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

Abstract

The invention relates to the technical field of talent management and data analysis, in particular to a multi-source data driven scientific and technological talent ability portrait accurate matching method. The method comprises the following steps: collecting and preprocessing multi-source talent data, and mining text information by applying a natural language processing technology; constructing a talent knowledge structure graph by identifying the relationship between the key entities and the extracted entities; determining a plurality of capability dimensions, and establishing a quantitative model or rule for each dimension to measure the capability; calculating scores of all dimensions by comprehensively considering personal differences of talents and working situation factors, and generating a visual ability portrait; a demand standard is determined by analyzing the situation of a demand side, the matching degree between talents and the demand standard is calculated after the talents are screened from a talent pool, the recommended talents are ranked according to the matching degree, and related parameters and rules are optimized according to feedback of the demand side. The talent ability can be comprehensively and accurately described, efficient and accurate matching is achieved, the method adapts to dynamic changes, and talent configuration in the science and technology service field is effectively optimized.

Description

Technical Field

[0001] The present invention relates to the field of talent management and data analysis technology, and specifically to a method for accurately matching scientific and technological talent capability portraits driven by multi-source data. Background Art

[0002] In today's era of rapid technological advancement, the technology services industry has become a key force driving innovation and progress across various sectors. This industry encompasses a wide range of sectors, including but not limited to information technology services, scientific research and technical support, and the commercialization of scientific and technological achievements. As the industry continues to expand and deepen, the demand for technology service professionals has exploded. Different technology service projects often require individuals with specific professional skills, comprehensive qualities, and innovative capabilities to ensure smooth project implementation and achieve their intended goals.

[0003] However, traditional talent evaluation and matching methods have exposed many shortcomings when faced with the complex and diverse group of science and technology service talents. On the one hand, most traditional evaluation methods focus on a single dimension, such as judging the suitability of talents based solely on academic background or work experience, ignoring the diversity of talent capabilities. On the other hand, traditional matching methods often lack systematicity and precision. Most of them are based on a rough comparison of simple job descriptions and talent resume information. They are unable to deeply analyze the fit between the talent's ability structure and the actual needs of the position. As a result, in the actual talent selection and job allocation process, there is a frequent mismatch between talent capabilities and job requirements. This not only makes it difficult for talents to fully realize their own value in the right position, but may also affect project progress, reduce work efficiency, and even lead to project failure, causing unnecessary losses to enterprises and related institutions.

[0004] Furthermore, the technology services industry itself is highly dynamic and innovative, with new technologies and applications constantly emerging, and industry trends and market demands constantly changing. This requires talent evaluation and matching mechanisms to keep pace with the times and flexibly adapt to these changes. However, existing methods are often rigid, lacking the ability to consider industry dynamics and adapt and optimize based on new circumstances. Summary of the Invention

[0005] The present invention provides a multi-source data-driven method for accurately matching the capability portraits of scientific and technological talents. By integrating multi-source data, constructing knowledge graphs, quantifying capability dimensions, comprehensively considering multi-factor scoring, and optimizing based on feedback, a series of steps are performed to accurately construct capability portraits of scientific and technological service talents, achieve efficient and accurate matching with the needs of the demand side, improve the utilization efficiency of talent resources, and optimize talent allocation in the field of scientific and technological services.

[0006] A multi-source data-driven method for accurately matching scientific and technological talent capability profiles, including: Data integration: Collect talent data containing multiple types of information, perform pre-processing operations, and use natural language processing technology to mine text data information; Graph construction: By identifying key entities in the data and extracting relationships between entities, we can build a talent knowledge structure graph and update the graph as needed; Dimension setting and quantification: Identify multiple capability dimensions of technology service talents and establish quantitative models or rules for each dimension to measure capability; Scoring calculation: Taking into account individual differences in talent and work situational factors, the scores of each capability dimension are calculated based on different weight settings; Profile generation: Integrate scores from various dimensions to generate a digital, visual profile that intuitively displays talent capabilities and characteristics; Demand analysis: Analyze the relevant situation of the demand side and determine the demand standards corresponding to each capability dimension; Talent screening: Screening out talents that initially meet the requirements from the existing talent database according to the demand standards; Matching assessment: Calculate the matching degree between the talent capability profile and the required standards, using appropriate similarity calculation methods and taking weights into account; Result recommendation: Talents are ranked based on their matching degree and recommended to the demander, with detailed profile information provided; Feedback optimization: Based on feedback from the demand side, use machine learning algorithms to optimize relevant weights, parameters, and rules.

[0007] Preferably, the data integration includes: The talent data collected covers educational background, work experience, professional skills, project achievements, certifications and qualifications, as well as social network-related data; The specific preprocessing operations are: Data cleaning: remove noise, duplication and invalid data records from the data; Data normalization: converting data of different formats and dimensions into a standard format to make them consistent and comparable; Feature extraction: Use relevant technologies to extract key feature information from the data. For text data, use natural language processing technology to conduct in-depth mining to determine the attitudes of talents towards different technical fields. Social network data is collected from mainstream professional social platforms and well-known industry forums. The collected content includes technical discussion records, industry insights shared, and interactions with peers.

[0008] Preferably, the construction of the talent knowledge structure map specifically includes: Entity recognition: Perform entity recognition on pre-processed multi-source data to identify key entities; Relationship extraction: Using natural language processing technology and machine learning algorithms, we extract relationships between entities from data, including but not limited to the "employed by" relationship between talents and work units, the "participated in" relationship between talents and projects, the "applied to" relationship between professional skills and projects, the "opened" relationship between the university they graduated from and the major they studied, and the "affiliated to" relationship between job titles and work units. We also continuously supplement and improve relationship types based on actual business needs. Graph construction: Identified entities are used as nodes in the graph, and extracted relationships are used as edges to construct a talent knowledge structure graph to show the relationship between various data and their impact on talent capabilities. A regular inspection mechanism is set up. When the amount of new data collected reaches a predetermined proportion or after a specific time period, the graph is updated in real time to ensure that it reflects the latest situation of the talent.

[0009] Preferably, the dimension setting and quantification include: Set capability dimensions: The set capability dimensions include basic technical capabilities, application development capabilities, industry insight capabilities, cross-domain integration capabilities and service innovation capabilities. Among them, basic technical capabilities are the foundation of professional and technical level, application development capabilities help business expansion, industry insight capabilities grasp industry trends, cross-domain integration capabilities reflect the ability to integrate cross-domain collaborative resources, and service innovation capabilities reflect the level of service innovation.

[0010] Capacity quantification: Basic technical capabilities are quantified through the scores of professional knowledge test questions, practical skills assessment items, and the problem-solving performance in actual projects. They are divided into entry-level, elementary, intermediate, advanced, and expert levels. Each level corresponds to a different score range for questions and a proportion of the score for practical skills assessment items. The professional knowledge test questions cover different sections and are weighted according to importance. The practical skills assessment items determine the scoring details based on the difficulty of the skills and the universality of the application scenarios. Application expansion capabilities are quantitatively evaluated according to preset rules based on the number of expanded application scenarios, innovation, and contribution to economic and social benefits. Innovation is scored based on the novelty of ideas and uniqueness of scenarios, and contribution to economic and social benefits is measured by comparing with the industry average and input-output ratio. Industry insight capabilities are quantified using a weighted summation system based on indicators such as the quality of industry analysis reports, accuracy of policy and regulatory interpretation, hit rate of trend assessment, and speed of regulatory response. Each indicator is evaluated and quantified from different dimensions. Cross-disciplinary integration capabilities are quantitatively assessed by combining the breadth and depth of interdisciplinary knowledge reserves, cross-disciplinary project collaboration performance, and feedback from team member collaboration and communication. Data for each component is obtained based on the corresponding measurement method and the scores are summarized by percentage. Service innovation capability is measured by counting the number of innovative measures, application effects, speed of new technology introduction and display of results, and calculating scores based on the corresponding proportions.

[0011] Preferably, in the scoring calculation step, the individual difference factors of talents considered include academic qualifications and years of work experience, and the work situation factors include the degree of competition in the industry and the stage of enterprise development; different weights are set for different types of data according to their importance to each capability dimension, and the weights are determined based on a large amount of historical data statistical analysis and expert experience. The scores of each capability dimension are calculated based on the above factors according to the established quantitative model or rules, and as the talent evaluation system is improved and new influencing factors emerge, the factors considered and the corresponding weight coefficients and data weight settings are added or adjusted in a timely manner to ensure the accuracy of the scoring.

[0012] Preferably, in the demand analysis step, an industry analysis model is used to comprehensively analyze the internal and external environment and competitive situation of the industry to determine the characteristics of the industry. At the same time, big data analysis technology is combined to collect industry-related data to enrich the analysis dimensions and improve accuracy. The project scale is determined by the project budget amount, the number of people involved and the coverage scope. The development stage is judged based on the project planning milestones, key nodes and actual progress, and the demand standards of each capability dimension are accurately determined.

[0013] Preferably, in the talent screening step, talents that preliminarily meet the requirements are screened from a talent database, and the talents in the database are all generated with capability profiles according to the above method.

[0014] Preferably, in the matching evaluation step, the matching degree is calculated using a weighted summation method, and different weights are set for the matching conditions of different capability dimensions; the matching condition of each capability dimension is calculated using a similarity calculation method, including but not limited to cosine similarity, Euclidean distance and Manhattan distance calculation methods.

[0015] Preferably, in the feedback optimization step, the weights of the ability dimensions and the demand standards are adjusted according to the feedback from the demander on the recommended talents; artificial intelligence technology is used to analyze and mine the feedback data to automatically optimize the weights of the ability dimensions, the parameters of the scoring model and the rules of the matching algorithm.

[0016] Compared with the prior art, the advantages of the present invention are: Comprehensive and accurate assessment of talent capabilities: By integrating multi-source data, building knowledge graphs, and multi-dimensional quantitative capabilities, it is possible to comprehensively, deeply, and accurately portray the ability characteristics of science and technology service talents, avoiding the one-sidedness of traditional evaluation methods, and more accurately reflecting the true ability level of talents. It helps to tap the potential of talents and provide a scientific basis for the rational use and training of talents.

[0017] Achieving efficient and precise matching: Based on the constructed ability profile and detailed demand analysis, using scientific matching degree calculation method and feedback optimization mechanism, it can efficiently and accurately match the demand side to the appropriate technology service talents, improve the matching degree of talents and positions, reduce the waste of resources and efficiency loss caused by improper selection of talents, optimize the allocation of human resources in the field of scientific and technological services, and promote the better development of scientific and technological service industry.

[0018] Adapting to dynamic changes: Considering the factors such as the development of talent ability, industry changes and demand changes, this method sets up corresponding dynamic adjustment mechanism, no matter the quantification of ability dimension, score calculation, or matching rule, etc., can be adjusted and optimized in time according to the actual situation, to ensure that the whole method always fits the actual application scene, maintains good adaptability and effectiveness. DETAILED DESCRIPTION

[0019] In order to better explain the invention, in order to facilitate understanding, a multi-source data driven scientific and technological talent ability profile precise matching method is described in detail below combined with actual operation process.

[0020] Data integration: Data collection: Form a professional data collection team, collect relevant data of technology service talents through various channels. Cooperate with major universities and vocational training institutions to obtain talent education background data, including major, course grades, degrees obtained, etc. Contact enterprise human resources departments to collect talent work experience data, such as employment time, leaving time, job title, job responsibilities, performance, etc. At the same time, obtain talent professional skill data from professional skill evaluation agencies, including programming language proficiency, software tool use ability, etc. Collect project achievement data of talents, such as project name, project size, project achievement, role, etc. For certificate qualification data, obtain professional certificates and qualification certificates of talents from relevant certificate issuing agencies. Use web crawler technology to collect social network data from mainstream professional social platforms and industry well-known forums, including technical discussion content published by talents, industry insight sharing, interaction records with peers, etc.

[0021] Data preprocessing: Data cleaning: Use data cleaning tools (such as Python's Pandas library) to clean the collected data. Write scripts to remove noise data in the data, such as format errors, duplicate records and invalid data. For education background data, check if the education information meets the standard, and remove records that do not meet the requirements. For work experience data, clean up unreasonable records such as leaving time earlier than employment time.

[0022] Data normalization: Data normalization methods are used to process data of varying formats and dimensions. For numerical data, such as course grades and project performance, the min-max normalization method is used to scale the data to the [0, 1] range. For textual data, such as job titles and professional skill descriptions, standardization is performed to unify wording and format.

[0023] Feature Extraction: Natural language processing techniques are used to extract features from text data. For social network data, the BERT pre-trained model is used to extract semantic features from text and analyze talent's attitudes towards different technical fields. Furthermore, the TF-IDF algorithm is used to extract keywords as talent profile information.

[0024] Graph construction: Entity Recognition: Use deep learning models (such as BiLSTM-CRF) to perform entity recognition on preprocessed multi-source data. The trained model identifies key entities such as individual talent, professional skills, project achievements, educational institutions, and work units. For educational background data, entities such as graduation school, major, enrollment date, and graduation date are identified. For project achievement data, entities such as project name, project type, project start and end dates, and project leader are identified. For work experience data, entities such as work unit name, employment date, departure date, and position title are identified.

[0025] Relationship Extraction: Relationship extraction utilizes natural language processing technology and machine learning algorithms. Dependency parsing and semantic role labeling methods are used to extract relationships between entities from the data. In addition to common relationships such as "employed at," "participated in," and "applied to," we also identify relationships between the university of graduation and the major studied, and between job titles and work units. We continuously supplement and refine relationship types based on actual business needs.

[0026] Graph Construction and Updates: Use a graph database (such as Neo4j) to construct a talent knowledge structure graph, using identified entities as nodes and extracted relationships as edges. Regularly check the collection of new data. When the amount of new data collected reaches a predetermined percentage (such as 20%) or a specific time period (such as every quarter), update the knowledge graph in real time to ensure that the graph reflects the latest talent capabilities and experience.

[0027] Dimension setting and quantification: Defining capability dimensions: From the perspective of the entire industry chain, identify the five capability dimensions of technology service professionals: basic technical capabilities, application development capabilities, industry insight, cross-domain integration capabilities, and service innovation capabilities. Organize discussions with industry experts and business representatives to clarify the definition and importance of each capability dimension.

[0028] Capacity quantification: Quantification of basic technical capabilities: Establish a quantitative model for basic technical capabilities, taking into account the scores of professional knowledge test questions, the scores of practical skills assessment items, and the problem-solving in actual projects. Design a professional knowledge test question bank that covers different sections such as professional basic knowledge and cutting-edge technical knowledge, and assign different score weights based on importance. Set up practical skills assessment items, and determine the scoring criteria based on the difficulty of the skills and the universality of the application scenarios. Divide basic technical capabilities into five levels: entry-level, elementary, intermediate, advanced, and expert. Each level corresponds to a different professional knowledge test question score range and practical skills assessment item score proportion.

[0029] Quantifying Application Expansion Capabilities: Develop quantitative criteria for application expansion capabilities, evaluating them based on the number of successfully expanded application scenarios, their innovativeness, and the resulting economic and social benefits. Establish an evaluation team to score the innovation of expanded application scenarios, evaluating them based on factors such as the novelty of innovative ideas and the uniqueness of application scenarios. Economic and social benefits will be measured by comparing them to industry averages and the input-output ratio.

[0030] Quantifying Industry Insight: A quantitative system for industry insight is constructed, using a weighted summation of indicators such as the quality of industry analysis reports, the accuracy of policy and regulatory interpretations, the hit rate of trend assessments, and the speed of regulatory response. Industry experts are invited to review industry analysis reports, evaluating their depth, logic, and data accuracy. The accuracy of policy and regulatory interpretations is judged by their alignment with official interpretations. The number of hits in trend assessments and the time it takes to respond to regulations are counted to calculate the corresponding scores.

[0031] Quantifying cross-disciplinary integration capabilities: Quantify cross-disciplinary integration capabilities by combining the breadth and depth of interdisciplinary knowledge reserves, performance evaluation results of cross-disciplinary project collaboration, and team member feedback on collaborative communication skills. Design a cross-disciplinary knowledge reserve questionnaire to understand the number of disciplines and core knowledge mastered by talent. Evaluate cross-disciplinary project collaboration performance using metrics such as project completion quality, timelines, and cost control. Use questionnaires and peer reviews to gather feedback from team members on collaborative communication skills.

[0032] Quantifying Service Innovation Capabilities: Service innovation capabilities are quantitatively scored by measuring the number of innovative service model and process initiatives, the effectiveness of innovative applications, the speed of new technology introduction, and the demonstration of innovative application results. An innovation assessment team will be established to compile statistics and evaluate the quantity and quality of innovative initiatives. The effectiveness of innovative application will be measured in terms of increased user satisfaction and business growth. The speed of new technology introduction will be assessed by comparing it to the industry average introduction time. The demonstration of innovative application results will be evaluated based on the number of demonstrations and the degree of recognition received.

[0033] Rating calculation: Factors to consider: Consider individual talent differences (such as education and years of experience) and contextual factors (such as industry competition and company development stage). Qualifications are categorized into doctoral, master's, bachelor's, and associate degrees, each with a different weighting coefficient. Years of experience are divided into five-year intervals, with each interval assigned a different coefficient. The intensity of industry competition is divided into three levels: intense, moderate, and moderate, each with a corresponding coefficient. The company's development stage (startup, growth, and mature) is also categorized into different coefficients.

[0034] Weighting: Different weights are assigned to different types of data based on their importance to each capability dimension. Through statistical analysis of a large amount of historical data and expert experience, the weight coefficients of each factor and the weights of different types of data in each capability dimension are determined.

[0035] Scoring Calculation: Taking the aforementioned factors into consideration, scores for each competency dimension are calculated according to established quantitative models or rules. As the talent evaluation system continues to improve and new influencing factors emerge, factors and corresponding weighting coefficients will be added or adjusted as appropriate to ensure the accuracy of scoring calculations.

[0036] Image generation: Use data visualization tools (such as Tableau and PowerBI) to integrate scores across competency dimensions and generate a digitally visualized talent competency profile. Use charts (such as radar charts and bar graphs) and score lists to intuitively display talent's competency characteristics across various dimensions, making them easier for users to review and compare.

[0037] Requirements analysis: Industry Characteristics Analysis: Utilizing industry analysis models such as SWOT analysis and Porter's Five Forces model, we comprehensively analyze the internal and external environment, competitive landscape, and other factors within the demand-side industry to determine industry characteristics. Incorporating big data analysis techniques, we collect relevant industry data, such as market share trends and new technology adoption, to further enrich our analysis and enhance our understanding of industry characteristics.

[0038] Project Scale: Project scale is determined based on factors such as the project budget, number of people involved, and scope. Project budgets are categorized into three levels: small (less than 1 million RMB), medium (between 1 million and 5 million RMB), and large (over 5 million RMB). Project scale is determined based on the number of people involved (using 10 or 50 people as thresholds) and scope (measured across regional, national, and multinational dimensions).

[0039] Development Stage Assessment: Determine the project's development stage based on the milestones and key milestones set in the project plan and actual progress. Accurately determine the project's development stage by combining factors such as project goals, resource input, and market feedback.

[0040] Determine requirements standards: Based on an analysis of the demander's industry characteristics, project scale, and development stage, accurately determine the requirements standards for each capability dimension. Organize discussions between the demander and industry experts to ensure the rationality and feasibility of the requirements standards.

[0041] Talent screening: Based on the determined demand criteria, we screen candidates from the talent database for those who preliminarily meet the requirements. All candidates in the talent database have their competency profiles pre-generated using the aforementioned methods. We establish a regular update mechanism to review the latest talent information quarterly. This update is based on information proactively submitted by candidates, feedback from their current employers, and publicly available information automatically collected by the system to ensure that the database information is consistent with the actual talent's situation.

[0042] Matching evaluation: Choosing a matching method: Select an appropriate similarity calculation method based on the data characteristics and actual needs. For vector data with relatively balanced dimensions, use cosine similarity. For data primarily in numerical form, use Euclidean distance. For data with a high degree of discrete data, use Manhattan distance.

[0043] Match Calculation: Calculate the match between the talent's ability profile and the required standards using a weighted summation approach, assigning different weights to the matching of different ability dimensions based on their importance in the requirements. Each ability dimension is evaluated using a corresponding similarity calculation method to ensure the accuracy and rationality of the match calculation.

[0044] Recommended results: Screened talents are ranked based on their matching scores, and highly compatible candidates are recommended to the demanders. At the same time, detailed profiles of each talent are provided to the demanders, including scores, strengths, and weaknesses across various competency dimensions, allowing them to further understand the talent's profile. Recommendations are presented through a combination of list display and report push, ensuring that demanders can easily access recommended information.

[0045] Feedback optimization: Feedback collection: Establish a feedback mechanism to promptly collect feedback from demanders on recommended talents, including evaluations of talent's capabilities, degree of fit with the position, and work performance.

[0046] Data mining and optimization: Utilize machine learning algorithms (such as neural networks and decision trees) to analyze and mine feedback data, automatically optimizing the weights of capability dimensions, scoring model parameters, and matching algorithm rules. Based on the feedback, the weights of capability dimensions and requirement criteria are adjusted to improve the accuracy and adaptability of subsequent matching.

[0047] Iterative Updates: We adhere to an iterative update mechanism and continuously adjust optimization strategies based on actual application results. We regularly evaluate optimization results and further refine methods and models based on the evaluation results to ensure continuous improvement in the performance of the entire matching system.

[0048] Through the above specific implementation methods, it is possible to achieve accurate construction and efficient matching of capability portraits of scientific and technological service talents, and provide strong support for talent management and allocation in the field of scientific and technological services.

[0049] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0050] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multi-source data-driven accurate matching method for scientific and technological talent capability portraits, characterized by: include: Data integration: Collect talent data containing multiple types of information, perform pre-processing operations, and use natural language processing technology to mine text data information; Graph construction: By identifying key entities in the data and extracting relationships between entities, we can build a talent knowledge structure graph and update the graph as needed; Dimension setting and quantification: Determine multiple capability dimensions of technology service talents, and establish quantitative models or rules to measure capability levels for each dimension; Scoring calculation: Taking into account individual differences in talent and work situational factors, the scores of each capability dimension are calculated based on different weight settings; Profile generation: Integrate scores from various dimensions to generate a digital, visual profile that intuitively displays talent capabilities and characteristics; Demand analysis: Analyze the relevant situation of the demand side and determine the demand standards corresponding to each capability dimension; Talent screening: Screening out talents that initially meet the requirements from the existing talent database according to the demand standards; Matching assessment: Calculate the matching degree between the talent capability profile and the required standards, using appropriate similarity calculation methods and taking weights into account; Result recommendation: Talents are ranked based on their matching degree and recommended to the demander, with detailed profile information provided; Feedback optimization: Based on feedback from the demand side, use machine learning algorithms to optimize relevant weights, parameters, and rules.

2. The multi-source data-driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: The data integration includes: The talent data collected covers educational background, work experience, professional skills, project achievements, certifications and qualifications, as well as social network-related data; The specific preprocessing operations are: Data cleaning: remove noise, duplication and invalid data records from the data; Data normalization: converting data of different formats and dimensions into a standard format to make them consistent and comparable; Feature extraction: Use relevant technologies to extract key feature information from the data. For text data, use natural language processing technology to conduct in-depth mining to determine the attitudes of talents towards different technical fields. Social network data is collected from mainstream professional social platforms and well-known industry forums. The collected content includes technical discussion records, industry insights shared, and interactions with peers.

3. The multi-source data-driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: The construction of the talent knowledge structure map specifically includes: Entity recognition: Perform entity recognition on pre-processed multi-source data to identify key entities; Relationship extraction: Using natural language processing technology and machine learning algorithms, we extract relationships between entities from data, including but not limited to the "employed by" relationship between talents and work units, the "participated in" relationship between talents and projects, the "applied to" relationship between professional skills and projects, the "opened" relationship between the university they graduated from and the major they studied, and the "affiliated to" relationship between job titles and work units. We also supplement and improve relationship types based on actual business needs. Graph construction: Identified entities are used as nodes in the graph, and extracted relationships are used as edges to construct a talent knowledge structure graph to show the relationship between various data and their impact on talent capabilities. A regular inspection mechanism is set up. When the amount of new data collected reaches a predetermined proportion or after a specific time period, the graph is updated in real time to ensure that it reflects the latest situation of the talent.

4. The multi-source data-driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: The dimension setting and quantification include: Setting capability dimensions: The set capability dimensions include basic technical capabilities, application development capabilities, industry insight capabilities, cross-domain integration capabilities, and service innovation capabilities; Capacity quantification: Basic technical capabilities are quantified through the scores of professional knowledge test questions, practical skills assessment items, and the problem-solving performance in actual projects. They are divided into entry-level, elementary, intermediate, advanced, and expert levels. Each level corresponds to a different score range for questions and a proportion of the score for practical skills assessment items. The professional knowledge test questions cover different sections and are weighted according to importance. The practical skills assessment items determine the scoring details based on the difficulty of the skills and the universality of the application scenarios. Application expansion capabilities are quantitatively evaluated according to preset rules based on the number of expanded application scenarios, innovation, and contribution to economic and social benefits. Innovation is scored based on the novelty of ideas and uniqueness of scenarios, and contribution to economic and social benefits is measured by comparing with the industry average and input-output ratio. Industry insight capabilities are quantified using a weighted summation system based on indicators such as the quality of industry analysis reports, accuracy of policy and regulatory interpretation, hit rate of trend assessment, and speed of regulatory response. Each indicator is evaluated and quantified from different dimensions. Cross-disciplinary integration capabilities are quantitatively assessed by combining the breadth and depth of interdisciplinary knowledge reserves, cross-disciplinary project collaboration performance, and feedback from team member collaboration and communication. Data for each component is obtained based on the corresponding measurement method and the scores are summarized by percentage. Service innovation capability is measured by counting the number of innovative measures, application effects, speed of new technology introduction and display of results, and calculating scores based on the corresponding proportions.

5. The multi-source data-driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: In the scoring calculation step, the individual talent difference factors considered include academic qualifications and years of work experience, and the work situation factors include the degree of competition in the industry and the stage of enterprise development; different weights are set for different types of data based on their importance to each capability dimension. The weights are determined based on a large amount of historical data statistical analysis and expert experience. The scores of each capability dimension are calculated based on the above factors according to the established quantitative model or rules. As the talent evaluation system is improved and new influencing factors emerge, the factors considered and the corresponding weight coefficients and data weight settings are added or adjusted to ensure the accuracy of the scoring.

6. The multi-source data-driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: In the demand analysis step, an industry analysis model is used to comprehensively analyze the internal and external environment and competitive situation of the industry to determine the characteristics of the industry. At the same time, big data analysis technology is combined to collect industry-related data to enrich the analysis dimensions and improve accuracy. The project scale is determined by the project budget amount, the number of people involved, and the coverage scope. The development stage is judged based on the project planning milestones, key nodes and actual progress, and the demand standards for each capability dimension are accurately determined.

7. The multi-source data driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: In the talent screening step, talents that preliminarily meet the requirements are screened from the talent database, and the talents in the database are all generated with ability profiles according to the above method.

8. The multi-source data-driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: In the matching evaluation step, the matching degree is calculated using a weighted summation method, with different weights set for the matching conditions of different capability dimensions; the matching condition of each capability dimension is calculated using a similarity calculation method, including but not limited to cosine similarity, Euclidean distance, and Manhattan distance calculation methods.

9. The multi-source data-driven accurate matching method for scientific and technological talent capability portraits according to claim 1 is characterized in that: In the feedback optimization step, the weights of the ability dimensions and the demand standards are adjusted based on the feedback from the demand side on the recommended talents; artificial intelligence technology is used to analyze and mine the feedback data to automatically optimize the weights of the ability dimensions, the parameters of the scoring model and the rules of the matching algorithm.