Scientific and technological innovation talent flow causal mechanism discovery method and device and electronic equipment

Abstract

The invention discloses a scientific and technological innovation talent flow causal mechanism discovery method and device and electronic equipment, and relates to the cross field of causal science and talent big data analysis, and the method comprises the steps: carrying out the calculation processing based on the talent flow historical data, and obtaining the transition probability between talent flow historical events; based on each transition probability and the talent flow historical data, adopting a Debrueine graph to construct a model, and obtaining a causal model for discovering a technological innovation talent flow causal mechanism; and performing intervention operation on the to-be-analyzed influence factors based on a preset talent flow probability model to obtain a target path of the causal model of the target talent flow influenced by the to-be-analyzed influence factors. According to the method, causal analysis and calculation are carried out on the influence factors and the talent flow by utilizing a causal inference method, the causal relationship between the influence factors and the flow results in the talent flow is quantified, and the scientificity and effectiveness of policy design are improved.

Description

Technical Field

[0001] This invention relates to the interdisciplinary field of causal science and talent big data analysis, and in particular to a method, apparatus and electronic device for discovering the causal mechanism of the flow of scientific and technological innovation talents. Background Technology

[0002] Due to the complexity and unexperimentability of social systems, research on talent mobility management strategies often follows empirical analysis, which limits its support for policy design and analysis. In addition, many phenomena in reality are characterized by ambiguity and uncertainty, and much decision-making information cannot be simply described using precise values, leading to inaccurate causal analysis of the mobility of scientific and technological innovation talents. Therefore, there is an urgent need for a discovery method that can construct the causal mechanism of talent mobility under ambiguity. Summary of the Invention

[0003] In view of this, the present invention provides a method, apparatus and electronic device for discovering the causal mechanism of the flow of scientific and technological innovation talents, the main purpose of which is to solve the problem of inaccurate causal analysis of the flow of scientific and technological innovation talents.

[0004] To address the aforementioned issues, this application provides a method for discovering the causal mechanism of the flow of scientific and technological innovation talent, comprising: Based on historical data on talent mobility, the probability of transition between historical talent mobility events is calculated and processed. Based on the aforementioned transition probabilities and the aforementioned historical data on talent mobility, a de Bruin graph is used to construct a model, resulting in a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. Intervention operations are performed on the influencing factors to be analyzed based on a preset talent mobility probability model to obtain the target path of the causal model of the influencing factors to be analyzed affecting the mobility of target talent.

[0005] Optionally, the method further includes: The pre-defined talent mobility probability model and the causal model are used to evaluate the factors that influence the mobility of innovative talents, and the average causal effect evaluation results corresponding to the factors are obtained.

[0006] Optionally, the calculation and processing based on historical talent mobility data to obtain the transfer probability between historical talent mobility events specifically includes: For a target talent mobility historical event, determine the first number of subsequent talent mobility historical events that occur after the target talent mobility historical event occurs; The transition probability between the historical talent mobility events is obtained by performing a division operation based on the first quantity and the second quantity of the target talent mobility historical events.

[0007] Optionally, the model is constructed using de Bruin graphs based on the transition probabilities and the historical data of talent mobility to obtain a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation, specifically including: Using each historical talent mobility event as the first node and the aforementioned transition probability as the first edge weight, a de Bruin graph is used to construct the model, resulting in a first-order causal model. An initial transition matrix is ​​constructed based on the aforementioned transition probabilities; Using the initial transition matrix as the first basis matrix, iterative multiplication operations are performed on the first basis matrix and the initial transition matrix to obtain a first steady-state probability transition matrix that satisfies the preset iteration termination condition. Using each edge in the first-order causal model as a second node, and the first steady-state probability transition matrix as the weight of the second edge, a de Bruin graph is used to construct the model, thus obtaining a second-order causal model.

[0008] Optionally, the method further includes: Using the steady-state probability transition matrix as the second basis matrix, iterative multiplication is performed on the second basis matrix and the steady-state probability transition matrix to obtain a second steady-state probability transition matrix that satisfies the preset iteration termination condition. By taking each edge in the second-order causal model as the third node and using the second steady-state probability transition matrix as the weight of the third edge, a de Bruin graph is used to construct the model, thus obtaining a third-order causal model.

[0009] Optionally, the intervention operation based on the preset talent mobility probability model to obtain the target path of the causal model of the influence of the influencing factors on the target talent mobility specifically includes: The preset talent mobility probability model is used to predict the influencing factors to be analyzed, and the first predicted probability is obtained. Interventions were performed on the influencing factors to be analyzed, resulting in intervention texts; The preset talent mobility probability model is used to predict the intervention text to obtain a second predicted probability; The total causal effect value is obtained by calculating based on the first predicted probability and the second predicted probability. When the total causal effect is non-zero, the mediating effect of the target talent mobility is analyzed to obtain the target path of the causal model of the influencing factors on the target talent mobility.

[0010] Optionally, the mediation effect analysis on the target talent mobility yields the target path of the causal model of the influencing factors on the target talent mobility. Specifically, this includes: Based on the causal model, the intermediary nodes for the flow of the target talent are identified; For the intermediary node, the preset talent mobility probability model is used to predict the original text of the influencing factors to be analyzed, and a third predicted probability corresponding to the intermediary node is obtained. The natural direct effect value is obtained by performing calculations based on the third predicted probability and the first predicted probability. For the intermediary node, the preset talent mobility probability model is used to predict the intervention text of the influencing factors to be analyzed, and a fourth prediction probability corresponding to the intermediary node is obtained. The natural indirect effect value is obtained by performing calculations based on the fourth predicted probability and the first predicted probability. Based on the natural direct effect value, the natural indirect effect value, and the total causal effect value, the causal model is located to obtain the target path of the influencing factors to be analyzed affecting the flow of target talent.

[0011] To address the aforementioned issues, this application provides a device for discovering the causal mechanism of the flow of scientific and technological innovation talent, comprising: The calculation module is used to perform calculations based on historical talent mobility data to obtain the transition probabilities between historical talent mobility events. A construction module is used to construct a model based on the aforementioned transition probabilities and the historical data of talent mobility using de Bruin graphs, thereby obtaining a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. The intervention module is used to intervene in the influencing factors to be analyzed based on a preset talent mobility probability model, so as to obtain the target path of the causal model of the influencing factors to be analyzed affecting the mobility of target talents.

[0012] To address the aforementioned problems, this application provides a storage medium storing a computer program that, when executed by a processor, implements the steps of the method for discovering the causal mechanism of the flow of scientific and technological innovation talents described above.

[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 discovering the causal mechanism of the flow of scientific and technological innovation talents.

[0014] The beneficial effects of this application are as follows: This application models the higher-order causality of the flow of scientific and technological innovation talents using de Bruin diagrams, visualizes the causal mechanism, and then uses an intervention-based interpretability model to accurately locate the causal path on the causal diagram, obtaining the target path of the causal model influencing the flow of target talents by the factors to be analyzed. This can be combined with the actual needs of macro-management to analyze the management strategies for talent flow and propose suggestions for improving and optimizing policies. The talent flow management decision analysis method promotes the integration of theory and practice, provides methodological tools to support macro-policy design and analysis, and thus improves the scientificity and effectiveness of talent flow management decisions.

[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 A flowchart illustrating a method for discovering the causal mechanism of the flow of scientific and technological innovation talents provided in an embodiment of this application is shown. Figure 2 A flowchart illustrating a method for discovering the causal mechanism of the flow of scientific and technological innovation talents according to another embodiment of this application is shown. Figure 3 This illustration shows a causal graph diagram of modeling higher-order causality using a de Bruin graph, according to another embodiment of this application. Figure 4 A structural block diagram of a device for discovering the causal mechanism of the flow of scientific and technological innovation talents according to another embodiment of this application is shown. 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 discovering the causal mechanism of the flow of scientific and technological innovation talents, such as... Figure 1 As shown, it includes: Step S101: Calculate and process historical talent mobility data to obtain the transition probability between historical talent mobility events; In the specific implementation process of this step, for the target talent mobility historical event, a first number of subsequent talent mobility historical events after the target talent mobility historical event occurs is determined; based on the first number and the second number of the target talent mobility historical events, a division operation is performed to obtain the transition probability between the talent mobility historical events.

[0026] Step S102: Based on the aforementioned transition probabilities and the historical data of talent mobility, a de Bruin diagram is used to construct a model, resulting in a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. In the specific implementation process of this step, a de Bruin diagram is used to construct a model based on the aforementioned transition probabilities and the historical data of talent mobility, resulting in a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. The causal model can be a second-order causal model, a third-order causal model, or a higher-order causal model.

[0027] Step S103: Based on the preset talent mobility probability model, intervene in the factors to be analyzed to obtain the target path of the causal model of the factors to be analyzed affecting the mobility of target talent.

[0028] In this step, the preset talent mobility probability model is used to predict the influencing factors to be analyzed, obtaining a first predicted probability; intervention is applied to the influencing factors to be analyzed, obtaining intervention text; the preset talent mobility probability model is used to predict the intervention text, obtaining a second predicted probability; based on the first and second predicted probabilities, a total causal effect value is obtained; when the total causal effect is non-zero, the mediating effect of the target talent mobility is analyzed to obtain the target path of the causal model influencing the target talent mobility by the influencing factors to be analyzed. The preset talent mobility probability model can be a pre-trained linear regression model, a deep learning model, etc.

[0029] This application uses de Bruin diagrams to model the higher-order causality of the flow of scientific and technological innovation talents, visualizing the causal mechanism. Secondly, it employs an intervention-based interpretability model to precisely locate the causal path on the causal diagram, obtaining the target path of the causal model influencing the flow of target talents. This allows for analysis of talent flow management strategies and the formulation of policy improvement and optimization recommendations, tailored to the actual needs of macro-management. This talent flow management decision analysis method promotes the integration of theory and practice, providing methodological tools to support macro-policy design and analysis, thereby enhancing the scientific rigor and effectiveness of talent flow management decisions.

[0030] Another embodiment of this application provides a method for discovering another causal mechanism of the flow of scientific and technological innovation talents, such as... Figure 2 As shown, it includes: Step S201: Calculate and process historical talent mobility data to obtain the transition probability between historical talent mobility events; In the specific implementation process of this step, for the target talent mobility historical event, a first number of subsequent talent mobility historical events after the target talent mobility historical event is determined; based on the first number and the second number of the target talent mobility historical events, a division operation is performed to obtain the transition probability between the talent mobility historical events. The mathematical formula for calculating the transition probability can be shown in the following formula (1): (1) Specifically, the transition probability between historical talent mobility events is calculated by the frequency of these events in the historical talent mobility data. For each event i, the number of occurrences of its subsequent event j is counted, and the weight, i.e., the transition probability, is obtained by dividing the frequency by the total number of occurrences of event i.

[0031] Step S202: Based on the aforementioned transition probabilities and the historical data of talent mobility, a de Bruin diagram is used to construct a model to obtain a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. In the specific implementation process of this step, each historical talent mobility event is taken as the first node and the transfer probability is taken as the first side weight. The de Bruin graph is used to construct the model to obtain a first-order causal model. Figure 3 This diagram illustrates a causal graph modeling of higher-order causality using de Bruin graphs. Figure 3 (a) is a first-order causal model. Each first node in the model represents an event. The edges in the graph represent the transition probabilities between events, which represent the probability that a downstream event will occur due to the occurrence of an event at the first node. An initial transition matrix is ​​constructed based on the aforementioned transition probabilities. Edge weights of the second-order model: If there is a continuous path X→Y→Z in the first-order model, then the edge weight from node X→Y to Y→Z in the second-order model is equal to the probability of Y→Z in the first-order model. For example, for the node C→A, since in the first-order model, the upstream change can only be B→C, and the subsequent change can only be A→B, it can be seen that the numbers on the edges are all 1. According to the above rules, an initial transition matrix is ​​constructed. The mathematical expression of the initial transition matrix can be expressed by the following formula (2): (2) Where n is the number of nodes in the second-order model, each This represents the transition probabilities between nodes in the second-order model. Using the initial transition matrix as the first basis matrix, iterative multiplication is performed on the first basis matrix and the initial transition matrix to obtain a first steady-state probability transition matrix that satisfies a preset iteration termination condition; specifically, the probabilities of each row are summed to equal 1. ,make The initial transition matrix is ​​shown in the following formula (3): (3) Through multiple matrix multiplications (exponentiation) ,when Stop iteration when the time is right, and put the last one As the steady-state probability transition matrix of the second-order model, each element represents the probability of transition between nodes in the model. Taking each edge of the first-order causal model as a second node, and using the first steady-state probability transition matrix as the weights of the second edges, a de Brouin graph is used to construct the model, resulting in the second-order causal model, as follows. Figure 3 As shown in (b), the causal model is obtained. The steady-state probability transition matrix is ​​used as the second basis matrix. Iterative multiplication is performed based on the second basis matrix and the steady-state probability transition matrix to obtain a second steady-state probability transition matrix that satisfies the preset iteration termination condition. Each edge in the second-order causal model is used as a third node. The second steady-state probability transition matrix is ​​used as the weight of the third edge, and a de Bruin graph is used to construct the model, resulting in a third-order causal model. The same operation is performed again on the second-order model to obtain the third-order causal graph as shown in (b). Figure 3 As shown in (c). In addition, the appearance of some non-existent edges is due to the existence of such connections in the first-order model network. For example, there should not be a connection from B→D→B to D→B→D, because the connection represented by these two points is a self-loop.

[0032] Step S203: Construct a pre-defined talent mobility probability model; In this step, historical factors influencing talent mobility are used as model inputs, and the probability of talent mobility occurrence is used as the model output. A linear regression model is employed to construct the model, resulting in a pre-defined talent mobility probability model. Specifically, historical data on factors influencing talent mobility are acquired. This historical data includes policy-related data and environmental data. Policy-related data includes salary subsidy intensity, settlement policy level, and tax incentive strength, while environmental data includes regional GDP, unemployment rate, and industry competitiveness index. The historical data on factors influencing talent mobility undergoes data preprocessing, including missing value handling, standardization, and categorical variable encoding. The historical data is then labeled to obtain a labeled dataset; labels are binary variables, with 0 representing no mobility and 1 representing mobility. The labeled dataset is divided into training, validation, and test sets. The linear regression model is trained using the training set to obtain an initial talent mobility probability model that meets the required accuracy. The trained initial talent mobility probability model is validated using the validation set, and its hyperparameters are tuned to obtain the pre-defined talent mobility probability model. The preset talent mobility probability model can also be obtained by training a deep learning network model. The deep learning network model can be a recurrent neural network model, a multilayer perceptron (MLP), a convolutional neural network (1D-CNN), a long short-term memory network (LSTM), or other deep learning network models.

[0033] Step S204: Use the preset talent mobility probability model and the causal model to evaluate the factors affecting the mobility of innovative talents, and obtain the average causal effect evaluation results corresponding to the factors to be analyzed; In the specific implementation process, this step involves determining intervention variables based on the influencing factors to be analyzed. Determine the intervention value corresponding to the intervention variable. Before intervention, the preset talent mobility probability model is used to predict the influencing factors to be analyzed for different talent samples, obtaining the predicted first probability value corresponding to different talent samples; the mean of each predicted probability value is calculated to obtain the predicted probability benchmark. The intervention variables in the influencing factors to be analyzed are replaced with intervention values ​​to obtain the intervention text; the preset talent mobility probability model is used to predict the intervention text for different talent samples to obtain the predicted second probability values ​​corresponding to different talent samples; the expected value of each predicted second probability value is calculated to obtain the mean of the predicted intervention probability. The average causal effect assessment result is obtained by subtracting the mean predicted intervention probability and the predicted probability benchmark. The mathematical expression of the average causal effect can be expressed as follows (4): (4) Here, ATE represents the average causal effect, indicating the input features. The average causal effect on the output is to Assignment The difference between the output and the benchmark output is the final result of ATE, representing the average causal effect of the intervention variable on the outcome variable. For example, if the intervention variable is the salary subsidy level, and the intervention value is to forcibly set the salary subsidy to a certain value (e.g., from 0.3 to 0.7), the final average causal effect assessment result is that increasing the salary subsidy from the current level to 0.7 will increase the probability of talent mobility by an average of 15%.

[0034] Step S205: Use the preset talent mobility probability model to predict the influencing factors to be analyzed, and obtain the first predicted probability; In the specific implementation process of this step, the preset talent mobility probability model is used to predict the influencing factors to be analyzed, and the first predicted probability is obtained. . i represents the i-th talent sample.

[0035] Step S206: Intervene in the influencing factors to be analyzed to obtain the intervention text; In this step, symbols in the input text that are irrelevant to talent policies are replaced with neutral symbols. For example, when analyzing the impact of talent policies on mobility, if the input text (such as policy documents, user comments, etc.) contains symbols that are irrelevant to the policy but may interfere with the model (such as special characters, irrelevant keywords, biased words, etc.), they need to be replaced with neutral symbols to eliminate noise. For instance, if the policy text of the influencing factors to be analyzed includes the word "holiday," which is irrelevant to talent mobility policies, the holiday should be replaced with a placeholder or a meaningless filler word such as "this factor" or "other conditions" to obtain the intervention text.

[0036] Step S207: Use the preset talent mobility probability model to predict the intervention text and obtain a second prediction probability; In the specific implementation process of this step, the preset talent mobility probability model is used to predict the intervention text to obtain the second prediction probability. .

[0037] Step S208: Calculate the total causal effect value based on the first predicted probability and the second predicted probability; In the specific implementation process of this step, for the target talent sample, the mathematical expression of the individual total causal effect value can be shown by the following formula (5): (5) For all talent samples, the mathematical expression for the total causal effect value can be expressed by the following formula (6): (6) in, E This represents the expected operation.

[0038] Step S209: When the total causal effect is non-zero, the mediating effect of the target talent mobility is analyzed to obtain the target path of the causal model of the influencing factors to be analyzed affecting the target talent mobility.

[0039] In the specific implementation of this step, when the total causal effect is non-zero, it indicates that the neutral sign of the substitution has an impact, and it is necessary to further locate the target path of the causal model in which the influencing factors to be analyzed affect the mobility of target talent. Based on the causal model, the mediating nodes of the mobility of target talent are determined. For the intermediary node, the preset talent mobility probability model is used to predict the original text of the influencing factors to be analyzed, and a third predicted probability corresponding to the intermediary node is obtained. The natural direct effect value is obtained by calculation based on the third predicted probability and the first predicted probability. The mathematical formula for calculating the natural direct effect value can be shown in the following formula (7): (7) In complex network models, intermediate nodes can represent various elements of the model, such as a hidden neuron or a weight in an attention mechanism. The value of this element only needs to be manually fixed to the value observed before intervening with the input text. Natural direct effects (NDEs) represent the effects of direct pathways.

[0040] For the intermediary node, the preset talent mobility probability model is used to predict the intervention text of the influencing factors to be analyzed, and a fourth predicted probability corresponding to the intermediary node is obtained. The natural indirect effect value is obtained by calculation based on the fourth predicted probability and the first predicted probability; the mathematical expression of the natural indirect effect value can be expressed by the following formula (8): (8) Natural indirect effects (NIEs) represent the effects of influence through indirect, mediated paths. Based on the natural direct effect value, the natural indirect effect value, and the total causal effect value, the causal model is located to obtain the target path of the influencing factor's impact on the flow of target talent. When the absolute value of the difference between the natural direct effect value and the total causal effect value is less than or equal to a first preset threshold, the direct path of the causal model is located to obtain the target path of the causal model of the influencing factor's impact on the flow of target talent. When the absolute value of the difference between the natural direct effect value and the total causal effect value is greater than a second preset threshold, a target intermediary node is determined based on a fourth predicted probability, and the indirect path containing the intermediary node with the highest fourth predicted probability is determined as the target path of the causal model of the influencing factor's impact on the flow of target talent.

[0041] This application models the higher-order causality of the flow of scientific and technological innovation talents using de Bruin diagrams, visualizing the causal mechanism. Secondly, it employs an intervention-based interpretability model to precisely locate the causal path on the causal diagram, obtaining the target path of the causal model influencing the flow of target talents. This approach can be combined with the actual needs of macro-management to analyze talent flow management strategies and propose policy improvement and optimization suggestions. The talent flow management decision analysis method promotes the integration of theory and practice, providing methodological tools to support macro-policy design and analysis, thereby enhancing the scientific nature and effectiveness of talent flow management decisions.

[0042] Another embodiment of this application provides a device for discovering the causal mechanism of the flow of scientific and technological innovation talents, such as... Figure 4 As shown, it includes: Calculation module 1 is used to perform calculations based on historical talent mobility data to obtain the transition probabilities between historical talent mobility events. Module 2 is used to construct a model based on the aforementioned transition probabilities and the historical data of talent mobility using de Brouin diagrams, thereby obtaining a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. Intervention module 3 is used to perform intervention operations on the influencing factors to be analyzed based on a preset talent mobility probability model, so as to obtain the target path of the causal model of the influencing factors to be analyzed affecting the mobility of target talents.

[0043] In the specific implementation process, the device further includes: an average causal effect evaluation module, which is specifically used to: evaluate the influencing factors to be analyzed that affect the flow of innovative talents using a preset talent mobility probability model and the causal model, and obtain the average causal effect evaluation result corresponding to the influencing factors to be analyzed.

[0044] In the specific implementation process, the calculation module 1 is specifically used to: determine the first number of subsequent talent mobility historical events after the occurrence of the target talent mobility historical event; and perform a division operation based on the first number and the second number of the target talent mobility historical events to obtain the transition probability between the talent mobility historical events.

[0045] In the specific implementation process, the construction module 2 is specifically used for: using each historical talent flow event as the first node and each of the aforementioned transition probabilities as the first edge weights to construct a model using a de Bruin graph, thereby obtaining a first-order causal model; constructing an initial transition matrix based on each of the aforementioned transition probabilities; using the initial transition matrix as the first basis matrix, performing iterative multiplication operations based on the first basis matrix and the initial transition matrix to obtain a first steady-state probability transition matrix that satisfies a preset iteration termination condition; using each edge in the first-order causal model as the second node, using the first steady-state probability transition matrix as the second edge weights to construct a model using a de Bruin graph, thereby obtaining a second-order causal model, thus obtaining the causal model.

[0046] In the specific implementation process, the construction module 2 is also used to: take the steady-state probability transition matrix as the second basis matrix, perform iterative multiplication operation based on the second basis matrix and the steady-state probability transition matrix to obtain a second steady-state probability transition matrix that satisfies the preset iteration termination condition; take each edge in the second-order causal model as the third node, use the second steady-state probability transition matrix as the third edge weight and use de Bruin graph to construct the model to obtain a third-order causal model, so as to obtain the causal model.

[0047] In the specific implementation process, the intervention module 3 is specifically used to: predict the influencing factor to be analyzed using the preset talent mobility probability model to obtain a first predicted probability; intervene in the influencing factor to be analyzed to obtain an intervention text; predict the intervention text using the preset talent mobility probability model to obtain a second predicted probability; calculate and process based on the first predicted probability and the second predicted probability to obtain a total causal effect value; when the total causal effect is non-zero, analyze the mediating effect of the target talent mobility to obtain the target path of the causal model of the influencing factor to be analyzed affecting the target talent mobility.

[0048] In the specific implementation process, the intervention module 3 is further used to: determine the intermediary node of the target talent flow based on the causal model; predict the original text of the influencing factor to be analyzed using the preset talent flow probability model for the intermediary node to obtain a third predicted probability corresponding to the intermediary node; calculate and process the third predicted probability and the first predicted probability to obtain the natural direct effect value; predict the intervention text of the influencing factor to be analyzed using the preset talent flow probability model for the intermediary node to obtain a fourth predicted probability corresponding to the intermediary node; calculate and process the fourth predicted probability and the first predicted probability to obtain the natural indirect effect value; locate the causal model based on the natural direct effect value, the natural indirect effect value and the total causal effect value to obtain the target path of the influencing factor to be analyzed affecting the target talent flow.

[0049] This application models the higher-order causality of the flow of scientific and technological innovation talents using de Bruin diagrams, visualizing the causal mechanism. Secondly, it employs an intervention-based interpretability model to precisely locate the causal path on the causal diagram, obtaining the target path of the causal model influencing the flow of target talents. This approach can be combined with the actual needs of macro-management to analyze talent flow management strategies and propose policy improvement and optimization suggestions. The talent flow management decision analysis method promotes the integration of theory and practice, providing methodological tools to support macro-policy design and analysis, thereby enhancing the scientific nature and effectiveness of talent flow management decisions.

[0050] Another embodiment of this application provides a storage medium storing a computer program, which, when executed by a processor, implements the following method steps: Step 1: Calculate and process historical talent mobility data to obtain the transfer probabilities between historical talent mobility events; Step 2: Based on the aforementioned transition probabilities and the historical data of talent mobility, a de Bruin diagram is used to construct a model, resulting in a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. Step 3: Based on the preset talent mobility probability model, intervene in the factors to be analyzed to obtain the target path of the causal model of the factors to be analyzed affecting the mobility of target talent.

[0051] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0052] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0053] The specific implementation process of the above method steps can be found in the embodiment of the above-mentioned method for discovering the causal mechanism of the flow of scientific and technological innovation talents, which will not be repeated here.

[0054] This application models the higher-order causality of the flow of scientific and technological innovation talents using de Bruin diagrams, visualizing the causal mechanism. Secondly, it employs an intervention-based interpretability model to precisely locate the causal path on the causal diagram, obtaining the target path of the causal model influencing the flow of target talents. This approach can be combined with the actual needs of macro-management to analyze talent flow management strategies and propose policy improvement and optimization suggestions. The talent flow management decision analysis method promotes the integration of theory and practice, providing methodological tools to support macro-policy design and analysis, thereby enhancing the scientific nature and effectiveness of talent flow management decisions.

[0055] 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. When the program is executed by the processor, it implements server-side functions or steps of a method for discovering the causal mechanism of technological innovation talent mobility.

[0056] 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. When the program of the electronic device is executed by the processor, it implements client-side functions or steps of a method for discovering the causal mechanism of technological innovation talent mobility.

[0057] 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: Calculate and process historical talent mobility data to obtain the transfer probabilities between historical talent mobility events; Step 2: Based on the aforementioned transition probabilities and the historical data of talent mobility, a de Bruin diagram is used to construct a model, resulting in a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. Step 3: Based on the preset talent mobility probability model, intervene in the factors to be analyzed to obtain the target path of the causal model of the factors to be analyzed affecting the mobility of target talent.

[0058] The specific implementation process of the above method steps can be found in the embodiment of the above-mentioned method for discovering the causal mechanism of the flow of scientific and technological innovation talents, which will not be repeated here.

[0059] This application models the higher-order causality of the flow of scientific and technological innovation talents using de Bruin diagrams, visualizing the causal mechanism. Secondly, it employs an intervention-based interpretability model to precisely locate the causal path on the causal diagram, obtaining the target path of the causal model influencing the flow of target talents. This approach can be combined with the actual needs of macro-management to analyze talent flow management strategies and propose policy improvement and optimization suggestions. The talent flow management decision analysis method promotes the integration of theory and practice, providing methodological tools to support macro-policy design and analysis, thereby enhancing the scientific nature and effectiveness of talent flow management decisions.

[0060] 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 discovering the causal mechanism of the flow of scientific and technological innovation talents, characterized in that, include: Based on historical data on talent mobility, the probability of transition between historical talent mobility events is calculated and processed. Based on the aforementioned transition probabilities and the aforementioned historical data on talent mobility, a de Bruin graph is used to construct a model, resulting in a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. Intervention operations are performed on the influencing factors to be analyzed based on a preset talent mobility probability model to obtain the target path of the causal model of the influencing factors to be analyzed affecting the mobility of target talent.

2. The method as described in claim 1, characterized in that, The method further includes: The pre-defined talent mobility probability model and the causal model are used to evaluate the factors that influence the mobility of innovative talents, and the average causal effect evaluation results corresponding to the factors are obtained.

3. The method as described in claim 1, characterized in that, The calculation and processing based on historical talent mobility data to obtain the transition probability between historical talent mobility events specifically includes: For a target talent mobility historical event, determine the first number of subsequent talent mobility historical events that occur after the target talent mobility historical event occurs; The transition probability between the historical talent mobility events is obtained by performing a division operation based on the first quantity and the second quantity of the target talent mobility historical events.

4. The method as described in claim 1, characterized in that, The de Bruin diagram is used to construct a model based on the aforementioned transition probabilities and historical talent mobility data, resulting in a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. Specifically, this model includes: Using each historical talent mobility event as the first node and the aforementioned transition probability as the first edge weight, a de Bruin graph is used to construct the model, resulting in a first-order causal model. An initial transition matrix is ​​constructed based on the aforementioned transition probabilities; Using the initial transition matrix as the first basis matrix, iterative multiplication operations are performed on the first basis matrix and the initial transition matrix to obtain a first steady-state probability transition matrix that satisfies the preset iteration termination condition. Using each edge in the first-order causal model as a second node, and the first steady-state probability transition matrix as the weight of the second edge, a de Bruin graph is used to construct the model, thus obtaining a second-order causal model.

5. The method as described in claim 4, characterized in that, The method further includes: Using the steady-state probability transition matrix as the second basis matrix, iterative multiplication is performed on the second basis matrix and the steady-state probability transition matrix to obtain a second steady-state probability transition matrix that satisfies the preset iteration termination condition. By taking each edge in the second-order causal model as the third node and using the second steady-state probability transition matrix as the weight of the third edge, a de Bruin graph is used to construct the model, thus obtaining a third-order causal model.

6. The method as described in claim 1, characterized in that, The intervention operation based on the preset talent mobility probability model to analyze the influencing factors, and to obtain the target path of the causal model of the influencing factors to analyze affecting the target talent mobility, specifically includes: The preset talent mobility probability model is used to predict the influencing factors to be analyzed, and the first predicted probability is obtained. Interventions were performed on the influencing factors to be analyzed, resulting in intervention texts; The preset talent mobility probability model is used to predict the intervention text to obtain a second predicted probability; The total causal effect value is obtained by calculating based on the first predicted probability and the second predicted probability. When the total causal effect is non-zero, the mediating effect of the target talent mobility is analyzed to obtain the target path of the causal model of the influencing factors on the target talent mobility.

7. The method as described in claim 6, characterized in that, The mediation effect analysis of the target talent mobility yields the target path of the causal model of the influencing factors on the target talent mobility, specifically including: Based on the causal model, the intermediary nodes for the flow of the target talent are identified; For the intermediary node, the preset talent mobility probability model is used to predict the original text of the influencing factors to be analyzed, and a third predicted probability corresponding to the intermediary node is obtained. The natural direct effect value is obtained by performing calculations based on the third predicted probability and the first predicted probability. For the intermediary node, the preset talent mobility probability model is used to predict the intervention text of the influencing factors to be analyzed, and a fourth prediction probability corresponding to the intermediary node is obtained. The natural indirect effect value is obtained by performing calculations based on the fourth predicted probability and the first predicted probability. Based on the natural direct effect value, the natural indirect effect value, and the total causal effect value, the causal model is located to obtain the target path of the influencing factors to be analyzed affecting the flow of target talent.

8. A device for discovering the causal mechanism of the flow of scientific and technological innovation talents, characterized in that, include: The calculation module is used to perform calculations based on historical talent mobility data to obtain the transition probabilities between historical talent mobility events. A construction module is used to construct a model based on the aforementioned transition probabilities and the historical data of talent mobility using de Bruin graphs, thereby obtaining a causal model for discovering the causal mechanism of talent mobility in scientific and technological innovation. The intervention module is used to intervene in the influencing factors to be analyzed based on a preset talent mobility probability model, so as to obtain the target path of the causal model of the influencing factors to be analyzed affecting the mobility of target talents.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for discovering the causal mechanism of the flow of scientific and technological innovation talents as described in any one of claims 1-7.

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 discovering the causal mechanism of the flow of scientific and technological innovation talents as described in any one of claims 1-7.

Images