Multi-dimensional intelligent analysis method and system for influence factors of scientific and technological achievement transformation in colleges and universities

Abstract

The invention relates to the technical field of big data analysis, in particular to a college scientific and technical achievement transformation influence factor multi-dimensional intelligent analysis method and a college scientific and technical achievement transformation influence factor multi-dimensional intelligent analysis system. The method comprises the following steps: obtaining an initial load matrix, setting a priori knowledge matrix to dynamically adjust the initial load matrix, accurately identifying a standardized path coefficient in combination with a preset model, further decomposing and calculating direct effect, indirect effect and total effect contribution degree of each factor, and meanwhile, calculating the direct effect, indirect effect and total effect contribution degree of each factor through the factor load matrix and the standardized path coefficient in combination with the preset model. And analyzing the time change trend of the importance of the influence factors, and finally realizing multi-dimensional scientific analysis of the influence factors of scientific and technological achievement transformation of colleges and universities.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, specifically to a multi-dimensional intelligent analysis method and system for factors influencing the transformation of scientific and technological achievements in universities. Background Technology

[0002] The commercialization of scientific and technological achievements from universities is a crucial link in bringing academic research and laboratory results to the market and transforming them into real productive forces. It is of great strategic significance for enhancing national innovation capabilities and driving economic development. However, the overall commercialization rate of scientific and technological achievements from Chinese universities is relatively low, and the benefits of commercialization do not match the investment in scientific research. One of the core reasons for this is the lack of systematic, accurate, and dynamic evaluation methods for the complex factors affecting commercialization performance.

[0003] Existing technologies for identifying factors influencing the transformation of scientific and technological achievements in universities are fragmented, resulting in data acquisition that cannot accurately reflect the complex interactions between various factors. Consequently, it is difficult to distinguish between the direct and indirect effects of these factors, leading to inaccurate and incomplete analysis results. Summary of the Invention

[0004] To address the problems in related technologies, this invention provides a multi-dimensional intelligent analysis method and system for the influencing factors of the transformation of scientific and technological achievements in universities, so as to overcome the aforementioned technical problems existing in the existing related technologies.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a multi-dimensional intelligent analysis method for factors influencing the transformation of scientific and technological achievements in universities, comprising the following steps:

[0006] S1. Obtain data on factors influencing the transformation of current university research results through a multi-source data open platform; define the types of factors influencing the transformation of research results, and quantify the data on factors influencing the transformation of current university research results according to the types of factors influencing the transformation of research results to obtain observation indicators;

[0007] S2. Dimensionality reduction is performed on the observation indicators, and a prior knowledge matrix is ​​set to optimize and adjust the dimension-reduced observation indicators to obtain factor loading matrix data.

[0008] S3. Analyze the factor loading matrix data using a preset final structural equation model to obtain standardized path coefficients. Calculate the direct and indirect effects of each factor on the conversion performance based on the standardized path coefficients, and summarize them to obtain the total effect contribution.

[0009] S4. Based on the factor loading matrix data and the standardized path coefficients, a pre-constructed latent variable growth curve model is used to perform a time span analysis of the importance of influencing factors, and obtain the data on the changes in the importance of influencing factors.

[0010] S5. Combine the total effect contribution and the change in the importance of the influencing factors, and push the data to the intelligent analysis platform for influencing factors for display via a communication network.

[0011] Preferably, the specific steps for obtaining data on factors influencing the commercialization of university research results through a multi-source data open platform, defining the types of factors influencing commercialization, and quantifying the data according to these types to obtain observation indicators are as follows:

[0012] S11. Collect data on the influencing factors of the transformation of scientific and technological achievements in universities online through a multi-source data open platform to obtain a dataset of the influencing factors of the transformation of scientific and technological achievements in universities.

[0013] S12. Define the types of factors influencing the transformation of research results;

[0014] S13. Using an NLP language model, extract the influencing factors of achievement transformation from each current university achievement transformation influencing factor data in the current university achievement transformation influencing factor dataset according to the type of influencing factors of achievement transformation, and obtain the current university achievement transformation influencing factor text dataset.

[0015] The text data on current university achievement transformation influencing factors refers to text information data in the current university achievement transformation influencing factor data that matches the type of achievement transformation influencing factor;

[0016] S14. Construct an assignment rule library, and use the assignment rule library to quantify and assign values ​​to the text data of the current university achievement transformation influencing factors in the current university achievement transformation influencing factors text dataset to obtain the current university achievement transformation influencing factors assignment dataset.

[0017] S15. The data of the current university achievement transformation influencing factors in the dataset is normalized by the Min-Max normalization method to obtain the observation index set.

[0018] Preferably, the specific steps for performing dimensionality reduction on the observation indicators and optimizing and adjusting the dimensionality-reduced observation indicators using a prior knowledge matrix to obtain the factor loading matrix data are as follows:

[0019] S21. Calculate the correlation coefficients of each observation index in the observation index set using principal component analysis to obtain the initial load matrix;

[0020] S22. Setting the Prior Knowledge Matrix The initial load matrix is ​​optimized and adjusted.

[0021] in, Indicates observation index With latent variables Strong correlation Indicates observation index With latent variables Moderately relevant Indicates observation index With latent variables Weak correlation Indicates observation index With latent variables No correlation; the dimension of the prior knowledge matrix is ​​consistent with the dimension of the initial load matrix;

[0022] S22 includes the following steps:

[0023] S221. Construct a rotation objective function based on the prior knowledge matrix; the formula for the rotation objective function is as follows:

[0024]

[0025] in, Represents the rotation objective function. Represents the first element in the initial load matrix. Line number The elements corresponding to the column, This indicates the total number of observed indicators. This represents the total number of latent variables. This represents the penalty function constructed based on the prior knowledge matrix; the penalty function is as follows:

[0026]

[0027] in, This represents the penalty intensity parameter. Represents the first in the prior knowledge matrix Line number The element corresponding to the column;

[0028] S222. Set a threshold for the objective function, and use the proximal gradient descent method to optimize and adjust the initial load matrix. Calculate the rotation objective function value of the optimized and adjusted initial load matrix. If the rotation objective function value is greater than or equal to the objective function threshold, the factor load matrix data is obtained; otherwise, continue optimization and adjustment until the rotation objective function value is greater than or equal to the objective function threshold.

[0029] By guiding factor rotation with a prior knowledge matrix, the obtained factor loading matrix data can more accurately reflect the relationship between common factors and each influencing factor, avoiding the occurrence of statistically significant but meaningless factors in the final result due to pure data-driven approaches. This achieves dynamic optimization of the factor loading matrix and provides an accurate and reliable data foundation for subsequent analysis.

[0030] Preferably, the specific steps for analyzing the factor loading matrix data using a preset final structural equation model to obtain standardized path coefficients, calculating the direct and indirect effects of each factor on the conversion performance based on the standardized path coefficients, and summarizing them to obtain the total effect contribution are as follows:

[0031] S31. The factor loading matrix data is analyzed using a preset final structural equation model to obtain standardized path coefficients;

[0032] The standardized path coefficient represents the path coefficient that measures the influence relationship between various factors, wherein the path coefficient includes direct path coefficient and indirect path coefficient;

[0033] The direct path coefficient represents the path coefficient of the influencing factor directly affecting conversion performance, while the indirect path coefficient represents the path coefficient of the influencing factor indirectly affecting conversion performance by acting on other influencing factors.

[0034] S32. Calculate the direct contribution of each factor to the conversion performance based on the standardized path coefficients. and indirect effect contribution The calculation formula is as follows:

[0035] ;

[0036] ;

[0037] in, Indicator Factors To factors The direct standardized path coefficients, Indicator Factors To factors The The first indirect path An indirect standardized path coefficient, Indicates the number of indirect paths. This represents the total number of indirect standardized path coefficients on the indirect path;

[0038] S33. The total effect contribution is obtained by summing the direct and indirect effect contributions using a summation formula. .

[0039] By combining the pre-set final structural equation model with the factor loading matrix data, we can accurately analyze the direct and indirect effects of various influencing factors in the transformation of scientific and technological achievements in universities, clarify the interaction paths between various factors, and provide a data foundation for accurately measuring the contribution of each influencing factor to the total effect of transformation performance.

[0040] Preferably, the specific steps for performing a time-span analysis of the importance of influencing factors based on the factor loading matrix data and the standardized path coefficients combined with a pre-constructed latent variable growth curve model to obtain the data on changes in the importance of influencing factors are as follows:

[0041] S41. Collect the influence factor importance change curve data corresponding to the historical achievement transformation data of several universities online through a multi-source data open platform to obtain the influence factor importance change curve dataset;

[0042] S42. Construct a final latent variable growth curve model based on the historical achievement transformation data;

[0043] S43. Input the current data on factors influencing the transformation of achievements and the standardized path coefficients into the final latent variable growth curve model to perform a time span analysis of the importance of the factors, and obtain the data on the changes in the importance of the factors.

[0044] The data on the changes in the importance of influencing factors represent curve data showing how the importance of influencing factors for the transformation of scientific and technological achievements of various universities changes over time.

[0045] By introducing a latent variable growth curve model to analyze the changes in the importance of each influencing factor over time, universities can accurately identify transformation bottlenecks and formulate targeted improvement strategies.

[0046] Preferably, the specific steps for combining the total effect contribution and the data on changes in the importance of the influencing factors, and then pushing them to the intelligent analysis platform for influencing factors via a communication network for display are as follows:

[0047] S51. Combine the total effect contribution and the change in importance of the influencing factors to generate an analysis report on the influencing factors of the transformation of scientific and technological achievements in universities;

[0048] S52. The report on the analysis of influencing factors of the transformation of scientific and technological achievements of universities is pushed to the intelligent analysis platform of influencing factors through the communication network and displayed on the screen.

[0049] This invention also includes a multi-dimensional intelligent analysis system for influencing factors of the transformation of scientific and technological achievements in universities, including a data quantification module for influencing factors, a dimensionality reduction module for observation indicators, a total effect contribution measurement module, an analysis module for changes in the importance of influencing factors, and an analysis report output and display module;

[0050] The influencing factor data quantification module obtains current university achievement transformation influencing factor data through a multi-source data open platform; it sets achievement transformation influencing factor types and quantifies the current university achievement transformation influencing factor data according to the achievement transformation influencing factor types to obtain observation indicators;

[0051] The observation index dimensionality reduction module performs dimensionality reduction processing on the observation indexes and sets a prior knowledge matrix to optimize and adjust the dimensionality-reduced observation indexes to obtain factor loading matrix data.

[0052] The total effect contribution measurement module analyzes the factor loading matrix data through a preset final structural equation model to obtain standardized path coefficients. Based on the standardized path coefficients, it calculates the direct and indirect effects of each factor on the conversion performance and summarizes them to obtain the total effect contribution.

[0053] The influencing factor importance change analysis module performs time span analysis of the influencing factor importance based on the factor loading matrix data and the standardized path coefficients combined with a pre-constructed latent variable growth curve model, and obtains the influencing factor importance change data.

[0054] The analysis report output and display module combines the total effect contribution and the change in the importance of the influencing factors data, and pushes them to the intelligent analysis platform for influencing factors for display via a communication network.

[0055] By employing the above technical solution, this invention provides a multi-dimensional intelligent analysis method and system for factors influencing the transformation of scientific and technological achievements in universities, which has at least the following beneficial effects:

[0056] 1. This invention quantifies the data on factors influencing the transformation of current university research results to obtain observation indicators. It then calculates the correlation coefficients of these indicators using principal component analysis to obtain an initial loading matrix. A prior knowledge matrix is ​​set to dynamically adjust the initial loading matrix. A preset model is used to accurately identify standardized path coefficients. The direct, indirect, and total effects of each factor are then decomposed and calculated. Furthermore, the temporal trends of the importance of influencing factors are analyzed using the factor loading matrix and standardized path coefficients combined with the preset model. Ultimately, this invention achieves a multi-dimensional scientific analysis of the factors influencing the transformation of university scientific and technological achievements.

[0057] 2. This invention guides factor rotation through a prior knowledge matrix, enabling the obtained factor loading matrix data to more accurately reflect the relationship between latent variables and various influencing factors. This avoids the occurrence of statistically significant but meaningless factors in the final results due to pure data-driven approaches, thereby achieving dynamic optimization of the factor loading matrix and providing an accurate and reliable data foundation for subsequent analysis.

[0058] 3. This invention uses a pre-defined final structural equation model combined with the factor loading matrix data to accurately analyze the direct and indirect effects of various influencing factors in the transformation of scientific and technological achievements in universities, clarifying the interaction paths between factors and providing a data foundation for accurately measuring the contribution of each influencing factor to the total effect of transformation performance; at the same time, it introduces a latent variable growth curve model to analyze the changes in the importance of each influencing factor over time, supporting universities to accurately identify transformation bottlenecks and formulate targeted improvement strategies. Attached Figure Description

[0059] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0060] Figure 1 A flowchart of the multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities provided by this invention;

[0061] Figure 2 A schematic diagram of the modules of the multi-dimensional intelligent analysis system for influencing factors of the transformation of scientific and technological achievements in universities provided by the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Example 1 is as follows:

[0064] To address the problem that existing technologies struggle to distinguish between the direct and indirect effects of factors influencing the commercialization of university scientific and technological achievements, leading to inaccurate and incomplete analysis results, this embodiment proposes a multi-dimensional intelligent analysis method for the influencing factors of university scientific and technological achievements commercialization. For example... Figure 1 As shown, the method includes the following steps:

[0065] S1. Obtain data on factors influencing the transformation of current university research results through a multi-source data open platform; define the types of factors influencing the transformation of research results, and quantify the data on factors influencing the transformation of current university research results according to the types of factors influencing the transformation of research results to obtain observation indicators;

[0066] S1 includes the following steps:

[0067] S11. Collect data on the influencing factors of the transformation of scientific and technological achievements in universities online through a multi-source data open platform to obtain a dataset of the influencing factors of the transformation of scientific and technological achievements in universities.

[0068] The data on factors influencing the transformation of scientific and technological achievements include data from university research management systems, patent data from the State Intellectual Property Office, technology contract registration data, industry-university-research cooperation agreement data, researcher questionnaire survey data, and policy text data.

[0069] S12. Define the types of factors influencing the commercialization of research results. These factors include, but are not limited to, the capabilities of the research team, the maturity of the technology, the strength of intellectual property protection, the support of the commercialization service platform, the strength of incentive policies, and the matching degree between the enterprise's capacity to undertake projects and market demand.

[0070] S13. Using an NLP language model, extract the influencing factors of achievement transformation from each current university achievement transformation influencing factor data in the current university achievement transformation influencing factor dataset according to the type of influencing factors of achievement transformation, and obtain the current university achievement transformation influencing factor text dataset.

[0071] The text data on current university achievement transformation influencing factors refers to text information data in the current university achievement transformation influencing factor data that matches the type of achievement transformation influencing factor;

[0072] S14. Construct an assignment rule library, and use the assignment rule library to quantify and assign values ​​to the text data of the current university achievement transformation influencing factors in the current university achievement transformation influencing factors text dataset to obtain the current university achievement transformation influencing factors assignment dataset.

[0073] S15. The data of the current university achievement transformation influencing factors in the dataset is normalized by the Min-Max normalization method to obtain the observation index set.

[0074] S2. Dimensionality reduction is performed on the observation indicators, and a prior knowledge matrix is ​​set to optimize and adjust the dimension-reduced observation indicators to obtain factor loading matrix data.

[0075] S2 includes the following steps:

[0076] S21. Calculate the correlation coefficients of each observation index in the observation index set using principal component analysis to obtain the initial load matrix;

[0077] S22. Setting the Prior Knowledge Matrix The initial load matrix is ​​optimized and adjusted.

[0078] in, Indicates observation index With latent variables Strong correlation Indicates observation index With latent variables Moderately relevant Indicates observation index With latent variables Weak correlation Indicates observation index With latent variables No correlation; the dimension of the prior knowledge matrix is ​​consistent with the dimension of the initial load matrix;

[0079] S22 includes the following steps:

[0080] S221. Construct a rotation objective function based on the prior knowledge matrix; the formula for the rotation objective function is as follows:

[0081]

[0082] in, Represents the rotation objective function. Represents the first element in the initial load matrix. Line number The elements corresponding to the column, This indicates the total number of observed indicators. This represents the total number of latent variables. This represents the penalty function constructed based on the prior knowledge matrix; the penalty function is as follows:

[0083]

[0084] in, This represents the penalty intensity parameter. Represents the first in the prior knowledge matrix Line number The element corresponding to the column;

[0085] S222. Set a threshold for the objective function. For example, use the proximal gradient descent method to optimize and adjust the initial load matrix. Calculate the rotation objective function value of the optimized and adjusted initial load matrix. If the rotation objective function value is greater than or equal to the objective function threshold, the factor load matrix data is obtained; otherwise, continue optimization and adjustment until the rotation objective function value is greater than or equal to the objective function threshold.

[0086] By guiding factor rotation with a prior knowledge matrix, the obtained factor loading matrix data can more accurately reflect the relationship between latent variables and various influencing factors, avoiding the occurrence of statistically significant but meaningless factors in the final results due to pure data-driven approaches. This achieves dynamic optimization of the factor loading matrix and provides an accurate and reliable data foundation for subsequent analysis.

[0087] S3. Analyze the factor loading matrix data using a preset final structural equation model to obtain standardized path coefficients. Calculate the direct and indirect effects of each factor on the conversion performance based on the standardized path coefficients, and summarize them to obtain the total effect contribution.

[0088] S3 includes the following steps:

[0089] S31. The factor loading matrix data is analyzed using a preset final structural equation model to obtain standardized path coefficients;

[0090] The standardized path coefficient represents the path coefficient that measures the influence relationship between various factors, wherein the path coefficient includes direct path coefficient and indirect path coefficient;

[0091] The direct path coefficient represents the path coefficient of the influencing factor directly affecting conversion performance, while the indirect path coefficient represents the path coefficient of the influencing factor indirectly affecting conversion performance by acting on other influencing factors.

[0092] The final structural equation model in S31 adopts the causal forest SEM model;

[0093] S31 includes the following steps:

[0094] S311. Collect historical achievement transformation data from several universities through a multi-source data open platform to obtain a historical achievement transformation dataset; the historical achievement transformation data includes factor loading matrix data and standardized path coefficients from different universities;

[0095] S312. Construct an initial structural equation model, set a first training data ratio, such as 8:2 or 7.5:2.5, which can be reasonably adjusted according to the actual situation. Divide the historical achievement transformation dataset according to the first training data ratio to obtain the first training dataset and the first test dataset.

[0096] S313. Set a first training error threshold, such as 5%-10%, which can be reasonably adjusted according to the actual situation. Input the first training data in the first training dataset into the initial structural equation model for training. Continuously adjust the parameters of the initial structural equation model according to the training results until the training error is less than the first training error threshold, and obtain the trained structural equation model.

[0097] S314. Set the first test precision, such as 90%-95%, which can be reasonably adjusted according to the actual situation. Input the first test data in the first test dataset into the trained structural equation model for testing, and calculate the accuracy of the test results. If the accuracy of the test results is greater than the first test precision, the final structural equation model is obtained; otherwise, return to S313 until the accuracy of the test results is greater than the first test precision.

[0098] The structure of the initial structural equation model can be seen in Table 1 below:

[0099] Table 1

[0100] Model Name Model type Model Structure Initial structural equation model Causal Forest SEM Model Number of trees: 500-1000 trees to ensure the stability of the estimation; Minimum leaf size: Each terminal node should contain at least 5-10 samples; Split feature ratio: p features are randomly selected for each split (p is the total number of features); Tree depth limit: Dynamically adjusted according to the sample size to avoid overfitting; Effect index: Conditional average treatment effect.

[0101] S32. Calculate the direct contribution of each factor to the conversion performance based on the standardized path coefficients. and indirect effect contribution The calculation formula is as follows:

[0102] ;

[0103] ;

[0104] in, Indicator Factors To factors The direct standardized path coefficients, Indicator Factors To factors The The first indirect path An indirect standardized path coefficient, Indicates the number of indirect paths. This represents the total number of indirect standardized path coefficients on the indirect path;

[0105] S33. The total effect contribution is obtained by summing the direct and indirect effect contributions using a summation formula. .

[0106] By introducing latent variable interaction terms in the process of constructing a structural equation model using a causal forest SEM model, we can examine the synergistic or substitution effects between factors such as research capabilities and policy support, and the maturity of research results and market demand, making the model analysis results more consistent with reality. At the same time, when testing the model, we can use the calculated composite reliability (CR) and mean variance extracted (AVE) as standards to measure the accuracy of the model test, which can better examine the convergent validity and discriminant validity of the final structural equation model.

[0107] By combining the pre-set structural equation model with the factor loading matrix data, the direct and indirect effects of various influencing factors in the transformation of scientific and technological achievements in universities are accurately analyzed, the interaction paths between various factors are clarified, and a data foundation is provided for accurately measuring the contribution of each influencing factor to the total effect of transformation performance.

[0108] S4. Based on the factor loading matrix data and the standardized path coefficients, a pre-constructed latent variable growth curve model is used to perform a time span analysis of the importance of influencing factors, and obtain the data on the changes in the importance of influencing factors.

[0109] S4 includes the following steps:

[0110] S41. Collect the influence factor importance change curve data corresponding to the historical achievement transformation data of several universities online through a multi-source data open platform to obtain the influence factor importance change curve dataset;

[0111] S42. Construct a final latent variable growth curve model based on the historical achievement transformation data;

[0112] The final latent variable growth curve model adopts a quantile regression forest model; the construction process of the final latent variable growth curve model includes the following steps:

[0113] S421. Construct an initial latent variable growth curve model and set a second training data ratio, such as 8:2 or 7.5:2.5. The specific ratio can be adjusted reasonably according to the actual situation.

[0114] S422. The historical achievement transformation dataset and the influencing factor importance change curve dataset are time-aligned and cleaned, and then the data is divided according to the second training data ratio to obtain the second training dataset and the second test dataset.

[0115] S423. Set a second training error threshold, such as 5%-10%, which can be reasonably adjusted according to the actual situation. Input the second training data in the second training dataset into the initial latent variable growth curve model for training. Continuously adjust the parameters of the initial latent variable growth curve model according to the training results until the training error is less than the second training error threshold, and obtain the trained latent variable growth curve model.

[0116] S424. Set the second test precision, such as 90%-95%, which can be reasonably adjusted according to the actual situation. Input the second test data in the second test dataset into the trained latent variable growth curve model for testing, and calculate the accuracy of the test results. If the accuracy of the test results is greater than the second test precision, the final latent variable growth curve model is obtained; otherwise, return to S423 until the accuracy of the test results is greater than the second test precision.

[0117] The structure of the initial latent variable growth curve model can be seen in Table 2 below:

[0118] Table 2

[0119] Model Name Model type Model Structure Initial latent variable growth curve model Latent variable growth mixture model Number of latent classes: 2-4 classes, automatically optimized based on the BIC criterion; Growth factor: two-factor model of random intercept and random slope; Time structure: linear growth basis, supporting quadratic term expansion; Maximum number of iterations: 500-800 rounds; Convergence threshold: log-likelihood change <1e-6; Early stopping mechanism: training terminates if no improvement is achieved in 8-10 consecutive rounds; Multi-starting point strategy: 30-50 random initializations are used to avoid local optima; Discriminative test: key influencing factors that show significant class differences in intercept / slope are automatically identified; Information criterion: the optimal number of classes is determined by combining BIC, aBIC, and entropy (≥0.8).

[0120] S43. Input the current data on factors influencing the transformation of achievements and the standardized path coefficients into the final latent variable growth curve model to perform a time span analysis of the importance of the factors, and obtain the data on the changes in the importance of the factors.

[0121] The data on the changes in the importance of influencing factors represent curve data showing how the importance of influencing factors for the transformation of scientific and technological achievements of various universities changes over time.

[0122] By introducing a latent variable growth curve model to analyze the changes in the importance of each influencing factor over time, universities can accurately identify transformation bottlenecks and formulate targeted improvement strategies.

[0123] S5. Combine the total effect contribution and the change in importance of the influencing factors, and push the data to the intelligent analysis platform for influencing factors for display via a communication network;

[0124] S5 includes the following steps:

[0125] S51. Combine the total effect contribution and the change in importance of the influencing factors to generate an analysis report on the influencing factors of the transformation of scientific and technological achievements in universities;

[0126] S52. The report on the analysis of influencing factors of the transformation of scientific and technological achievements of universities is pushed to the intelligent analysis platform of influencing factors through the communication network and displayed on the screen.

[0127] Example 2 is as follows:

[0128] Please see Figure 2 A multi-dimensional intelligent analysis system for the influencing factors of the transformation of scientific and technological achievements in universities includes a data quantification module for influencing factors, a dimensionality reduction module for observation indicators, a total effect contribution measurement module, an analysis module for changes in the importance of influencing factors, and an analysis report output and display module.

[0129] The influencing factor data quantification module obtains current university achievement transformation influencing factor data through a multi-source data open platform; it sets achievement transformation influencing factor types and quantifies the current university achievement transformation influencing factor data according to the achievement transformation influencing factor types to obtain observation indicators;

[0130] The observation index dimensionality reduction module performs dimensionality reduction processing on the observation indexes and sets a prior knowledge matrix to optimize and adjust the dimensionality-reduced observation indexes to obtain factor loading matrix data.

[0131] The total effect contribution measurement module analyzes the factor loading matrix data through a preset final structural equation model to obtain standardized path coefficients. Based on the standardized path coefficients, it calculates the direct and indirect effects of each factor on the conversion performance and summarizes them to obtain the total effect contribution.

[0132] The influencing factor importance change analysis module performs time span analysis of the influencing factor importance based on the factor loading matrix data and the standardized path coefficients combined with a pre-constructed latent variable growth curve model, and obtains the influencing factor importance change data.

[0133] The analysis report output and display module combines the total effect contribution and the change in the importance of the influencing factors data, and pushes them to the intelligent analysis platform for influencing factors for display via a communication network.

[0134] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0135] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0136] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A multi-dimensional intelligent analysis method for factors influencing the transformation of scientific and technological achievements in universities, characterized in that, Includes the following steps: S1. Obtain data on factors influencing the transformation of current university research results through a multi-source data open platform; define the types of factors influencing the transformation of research results, and quantify the data on factors influencing the transformation of current university research results according to the types of factors influencing the transformation of research results to obtain observation indicators; S2. Dimensionality reduction is performed on the observation indicators, and a prior knowledge matrix is ​​set to optimize and adjust the dimension-reduced observation indicators to obtain factor loading matrix data. S3. Analyze the factor loading matrix data using a preset final structural equation model to obtain standardized path coefficients. Calculate the direct and indirect effects of each factor on the conversion performance based on the standardized path coefficients, and summarize them to obtain the total effect contribution. S4. Based on the factor loading matrix data and the standardized path coefficients, a pre-constructed latent variable growth curve model is used to perform a time span analysis of the importance of influencing factors, and obtain the data on the changes in the importance of influencing factors. S5. Combine the total effect contribution and the change in the importance of the influencing factors, and push the data to the intelligent analysis platform for influencing factors for display via a communication network.

2. The multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities according to claim 1, characterized in that, S1 includes the following steps: S11. Collect data on the influencing factors of the transformation of scientific and technological achievements in universities online through a multi-source data open platform to obtain a dataset of the influencing factors of the transformation of scientific and technological achievements in universities. S12. Define the types of factors influencing the transformation of research results; S13. Using an NLP language model, extract the influencing factors of achievement transformation from each current university achievement transformation influencing factor data in the current university achievement transformation influencing factor dataset according to the type of influencing factors of achievement transformation, and obtain the current university achievement transformation influencing factor text dataset. The text data on current university achievement transformation influencing factors refers to text information data in the current university achievement transformation influencing factor data that matches the type of achievement transformation influencing factor; S14. Construct an assignment rule library, and use the assignment rule library to quantify and assign values ​​to the text data of the current university achievement transformation influencing factors in the current university achievement transformation influencing factors text dataset to obtain the current university achievement transformation influencing factors assignment dataset. S15. The data of the current university achievement transformation influencing factors in the dataset is normalized by the Min-Max normalization method to obtain the observation index set.

3. The multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities, as described in claim 2, is characterized in that... S2 includes the following steps: S21. Calculate the correlation coefficients of each observation index in the observation index set using principal component analysis to obtain the initial load matrix; S22. Setting the Prior Knowledge Matrix The initial load matrix is ​​optimized and adjusted. in, Indicates observation index With latent variables Strong correlation Indicates observation index With latent variables Moderately relevant Indicates observation index With latent variables Weak correlation Indicates observation index With latent variables No correlation; the dimension of the prior knowledge matrix is ​​consistent with the dimension of the initial load matrix.

4. The multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities, as described in claim 3, is characterized in that... S22 includes the following steps: S221. Construct a rotation objective function based on the prior knowledge matrix; the formula for the rotation objective function is as follows: in, Represents the rotation objective function. Represents the first element in the initial load matrix. Line number The elements corresponding to the column, This indicates the total number of observed indicators. This represents the total number of latent variables. This represents the penalty function constructed based on the prior knowledge matrix; the penalty function is as follows: in, This represents the penalty intensity parameter. Represents the first in the prior knowledge matrix Line number The element corresponding to the column; S222. Set a threshold for the objective function, and use the proximal gradient descent method to optimize and adjust the initial load matrix. Calculate the rotation objective function value of the optimized and adjusted initial load matrix. If the rotation objective function value is greater than or equal to the objective function threshold, the factor load matrix data is obtained; otherwise, continue optimization and adjustment until the rotation objective function value is greater than or equal to the objective function threshold.

5. The multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities according to claim 4, characterized in that, S3 includes the following steps: S31. The factor loading matrix data is analyzed using a preset final structural equation model to obtain standardized path coefficients; The standardized path coefficient represents the path coefficient that measures the influence relationship between various factors, wherein the path coefficient includes direct path coefficient and indirect path coefficient; The direct path coefficient represents the path coefficient of the influencing factor directly affecting conversion performance, while the indirect path coefficient represents the path coefficient of the influencing factor indirectly affecting conversion performance by acting on other influencing factors. S32. Calculate the direct contribution of each factor to the conversion performance based on the standardized path coefficients. and indirect effect contribution The calculation formula is as follows: ； ； in, Indicator Factors To factors The direct standardized path coefficients, Indicator Factors To factors The The first indirect path An indirect standardized path coefficient, Indicates the number of indirect paths. This represents the total number of indirect standardized path coefficients on the indirect path; S33. The total effect contribution is obtained by summing the direct and indirect effect contributions using a summation formula. .

6. The multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities according to claim 5, is characterized in that, The final structural equation model described in S31 adopts the causal forest SEM model.

7. The multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities according to claim 6, characterized in that, S4 includes the following steps: S41. Collect the influence factor importance change curve data corresponding to the historical achievement transformation data of several universities online through a multi-source data open platform to obtain the influence factor importance change curve dataset; S42. Construct a final latent variable growth curve model based on the historical achievement transformation data; S43. Input the current data on factors influencing the transformation of achievements and the standardized path coefficients into the final latent variable growth curve model to perform a time span analysis of the importance of the factors, and obtain the data on the changes in the importance of the factors. The data on the changes in the importance of influencing factors represent curve data showing how the importance of influencing factors for the transformation of scientific and technological achievements of various universities changes over time.

8. The multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities according to claim 7, characterized in that, S5 includes the following steps: S51. Combine the total effect contribution and the change in importance of the influencing factors to generate an analysis report on the influencing factors of the transformation of scientific and technological achievements in universities; S52. The report on the analysis of influencing factors of the transformation of scientific and technological achievements of universities is pushed to the intelligent analysis platform of influencing factors through the communication network and displayed on the screen.

9. A computer device, characterized in that, It includes a processor and a memory, the memory being used to store a computer program, which, when executed by the processor, implements the multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities as described in any one of claims 1-8.

10. A system for implementing the multi-dimensional intelligent analysis method for influencing factors of the transformation of scientific and technological achievements in universities, as described in any one of claims 1-8, characterized in that, The system includes a data quantification module for influencing factors, a dimensionality reduction module for observation indicators, a contribution measurement module for total effect, a module for analyzing changes in the importance of influencing factors, and a module for outputting and displaying analysis reports. The influencing factor data quantification module obtains current university achievement transformation influencing factor data through a multi-source data open platform; it sets achievement transformation influencing factor types and quantifies the current university achievement transformation influencing factor data according to the achievement transformation influencing factor types to obtain observation indicators; The observation index dimensionality reduction module performs dimensionality reduction processing on the observation indexes and sets a prior knowledge matrix to optimize and adjust the dimensionality-reduced observation indexes to obtain factor loading matrix data. The total effect contribution measurement module analyzes the factor loading matrix data through a preset final structural equation model to obtain standardized path coefficients. Based on the standardized path coefficients, it calculates the direct and indirect effects of each factor on the conversion performance and summarizes them to obtain the total effect contribution. The influencing factor importance change analysis module performs time span analysis of the influencing factor importance based on the factor loading matrix data and the standardized path coefficients combined with a pre-constructed latent variable growth curve model, and obtains the influencing factor importance change data. The analysis report output and display module combines the total effect contribution and the change in the importance of the influencing factors data, and pushes them to the intelligent analysis platform for influencing factors for display via a communication network.

Images