamically adjust the growth potential level; set the collection parameters of different data sources based on the growth potential level; Multi-dimensional potential data collection module: used to obtain the multi-dimensional potential data of candidates from multiple data sources according to the collection parameters, pre-process and structure the multi-dimensional potential data, and extract the multi-dimensional potential characteristics of the candidates; Association graph construction module: used to associate and match records from different data sources of the same candidate to form a unified view, mine association rules between different candidates, form an association network, integrate the unified view and association network, and construct an association graph for multi-source heterogeneous data; High-order potential feature extraction module: used to extract the network potential features of candidates based on the association map, explore the combined associations between the multi-dimensional potential features of candidates, form combined potential features, fuse network potential features and combined potential features, and generate high-order potential features; Portrait generation module: used to generate comprehensive quality scores of candidates based on multi-dimensional potential characteristics, build a growth potential assessment model, generate growth potential scores and development momentum scores, and combine comprehensive quality scores, growth potential scores and development momentum scores to generate growth potential portraits of candidates and store them in the candidate portrait library; Talent recommendation module: used to retrieve candidates matching the job requirements from the candidate portrait library; screen the candidates matching the job requirements, generate a candidate recommendation report, and present key recommendation information of the candidates.

9. An electronic device comprising a memory, a central processing unit, and a computer program stored in the memory and executable on the central processing unit, characterized in that: When the central processor executes the computer program, the multi-source data-driven talent potential analysis and demand matching method described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed, implements the multi-source data-driven talent potential analysis and demand matching method according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • Personal value calculation method based on big data

    CN105160498A

  • Teacher portrait construction method based on teaching and research data

    CN113656687A

Cited By

  • Talent information processing method and system based on occupational attainment and big data analysis

    CN119784346A

  • A talent information processing method and system based on professional quality and big data analysis

    CN119784346B

  • Multi-source data driven scientific and technological talent ability portrait accurate matching method

    CN120781094A

  • People and post matching method and system and storage medium

    CN121279653A

  • Talent resource information processing method and platform based on big data analysis

    CN121388019A