Scientific and technological innovation talent flow analysis method and device based on multilayer heterogeneous network

Abstract

The invention discloses a scientific and technological innovation talent flow analysis method and device based on a multi-layer heterogeneous network. The method relates to the field of scientific and technological innovation talent big data analysis and complex network modeling, and comprises the following steps: constructing a multilayer heterogeneous network model based on talent flow data; performing node space-time distribution characteristic analysis based on the node distribution characteristics of the multi-layer heterogeneous network model to obtain a node space-time distribution index; based on the node distribution characteristics and the edge distribution characteristics of the multi-layer heterogeneous network model, evolution measurement characteristic analysis is carried out, and an evolution measurement index is obtained; and based on the node space-time distribution index and the evolution measurement index, carrying out scientific and technological innovation talent flow analysis by adopting a preset correlation analysis model to obtain a scientific and technological innovation talent flow analysis result. According to the method, the accuracy of flow analysis of scientific and technological innovation talents can be improved, and a foundation is laid for making talent strategies and optimizing talent resource allocation.

Description

Technical Field

[0001] This invention relates to the field of big data analysis and complex network modeling of scientific and technological innovation talents, and in particular to a method and apparatus for analyzing the flow of scientific and technological innovation talents based on multi-layer heterogeneous networks. Background Technology

[0002] Current technology transfer for scientific and technological innovation talent suffers from the following technical shortcomings: Existing methods can only process single types of data (such as co-authored papers or patent collaborations) and cannot simultaneously integrate multi-dimensional flow trajectories across academia, technology, and career. Traditional static network models struggle to depict the dynamic characteristics of talent flow over time (such as the phased bursts of cross-regional mobility), and mainstream analytical tools do not consider the cross-level coupling effects between macro-policy layers (such as talent introduction programs) and micro-individual layers (such as career choices). Existing heterogeneous network construction methods (such as adjacency matrix-based splicing techniques) suffer from exponentially increasing computational complexity due to data sparsity. Summary of the Invention

[0003] In view of this, the present invention provides a method, device and electronic device for analyzing the flow of scientific and technological innovation talents based on multi-layer heterogeneous networks, the main purpose of which is to solve the problem of inaccurate analysis of the flow of scientific and technological innovation talents.

[0004] To address the aforementioned issues, this application provides a method for analyzing the flow of scientific and technological innovation talent based on multi-layer heterogeneous networks, comprising: Constructing a multi-layered heterogeneous network model based on talent mobility data; Based on the node distribution characteristics of the multi-layer heterogeneous network model, the spatiotemporal distribution characteristics of nodes are analyzed to obtain the spatiotemporal distribution index of nodes. Evolutionary metric features are analyzed based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model to obtain evolutionary metric indices. Based on the spatiotemporal distribution index of the nodes and the evolution measurement index, a preset correlation analysis model is used to analyze the flow of scientific and technological innovation talents, and the analysis results of the flow of scientific and technological innovation talents are obtained.

[0005] Optionally, the construction of a multi-layer heterogeneous network model based on talent mobility data specifically includes: Data transformation is performed on talent mobility data to obtain the point elements and edge elements of a multi-layered heterogeneous network; Based on the point elements and edge elements, the target graph data structure is used to construct models for different network layers to obtain the multi-layer heterogeneous network model.

[0006] Optionally, the step of performing spatiotemporal distribution feature analysis on the node distribution characteristics based on the multi-layer heterogeneous network model to obtain node spatiotemporal distribution indices specifically includes: Based on the multi-layer heterogeneous network model, the first number of talents per unit area of ​​the evaluation region and the first area of ​​the evaluation region are calculated to obtain the talent density index. The talent agglomeration index is obtained by calculating based on the talent density index, the total number of talents in the multi-layer heterogeneous network model, and the total evaluation area of ​​the multi-layer heterogeneous network model. The uniformity index is obtained by calculating based on the first number of talents, the first area, the total number of talents, and the total evaluation area. The organization nodes in different evaluation regions of the node distribution characteristics of the multi-layer heterogeneous network model are screened, and the primacy index is obtained by calculation based on the number of organization nodes in different evaluation regions after screening. The total number of talents, the ranking of the evaluation region, and the number of ranking scale dimensions in different evaluation regions are calculated based on the node distribution characteristics of the multi-layer heterogeneous network model to obtain the ranking scale index. The proportion index is obtained by calculating the number of second talents in the evaluation area per unit area based on the node distribution characteristics of the multi-layer heterogeneous network model. The Gini coefficient index is obtained by calculating the number of regions in different evaluation areas, the average number of talents in different evaluation areas, and the number of second talents in different evaluation areas. The spatiotemporal distribution characteristics of the nodes include talent density indicators, talent agglomeration indicators, uniformity index indicators, primacy indicators, rank size indicators, proportion indicators, and Gini coefficient indicators.

[0007] Optionally, the evolutionary metric analysis based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model to obtain evolutionary metric indices specifically includes: Extract the key nodes of the multi-layer heterogeneous network model; Evaluation metrics are calculated based on the key nodes to obtain key node measurement metrics, including degree centrality, betweenness centrality, and PageRank index. Evaluation indicators are calculated based on the edge distribution characteristics of the multi-layer heterogeneous network model to obtain a talent mobility tendency measure index, which includes a clustering coefficient index and a matching coefficient index. Based on the node distribution characteristics and edge distribution characteristics in the multi-layer heterogeneous network model, a preset intervention opportunity model is used to calculate the evaluation index, and the talent mobility attraction measurement index is obtained. The talent mobility attraction measurement index includes the talent transit rate index and the talent attraction index. The evolution measurement indicators include key node measurement indicators, talent mobility tendency measurement indicators, and talent mobility attraction measurement indicators.

[0008] Optionally, the step of calculating evaluation indicators based on the key nodes to obtain key node measurement indicators specifically includes: The in-degree and out-degree parameters of the key nodes are calculated based on their adjacency matrices to obtain the degree centrality index of the key nodes. The betweenness centrality index of the target key node is obtained by calculating the number of first shortest paths between any two key nodes and the number of second shortest paths through any target key node between the two key nodes. A random walk model consisting of a follow-link propagation network and a random jump fairness network is used to calculate the evaluation index for the key node, and the PageRank index of the key node at different times is obtained.

[0009] Optionally, the step of calculating evaluation indicators based on the edge distribution characteristics of the multi-layer heterogeneous network model to obtain a talent mobility tendency measure index specifically includes: The clustering coefficient is calculated based on the actual number of edges between the current node element and its neighboring node elements and the number of neighboring node elements. The evaluation index is calculated based on the joint probability distribution of any two key nodes and the redundancy distribution of each key node, resulting in the matching coefficient index.

[0010] Optionally, the step of calculating evaluation indicators based on the node distribution characteristics and edge distribution characteristics in the multi-layer heterogeneous network model using a preset intervention opportunity model to obtain a measure of talent mobility attraction specifically includes: The global talent flow index is obtained by calculating the evaluation index based on the number of third-party talents flowing through any two of the key nodes and the total number of flowing talents. Based on the number of third-party talents and the total number of talents in the multi-layer heterogeneous network model, evaluation indicators are calculated to obtain local talent flow indicators. For any scientific and technological innovation talent, a decision is made using the preset intervention opportunity model that integrates radiation-type opportunity model and exploratory-type opportunity model to obtain an attractiveness index of the target location to the scientific and technological innovation talent relative to the starting location. The talent throughput rate indicator includes the global talent flow indicator and the local talent flow indicator.

[0011] Optionally, the analysis of the flow of scientific and technological innovation talents is conducted using a preset correlation analysis model based on the spatiotemporal distribution index of the nodes and the evolutionary measurement index, to obtain the analysis results of the flow of scientific and technological innovation talents, specifically including: The spatiotemporal distribution index of the nodes and the evolutionary measurement index are preprocessed to obtain a first standard feature index corresponding to the spatiotemporal distribution index of the nodes and a second standard feature index corresponding to the evolutionary measurement index. The first standard feature index and the second standard feature index are concatenated to obtain the feature vector of each node; Based on the aforementioned feature vectors, the preset correlation analysis model is used to conduct correlation analysis on the talent mobility results, thereby obtaining the analysis results of the talent mobility of scientific and technological innovation talents.

[0012] To address the aforementioned issues, this application provides a device for analyzing the flow of scientific and technological innovation talent based on a multi-layer heterogeneous network, comprising: The building module is used to construct multi-layer heterogeneous network models based on talent mobility data; The first analysis module is used to perform spatiotemporal distribution feature analysis of nodes based on the node distribution characteristics of the multi-layer heterogeneous network model, and obtain spatiotemporal distribution indicators of nodes. The second analysis module is used to perform evolutionary measure feature analysis based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model, and obtain evolutionary measure indices. The correlation analysis module is used to analyze the flow of scientific and technological innovation talents based on the spatiotemporal distribution index of the nodes and the evolution measurement index using a preset correlation analysis model, and obtain the analysis results of the flow of scientific and technological innovation talents.

[0013] To address the aforementioned problems, this application provides an electronic device, comprising at least a memory and a processor. The memory stores a computer program, and the processor, when executing the computer program in the memory, implements the steps of the aforementioned method for analyzing the flow of scientific and technological innovation talents based on multi-layer heterogeneous networks.

[0014] The beneficial effects of this application include: A systematic analysis of key elements of talent mobility reveals the core mechanisms influencing technological innovation and provides empirical support for policy making. The methodology of this application helps optimize the allocation of human resources, promotes the effective flow of talent between different research institutions and industries, and also drives the construction of regional innovation networks, thereby enhancing innovation capabilities and competitiveness.

[0015] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0016] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 The illustration shows a flowchart of a method for analyzing the flow of scientific and technological innovation talents based on a multi-layer heterogeneous network, as provided in an embodiment of this application. Figure 2 This illustration shows a flowchart of a method for analyzing the flow of scientific and technological innovation talents based on a multi-layer heterogeneous network, according to another embodiment of this application. Figure 3 This illustration shows a partial structural diagram of the multi-layer heterogeneous network model for talent mobility provided in an embodiment of this application. Figure 4 The diagram shows a structural block diagram of a technology innovation talent mobility analysis device based on a multi-layer heterogeneous network, according to another embodiment of this application. Detailed Implementation

[0017] Various embodiments and features of this application are described herein with reference to the accompanying drawings.

[0018] It should be understood that various modifications can be made to the embodiments described herein. Therefore, the above description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope and spirit of this application will be apparent to those skilled in the art.

[0019] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0020] These and other features of this application will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.

[0021] It should also be understood that although this application has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of this application.

[0022] The above and other aspects, features and advantages of this application will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.

[0023] Specific embodiments of this application are described thereafter with reference to the accompanying drawings; however, it should be understood that the claimed embodiments are merely examples of this application, which can be implemented in various ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that could obscure the application. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely serve as the basis and representative basis for the claims to teach those skilled in the art to use this application in a variety of substantially any suitable detailed structures.

[0024] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in other embodiments,” all of which may refer to one or more of the same or different embodiments according to this application.

[0025] This application provides a method for analyzing the flow of scientific and technological innovation talent based on multi-layer heterogeneous networks, such as... Figure 1 As shown, it includes: Step S101: Construct a multi-layer heterogeneous network model based on talent mobility data; In this step, the talent mobility data is transformed to obtain the point elements and edge elements of a multi-layer heterogeneous network. Based on the point elements and edge elements, a model is constructed for different network layers using the target graph data structure to obtain the multi-layer heterogeneous network model.

[0026] Step S102: Based on the node distribution characteristics of the multi-layer heterogeneous network model, perform spatiotemporal distribution characteristic analysis of nodes to obtain spatiotemporal distribution indices of nodes; In this step, the talent density index is obtained by calculating the number of talents per unit area of ​​the evaluation region and the area of ​​the first region in the evaluation region of the multi-layer heterogeneous network model. The talent clustering index is then obtained by calculating the talent density index, the total number of talents in the multi-layer heterogeneous network model, and the total evaluation region area of ​​the multi-layer heterogeneous network model. The uniformity index is obtained by calculating the number of organizational nodes in different evaluation regions based on the first number of talents, the first region area, the total number of talents, and the total evaluation region area. Finally, the organizational nodes in different evaluation regions of the node distribution characteristics of the multi-layer heterogeneous network model are filtered, and the number of organizational nodes in the filtered different evaluation regions is calculated. The primacy index is obtained; the rank of the evaluation region, the rank size dimension, and the total number of talents in different evaluation regions based on the node distribution characteristics of the multi-layer heterogeneous network model are calculated to obtain the rank size index; the number of second talents per unit area in the evaluation region based on the node distribution characteristics of the multi-layer heterogeneous network model are calculated to obtain the proportion index; the Gini coefficient is obtained by calculating the number of regions in different evaluation regions, the average number of talents in different evaluation regions, and the number of second talents in different evaluation regions; wherein, the spatiotemporal distribution characteristics of the nodes include the talent density index, the talent clustering index, the uniformity index, the primacy index, the rank size index, the proportion index, and the Gini coefficient index.

[0027] Step S103: Based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model, perform evolutionary measure feature analysis to obtain evolutionary measure indices; In the specific implementation process, this step extracts the key nodes of the multi-layer heterogeneous network model; Evaluation indicators are calculated based on the key nodes to obtain key node measurement indicators, including degree centrality, betweenness centrality, and PageRank index. Evaluation indicators are also calculated based on the edge distribution characteristics of the multi-layer heterogeneous network model to obtain talent mobility tendency measurement indicators, including clustering coefficient and isomatch coefficient. Furthermore, evaluation indicators are calculated using a preset intervention opportunity model based on the node and edge distribution characteristics of the multi-layer heterogeneous network model to obtain talent mobility attraction measurement indicators, including talent transit rate and talent attractiveness. The evolutionary measurement indicators include key node measurement indicators, talent mobility tendency measurement indicators, and talent mobility attraction measurement indicators.

[0028] Step S104: Based on the spatiotemporal distribution index of the nodes and the evolution measurement index, a preset correlation analysis model is used to analyze the flow of scientific and technological innovation talents, and the analysis results of the flow of scientific and technological innovation talents are obtained.

[0029] In this step, the spatiotemporal distribution index of the nodes and the evolution measurement index are preprocessed to obtain a first standard feature index corresponding to the spatiotemporal distribution index of the nodes and a second standard feature index corresponding to the evolution measurement index; the first standard feature index and the second standard feature index are concatenated to obtain the feature vector of each node; based on each feature vector, the preset correlation analysis model is used to perform correlation analysis on the talent mobility results to obtain the analysis results of the mobility of scientific and technological innovation talents.

[0030] This application, through a systematic analysis of key elements of talent mobility, reveals the core mechanisms influencing technological innovation and provides empirical support for policy making. This approach helps optimize the allocation of human resources, promotes the effective flow of talent between different research institutions and industries, and also drives the construction of regional innovation networks, thereby enhancing innovation capabilities and competitiveness.

[0031] Another embodiment of this application provides a different method for analyzing the flow of scientific and technological innovation talent based on multi-layer heterogeneous networks, such as... Figure 2 As shown, it includes: Step S201: Transform the talent mobility data to obtain the point elements and edge elements of the multi-layer heterogeneous network; In this step, the collected data is cleaned and integrated, then transformed into nodes and edges in a multi-layered heterogeneous network. Node elements can include individual nodes (such as scientific and technological innovation talents), organizational nodes (such as universities and research institutes), and geographical nodes (such as cities). The node element type table is shown in Table 1.

[0032] Edge nodes can represent different types of relationships, such as geographical flow relationships, spatial association relationships, cooperative relationships, and subordinate relationships. The edge element type table is shown in Table 2:

[0033] Step S202: Based on the point elements and the edge elements, the target graph data structure is used to construct models for different network layers to obtain the multi-layer heterogeneous network model; In the specific implementation process of this step, based on the talent mobility element set and the constructed dataset, and combined with the content of talent mobility network characteristic analysis, the sub-network layers in the multi-layer heterogeneous network model are clarified, as well as the physical meaning of the nodes and network edges in the sub-network layers, providing a foundation for subsequent dynamic evolution characteristic analysis. Considering the needs of talent mobility network characteristic analysis and influencing factor discovery, the main network layers constructed are shown in Table 3:

[0034] Based on the types of nodes and edges, multiple sub-networks are constructed, each representing a specific relationship or attribute. A graph data structure is used to represent the network, where different types of nodes and edges correspond to different network layers. The graph data structure can be an adjacency matrix or an adjacency list, etc. When constructing the network structure, the connection relationships and weight information between nodes and edges need to be considered to accurately reflect the flow of scientific and technological innovation talent in the real world. Connections between nodes can be established based on work experience, academic cooperation, etc. For example, in a talent association network, scientific and technological innovation talent can be regarded as nodes, and an individual association network can be established through basic information. The nodes in the established multiple sub-network layers are analyzed. Individual, city, and organization-based standardized nodes are used as nodes in the multi-layered heterogeneous network. The relationships between edges in the sub-network layers are used to describe the node relationships in the multi-layered heterogeneous network. Following the multi-layered network modeling method, a multi-layered heterogeneous network of talent flow is constructed, providing a model foundation for subsequent analysis of evolutionary characteristics and influencing factors. Figure 3 The diagram shows a partial structural schematic of a multi-layered heterogeneous network model for talent mobility. The multi-layered heterogeneous network is based on a dynamic network model and describes the distribution and mobility characteristics of talent from different perspectives such as cities and organizations.

[0035] Step S203: Based on the node distribution characteristics of the multi-layer heterogeneous network model, perform spatiotemporal distribution characteristic analysis of nodes to obtain spatiotemporal distribution indices of nodes; In the specific implementation process of this step, the first number of talents per unit area of ​​the evaluation region and the first area of ​​the evaluation region are calculated based on the multi-layer heterogeneous network model to obtain the talent density index; the talent density is used to represent the number of talents per unit land area / field / ability, and the calculation formula is as follows (1):

[0036] in, Indicators representing talent density This indicates the number of people with a specific skill or ability within a particular administrative region / field. This represents the area of ​​a city's administrative division / the number of secondary domains contained in a certain primary domain / the number of dimensions of a certain capability.

[0037] Based on the talent density index, the total number of talents in the multi-layer heterogeneous network model, and the total evaluation area of ​​the multi-layer heterogeneous network model, a talent agglomeration index is obtained. Talent agglomeration represents the degree of agglomeration of a region relative to the uniform distribution of talents nationwide, and can be expressed as the proportion of talents a certain region has agglomerated on the national land area. The mathematical expression is shown in the following formula (2):

[0038] in, yes The concentration of talent in the region; yes The number of talented people in the region; yes The land area of ​​the region; It represents the total number of talents in multi-layer heterogeneous network models; It is the total evaluation region area of ​​the multi-layer heterogeneous network model. When talent concentration... A talent concentration index (TCI) indicates that the proportion of talent concentrated in a certain region per unit of land area is higher than the national average, indicating that the talent resources are relatively abundant. When the talent concentration index is less than 1, it indicates that the region is short of talent.

[0039] The uniformity index is calculated based on the first number of talents, the first area, the total number of talents, and the total evaluation area. The uniformity index quantitatively represents the dispersion and concentration of talents within a certain area. This method is used to characterize the degree of balance in the distribution of talents in cities across the country. The mathematical expression is shown in the following formula (3):

[0040] in, This represents the proportion of each unit indicator to the total indicator. The mathematical expression can be shown in the following formula (4):

[0041] in, The uniformity index, Indicates region Number of talents This indicates the area of ​​the city's administrative division. It represents the total number of talents in multi-layer heterogeneous network models; It represents the total evaluation region area of ​​the multi-layer heterogeneous network model; it also represents the proportion of each unit indicator to the total indicator. The value range is (0,1). The closer it is to 1, the more evenly the talent is distributed. The smaller the value, the more concentrated the talent is.

[0042] The organizational nodes in different evaluation regions of the node distribution characteristics of the multi-layer heterogeneous network model are screened, and the primacy index is obtained based on the number of organizational nodes in different evaluation regions after screening. The primacy of talents in geographical regions, fields and abilities is described to characterize the concentration of different talents in the largest city / largest field / organization, reflecting the degree of the city / field / ability with the highest concentration of talents in the country compared with the second and fourth concentrated cities / fields / organizations. The calculation is performed using the two-city method and the four-city method. The mathematical expression for calculating the primacy using the two-city method can be shown in the following formula (5):

[0043] The mathematical expression for calculating the primacy using the four-city method can be shown in the following formula (6):

[0044] in, and These represent the primacy of two cities / two sectors / two organizations and the primacy of four cities / four sectors / four organizations, respectively. The city / sector / organization ranked first in terms of the number of representative talents. , , The cities, fields, and organizations that rank 2nd, 3rd, and 4th in terms of the number of representative talents.

[0045] Based on the node distribution characteristics of the multi-layer heterogeneous network model, the total number of talents in different evaluation regions, the ranking of the evaluation regions, and the number of ranking scale dimensions are calculated to obtain the ranking scale index. The ranking scale measure examines the scale distribution of an indicator from the perspective of the relationship between the talent scale of a node (geography, field, and organization) and its ranking. The mathematical expression of talent / field / organization ranking and scale can be shown by the following formula (7):

[0046] The mathematical expression for the logarithmic rank index can also be expressed as shown in the following formula (8):

[0047] in, Represents the number of talents at a certain node. It represents its position or order. It is a constant. The dimension representing the positional-scale is usually considered to be when When the distribution structure is relatively balanced, it indicates that the distribution structure is relatively balanced, while when... The distribution is relatively dispersed at certain times, while when When the clustering is high, it indicates that the node has a high degree of clustering, which is used to indicate the structural rationality of the node.

[0048] The number of second talents in the evaluation area per unit area based on the node distribution characteristics of the multi-layer heterogeneous network model is calculated to obtain the proportion index; the proportion is used to represent the ratio of the number of talents in a certain node to the total number of talents, reflecting the concentration of talents in a certain city, field, or organization, and its mathematical expression is shown in the following formula (9):

[0049] in, Representing proportion, yes The number of talents at the node It represents the total number of talents in multi-layer heterogeneous network models.

[0050] The Gini coefficient is calculated based on the number of regions in different evaluation areas, the average number of talents in different evaluation areas, and the number of second-level talents in different evaluation areas. The Gini coefficient is used to measure the degree of uneven distribution of talents in urban nodes in the talent mobility network. The mathematical expression can also be shown in the following formula (10):

[0051] in, Number of cities; The average number of talents in a representative city; These represent the talent density of any two cities. The value ranges from 0 to 1, where 0 represents a geographically uniform distribution and 1 represents a geographically concentrated distribution. A value between 0.2 and 0.4 indicates a relatively reasonable geographical distribution; A value between 0.4 and 0.5 indicates a significant geographical variation in distribution; A value greater than 0.5 indicates a significant geographical disparity. The spatiotemporal distribution characteristics of the nodes include talent density, talent agglomeration, uniformity index, primacy index, rank size index, proportion index, and Gini coefficient index.

[0052] Step S204: Based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model, perform evolutionary measure feature analysis to obtain evolutionary measure indices; In the specific implementation process, this step extracts the key nodes of the multi-layer heterogeneous network model; Evaluation metrics are calculated based on the key nodes to obtain key node metrics, including degree centrality, betweenness centrality, and PageRank index. Specifically, the in-degree and out-degree parameters of the key nodes are calculated based on their adjacency matrices to obtain the degree centrality index. Degree centrality reflects the concentration of talent. In directed networks, the degree of a node is divided into in-degree and out-degree parameters. in-degree parameter This refers to starting from other nodes and pointing to... The number of edges; nodes Out-degree parameter It refers to from The number of edges originating from other nodes. For an adjacency matrix... For the network The mathematical expression for the in-degree parameter in the degree centrality index can be expressed by the following formula (11):

[0053] The mathematical expression for the out-degree parameter can be expressed as follows (12):

[0054] The betweenness centrality index of the target key node is obtained by calculating the number of first shortest paths between any two key nodes and the number of second shortest paths through any target key node between the two key nodes. Betweenness centrality reflects the centrality of the flow nodes in talent mobility. It describes the number of shortest paths through a node and numerically describes the node's control over the information transmission on the shortest paths through it. The mathematical expression of the betweenness centrality index can be shown by the following formula (13):

[0055] in, For the node To the node The total number of shortest paths, For the node To the node The shortest path passes through the nodes The number. Assuming that information always travels along the shortest path between two points, if... If it is 0, it means that the node For nodes With nodes There is no direct control over information transmission between them.

[0056] A random walk model, consisting of a follow-link propagation network and a random jump fairness network, is used to calculate evaluation metrics for the key nodes, resulting in PageRank indexes for the key nodes at different times. PageRank comprehensively considers both talent flow and talent quality to evaluate key cities, organizations, and fields. Its basic idea is that the importance of a node on a network depends on the quantity and quality of other pages pointing to it. For general directed networks, PageRank first assigns initial values ​​to all nodes. , PageRank introduces a constant. Each node The probability is to randomly select one of the outgoing edges and move along that edge to the next node, so that... The probability is that a node is randomly selected on the network. The algorithm iterates continuously according to a certain correction rule, and at the th... When walking, The mathematical formula for calculating the value is shown in the following formula (14):

[0057] Evaluation indices are calculated based on the edge distribution characteristics of the multi-layer heterogeneous network model to obtain a measure of talent mobility tendency. These indices include clustering coefficients and assortment coefficients. Specifically, the clustering coefficient is calculated based on the actual number of edges between the current node and its neighboring nodes, and the number of neighboring nodes. The clustering coefficient is used to measure the tightness of talent mobility. In a talent mobility network, the degree of talent mobility between adjacent institutions or fields within an organization or field reflects the tightness of their communication network. Let the nodes in the network... Degree is Then it has Given a set of neighboring nodes, assuming all neighboring nodes are also neighbors of each other, then there exists a relationship between these adjacent nodes. Edges. However, in reality, such a situation is rare. The mathematical formula for calculating the clustering coefficient index is as follows (15):

[0058] in, express of The actual number of edges connecting each neighbor node. If or ,but .

[0059] The matching coefficient is calculated based on the joint probability distribution of any two key nodes and the redundancy distribution of each key node. The matching coefficient reflects whether talent exchange exhibits a homogeneous tendency, i.e., whether talent is more likely to migrate between similar institutions, cities, or other entities. If the presence or absence of an edge connecting two nodes in the network is irrelevant to the degree values ​​of those two nodes, the network is said to be degree-independent, or neutral; otherwise, the network is degree-dependent.

[0060] in, Is the degree as , The joint probability distribution of edges connecting nodes. It is a node with degree . The redundancy distribution. Taking degree-matching as an example, in degree-related networks... and The difference between the two values ​​is not identical; it is used to characterize the degree of homo-matching or hetero-matching in the network. The mathematical expression for the homo-matching coefficient index can be shown in the following formula (16):

[0061] in, Redundancy distribution The variance. Clearly, ,like Then the network is compatible; if If so, the network is heterogeneous.

[0062] Based on the node distribution characteristics and edge distribution characteristics in the multi-layer heterogeneous network model, a preset intervention opportunity model is used to calculate evaluation indicators, resulting in a talent mobility attraction measure index. This index includes a talent transit rate index and a talent attraction index. The number of third-party talents flowing through any two key nodes is also considered. and total number of mobile talents The evaluation indicators are calculated to obtain the global talent flow index; the mathematical formula for calculating the global talent flow index can be shown in the following formula (17):

[0063] Evaluation metrics are calculated based on the number of third-party talents and the total number of talents in the multi-layer heterogeneous network model to obtain a local talent flow index; the local talent flow index is defined as the number of people flowing along the path. Total number of people at the starting point The proportion of local talent flow indicators can be calculated using the following formula (18):

[0064] For any scientific and technological innovation talent, a pre-defined intervention opportunity model, which integrates the radial opportunity model and the exploratory opportunity model, is used for decision-making to obtain an attractiveness index of the destination relative to the origin for the talent. This is combined with the intervention opportunity model in population mobility prediction to measure the intensity of personnel mobility in the talent mobility network. Based on the characteristics of talent mobility behavior, a unified opportunity model integrating the radial and exploratory opportunity models is proposed. It is assumed that individuals compare the magnitude of the origin benefit, intervention opportunity benefit, and destination benefit when choosing a destination. In the radial model, individuals tend to choose a destination with a higher return than the origin benefit and a lower intervention opportunity benefit than the origin benefit, reflecting a cautious tendency. In the exploratory model, individuals exhibit a strong exploratory tendency. The unified opportunity model uses these two tendencies as parameters, with the exploratory tendency represented by a parameter and the cautious tendency represented by a parameter. The model's approach is that individuals first evaluate the benefit value of the destination opportunity when choosing a destination. After evaluating the opportunity benefits of different locations, individuals comprehensively compare the magnitude of the destination benefit, origin benefit, and intervention opportunity benefit, and then choose a specific location as their destination. Based on the rules of the above model, the attractiveness of an entity to individuals within that entity can be obtained. The mathematical formula for calculating the attractiveness index is shown in the following formula (19):

[0065] in, yes Opportunity; Location Opportunity; It is an opportunity for intervention, that is, between the destination and the destination. From the starting point Total opportunities across all locations; , There are two non-negative parameters, and they have The talent flow rate indicator includes both the global talent flow indicator and the local talent flow indicator. The evolutionary measurement indicators include key node measurement indicators, talent mobility tendency measurement indicators, and talent mobility attraction measurement indicators.

[0066] Step S205: Based on the spatiotemporal distribution index of the nodes and the evolution measurement index, a preset correlation analysis model is used to analyze the flow of scientific and technological innovation talents, and the analysis results of the flow of scientific and technological innovation talents are obtained.

[0067] In this step, the spatiotemporal distribution indicators and evolutionary measurement indicators of the nodes are preprocessed to obtain a first standard feature indicator corresponding to the spatiotemporal distribution indicators of the nodes and a second standard feature indicator corresponding to the evolutionary measurement indicators. The first and second standard feature indicators are then concatenated to obtain feature vectors for each node. Based on these feature vectors, a pre-defined correlation analysis model is used to perform correlation analysis on the talent mobility results, yielding the analysis results of the flow of scientific and technological innovation talent. Correlation analysis is a further analysis of the association rules, aiming to explain the reasons and mechanisms behind them. Statistical analysis and visualization analysis methods are used to analyze the model's output results to explore the correlation between network evolutionary characteristics and talent mobility results. The pre-defined correlation analysis model can be a model trained using the Apriori algorithm or the FP-Growth algorithm, with model parameters set (such as minimum support and minimum confidence). The model is trained using training data to generate association rules, and based on these rules, the evolutionary feature set is used as the independent variable, and the talent mobility results (inflow, outflow, and personnel classification flow probability statistics) are used as the dependent variable to construct the correlation analysis model. Finally, the correlation between network evolution characteristics and talent mobility is calculated. Furthermore, the model is evaluated and optimized during training. Specifically, cross-validation and other methods are used to evaluate the established association analysis model, and its performance is assessed based on relevant metrics (such as accuracy, recall, coverage, and boost). Based on the evaluation results, the model is optimized in several ways, including adjusting model parameters (such as minimum support and minimum confidence) to improve performance, improving the Apriori and FP-Growth algorithms to increase efficiency, selecting different features to reduce model complexity, or expanding the training dataset to improve robustness.

[0068] This application, through a systematic analysis of key elements of talent mobility, reveals the core mechanisms influencing technological innovation and provides empirical support for policy making. This approach helps optimize the allocation of human resources, promotes the effective flow of talent between different research institutions and industries, and also drives the construction of regional innovation networks, thereby enhancing innovation capabilities and competitiveness.

[0069] Another embodiment of this application provides a device for analyzing the flow of scientific and technological innovation talent based on a multi-layer heterogeneous network, such as... Figure 4 As shown, it includes: Module 1 is used to build a multi-layer heterogeneous network model based on talent mobility data; The first analysis module 2 is used to perform spatiotemporal distribution feature analysis of nodes based on the node distribution characteristics of the multi-layer heterogeneous network model, and obtain spatiotemporal distribution indicators of nodes. The second analysis module 3 is used to perform evolutionary measure feature analysis based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model, and obtain evolutionary measure indices. The correlation analysis module 4 is used to perform a correlation analysis analysis of the flow of scientific and technological innovation talents based on the spatiotemporal distribution index of the nodes and the evolution measurement index, and to obtain the analysis results of the flow of scientific and technological innovation talents.

[0070] In the specific implementation process, the construction module 1 is specifically used to: transform talent flow data to obtain point elements and edge elements of a multi-layer heterogeneous network; and construct models for different network layers based on the point elements and edge elements using the target graph data structure to obtain the multi-layer heterogeneous network model.

[0071] In specific implementation, the first analysis module 2 is specifically used to calculate the first number of talents per unit area of ​​the evaluation region and the first area of ​​the evaluation region based on the multi-layer heterogeneous network model to obtain a talent density index; to calculate the talent agglomeration index based on the talent density index, the total number of talents in the multi-layer heterogeneous network model, and the total evaluation region area of ​​the multi-layer heterogeneous network model; to calculate the uniformity index based on the first number of talents, the first area, the total number of talents, and the total evaluation region area; and to filter the organizational nodes of different evaluation regions in the node distribution characteristics of the multi-layer heterogeneous network model, and to perform calculations based on the number of organizational nodes in the filtered different evaluation regions. The calculation process yields the primacy index; based on the node distribution characteristics of the multi-layer heterogeneous network model, the total number of talents in different evaluation regions, the rank of the evaluation region, and the rank size dimension are calculated to obtain the rank size index; based on the node distribution characteristics of the multi-layer heterogeneous network model, the number of second talents per unit area in the evaluation region is calculated to obtain the proportion index; based on the number of regions in different evaluation regions, the average number of talents in different evaluation regions, and the number of second talents in different evaluation regions, the calculation process yields the Gini coefficient index; wherein, the spatiotemporal distribution characteristics of the nodes include the talent density index, talent clustering index, uniformity index, primacy index, rank size index, proportion index, and Gini coefficient index.

[0072] In the specific implementation process, the second analysis module 3 is specifically used for: extracting key nodes of the multi-layer heterogeneous network model; calculating evaluation indicators based on the key nodes to obtain key node measurement indicators, including degree centrality, betweenness centrality, and PageRank index; calculating evaluation indicators based on the edge distribution characteristics of the multi-layer heterogeneous network model to obtain talent mobility tendency measurement indicators, including clustering coefficient and isomatch coefficient; calculating evaluation indicators based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model using a preset intervention opportunity model to obtain talent mobility attraction measurement indicators, including talent transit rate and talent attraction index; wherein, the evolution measurement indicators include key node measurement indicators, talent mobility tendency measurement indicators, and talent mobility attraction measurement indicators.

[0073] In the specific implementation process, the second analysis module 3 is also used to: calculate the in-degree parameter and out-degree parameter of the key node based on the adjacency matrix of the key node to obtain the degree centrality index of the key node; calculate the betweenness centrality index of the target key node by performing calculation processing based on the number of first shortest paths between any two key nodes and the number of second shortest paths through any target key node between the two key nodes; and use a random walk model composed of a follow-link propagation network and a random jump fairness network to calculate the evaluation index of the key node to obtain the PageRank index of the key node at different times.

[0074] In the specific implementation process, the second analysis module 3 is also used to: calculate the evaluation index based on the actual number of edges between the current node element and the neighboring node elements and the number of neighboring node elements to obtain the clustering coefficient index; and calculate the evaluation index based on the joint probability distribution of any two key nodes and the redundancy distribution of each key node to obtain the isomatch coefficient index.

[0075] In the specific implementation process, the second analysis module 3 is also used to: calculate the evaluation index based on the number of third talents flowing through any two key nodes and the total number of flowing talents to obtain the global talent flow index; calculate the evaluation index based on the number of third talents and the total number of talents in the multi-layer heterogeneous network model to obtain the local talent flow index; and make a decision for any scientific and technological innovation talent by using the preset intervention opportunity model that integrates the radiation-type opportunity model and the exploration-type opportunity model to obtain the attraction index of the destination relative to the starting point for the scientific and technological innovation talent; wherein, the talent transit rate index includes the global talent flow index and the local talent flow index.

[0076] In the specific implementation process, the correlation analysis module 4 is specifically used for: preprocessing the spatiotemporal distribution index of the nodes and the evolution measurement index to obtain a first standard feature index corresponding to the spatiotemporal distribution index of the nodes and a second standard feature index corresponding to the evolution measurement index; concatenating the first standard feature index and the second standard feature index to obtain the feature vector of each node; and performing correlation analysis on the talent mobility results based on each feature vector using the preset correlation analysis model to obtain the analysis results of the flow of scientific and technological innovation talents.

[0077] This application, through a systematic analysis of key elements of talent mobility, reveals the core mechanisms influencing technological innovation and provides empirical support for policy making. This approach helps optimize the allocation of human resources, promotes the effective flow of talent between different research institutions and industries, and also drives the construction of regional innovation networks, thereby enhancing innovation capabilities and competitiveness.

[0078] Another embodiment of this application provides an electronic device, which can be a server. The electronic device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. The program executed by the processor can implement the server-side functions or steps of a method for analyzing the flow of scientific and technological innovation talent based on a multi-layered heterogeneous network.

[0079] In one embodiment, an electronic device is provided, which can be a client. The electronic device includes a processor, memory, a network interface, a display screen, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with an external server via a network connection. The program of the electronic device, executed by the processor, can implement client-side functions or steps of a method for analyzing the flow of technological innovation talent based on a multi-layered heterogeneous network.

[0080] Another embodiment of this application provides an electronic device, including at least a memory and a processor. The memory stores a computer program, and the processor, when executing the computer program in the memory, performs the following method steps: Step 1: Construct a multi-layered heterogeneous network model based on talent mobility data; Step 2: Based on the node distribution characteristics of the multi-layer heterogeneous network model, perform spatiotemporal distribution characteristic analysis of nodes to obtain spatiotemporal distribution indices of nodes; Step 3: Based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model, perform evolutionary measure characteristic analysis to obtain evolutionary measure indices; Step 4: Based on the spatiotemporal distribution index of the nodes and the evolution measurement index, a preset correlation analysis model is used to analyze the flow of scientific and technological innovation talents, and the analysis results of the flow of scientific and technological innovation talents are obtained.

[0081] The specific implementation process of the above method steps can be found in any of the above embodiments of the analysis method for the flow of scientific and technological innovation talents based on multi-layer heterogeneous networks, and will not be repeated here.

[0082] This application, through a systematic analysis of key elements of talent mobility, reveals the core mechanisms influencing technological innovation and provides empirical support for policy making. This approach helps optimize the allocation of human resources, promotes the effective flow of talent between different research institutions and industries, and also drives the construction of regional innovation networks, thereby enhancing innovation capabilities and competitiveness.

[0083] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.

Claims

1. A method for analyzing the flow of scientific and technological innovation talent based on multi-layer heterogeneous networks, characterized in that, include: Constructing a multi-layered heterogeneous network model based on talent mobility data; Based on the node distribution characteristics of the multi-layer heterogeneous network model, the spatiotemporal distribution characteristics of nodes are analyzed to obtain the spatiotemporal distribution index of nodes. Evolutionary metric features are analyzed based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model to obtain evolutionary metric indices. Based on the spatiotemporal distribution index of the nodes and the evolution measurement index, a preset correlation analysis model is used to analyze the flow of scientific and technological innovation talents, and the analysis results of the flow of scientific and technological innovation talents are obtained.

2. The method as described in claim 1, characterized in that, The construction of a multi-layer heterogeneous network model based on talent mobility data specifically includes: Data transformation is performed on talent mobility data to obtain the point elements and edge elements of a multi-layered heterogeneous network; Based on the point elements and edge elements, the target graph data structure is used to construct models for different network layers to obtain the multi-layer heterogeneous network model.

3. The method as described in claim 1, characterized in that, The node spatiotemporal distribution feature analysis based on the node distribution characteristics of the multi-layer heterogeneous network model yields node spatiotemporal distribution indices, specifically including: Based on the multi-layer heterogeneous network model, the first number of talents per unit area of ​​the evaluation region and the first area of ​​the evaluation region are calculated to obtain the talent density index. The talent agglomeration index is obtained by calculating based on the talent density index, the total number of talents in the multi-layer heterogeneous network model, and the total evaluation area of ​​the multi-layer heterogeneous network model. The uniformity index is obtained by calculating based on the first number of talents, the first area, the total number of talents, and the total evaluation area. The organization nodes in different evaluation regions of the node distribution characteristics of the multi-layer heterogeneous network model are screened, and the primacy index is obtained by calculation based on the number of organization nodes in different evaluation regions after screening. The total number of talents, the ranking of the evaluation region, and the number of ranking scale dimensions in different evaluation regions are calculated based on the node distribution characteristics of the multi-layer heterogeneous network model to obtain the ranking scale index. The proportion index is obtained by calculating the number of second talents in the evaluation area per unit area based on the node distribution characteristics of the multi-layer heterogeneous network model. The Gini coefficient index is obtained by calculating the number of regions in different evaluation areas, the average number of talents in different evaluation areas, and the number of second talents in different evaluation areas. The spatiotemporal distribution characteristics of the nodes include talent density indicators, talent agglomeration indicators, uniformity index indicators, primacy indicators, rank size indicators, proportion indicators, and Gini coefficient indicators.

4. The method as described in claim 1, characterized in that, The evolutionary metric is obtained by performing evolutionary metric feature analysis based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model, specifically including: Extract the key nodes of the multi-layer heterogeneous network model; Evaluation metrics are calculated based on the key nodes to obtain key node measurement metrics, including degree centrality, betweenness centrality, and PageRank index. Evaluation indicators are calculated based on the edge distribution characteristics of the multi-layer heterogeneous network model to obtain a talent mobility tendency measure index, which includes a clustering coefficient index and a matching coefficient index. Based on the node distribution characteristics and edge distribution characteristics in the multi-layer heterogeneous network model, a preset intervention opportunity model is used to calculate the evaluation index, and the talent mobility attraction measurement index is obtained. The talent mobility attraction measurement index includes the talent transit rate index and the talent attraction index. The evolution measurement indicators include key node measurement indicators, talent mobility tendency measurement indicators, and talent mobility attraction measurement indicators.

5. The method as described in claim 4, characterized in that, The step of calculating evaluation indicators based on the key nodes to obtain key node measurement indicators specifically includes: The in-degree and out-degree parameters of the key nodes are calculated based on their adjacency matrices to obtain the degree centrality index of the key nodes. The betweenness centrality index of the target key node is obtained by calculating the number of first shortest paths between any two key nodes and the number of second shortest paths through any target key node between the two key nodes. A random walk model consisting of a follow-link propagation network and a random jump fairness network is used to calculate the evaluation index for the key node, and the PageRank index of the key node at different times is obtained.

6. The method as described in claim 4, characterized in that, The evaluation index is calculated based on the edge distribution characteristics of the multi-layer heterogeneous network model to obtain a measure of talent mobility tendency, specifically including: The clustering coefficient is calculated based on the actual number of edges between the current node element and its neighboring node elements and the number of neighboring node elements. The evaluation index is calculated based on the joint probability distribution of any two key nodes and the redundancy distribution of each key node, resulting in the matching coefficient index.

7. The method as described in claim 4, characterized in that, The evaluation index is calculated using a preset intervention opportunity model based on the node distribution characteristics and edge distribution characteristics in the multi-layer heterogeneous network model, resulting in a measure of talent mobility attraction, specifically including: The global talent flow index is obtained by calculating the evaluation index based on the number of third-party talents flowing through any two of the key nodes and the total number of flowing talents. Based on the number of third-party talents and the total number of talents in the multi-layer heterogeneous network model, evaluation indicators are calculated to obtain local talent flow indicators. For any science and technology innovation talent, a decision is made using the preset intervention opportunity model that integrates radiation-type opportunity model and exploratory-type opportunity model to obtain an attractiveness index of the target location relative to the starting location for the science and technology innovation talent. The talent throughput rate indicator includes the global talent flow indicator and the local talent flow indicator.

8. The method as described in claim 1, characterized in that, The analysis of the flow of scientific and technological innovation talents is conducted using a preset correlation analysis model based on the spatiotemporal distribution index of the nodes and the evolutionary measure index, resulting in the analysis of the flow of scientific and technological innovation talents, specifically including: The spatiotemporal distribution index of the nodes and the evolutionary measurement index are preprocessed to obtain a first standard feature index corresponding to the spatiotemporal distribution index of the nodes and a second standard feature index corresponding to the evolutionary measurement index. The first standard feature index and the second standard feature index are concatenated to obtain the feature vector of each node; Based on the aforementioned feature vectors, the preset correlation analysis model is used to conduct correlation analysis on the talent mobility results, thereby obtaining the analysis results of the talent mobility of scientific and technological innovation talents.

9. A device for analyzing the flow of scientific and technological innovation talent based on a multi-layer heterogeneous network, characterized in that, include: The building module is used to construct multi-layer heterogeneous network models based on talent mobility data; The first analysis module is used to perform spatiotemporal distribution feature analysis of nodes based on the node distribution characteristics of the multi-layer heterogeneous network model, and obtain spatiotemporal distribution indicators of nodes. The second analysis module is used to perform evolutionary measure feature analysis based on the node distribution characteristics and edge distribution characteristics of the multi-layer heterogeneous network model, and obtain evolutionary measure indices. The correlation analysis module is used to analyze the flow of scientific and technological innovation talents based on the spatiotemporal distribution index of the nodes and the evolution measurement index using a preset correlation analysis model, and to obtain the analysis results of the flow of scientific and technological innovation talents.

10. An electronic device, characterized in that, It includes at least a memory and a processor, wherein the memory stores a computer program, and the processor, when executing the computer program in the memory, implements the steps of the method for analyzing the flow of scientific and technological innovation talents based on a multi-layer heterogeneous network as described in any one of claims 1-8.

Images