Four-dimensional talent portrait construction method and device, equipment and storage medium

Abstract

The invention relates to a four-dimensional talent portrait construction method, and the method comprises the steps: obtaining multi-dimensional talent data, including basic data, professional behavior data, network association data and network preference data, carrying out the tagging processing of the data, extracting and weighting various feature tags, and obtaining a four-dimensional talent portrait. And constructing a basic portrait, a behavior portrait, an associated portrait and a psychological portrait of the talent. And combining the portraits, carrying out comprehensive analysis, and generating a complete four-dimensional talent portrait. Through comprehensive analysis and dynamic updating of the multi-dimensional data, the accuracy and comprehensiveness of the talent portrait are improved, reliable technical support is provided for comprehensive evaluation and development prediction of talents, and the problems that in the prior art, the talent portrait is low in accuracy, incomplete in data coverage and not timely in updating are effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and in particular to a method, device, equipment and storage medium for constructing a four-dimensional talent portrait. Background Art

[0002] In the modern financial industry, talent is a key factor in promoting innovation, managing risks and achieving corporate strategic goals. With the complexity and intensified competition in the financial market, financial institutions are increasingly demanding high-quality talent. The selection, training and management of financial talent has become an important link that affects the core competitiveness of enterprises.

[0003] Traditional talent management methods often rely on static data and experience judgment, which makes it difficult to cope with the rapidly changing market environment and increasingly complex business needs. With the development of big data and artificial intelligence technology, talent profiling, as an emerging technical means, has gradually been introduced into the financial industry. Talent profiling can provide financial institutions with accurate talent assessment and personalized management solutions by analyzing multi-dimensional data such as personal basic information, work experience, professional skills, social networks, and behavioral data.

[0004] However, the application of existing talent profiling technology in the financial field still faces some challenges. First, data acquisition and processing is a complex and time-consuming process, especially in the financial industry, where data is often highly sensitive and distributed across multiple systems. Second, the construction of talent profiling often relies on historical data, while the talent needs of the financial industry are constantly changing with market dynamics, making it difficult for traditional talent profiling methods to provide real-time and accurate decision support. Finally, the financial industry's talent requirements are highly professional and complex, and existing models and algorithms are difficult to fully capture these characteristics.

[0005] In order to solve these problems, more and more financial institutions have begun to adopt talent profiling technology based on big data and artificial intelligence to improve the efficiency and effectiveness of talent management. For example, by integrating multi-source data, applying natural language processing technology, and combining machine learning algorithms, financial institutions can update talent profiles in real time and provide more targeted management decisions. This not only helps to improve the efficiency of talent selection and training, but also helps institutions better control risks and innovate businesses.

[0006] In this context, how to effectively construct and apply talent portraits in the financial field has become one of the important means for financial institutions to enhance their competitiveness and achieve sustainable development. Summary of the invention

[0007] The main purpose of the present invention is to provide a four-dimensional talent portrait construction method, device, equipment and storage medium, aiming to solve the technical problems of low accuracy and incomplete data coverage of talent portraits in the prior art.

[0008] To achieve the above object, the present invention provides a method for constructing a four-dimensional talent portrait, comprising:

[0009] Acquire multi-dimensional talent data, wherein the multi-dimensional talent data includes talent basic data, professional behavior data, network association data, and network preference data;

[0010] Labeling the multi-dimensional talent data, extracting various characteristic labels of each talent, the characteristic labels including basic data labels, behavioral data labels, partnership labels and network preference labels, and assigning weights to each characteristic label;

[0011] Construct a basic talent portrait based on the basic data tags and corresponding weights, construct a talent behavior portrait based on the behavior data tags and corresponding weights, construct a talent association portrait based on the partnership tags and corresponding weights, and construct a talent psychological portrait based on the network preference tags and corresponding weights;

[0012] The basic talent portrait, talent behavior portrait, talent association portrait and talent psychological portrait are comprehensively analyzed to generate a complete four-dimensional talent portrait for comprehensive evaluation and prediction of talents.

[0013] In one embodiment, constructing a basic talent profile based on the basic data tags and corresponding weights includes:

[0014] Extracting basic information of talents from the multi-dimensional talent data based on the basic data tags, wherein the basic information includes the name, education background and position of the talents;

[0015] Perform text processing on the extracted basic information and extract multiple core phrases from the basic information of the talent through natural language processing technology;

[0016] Assign each core phrase the same weight as the corresponding basic data label;

[0017] Associating the core phrases with basic information of talents, and marking and storing them according to the phrase weights to form a structured talent information corpus;

[0018] Based on the core phrases and corresponding phrase weights in the talent information corpus, a basic talent profile is constructed.

[0019] In one embodiment, constructing a talent behavior profile based on the behavior data tags and corresponding weights includes:

[0020] Extracting talent behavior information from multi-dimensional talent data based on the behavior data tags, the behavior information including the number of papers, citations, H-Index, G-Index, research field, publication year and frequency of the talent;

[0021] Standardize the extracted behavior information, and extract core behavior features based on the standardized behavior information;

[0022] Assign the same weight to the core behavior features as to the corresponding behavior data labels, and perform weighted processing on the core behavior features;

[0023] Integrate the weighted core behavioral characteristics to construct a talent behavior profile.

[0024] In one embodiment, constructing a talent association profile based on the partnership labels and corresponding weights includes:

[0025] Extracting talent cooperation relationship information from multi-dimensional talent data based on the cooperation relationship label, wherein the cooperation relationship information includes the relationship between the talent and the mentor, the student and the collaborator;

[0026] Clean and standardize the extracted cooperative relationship information, and extract network relationship features based on the standardized cooperative relationship data;

[0027] Assign the same weights as the corresponding partnership labels to the extracted network relationship features, and perform weighted processing on these features;

[0028] Use solid dots to represent the relationship between talents and mentors, use gray dots to represent the relationship between talents and students, and use white dots to represent the relationship between talents and collaborators. Distribute solid dots, gray dots, and white dots in the figure according to the relationship type.

[0029] According to the actual contact frequency and number of cooperation, use lines to connect the solid dots, gray dots and white dots to form a hedgehog diagram;

[0030] Integrate the weighted network relationship features in the hedgehog diagram to construct a talent association portrait.

[0031] In one embodiment, constructing a talent psychological profile based on the network preference tags and corresponding weights includes:

[0032] Obtain text information of sample users from online social platforms and perform data preprocessing on the extracted text information;

[0033] Based on the text information of sample users, network preference data is generated and structured, converting unstructured text data into structured information including keywords, topic tags, and sentiment analysis results;

[0034] Extracting the behavioral characteristics of sample users from structured network preference data, constructing behavioral representations, and establishing a personality characteristic model for a preset number of groups based on the behavioral representations;

[0035] Utilize the established personality trait model of preset number groups to analyze the online behavior of talents, extract the corresponding personality traits, and generate the psychological trait value of talents based on the network preference tags;

[0036] Multiply each psychological feature value by the weight of the corresponding network preference label to generate a weighted psychological feature value;

[0037] Integrate the weighted psychological characteristic values ​​to generate a complete psychological portrait of the talent.

[0038] In one embodiment, the talent basic profile, talent behavior profile, talent association profile and talent psychological profile are comprehensively analyzed to generate a complete four-dimensional talent profile, including:

[0039] The features in each portrait are standardized so that features of different dimensions can be compared and analyzed on the same scale;

[0040] Combine the key features in the talent basic profile, talent behavior profile, talent association profile, and talent psychological profile with the weights assigned to the feature labels;

[0041] The weighted key features are integrated to generate a unified feature set, so that the feature data of each dimension can reflect the comprehensive characteristics of the talent as a whole after integration;

[0042] Use multivariate statistical analysis or machine learning algorithms to conduct a comprehensive analysis of the fused feature set to identify the main feature patterns of talents in various dimensions;

[0043] Visualize the results of the comprehensive analysis to generate a complete four-dimensional talent portrait.

[0044] In one embodiment, the multi-dimensional talent data is labeled to extract various characteristic labels of each talent, including:

[0045] extracting key features from the multi-dimensional talent data;

[0046] Using statistical analysis methods or machine learning algorithms, a label generation method is determined based on the distribution characteristics of the extracted feature data or pattern recognition results;

[0047] According to the determined label generation method, corresponding label generation rules are formulated, and characteristic labels are automatically assigned to the data points of each talent in the multi-dimensional talent data through the set algorithm.

[0048] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a four-dimensional talent portrait construction device, which includes a memory, a processor, and a four-dimensional talent portrait construction program stored in the memory and executable on the processor, and the four-dimensional talent portrait construction program, when executed by the processor, implements the steps of the four-dimensional talent portrait construction method as described above.

[0049] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer storage medium, on which a four-dimensional talent portrait construction program is stored, and when the four-dimensional talent portrait construction program is executed by a processor, the steps of the four-dimensional talent portrait construction method as described above are implemented.

[0050] Beneficial effects: The present invention relates to a method for constructing a four-dimensional talent portrait, which obtains multi-dimensional talent data, including basic data, professional behavior data, network association data, and network preference data, labels these data, extracts and weights various feature labels, and then constructs a basic portrait, behavior portrait, association portrait, and psychological portrait of the talent. Combined with these portraits, a comprehensive analysis is performed to generate a complete four-dimensional talent portrait. The present invention improves the accuracy and comprehensiveness of talent portraits through comprehensive analysis and dynamic updating of multi-dimensional data, provides reliable technical support for comprehensive evaluation and development prediction of talents, and effectively solves the problems of low accuracy of talent portraits, incomplete data coverage, and untimely updates in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0052] Figure 1 It is a flowchart of an embodiment of a method for constructing a four-dimensional talent portrait according to the present invention;

[0053] Figure 2 A schematic diagram of functional modules of a preferred embodiment of the apparatus for constructing a four-dimensional talent portrait of the present invention;

[0054] Figure 3 A schematic diagram of the structure of the device hardware operating environment involved in the embodiment of the device for constructing a four-dimensional talent portrait of the present invention. DETAILED DESCRIPTION

[0055] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0056] In the modern financial industry, talent is a key factor in promoting innovation, managing risks and achieving corporate strategic goals. With the complexity and intensified competition in the financial market, financial institutions are increasingly demanding high-quality talent. The selection, training and management of financial talent has become an important link that affects the core competitiveness of enterprises.

[0057] Traditional talent management methods often rely on static data and experience judgment, which makes it difficult to cope with the rapidly changing market environment and increasingly complex business needs. With the development of big data and artificial intelligence technology, talent profiling, as an emerging technical means, has gradually been introduced into the financial industry. Talent profiling can provide financial institutions with accurate talent assessment and personalized management solutions by analyzing multi-dimensional data such as personal basic information, work experience, professional skills, social networks, and behavioral data.

[0058] However, the application of existing talent profiling technology in the financial field still faces some challenges. First, data acquisition and processing is a complex and time-consuming process, especially in the financial industry, where data is often highly sensitive and distributed across multiple systems. Second, the construction of talent profiling often relies on historical data, while the talent needs of the financial industry are constantly changing with market dynamics, making it difficult for traditional talent profiling methods to provide real-time and accurate decision support. Finally, the financial industry's talent requirements are highly professional and complex, and existing models and algorithms are difficult to fully capture these characteristics.

[0059] In order to solve these problems, more and more financial institutions have begun to adopt talent profiling technology based on big data and artificial intelligence to improve the efficiency and effectiveness of talent management. For example, by integrating multi-source data, applying natural language processing technology, and combining machine learning algorithms, financial institutions can update talent profiles in real time and provide more targeted management decisions. This not only helps to improve the efficiency of talent selection and training, but also helps institutions better control risks and innovate businesses.

[0060] In this context, how to effectively construct and apply talent portraits in the financial field has become one of the important means for financial institutions to enhance their competitiveness and achieve sustainable development.

[0061] See also Figure 1 , Figure 1 This is a flow chart of an embodiment of a method for constructing a four-dimensional talent profile provided by the present invention. It should be noted that although a logical sequence is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0062] like Figure 1 As shown, the four-dimensional talent portrait construction method proposed in the present invention includes the following steps:

[0063] S10, obtaining multi-dimensional talent data, wherein the multi-dimensional talent data includes talent basic data, professional behavior data, network association data, and network preference data;

[0064] In this embodiment, multi-dimensional data related to talents are collected from multiple data sources. These data sources may include internal databases, social media platforms, academic publications, internal enterprise systems, etc. The types of data collected are extensive, covering basic information, behavior records, social relationships, and preference habits. The data acquisition method may include API calls, data crawling, file import and other technical means, the purpose of which is to ensure the comprehensiveness and diversity of the data.

[0065] Talent basic data refers to static information related to the individual identity of talents, such as name, age, education, position, work experience, etc. Basic data usually comes from the company's human resources system, government database or public professional social platform.

[0066] Professional behavior data refers to the behavior of talents in their professional fields, such as paper publication records, patent application records, project participation, performance, etc. These data can be obtained through academic databases, corporate performance management systems or industry reports.

[0067] Network association data involves the relationships of talents in social networks or professional networks, such as mentor relationships, colleague relationships, partnerships, etc. This type of data is usually obtained through social media analysis, internal communication records, or cooperative project records, reflecting the social network structure and influence of talents.

[0068] Network preference data refers to the interests and behavioral preferences of talents on the Internet, such as likes, shares, and comments on social media, and the topics or fields they follow. This type of data can be obtained by analyzing data on social media platforms, web browsing records, or content consumption habits, and reflects the interests and behavioral tendencies of talents.

[0069] In one specific implementation, when acquiring multi-dimensional talent data, the basic data, behavioral data, network association data, and network preference data from different data sources are integrated by integrating with multiple data sources in the financial industry and using cross-platform data fusion technology. This process uses ETL (Extract, Transform, Load) tools to extract, transform, and load data. In order to ensure the real-time nature of the data, real-time data stream processing technology (such as Apache Kafka) is used to process the data to ensure that the data is preprocessed and cleaned when it is collected. By using machine learning technology, redundant information in the basic data is automatically identified and eliminated, further improving the quality and accuracy of the data. Finally, after these data are standardized, a talent data set is generated for subsequent analysis.

[0070] By acquiring multi-dimensional talent data and integrating basic information, professional behavior, social networks and network preferences, a more accurate and comprehensive talent portrait model is constructed.

[0071] S20, labeling the multi-dimensional talent data, extracting various feature tags for each talent, the feature tags including basic data tags, behavior data tags, partnership tags, and network preference tags, and assigning a weight to each feature tag;

[0072] In this embodiment, the acquired multi-dimensional talent data is processed and analyzed, and each data point (such as a certain characteristic of the talent) is assigned one or more characteristic tags. The labeling process converts the original data into more structured and operational tags through specific algorithms or rules. These tags reflect different aspects of the talent, such as basic information, behavior patterns, social relationships and network preferences.

[0073] In labeling, labels that represent talent characteristics are identified and extracted from multi-dimensional data. This process may involve natural language processing, statistical analysis, machine learning and other technologies to identify and classify key information in various types of data. Feature labels may include but are not limited to: education, position, project participation, partnership, social media interaction preferences, etc.

[0074] Basic data tags cover basic static information of talents, such as name, age, gender, education background, work experience, etc. Basic data tags reflect the identity information and basic career status of talents, and are one of the core elements of building talent portraits.

[0075] Behavioral data tags reflect the dynamic behavioral characteristics of talents in their careers, such as work performance, academic achievements, project completion, transaction records, etc. Behavioral data tags help evaluate the professional capabilities and growth potential of talents.

[0076] The partnership label describes the social and cooperative relationships of talents in their professional networks, such as mentors, colleagues, partners, etc. The partnership label can reflect the social network and influence of talents and is an important indicator of social ability and cooperation potential.

[0077] Network preference tags are based on the talent’s online behavior data, reflecting their interests and preferences, such as topics they follow, discussions they participate in, content they like and share, etc. Network preference tags can provide in-depth insights into the talent’s personality, psychological characteristics, and areas of interest.

[0078] Assign a weight to each extracted feature label to indicate the importance of the label in the overall talent portrait. The weight assignment can be based on statistical analysis, expert evaluation, or determined by machine learning algorithms. The size of the weight determines the degree of influence of the label on the final portrait.

[0079] In one specific implementation, a hybrid label generation method is used when labeling multi-dimensional talent data. First, the basic data labels are processed using a rule engine, and the basic data labels are automatically generated through preset rules (such as educational level, position level). Next, for behavioral data labels, a time series-based clustering algorithm is used to identify the behavioral patterns of talents in different periods (such as performance change trends, project participation), extract key behavioral features from them and generate labels. In the generation of partnership labels, the graph convolutional network (GCN) is combined to analyze the topological structure of the social network, identify the core relationship network, and then generate labels. In the process of generating network preference labels, an emotional computing model (such as a BERT variant) is applied to deeply analyze the interactive content of social media, extract emotional tendencies and interest characteristics, and generate corresponding labels. All labels are automatically weighted according to their importance in the algorithm.

[0080] By labeling multi-dimensional talent data, combined with basic information, behavioral characteristics, social relationships and network preferences, we can accurately build talent portraits. By assigning weights to each feature label, we can better reflect the importance of each label in the portrait, ensuring the accuracy and comprehensiveness of the portrait.

[0081] S30, constructing a basic talent portrait based on the basic data tags and corresponding weights, constructing a talent behavior portrait based on the behavior data tags and corresponding weights, constructing a talent association portrait based on the partnership tags and corresponding weights, and constructing a talent psychological portrait based on the network preference tags and corresponding weights;

[0082] In this embodiment, basic data labels (such as education, position, age, etc.) extracted from multi-dimensional data are combined with the weight of each label to construct a basic portrait of the talent. The size of the weight reflects the importance of the label in the overall portrait, ensuring that the basic data can accurately reflect the identity information and professional background of the talent when constructing the portrait. The construction process of the basic portrait usually includes the classification of labels, weighted calculation, and comprehensive scoring based on weights, and finally generates a portrait that can represent the overall basic characteristics of the talent.

[0083] The construction of behavioral portraits is based on labels extracted from the behavioral data of talents (such as project participation, performance records, academic achievements, etc.). These labels reflect the dynamic behavioral characteristics of talents in their careers. The weights of behavioral data labels reflect the importance of different behaviors in the overall portrait. For example, frequent project participation may be more influential than a single academic achievement. By weighting and comprehensively analyzing behavioral labels, behavioral portraits that reflect the professional behavioral characteristics of talents are generated.

[0084] The association portrait is based on the partnership labels (such as mentorship, partnership, social circle, etc.) extracted from social network data. These labels reflect the position and relationship strength of talents in their social networks. By combining the weights of these labels, an association portrait that can reflect the social relationships and professional network influence of talents is generated. This process may involve social network analysis, graph analysis algorithms, and weighted processing of social data to ensure that the association portrait can accurately reflect the social structure and cooperation potential of talents.

[0085] Psychological portraits are based on labels extracted from the online behavior data of talents (such as social media interactions, content consumption habits, hobbies, etc.). These labels reflect the personality, interests, and psychological tendencies of talents. The weight of network preference labels determines the degree of influence of different network behaviors on the overall psychological portrait. By weighting these labels, a psychological portrait that can fully reflect the psychological characteristics of talents is constructed. This process may include technical means such as sentiment analysis, interest mining, and preference modeling to ensure the accuracy and representativeness of the portrait.

[0086] In another specific implementation, when constructing a basic portrait, the analytic hierarchy process (AHP) is used. By analyzing the hierarchical structure of each basic data label, the weight of each label is determined, and a comprehensive score is performed to generate a basic portrait. In the construction of the behavioral portrait, the fuzzy comprehensive evaluation method is used. By fuzzifying the behavioral data labels, the weights of each fuzzy set are integrated to generate a behavioral portrait. In the construction of the association portrait, the multidimensional scaling method (MDS) is used. By analyzing the similarity of the cooperative relationship labels, a multidimensional association portrait reflecting the social relationship structure of the talent is generated. In the construction of the psychological portrait, the Bayesian network is applied. By calculating the prior probability and conditional probability of the network preference label, a probability model that can reflect the psychological state of the talent is constructed, and a psychological portrait is generated.

[0087] Example description:

[0088] In a large financial institution, it was decided to conduct a comprehensive talent profiling analysis of employees in its asset management department to better assess the potential of employees and develop personalized career development plans.

[0089] Constructing a basic profile of talent: Extract key tags from the basic data of employees, such as education, professional title, years of experience in the financial industry, and professional certificates (such as CFA, FRM). After these tags are weighted, a basic profile of each employee is constructed. For example, an asset manager with 10 years of experience in the financial industry and a CFA certificate will be given a higher basic data weight to reflect his professional status in the industry.

[0090] Constructing talent behavior profiles: Extract labels from employee behavior data, such as fund management scale, average annual investment return, risk-adjusted return (such as Sharpe ratio), and customer satisfaction. Based on the weights of these behavior labels, the employee's behavior profile is constructed. For example, a fund manager who can maintain a high investment return rate in market fluctuations will have a behavior profile that highlights his excellent risk management capabilities.

[0091] Build talent association portraits: By analyzing employees' partnership data, such as the frequency of collaboration with colleagues, mentorship, and cross-departmental cooperation projects, we extract partnership labels. Based on the weights of these labels, we generate an employee association portrait. For example, a manager who often leads cross-departmental projects will have a wide range of internal company connections and good teamwork skills in his association portrait.

[0092] Constructing psychological portraits of talents: Extract labels from employees' network preference data, such as their activity in financial forums, their attention to market hot spots, and the types of content shared on social media (such as risk management, technical analysis, etc.). By analyzing the weights of these preference labels, a psychological portrait of the employee is constructed. For example, the psychological portrait of an employee who frequently shares risk management skills on social media may show that he has a particularly keen insight into risk factors in the financial market.

[0093] A more accurate and comprehensive talent portrait is constructed by using feature labels and their weights based on four dimensions: basic data, behavioral data, partnerships, and network preferences. The weight distribution of each portrait dimension ensures the personalization and representativeness of the portrait, and can more accurately reflect the multi-dimensional characteristics of talents. The accuracy and wide application of the portrait are significantly improved, and it is suitable for talent assessment and decision support in a variety of scenarios.

[0094] S40, comprehensively analyzing the talent basic portrait, talent behavior portrait, talent association portrait and talent psychological portrait to generate a complete four-dimensional talent portrait for comprehensive evaluation and prediction of talents.

[0095] In this embodiment, the four types of talent profiles (basic profile, behavioral profile, associated profile and psychological profile) that have been constructed are integrated and analyzed in depth. Comprehensive analysis usually uses a variety of data processing and analysis techniques, such as multivariate statistical analysis, machine learning algorithms or fusion models, with the aim of processing the profile data of different dimensions in a unified manner and finding the correlations and patterns between the dimensions.

[0096] After completing the comprehensive analysis, the system generates a complete and unified four-dimensional talent portrait. This portrait combines data from four dimensions, provides a comprehensive perspective, and shows the overall characteristics of the talent. The four-dimensional portrait can help decision makers understand the comprehensive ability, social influence, behavioral patterns, and psychological tendencies of talents.

[0097] The generated four-dimensional talent portrait is used to comprehensively evaluate talents and predict future performance through analysis of historical data and portraits. This prediction can help organizations make more targeted decisions in selection, training, promotion, and team building.

[0098] In one specific implementation, during the comprehensive analysis process, the financial institution used multiple regression analysis, taking the key characteristic variables in the basic profile, behavioral profile, association profile and psychological profile as input, and identifying the interrelationships and influencing factors between the dimensions through the regression analysis model. The resulting four-dimensional talent profile contains a comprehensive score for each employee, which is used for role allocation in the financial project team. The four-dimensional profile is further used to predict the performance of employees under specific market conditions, such as predicting risk tolerance in market fluctuations based on historical behavior patterns.

[0099] In this embodiment, the construction of talent portraits relies on multi-dimensional data input, which can be continuously updated through real-time collection. For example, users' behavior data, social relationship data, and preference data can be dynamically obtained from social media, internal project management systems, academic databases, and other channels.

[0100] When new data enters the system, the solution can immediately pre-process it, including data cleaning, standardization and structuring. This means that the system can dynamically process and integrate new data, so that the portrait data is always up to date.

[0101] This embodiment allows for the dynamic generation or adjustment of labels based on the latest data through statistical analysis methods or machine learning algorithms. For example, as the behavior or social network of talents changes, the system can dynamically adjust their behavior labels, association labels, etc. to reflect the latest characteristics.

[0102] By introducing adaptive algorithms, the system can automatically optimize label generation rules according to changes in data distribution and patterns, so that the labeling process can be dynamically adjusted as data changes. This dynamic generation mechanism ensures the timeliness and accuracy of labels.

[0103] As new data and labels are generated, the system can automatically update the four-dimensional portrait of talents. This means that whenever new behavioral data, partnerships, or network preference data are input, the system can immediately recalculate and generate the latest portrait, ensuring that the portrait always reflects the current status of the talent.

[0104] In the process of building a portrait, by dynamically integrating the features of each dimension and adjusting the weights in real time, the weighted comprehensive analysis can reflect the latest feature patterns. Whether it is a change in key features or an improvement in the analysis algorithm, the system can dynamically update the portrait.

[0105] Because the system can update the portrait in real time, it also has the ability to dynamically predict talent development trends or behavioral changes. The system can use the latest data and portraits to perform trend analysis, generate real-time prediction results, and provide dynamic support for organizational decision-making. The system can also use the difference between the prediction results and actual performance as feedback to dynamically adjust the analysis and label generation algorithms, so that the portrait generation process is continuously optimized and the accuracy is gradually improved.

[0106] The system can automatically adjust and optimize the generation process of labels and portraits as data changes and inputs are updated, ensuring that the final output talent portrait is always accurate, timely, and has high predictive power. This dynamic nature makes the solution particularly suitable for rapidly changing environments and can provide organizations with flexible and real-time decision support.

[0107] By comprehensively analyzing the basic, behavioral, relational, and psychological portraits of talents, a complete four-dimensional talent portrait can be generated, providing strong support for the comprehensive evaluation and prediction of talents. Comprehensive analysis can not only reveal the relationship between the portraits of each dimension, but also effectively integrate information from different dimensions, making the portrait more holistic and predictive. It can more accurately identify the multi-dimensional characteristics of talents, and through advanced analysis and prediction technology, help organizations make more targeted decisions in selection, training, promotion, and team building, thereby improving the efficiency and accuracy of human resource management.

[0108] The present invention relates to a method for constructing a four-dimensional talent portrait, which obtains multi-dimensional talent data, including basic data, professional behavior data, network association data and network preference data, labels these data, extracts and weights various feature labels, and then constructs a basic portrait, behavior portrait, association portrait and psychological portrait of the talent. Combined with these portraits, a comprehensive analysis is performed to generate a complete four-dimensional talent portrait. The present invention improves the accuracy and comprehensiveness of talent portraits through comprehensive analysis and dynamic updating of multi-dimensional data, provides reliable technical support for comprehensive evaluation and development prediction of talents, and effectively solves the problems of low accuracy, incomplete data coverage and untimely updating of talent portraits in the prior art.

[0109] In one embodiment, in the above S30, constructing a basic talent profile based on the basic data tags and corresponding weights includes:

[0110] S301, extracting basic information of talents from multi-dimensional talent data based on the basic data tags, wherein the basic information includes the name, education background and position of the talents;

[0111] S302, performing text processing on the extracted basic information, and extracting multiple core phrases from the basic information of the talent through natural language processing technology;

[0112] S303, assigning each core phrase the same weight as the corresponding basic data label;

[0113] S304, associating the core phrases with basic information of talents, and marking and storing them according to the phrase weights to form a structured talent information corpus;

[0114] S305, constructing a basic profile of the talent based on the core phrases and corresponding phrase weights in the talent information corpus.

[0115] In this embodiment, information related to basic data tags is extracted from a multi-dimensional talent data set. This information includes but is not limited to basic personal information such as the talent's name, education, and position. The extraction of this data may involve obtaining relevant information from multiple data sources (such as HR systems, education background databases, etc.). The extracted basic information is sorted and standardized by text processing technology to ensure data consistency and availability. The content of text processing may include data cleaning, format unification, text normalization, etc. Using natural language processing (NLP) technology, multiple core phrases are automatically identified and extracted from the processed basic information. Core phrases refer to information units that can reflect the key characteristics of talents, such as specific degrees, titles, key descriptions in work experience, etc. Weights are assigned to each extracted core phrase, and these weights are the same as the weights of the corresponding basic data tags. The weight allocation is intended to reflect the importance of each phrase in constructing a basic portrait. The core phrases that have been assigned weights are associated with the original basic information. Subsequently, according to the weights of the phrases, the information is marked and stored in a structured corpus for subsequent processing and analysis. By processing, extracting and marking the basic information, a structured corpus is finally generated. This corpus contains all the core phrases that have been processed and weighted, and is the basis for building a basic profile of talents. By using the core phrases in the corpus and their corresponding weights, a basic profile of talents is constructed. This profile reflects the main characteristics of talents in the basic data dimension and provides a clear summary of personal information.

[0116] Example description:

[0117] In a large financial institution, the department responsible for talent management hopes to more accurately evaluate and manage its executive team by building a basic profile. The following are the specific steps to build a basic talent profile based on basic data tags and corresponding weights.

[0118] Extracting basic information of talents: The financial institution first extracts basic information of executives from the internal HR system and public professional social platforms (such as LinkedIn). This information includes each executive's name, education (such as MBA), position (such as CFO), and professional certification (such as CFA, CPA certification). After data aggregation, this information forms a basic data set for each executive.

[0119] Text processing and core phrase extraction: For the extracted basic information, financial institutions use natural language processing technology (such as TF-IDF) to process text data. For example, by analyzing job descriptions, core phrases such as "Chief Financial Officer", "MBA", and "CFA" are extracted. The text processing process also includes cleaning and standardizing the data to ensure consistency and accuracy.

[0120] Weight assignment and tag storage: To quantify the importance of this information, financial institutions assign each core phrase the same weight as its corresponding underlying data label. For example, "CFA" as a professional certification label has a higher weight because it is very important to the financial decision-making role. All these core phrases are labeled according to their weights and stored in a structured talent information corpus.

[0121] Constructing basic talent portraits: Finally, based on the core phrases and their weights in the corpus, a basic portrait of each executive was constructed. For example, the basic portrait of a certain executive shows that he has an MBA degree, CFA certification, and has served as chief financial officer for more than 5 years. These portraits can help financial institutions fully understand the educational background and professional qualifications of each executive, and provide data support for subsequent performance evaluation, promotion decisions, and succession planning.

[0122] This embodiment can construct a more accurate and personalized basic talent portrait through multi-dimensional extraction of basic data, text processing, core phrase extraction and weighting. Using a structured corpus, the core characteristics of talents are effectively captured and expressed, ensuring that the basic portrait can reflect important information about employees in terms of education, position, professional qualifications and career development path.

[0123] In one embodiment, in the above S30, constructing a talent behavior profile based on the behavior data labels and corresponding weights includes:

[0124] S306, extracting talent behavior information from the multi-dimensional talent data based on the behavior data tag, the behavior information including the number of papers, number of citations, H-Index, G-Index, research field, publication year and frequency of the talent;

[0125] S307, performing standardization processing on the extracted behavior information, and extracting core behavior features based on the standardized behavior information;

[0126] S308, assigning the same weight as the corresponding behavior data label to the core behavior feature, and performing weighted processing on the core behavior feature;

[0127] S309, integrate the weighted core behavioral characteristics to construct a talent behavior profile.

[0128] In this embodiment, behavior-related information is extracted from multi-dimensional talent data. This behavior information may include various indicators of academic or professional performance, such as the number of papers, citations, H-Index, G-Index, research field, publication year and frequency, etc. This data usually comes from academic databases, scientific research management systems or industry databases.

[0129] Standardization is to ensure that behavioral information from different data sources or indicators can be compared on the same scale. Standardization may include converting behavioral data of different units, ranges or formats into a unified scale or range to facilitate subsequent feature extraction and analysis.

[0130] After standardization, the key features that best reflect the talent's behavioral patterns or performance are extracted from the behavioral data. Core behavioral features refer to those behavioral data points with high differentiation or representativeness, such as frequently published research fields, significant academic contributions, etc.

[0131] Each core behavior feature is bound to the corresponding behavior data label and assigned a corresponding weight. The weight indicates the importance of the feature in the behavior profile. Subsequently, weighted processing is performed so that when constructing the behavior profile, features with higher weights can have a greater impact on the profile.

[0132] All weighted core behavioral characteristics are integrated to generate a comprehensive behavioral profile, which can fully display the behavioral patterns and performance of talents in their fields.

[0133] In one specific implementation, a financial research institution extracts behavioral information of financial scholars from multiple academic databases (such as Web of Science, Google Scholar), including the number of their papers, citations, H-Index, and G-Index. Then, the Z-score standardization method is used to standardize this behavioral information so that all indicators can be compared on the same scale. On this basis, the core behavioral characteristics of each scholar are extracted, such as the main research field, the most influential papers, etc. These core behavioral characteristics are assigned the same weights as their corresponding labels. For example, scholars with higher H-Index will receive higher weights. Finally, by integrating these weighted features, a behavioral portrait reflecting the academic influence of each scholar is generated.

[0134] H-Index is an indicator to measure the academic influence of scholars or scientists. The definition of H-index is that a scholar has H papers that are cited at least H times. H-Index can comprehensively consider the number and quality (number of citations) of scholars' papers, and relatively robustly reflect the academic influence of scholars.

[0135] G-Index means that the total number of citations of a scholar's first G papers is at least G2. Compared with H-Index, G-Index pays more attention to highly cited papers and can better reflect the academic contributions of researchers who have published a small number of high-impact papers. G-Index alleviates the problem of H-Index "discounting" highly cited papers to some extent.

[0136] This embodiment builds a talent behavior profile based on behavior data tags and their corresponding weights, which can provide a systematic way to accurately identify and display the behavioral characteristics of talents in their fields. Standardization and weighted integration ensure the comparability and accuracy of behavioral data in different dimensions, so that the behavior profile can fully reflect the actual performance and potential of talents.

[0137] In one embodiment, in the above S30, constructing a talent association profile based on the partnership labels and corresponding weights includes:

[0138] S310, extracting talent cooperation relationship information from the multi-dimensional talent data based on the cooperation relationship tag, wherein the cooperation relationship information includes the relationship between the talent and the mentor, the student, and the collaborator;

[0139] S311, cleaning and standardizing the extracted cooperation relationship information, and extracting network relationship features based on the standardized cooperation relationship data;

[0140] S312, assigning the same weight as the corresponding cooperation relationship label to the extracted network relationship features, and performing weighted processing on these features;

[0141] S313, use solid dots to represent the relationship between talents and mentors, use gray dots to represent the relationship between talents and students, and use white dots to represent the relationship between talents and collaborators. The solid dots, gray dots, and white dots are distributed in the figure according to the relationship type;

[0142] S314, according to the actual contact frequency and cooperation times, use lines to connect the solid dots, gray dots and white dots to form a hedgehog diagram;

[0143] S315, integrate the weighted network relationship features in the hedgehog graph to construct a talent association portrait.

[0144] In this embodiment, information related to the partnership is extracted from multi-dimensional talent data. This information may include relationship data between the talent and his mentor, students, and partners. The data sources may include academic cooperation records, project management systems, social network analysis, etc.

[0145] Data cleaning involves removing redundant, erroneous or incomplete data, while standardization unifies the data into a comparable format for subsequent analysis.

[0146] Key features that represent talent collaborations are extracted from the standardized data. These network relationship features may include collaboration frequency, connection strength, and position in the social network, which help describe the role and influence of talent in their social and professional networks.

[0147] Each network relationship feature is assigned a weight that is the same as the weight of the corresponding partnership label. By weighting, the features that have a significant impact on the partnership can be highlighted.

[0148] In the visualization, dots of different colors and types are used to represent the relationship between talents and different roles. Solid dots, gray dots, and white dots represent the relationship with mentors, students, and collaborators, respectively. These dots are distributed in the figure according to the relationship type to clearly show the collaborative relationship of talents.

[0149] By connecting these dots, a hedgehog diagram is formed to represent the strength of social relationships. The length and thickness of the lines may be based on the frequency of contact and the number of collaborations, helping to visually display the core relationships and network density of talents in their professional networks.

[0150] The weighted network relationship features are integrated to generate a comprehensive talent association portrait, which can fully display the talent's cooperative relationships, network structure and influence in the social network.

[0151] In another specific implementation, an investment company extracts employee partnership information from employees' internal collaboration records and external social platforms, including interaction records with mentors, students, and collaborators. Machine learning algorithms (such as K-means clustering) are used to perform data cleaning and cluster analysis on the extracted information, and highly similar partnerships are grouped together, and the data is standardized. On this basis, network relationship mining technology is used to extract network relationship features, such as the depth and breadth of partnerships. Weights corresponding to the partnership labels are assigned to these features, and weighted processing is performed on them. In the visual chart, dots and lines of different colors are used to represent the frequency of employees' collaboration with different relationship objects, forming a hedgehog diagram. These weighted network relationship features are integrated to generate a comprehensive association portrait, helping companies better understand employees' collaboration networks and their core role in projects.

[0152] This embodiment can systematically identify and display the talent's cooperative relationships and their importance in their professional networks by constructing talent association portraits based on cooperative relationship labels and their corresponding weights. By cleaning, standardizing and weighting cooperative relationship information and displaying it graphically, the constructed association portrait can intuitively reflect the talent's social network structure, cooperative intensity and influence, improving the accuracy and operability of the analysis.

[0153] In one embodiment, in the above S30, constructing a talent psychological profile based on the network preference tags and corresponding weights includes:

[0154] S316, obtaining text information of sample users from the online social platform, and performing data preprocessing on the extracted text information;

[0155] S317, based on the text information of the sample users, network preference data is generated, and structured processing is performed to convert the unstructured text data into structured information including keywords, topic tags, and sentiment analysis results;

[0156] S318, extracting behavioral characteristics of sample users from the structured network preference data, constructing behavioral representations, and establishing personality characteristic models for a preset number of groups based on the behavioral representations;

[0157] S319, using the established personality characteristic model of the preset number of groups, analyzing the network behavior of the talent, extracting the corresponding personality characteristics, and generating the psychological characteristic value of the talent based on the network preference label;

[0158] S320, multiplying each psychological feature value by the weight of the corresponding network preference label to generate a weighted psychological feature value;

[0159] S321, integrate the weighted psychological characteristic values ​​to generate a complete psychological portrait of the talent.

[0160] In this embodiment, sample user text information is collected from social media or other network platforms, such as user posts, comments, sharing, etc. Data preprocessing includes text cleaning (such as removing noise words and HTML tags), word segmentation, stop word filtering, etc. to ensure the quality of the data and the accuracy of subsequent processing.

[0161] Natural language processing (NLP) technology is used to analyze the preprocessed text information, extract core keywords and topic tags, and perform sentiment analysis. Through these steps, the originally unstructured text data is converted into a structured data form that is easy to analyze.

[0162] By analyzing structured network preference data, we can extract features that reflect user behavior, such as frequently discussed topics, emotional tendencies, interaction patterns, etc. These behavioral features constitute the user's behavioral representation. Then, based on these behavioral representations, we build a certain number of models to describe the user's various personality traits.

[0163] Using the personality trait model established in the previous step, we conduct an in-depth analysis of users’ online behaviors, identify their main personality traits (such as extroversion, openness, etc.), and combine these traits with network preference labels to generate psychological trait values.

[0164] The generated psychological feature values ​​will be weighted according to the weights of the network preference labels associated with them to obtain weighted psychological feature values ​​that are more in line with the actual situation and reflect the importance of each feature.

[0165] All weighted psychological characteristic values ​​are integrated to generate a comprehensive psychological portrait of the talent, which can fully reflect the psychological characteristics and online behavior preferences of the talent.

[0166] In one specific implementation, a text preprocessing technology based on a semantic network is used to obtain text information of sample users from a network social platform. This technology not only performs conventional denoising and word segmentation on the text, but also identifies potential semantic relationships and contextual associations in the text by constructing a semantic network. Subsequently, when generating network preference data, a two-layer processing method based on a topic model and sentiment vectorization is adopted. First, the topic of the text is identified by a topic model (such as LDA), and then the sentiment in each topic is quantified and vectorized by a sentiment analysis model (such as VADER). The structured processing uses a structured technology based on a knowledge graph to convert unstructured text data into structured information, and at the same time embeds the relationship of the knowledge graph in the data to enhance the semantic understanding ability of the data. When extracting behavioral features, a behavioral pattern mining algorithm is used to automatically identify repeated patterns and abnormal patterns in user behavior, and to construct behavioral representations based on these patterns. The final personality trait model is established through deep reinforcement learning, which can dynamically adjust and optimize the weights of feature extraction, thereby improving the accuracy of personality trait recognition.

[0167] This embodiment can deeply explore and analyze the psychological characteristics of talents in online behavior by constructing talent psychological portraits based on network preference tags and their corresponding weights. By using the preprocessing of text information, the generation of structured data and the extraction of behavioral characteristics, combined with the preset personality characteristic model, weighted psychological characteristic values ​​are finally generated and integrated into a complete psychological portrait. This not only improves the accuracy and interpretability of psychological portraits, but also enables organizations to more accurately understand and predict the psychological tendencies and behavioral patterns of talents, providing a scientific basis for personalized management and decision-making.

[0168] In one embodiment, in the above S40, the talent basic profile, talent behavior profile, talent association profile and talent psychological profile are comprehensively analyzed to generate a complete four-dimensional talent profile, including:

[0169] S401, performing unified standardization processing on the features in each portrait, so that the features of different dimensions can be compared and analyzed on the same scale;

[0170] S402, combining the key features in the talent basic profile, talent behavior profile, talent association profile, and talent psychological profile with the weights assigned to the feature labels;

[0171] S403, fusing the weighted key features to generate a unified feature set, so that the feature data of each dimension can reflect the comprehensive characteristics of the talent as a whole after fusion;

[0172] S404, using multivariate statistical analysis or machine learning algorithms to comprehensively analyze the fused feature set to identify the main feature patterns of the talent in each dimension;

[0173] S405, visualize the results of the comprehensive analysis to generate a complete four-dimensional talent portrait.

[0174] In this embodiment, the features in the basic profile, behavioral profile, association profile and psychological profile of the talent are standardized. The purpose of standardization is to convert the features of different dimensions into the same scale so that these features can be compared and comprehensively analyzed on the same platform. This processing method usually includes data normalization, standard score (Z-score) calculation, Min-Max scaling, etc.

[0175] The weight of the feature label will be combined with the corresponding key feature. By assigning weights to each key feature, it ensures that important features can receive greater attention and influence in the subsequent analysis and fusion process.

[0176] By fusing the weighted key features of all dimensions, a unified feature set is formed. This fusion process can be done in a variety of ways, such as weighted averaging, feature selection or feature combination, so that the features of each dimension can comprehensively reflect the overall characteristics of the talent.

[0177] The fused feature set will be processed through multivariate statistical analysis or machine learning algorithms to identify the main feature patterns of talents in various dimensions. This may include algorithms such as principal component analysis (PCA), cluster analysis, support vector machine (SVM), neural network, etc.

[0178] The results of the comprehensive analysis are presented through visualization technology to generate a comprehensive, visualized four-dimensional talent portrait. This portrait can clearly show the comprehensive characteristics of the talent and their performance in various dimensions, providing a basis for decision-making.

[0179] In another specific implementation, a financial technology company uses Min-Max scaling to standardize the features of each dimension when processing the multi-dimensional portraits of employees, so that all features are scaled to the same range (such as 0 to 1). On this basis, key features such as job level in the basic talent portrait, innovation ability in the behavioral portrait, network centrality in the association portrait, and decision-making tendency in the psychological portrait are combined with pre-assigned weights. The fusion step adopts a feature combination method to generate a richer feature set by combining multi-dimensional features. In the comprehensive analysis stage, the support vector machine (SVM) algorithm is used to classify and pattern recognize the fused feature set to find the performance patterns of employees in different work scenarios. Finally, a four-dimensional talent portrait is generated through visualization technology, which graphically displays the comprehensive characteristics of employees, their potential in the company, and applicable positions.

[0180] This embodiment generates a complete four-dimensional talent portrait by comprehensively analyzing and weighted fusion of talent portraits in four dimensions. Standardization ensures the comparability of features in different dimensions, while the weighted fusion step enhances the influence of key features in the final portrait. By using multivariate statistical analysis or machine learning algorithms, the main characteristic patterns of talents are identified, and intuitive portrait displays are generated through visualization technology, providing a scientific basis for organizational decision support and talent management.

[0181] In one embodiment, in the above S20, the multi-dimensional talent data is labeled to extract various feature labels of each talent, including:

[0182] S201, extracting key features from the multi-dimensional talent data;

[0183] S202, using a statistical analysis method or a machine learning algorithm to determine a label generation method based on the distribution characteristics of the extracted feature data or the pattern recognition results;

[0184] S203, according to the determined label generation method, formulate corresponding label generation rules, and automatically assign feature labels to the data points of each talent in the multi-dimensional talent data through a set algorithm.

[0185] In this embodiment, the most representative or influential features are identified from multi-dimensional talent data. Key features refer to data points that can fully reflect the talent's ability, behavior, relationship or preference. For example, academic background, work experience, professional skills, cooperative relationships, etc. are all potential key features. Extracting key features usually involves filtering, selecting and preprocessing data.

[0186] After extracting key features, the next step is to determine the method for generating feature labels. This step can use statistical analysis methods (such as cluster analysis, regression analysis) or machine learning algorithms (such as decision trees, support vector machines, neural networks). These methods automatically select the most suitable strategy for generating labels by analyzing the distribution pattern of feature data or the identified pattern.

[0187] Once the label generation method is determined, the next step is to develop specific label generation rules. These rules can be logic rules based on thresholds, classification rules based on pattern recognition results, or decision rules based on statistical analysis. The formulation of rules should ensure that the generated labels can accurately reflect the characteristics of talents and are consistent across different data sets and scenarios.

[0188] Finally, according to the established label generation rules, the characteristic labels are automatically assigned to each talent’s data point through the set algorithm. This process can be automated to ensure that all data points can be accurately labeled, so that multi-dimensional data can be effectively used in subsequent analysis and portrait construction.

[0189] Example: In a large financial institution, the HR department hopes to label employees’ multi-dimensional data through a data-driven approach to better identify and manage different types of talents. Specific steps may include:

[0190] Extracting key features: The HR department extracted multi-dimensional data of employees from the company's data warehouse, including education, years of work experience, project participation, performance ratings, etc. In this data, the HR department used data mining technology to identify several key features, such as the length of work experience, the frequency of project participation, and the stability of performance ratings. These key features are considered to be important indicators of employee capabilities and potential.

[0191] Determine the label generation method: To further analyze these key features, the HR department decided to use the K-means clustering algorithm. By analyzing the distribution of these feature data, the algorithm divides employees into several different groups, such as "high-performance stable type" and "fast-growing type". The division of these groups helps the HR department identify different types of employees and tailor different career development paths for them.

[0192] Formulate label generation rules: Based on the results of K-means clustering, the HR department formulated label generation rules. For example, employees with stable performance scores and long working years are labeled as "high performance stable type", while employees with short working years but frequent project participation are labeled as "fast growth type". These label generation rules ensure that employees can be accurately classified according to their actual performance.

[0193] Automatic label assignment: Finally, the HR department automatically assigns these labels to each employee's data point through a set algorithm. For example, employee A has a high performance score and a long working experience, so he is automatically labeled as "high performance and stable type"; while employee B, although he has a short working experience, has participated in many important projects, so he is labeled as "fast growth type". These labels provide a scientific basis for subsequent talent management, evaluation and training.

[0194] This embodiment can effectively extract and mark the labels that best represent the characteristics of talents by labeling multi-dimensional talent data. Statistical analysis methods or machine learning algorithms are used to ensure the scientificity and accuracy of label generation. The feature labels finally generated can not only reflect the core characteristics of talents, but also provide an accurate data basis for subsequent talent portrait construction and analysis.

[0195] The present invention also provides a four-dimensional talent portrait construction device, referring to Figure 2 , Figure 2 This is a functional module diagram of a preferred embodiment of the four-dimensional talent portrait construction device of the present invention. The four-dimensional talent portrait construction device includes:

[0196] A data acquisition module, which acquires multi-dimensional talent data, wherein the multi-dimensional talent data includes talent basic data, professional behavior data, network association data, and network preference data;

[0197] A labeling processing module performs labeling processing on the multi-dimensional talent data, extracts various characteristic labels of each talent, the characteristic labels include basic data labels, behavioral data labels, partnership labels and network preference labels, and assigns a weight to each characteristic label;

[0198] A portrait construction module, which constructs a basic talent portrait based on the basic data tags and corresponding weights, a talent behavior portrait based on the behavior data tags and corresponding weights, a talent association portrait based on the partnership tags and corresponding weights, and a talent psychological portrait based on the network preference tags and corresponding weights;

[0199] The four-dimensional talent portrait generation module conducts a comprehensive analysis of the talent basic portrait, talent behavior portrait, talent association portrait and talent psychological portrait to generate a complete four-dimensional talent portrait for comprehensive evaluation and prediction of talents.

[0200] The specific implementation of the four-dimensional talent portrait construction device of the present invention is basically the same as the various embodiments of the above-mentioned four-dimensional talent portrait construction method, and will not be repeated here.

[0201] The present invention also provides a four-dimensional talent portrait construction device, such as Figure 3As shown, the four-dimensional talent portrait construction device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0202] Those skilled in the art will understand that Figure 3 The hardware structure of the four-dimensional talent portrait construction device shown in does not constitute a limitation of the four-dimensional talent portrait construction device, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0203] like Figure 3 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a four-dimensional talent portrait construction program. The operating system is a program for managing and controlling the four-dimensional talent portrait construction equipment and software resources, and supports the operation of the network communication module, the user interface module, the four-dimensional talent portrait construction program, and other programs or software; the network communication module is used to manage and control the network interface 1004; the user interface module is used to manage and control the user interface 1003.

[0204] exist Figure 3 In the hardware structure of the four-dimensional talent portrait construction device shown, the network interface 1004 is mainly used to connect to the background server and communicate data with the background server; the user interface 1003 is mainly used to connect to the client and communicate data with the client; the processor 1001 can call the four-dimensional talent portrait construction program stored in the memory 1005, and perform the same operations as the four-dimensional talent portrait construction method.

[0205] The specific implementation of the four-dimensional talent portrait construction device of the present invention is basically the same as the various embodiments of the above-mentioned four-dimensional talent portrait construction method, and will not be repeated here.

[0206] In addition, an embodiment of the present invention further proposes a computer storage medium on which a four-dimensional talent portrait construction program is stored. When the four-dimensional talent portrait construction program is executed by a processor, the steps of the four-dimensional talent portrait construction method as described above are implemented.

[0207] The specific implementation methods of the computer storage medium of the present invention are basically the same as the above-mentioned embodiments of the four-dimensional talent portrait construction method, and will not be repeated here.

Claims

1. A method for constructing a four-dimensional talent portrait, characterized in that: The following steps are involved: Acquire multi-dimensional talent data, wherein the multi-dimensional talent data includes talent basic data, professional behavior data, network association data, and network preference data; Labeling the multi-dimensional talent data, extracting various characteristic labels of each talent, the characteristic labels including basic data labels, behavioral data labels, partnership labels and network preference labels, and assigning weights to each characteristic label; Construct a basic talent portrait based on the basic data tags and corresponding weights, construct a talent behavior portrait based on the behavior data tags and corresponding weights, construct a talent association portrait based on the partnership tags and corresponding weights, and construct a talent psychological portrait based on the network preference tags and corresponding weights; The basic talent portrait, talent behavior portrait, talent association portrait and talent psychological portrait are comprehensively analyzed to generate a complete four-dimensional talent portrait for comprehensive evaluation and prediction of talents.

2. The method for constructing a four-dimensional talent portrait according to claim 1, characterized in that: A basic talent profile is constructed based on the basic data labels and corresponding weights, including: Extracting basic information of talents from the multi-dimensional talent data based on the basic data tags, wherein the basic information includes the name, education background and position of the talents; Perform text processing on the extracted basic information and extract multiple core phrases from the basic information of the talent through natural language processing technology; Assign each core phrase the same weight as the corresponding basic data label; Associating the core phrases with basic information of talents, and marking and storing them according to the phrase weights to form a structured talent information corpus; Based on the core phrases and corresponding phrase weights in the talent information corpus, a basic talent profile is constructed.

3. The method for constructing a four-dimensional talent portrait according to claim 1, characterized in that: Constructing a talent behavior profile based on the behavior data labels and corresponding weights, including: Extracting talent behavior information from multi-dimensional talent data based on the behavior data tags, the behavior information including the number of papers, citations, H-Index, G-Index, research field, publication year and frequency of the talent; Standardize the extracted behavior information, and extract core behavior features based on the standardized behavior information; Assign the same weight to the core behavior features as to the corresponding behavior data labels, and perform weighted processing on the core behavior features; Integrate the weighted core behavioral characteristics to construct a talent behavior profile.

4. The method for constructing a four-dimensional talent portrait according to claim 1, characterized in that: A talent association profile is constructed based on the partnership labels and corresponding weights, including: Extracting talent cooperation relationship information from multi-dimensional talent data based on the cooperation relationship label, wherein the cooperation relationship information includes the relationship between the talent and the mentor, the student and the collaborator; Clean and standardize the extracted cooperative relationship information, and extract network relationship features based on the standardized cooperative relationship data; Assign the same weights as the corresponding partnership labels to the extracted network relationship features, and perform weighted processing on these features; Use solid dots to represent the relationship between talents and mentors, use gray dots to represent the relationship between talents and students, and use white dots to represent the relationship between talents and collaborators. Distribute solid dots, gray dots, and white dots in the figure according to the relationship type. According to the actual contact frequency and number of cooperation, use lines to connect the solid dots, gray dots and white dots to form a hedgehog diagram; Integrate the weighted network relationship features in the hedgehog diagram to construct a talent association portrait.

5. The method for constructing a four-dimensional talent portrait as claimed in claim 1, characterized in that: Constructing a talent psychological profile based on the network preference labels and corresponding weights, including: Obtain text information of sample users from online social platforms and perform data preprocessing on the extracted text information; Based on the text information of sample users, network preference data is generated and structured, converting unstructured text data into structured information including keywords, topic tags, and sentiment analysis results; Extracting the behavioral characteristics of sample users from structured network preference data, constructing behavioral representations, and establishing a personality characteristic model for a preset number of groups based on the behavioral representations; Utilize the established personality trait model of preset number groups to analyze the online behavior of talents, extract the corresponding personality traits, and generate the psychological trait value of talents based on the network preference tags; Multiply each psychological feature value by the weight of the corresponding network preference label to generate a weighted psychological feature value; Integrate the weighted psychological characteristic values ​​to generate a complete psychological portrait of the talent.

6. The method for constructing a four-dimensional talent portrait according to claim 1, characterized in that: Comprehensively analyze the basic talent portrait, talent behavior portrait, talent association portrait and talent psychological portrait to generate a complete four-dimensional talent portrait, including: The features in each portrait are standardized so that features of different dimensions can be compared and analyzed on the same scale; Combine the key features in the talent basic profile, talent behavior profile, talent association profile, and talent psychological profile with the weights assigned to the feature labels; The weighted key features are integrated to generate a unified feature set, so that the feature data of each dimension can reflect the comprehensive characteristics of the talent as a whole after integration; Use multivariate statistical analysis or machine learning algorithms to conduct a comprehensive analysis of the fused feature set to identify the main feature patterns of talents in various dimensions; Visualize the results of the comprehensive analysis to generate a complete four-dimensional talent portrait.

7. The method for constructing a four-dimensional talent portrait according to claim 1, characterized in that: The multi-dimensional talent data is labeled to extract various characteristic labels of each talent, including: extracting key features from the multi-dimensional talent data; Using statistical analysis methods or machine learning algorithms, a label generation method is determined based on the distribution characteristics of the extracted feature data or pattern recognition results; According to the determined label generation method, corresponding label generation rules are formulated, and characteristic labels are automatically assigned to the data points of each talent in the multi-dimensional talent data through the set algorithm.

8. A four-dimensional talent portrait construction device, characterized in that: The four-dimensional talent portrait construction device includes: A data acquisition module, which acquires multi-dimensional talent data, wherein the multi-dimensional talent data includes talent basic data, professional behavior data, network association data, and network preference data; A labeling processing module performs labeling processing on the multi-dimensional talent data, extracts various characteristic labels of each talent, the characteristic labels include basic data labels, behavioral data labels, partnership labels and network preference labels, and assigns a weight to each characteristic label; A portrait construction module, which constructs a basic talent portrait based on the basic data tags and corresponding weights, a talent behavior portrait based on the behavior data tags and corresponding weights, a talent association portrait based on the partnership tags and corresponding weights, and a talent psychological portrait based on the network preference tags and corresponding weights; The four-dimensional talent portrait generation module conducts a comprehensive analysis of the talent basic portrait, talent behavior portrait, talent association portrait and talent psychological portrait to generate a complete four-dimensional talent portrait for comprehensive evaluation and prediction of talents.

9. A four-dimensional talent portrait construction device, characterized in that: The four-dimensional talent portrait construction device includes a memory, a processor, and a four-dimensional talent portrait construction program stored in the memory and executable on the processor. When the four-dimensional talent portrait construction program is executed by the processor, the steps of the four-dimensional talent portrait construction method as described in any one of claims 1 to 7 are implemented.

10. A computer storage medium, characterized in that: The storage medium stores a four-dimensional talent portrait construction program, which, when executed by a processor, implements the steps of the four-dimensional talent portrait construction method as described in any one of claims 1-7.

Images