A method for measuring the collaborative correlation of production and teaching integration based on multi-dimensional feature mapping

Abstract

This invention discloses a method for measuring the collaborative correlation of industry-education integration based on multi-dimensional feature mapping, belonging to the field of collaborative management technology. This method collects initial industry-education integration data from multiple data sources via an API interface, obtains industry-education integration data after preprocessing, constructs a judgment matrix for proportional scaling scoring, extracts the largest eigenvalue of the judgment matrix to calculate the consistency ratio to complete the initial weight assessment, and simultaneously calculates the entropy values ​​of each dimension for a secondary weight assessment. A comprehensive weight is generated when the initial weight assessment meets the consistency compliance requirement and the secondary weight assessment meets the current weight standard. A correlation matrix is ​​constructed and regression coefficients and their p-values ​​are calculated to determine statistical causality, and then a collaborative measurement value is constructed and corrected. The comprehensive weight is adjusted using the collaborative measurement value as an adjustment coefficient, and the adjusted weight is used to perform vector operations with the eigenvalues ​​of each dimension to generate the industry-education integration measurement value and perform interval matching evaluation.

Description

Technical Field

[0001] This invention relates to the field of collaborative management technology, specifically to a method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping. Background Technology

[0002] Once two systems begin to collaborate, their information structures rapidly become more complex; industry-education integration is a prime example of this structural complexity. On one side is the curriculum, talent development pathways, and research output structure within the education system; on the other is the technological demands, job competency structure, and production resource allocation within the industrial system. When these two systems continuously interact, the data ceases to be simple statistics but forms a set of interconnected data with multidimensional structural characteristics, such as the depth of cooperation, resource allocation relationships, technology transfer paths, and employment matching relationships. These data exhibit significant multidimensional structural features, and there are often interactive relationships between different dimensions. With the development of information technology, multi-source data can be continuously aggregated through API interfaces, gradually making industry-education integration data computable. In this context, mapping complex data structures to a unified feature space through multidimensional feature mapping, and then constructing a collaborative correlation measurement mechanism for industry-education integration based on this space, is gradually becoming a technical approach for quantitatively analyzing the operational status of industry-education integration.

[0003] In practical applications, the measurement of collaborative relationships in industry-education integration often faces the problem of insufficient structural identification. Existing evaluation methods mostly rely on single statistical indicators or simple weighting methods, such as statistical evaluation based solely on indicators like the number of collaborations, projects, or employment rates. While these methods can generate basic statistical results, they struggle to reveal the collaborative relationships between different dimensions and fail to reflect the degree of influence of each dimension on the overall integration structure. Furthermore, in multi-source data environments, differences in scale, information density, and complex relationships exist between different data sources. Using only a single indicator or fixed weighting methods for evaluation makes it difficult to reflect the operational status of industry-education integration. In addition, regarding dimensional relationship analysis, traditional methods typically lack mechanisms for identifying correlations and causal relationships between dimensions, making it difficult to effectively measure interactions between dimensions during actual analysis, thus making it difficult for evaluation results to reflect the true collaborative structure. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping, thus solving the problems mentioned in the background technology.

[0005] To achieve the above objectives, this invention provides the following technical solution: a method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping, comprising the following steps: S1. After obtaining legal access authorization through the data source, collect initial industry-education integration data from multiple data sources through the API interface, and obtain industry-education integration data after preprocessing. S2. Classify the industry-education integration data by similarity pattern dimension, and construct a judgment matrix to score the dimension comparison results; S3. Extract the largest eigenvalue of the judgment matrix and construct a consistency ratio for preliminary weight evaluation. At the same time, construct the entropy values ​​of each dimension for secondary weight evaluation. When the preliminary weight evaluation meets the consistency compliance and the secondary weight evaluation meets the current weight standard, construct the comprehensive weight of each dimension through the proportional scale. S4. Construct a correlation matrix and select the dependent variable to calculate the regression coefficient and its p-value, determine the statistical causal relationship, and then perform a nonlinear function transformation on the dimensional relationship value to obtain the synergistic measure value and make corrections. S5. Adjust the overall weights by using the collaborative metric as an adjustment coefficient, fit the adjusted dimensional weights with the corresponding feature values ​​of each dimension, and construct the industry-education integration metric for weight matching status evaluation.

[0006] Preferably, S1 includes S11; S11. After obtaining legal access authorization from the data source, collect initial industry-education data from multiple data sources through the API interface; The data sources include school academic affairs management systems, enterprise ERP systems, and government public data platforms; The initial industry-education data includes the depth of cooperation, resource matching degree, output results, and industry adaptability; The depth of cooperation includes the quantity, form, and stability of university-enterprise cooperation; the degree of resource matching includes the funds and resources invested by enterprises; the output results include the technology transfer of cooperation results; and the industry fit includes the employment quality and job placement rate of students.

[0007] Preferably, S1 further includes S12; S12. Process the acquired initial industry-education integration data to obtain industry-education integration data; The data processing includes data cleaning, format conversion, noise reduction, and standardization. The data cleaning is used to remove invalid data, duplicate data, and missing values ​​from the initial data of the industry-education integration program based on the data cleaning algorithm. The format conversion is used to convert different data in the initial data of industry-education integration into a standard format using a data format conversion tool. The denoising is used to filter irrelevant noise in the initial data of the production and education program based on a data denoising algorithm. The standardization process is used to convert raw data with different dimensions and units into dimensionless standardized data points according to the standardization algorithm, and to organize the dimensionless standardized data points to obtain industry-education integration data.

[0008] Preferably, S2 includes S21 and S22; S21. Randomly initialize all dimensionless standardized data points in the industry-education integration data into several cluster centers. Use the K-means clustering algorithm to calculate the Euclidean distance from each data point to each cluster center and assign the data points to the nearest cluster center. After each round of clustering assignment, recalculate the mean coordinates of the data points in each cluster based on the coordinates of all data points in the current cluster as the new cluster center. Compare the cluster labels of each data point in this round with the cluster labels of each data point in the previous round one by one. When the cluster labels of all data points in this round are completely consistent with the corresponding labels of the clusters of the data points in the previous round, it is determined that the clustering result of this round is consistent with the clustering result of the previous round, and the assignment is stopped. The data points assigned to the same cluster center are classified into the dimension of similar patterns. S22. Perform pairwise comparisons on any two dimensions in all dimensions and construct a judgment matrix. Preset dimension evaluation rules and assign scale values ​​to each dimension in the judgment matrix based on the comparison results of each dimension with the indicator evaluation rules. Score the comparison results of each pair of dimensions in the judgment matrix using a scale from 1 to 9, where 1 indicates that the data features corresponding to the two dimensions are at the same evaluation level, and 9 indicates that the data features corresponding to the current dimension are at a higher evaluation level than the other dimension. The preset dimension evaluation rules read the historical industry-education integration data corresponding to each dimension in the historical industry-education integration database, count the number of samples in the historical industry-education integration data of each dimension and divide them into intervals to generate several continuous evaluation intervals. Then, according to the order of the number of samples in each evaluation interval, a level identifier is written for each evaluation interval to form a dimension evaluation level sequence. The dimension name, corresponding evaluation interval and level identifier are written into the industry-education integration indicator library as dimension evaluation rules.

[0009] Preferably, S3 includes S31; S31. Read the number of dimensions in the judgment matrix, and call the matrix eigenvalue solving algorithm to solve the characteristic equation of the judgment matrix to obtain the eigenvalue set of the judgment matrix. Then, extract the largest eigenvalue from the eigenvalue set, and perform consistency index calculation based on the largest eigenvalue and the number of dimensions of the judgment matrix. Specifically: Then, the consistency index obtained in real time is divided by the average historical consistency index to obtain the consistency ratio. The judgment matrix is ​​then preliminarily weighted based on the consistency ratio, as follows. Within a historically stable operating range, a sequence of consistency ratios is statistically analyzed, and the median value is set as the consistency threshold. When the consistency ratio is less than or equal to the consistency threshold, it means that the current judgment matrix is ​​consistent and compliant, and the current ratio scale is retained. When the consistency ratio is greater than the consistency threshold, it indicates that the current judgment matrix is ​​not compliant. At this time, the current ratio scale is increased by 1, and the dimensions in the judgment matrix are adjusted for iterative analysis until the current judgment matrix is ​​compliant.

[0010] Preferably, S3 further includes S32 and S33; S32. Extract the data column corresponding to each dimension in the judgment matrix, perform probability distribution calculation on the current data column to generate a dimension probability distribution sequence, perform information entropy calculation based on the probability distribution sequence to obtain the entropy value of the corresponding dimension, and form an entropy value sequence by combining the entropy values ​​of each dimension. Then, calculate the mean and standard deviation of the entropy value sequence using statistical methods, set the sum of the mean and standard deviation as the entropy value critical threshold, and compare it with the entropy values ​​of each dimension. Generate a secondary weight evaluation based on the comparison results, as follows. When the entropy value is less than the entropy threshold, it means that the data points in the current dimension are concentrated and the information content of the current data point in the current dimension does not meet the current weight standard. At this time, the current scale is increased by 1 and the dimensions in the judgment matrix are adjusted for iterative analysis until the dimension meets the current weight standard. When the entropy value is greater than or equal to the entropy threshold, it means that the data points in the current dimension are evenly distributed and the information content of the current dimension meets the requirements for weight calculation. At this time, the current scale is retained. S33. After the preliminary weight assessment and the secondary weight assessment, when the preliminary weight assessment shows that the judgment matrix is ​​consistent and compliant and the secondary weight assessment shows that it meets the current weight standard, the adjusted judgment matrix is ​​read, and the weight vector solving algorithm is called to perform a multiplication calculation on each column element of the judgment matrix to generate a row product value. The dimension root operation is performed on each row product value to obtain the row feature value, and then all row feature values ​​are normalized to generate a dimension weight vector. Subsequently, the entropy value corresponding to each dimension is read and the entropy weight algorithm is called to divide the entropy value of each dimension by the sum of the entropy values ​​to obtain the objective weight sequence of the dimension. Then, the weight fusion algorithm is called to perform dimension-by-dimensional summation calculation on the dimension weight vector and the objective weight sequence of the dimension in the order of the dimensions, and the summation result is normalized to generate the comprehensive weight sequence of each dimension.

[0011] Preferably, S4 includes S41; S41. After obtaining the comprehensive weights, read the scale values ​​between each dimension in the judgment matrix, construct a dimension relationship matrix based on the scale values, and then perform relationship measurement calculations based on the scale values ​​and the comprehensive weights of each dimension to generate the dimension relationship value Wg, specifically: , where C ijW represents the dimensional relationship value between the i-th and j-th dimensions. i and W j Let a represent the combined weights of the i-th and j-th dimensions, respectively. ij This represents the scale value between the i-th and j-th dimensions in the judgment matrix.

[0012] Preferably, S4 further includes S42 and S43; S42. Select one dimension as the dependent variable and the other dimensions as independent variables. Use the least squares algorithm to calculate the regression coefficient of each independent variable on the dependent variable. After obtaining the regression coefficients, calculate the p-value of each regression coefficient. If the p-value is less than 0.05, it indicates that there is a statistical causal relationship and a causal relationship report is generated. If the p-value is greater than or equal to 0.05, it indicates that there is no statistical causal relationship and a no-causal relationship report is generated. S43. Calculate the synergistic metric value by performing a nonlinear function transformation based on the dimensional relationship values ​​of each pair of dimensions in the correlation matrix. When calculating the synergistic metric value, if the causal relationship is statistically causal, then introduce a causal relationship correction factor to adjust the synergistic metric value.

[0013] Preferably, S5 includes S51; S51. Adjust the overall weight using the synergy metric as an adjustment factor. The specific adjustment scheme is as follows: Statistical analysis of historical collaborative metrics is performed using statistical methods to calculate the mean and standard deviation of collaborative metrics for each dimension in historical data, and the sum of the mean and standard deviation is preset as the interactive judgment threshold. When the collaborative metric value is less than or equal to the interaction judgment threshold, it indicates that the interaction effect of the current dimension is significant, and the overall weight is increased accordingly. When the collaborative metric value is greater than the interaction judgment threshold, it indicates that the interaction effect of the current dimension is not significant, and the overall weight is reduced accordingly.

[0014] Preferably, S5 further includes S52; S52. Read the feature values ​​corresponding to each dimension, and call the vector weighting algorithm to perform a dimension-by-dimensional product operation on the feature values ​​of each dimension and the adjusted dimension weights to generate a sequence of score values ​​for each dimension. Then, perform interval normalization processing on the score value sequence of the same dimension to obtain the dimension evaluation value of each dimension. Perform cumulative aggregation operation on the dimension evaluation values ​​according to the dimension order to generate the industry-education integration measurement value. The system reads the sample sequence of industry-education integration metrics from the historical evaluation database, sorts them according to their numerical values ​​to generate an ordered sample sequence, calls the quantile interval partitioning algorithm, and divides the ordered sample sequence into several continuous intervals according to the preset number of intervals. The starting sample value of each interval is read as the lower bound value, and the ending sample value of each interval is read as the upper bound value. Finally, a weighted matching status evaluation is performed with the industry-education integration metrics, as detailed below: When the industry-education integration metric value is greater than or equal to the lower bound of the interval and less than the upper bound of the interval, it is determined as an interval hit, indicating that the current dimension weight matches, and the status level identifier corresponding to the current interval is determined as the current industry-education integration status level. When the industry-education integration metric value equals a certain boundary value, it indicates that the current dimension weight is matched. Based on the boundary inclusion rule, its belonging interval is selected, and the status level identifier corresponding to the belonging interval is determined as the current industry-education integration status level. When the industry-education integration metric does not meet any interval inclusion rule, it is determined as an interval miss. An interval miss indicates that the comprehensive index has exceeded the limit, which means that the current dimension weights are mismatched, and an out-of-bounds direction determination is performed, as follows: When the industry-education integration metric value is less than the lower bound of the minimum interval, the out-of-bounds direction is marked as a lower out-of-bounds, and the lower out-of-bounds handling process is triggered. The status level identifier is set to the out-of-bounds status code, and the weight is reduced through S4 iterative analysis. The weight matching status is re-evaluated. When the industry-education integration metric value is greater than or equal to the upper limit of the maximum interval, the out-of-bounds direction is marked as an upper out-of-bounds event, and the upper out-of-bounds handling process is triggered. The status level identifier is set to the out-of-bounds status code, and the weight is increased through S4 iterative analysis, and the weight matching status assessment is performed again.

[0015] This invention provides a method for measuring the collaborative relationship between industry and education based on multi-dimensional feature mapping. It has the following beneficial effects: (1) After obtaining legal access authorization from the data source, method S1 collects initial industry-education integration data such as cooperation depth, resource matching degree, output results, and industry adaptation from the school's academic affairs management system, enterprise ERP system, and government public data platform through API interface. Then, it generates industry-education integration data with a unified structure through data cleaning, format conversion, noise reduction, and standardization. On this basis, S2 calculates the Euclidean distance from each data point to the cluster center through the K-means clustering algorithm, and classifies data points with similar patterns into the same dimension. Then, it constructs a judgment matrix through the analytic hierarchy process and scores the comparison results between each dimension using a scale from 1 to 9. This transforms the industry-education integration data from the original discrete data into an evaluation data system with a clear dimensional structure, completing the task of dimensional organization and comparison relationship construction of industry-education integration data.

[0016] (2) Method S3 calculates the eigenvalues ​​of each dimension in the judgment matrix using regression analysis and extracts the largest eigenvalue of the judgment matrix. It then calculates the consistency ratio by comparing the consistency index with the random consistency index to perform a preliminary weight assessment of the judgment matrix. Subsequently, it calculates the entropy value of each dimension based on the Shannon entropy formula to perform a secondary weight assessment of the data distribution status of each dimension. When the preliminary weight assessment indicates that the judgment matrix is ​​consistent and the secondary weight assessment indicates that it meets the current weight standard, the comprehensive weight of each dimension is obtained by comprehensively calculating the scale using the weighted average method. This ensures that the dimension weights simultaneously meet the condition of consistency of the judgment matrix structure and the information content condition of the dimension data distribution, thus completing the comprehensive calculation task of the dimension weights.

[0017] (3) In this method, S4 reads the scale between each dimension in the judgment matrix and performs relationship measurement calculation based on the scale and the comprehensive weight of each dimension to generate the dimension relationship value Wg. By calculating the dimension relationship values ​​between all dimensions, a correlation matrix is ​​generated. One dimension is selected as the dependent variable and the other dimensions are selected as independent variables. The least squares algorithm is used to calculate the regression coefficient and its p value to determine whether there is a statistical causal relationship between the dimensions. Then, the synergistic measurement value is calculated by nonlinear function transformation based on the dimension relationship value in the correlation matrix. When there is a statistical causal relationship, the synergistic measurement value is adjusted by the causal relationship correction factor to obtain the adjusted synergistic measurement value. S5 uses the synergy metric as an adjustment coefficient to adjust the overall weights. It then performs vector weighting operations on the adjusted dimensional weights and feature values ​​of each dimension to generate a sequence of dimensional score values. After interval normalization, it obtains the dimensional evaluation value for each dimension and aggregates these values ​​to generate the industry-education integration metric. Subsequently, it reads the sample sequence of industry-education integration metrics from the historical evaluation database, generates lower and upper bounds for each interval using a quantile interval partitioning algorithm, and performs weight matching status assessment and boundary crossing direction determination on the industry-education integration metrics. This completes the tasks of identifying the industry-education integration status level and analyzing the weight matching status. Through this process, the industry-education integration evaluation is not only based on dimensional weights but also adjusts the weights by considering the correlation structure and synergy between dimensions, enabling the industry-education integration metric to reflect the synergistic relationship structure between multi-dimensional features. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the steps of the industry-education integration collaborative correlation measurement method based on multidimensional feature mapping of the present invention; Figure 2 This is a block diagram illustrating the principle and logic of the industry-education integration collaborative correlation measurement method based on multidimensional feature mapping of this invention. Detailed Implementation

[0019] 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.

[0020] Example 1

[0021] Please see Figure 1 This invention provides a method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping. To achieve the above objectives, this invention is implemented through the following technical solution, including the following steps: S1. After obtaining legal access authorization through the data source, initial industry-education integration data is collected from multiple data sources through the API interface and obtained after preprocessing. S2. Classify the industry-education integration data by similarity pattern dimension, and construct a judgment matrix to score the dimension comparison results; S3. Extract the largest eigenvalue of the judgment matrix and construct a consistency ratio for preliminary weight evaluation. At the same time, construct the entropy values ​​of each dimension for secondary weight evaluation. When the preliminary weight evaluation meets the consistency compliance and the secondary weight evaluation meets the current weight standard, construct the comprehensive weight of each dimension through the proportional scale. S4. Construct a correlation matrix and select the dependent variable to calculate the regression coefficient and its p-value, determine the statistical causal relationship, and then perform a nonlinear function transformation on the dimensional relationship value to obtain the synergistic measure value and make corrections. S5. Adjust the overall weights by using the collaborative metric as an adjustment coefficient, fit the adjusted dimensional weights with the corresponding feature values ​​of each dimension, and construct the industry-education integration metric for weight matching status evaluation.

[0022] In this embodiment, after obtaining legal access authorization from the data source, S1 collects initial industry-education integration data from multiple data sources using an API interface. Through preprocessing, it obtains industry-education integration data with a unified structure, enabling data from school academic affairs management systems, enterprise ERP systems, and government public data platforms to be analyzed under the same data structure. Based on this, S2 performs similarity pattern dimensional classification on the industry-education integration data and constructs a judgment matrix to score the dimensional comparison results, reorganizing industry-education integration data from different sources and of different types into a quantifiable and comparable dimensional structure. This process transforms the originally scattered data set into a data structure with clear dimensional relationships, completing the task of unified organization and dimensional relationship construction of industry-education integration data, and establishing a basic data structure for subsequent weight evaluation and measurement calculation. S3 extracts the largest eigenvalue of the judgment matrix to construct a consistency ratio, performs preliminary weight evaluation on the dimensional comparison results, and judges whether the dimensional comparison structure meets consistency requirements through consistency compliance conditions. Simultaneously, it constructs entropy values ​​for each dimension to analyze the dimensional data distribution and judges the information distribution status of the dimensional data through secondary weight evaluation. When the preliminary weight evaluation meets consistency compliance and the secondary weight evaluation conforms to the current weight standard, a comprehensive weight for each dimension is constructed using a proportional scale. This processing method ensures that the construction of dimensional weights considers both the consistency condition of the judgment matrix structure and the distribution characteristics of the data in each dimension, thus completing the task of calculating dimensional weights. Compared with techniques that only set weights through a single score or a fixed ratio, this method uses both the consistency ratio and entropy value in the weight evaluation, making the calculation process of dimensional weights subject to both structural constraints and data distribution characteristics. S4 constructs a correlation matrix and selects the dependent variable. Regression coefficients and their p-values ​​are calculated through regression analysis to determine the statistical causal relationship between dimensions. A nonlinear function transformation is performed based on the dimensional relationship values ​​to obtain the synergistic metric value, which is then corrected when the statistical causal relationship is established. S5 uses the synergistic metric value as an adjustment coefficient to adjust the comprehensive weights. The adjusted dimensional weights are then fitted with the corresponding feature values ​​of each dimension to construct the industry-education integration metric value, and a weight matching status evaluation is performed on the industry-education integration metric value. Through the above processing flow, the industry-education integration metric value not only reflects the importance of each dimension itself but also reflects the correlation structure between dimensions, enabling the industry-education integration metric value to be used to identify the dimensional weight matching status and determine the structural status of industry-education integration. Compared to techniques that calculate evaluation results using only fixed weights, this method introduces dimensional correlation analysis during the weight calculation process, enabling the measurement results to reflect the correlation and change structure between multidimensional data.

[0023] Example 2

[0024] Please refer to Figure 2 Specifically: S1 includes S11; S11. After obtaining legal access authorization from the data source, collect initial industry-education data from multiple data sources through the API interface; The data sources include school academic affairs management systems, enterprise ERP systems, and government public data platforms; The initial industry-education data includes the depth of cooperation, resource matching degree, output results, and industry adaptability; The depth of cooperation includes the quantity, form, and stability of university-enterprise cooperation; the degree of resource matching includes the funds and resources invested by enterprises; the output results include the technology transfer of cooperation results; and the industry fit includes the employment quality and job placement rate of students.

[0025] S1 further includes S12; S12. Process the acquired initial industry-education integration data to obtain industry-education integration data; The data processing includes data cleaning, format conversion, noise reduction, and standardization. The data cleaning is used to remove invalid data, duplicate data, and missing values ​​from the initial data of the industry-education integration program based on the data cleaning algorithm. The format conversion is used to convert different data in the initial data of industry-education integration into a standard format using a data format conversion tool. The denoising is used to filter irrelevant noise in the initial data of the production and education program based on a data denoising algorithm. The standardization process is used to convert raw data with different dimensions and units into dimensionless standardized data points according to the standardization algorithm, and to organize the dimensionless standardized data points to obtain industry-education integration data.

[0026] In this embodiment, after completing the legal access authorization for the data source, S11 uses the API interface to collect initial industry-education data from the school's academic affairs management system, the enterprise ERP system, and the government's public data platform. The collected data covers information such as cooperation depth, resource matching degree, output results, and industry adaptability. Among them, cooperation depth corresponds to the number, form, and stability of school-enterprise cooperation; resource matching degree corresponds to the scale of enterprise investment and resource allocation; output results correspond to the technology transfer results generated during school-enterprise cooperation; and industry adaptability corresponds to the student employment quality and job matching rate. This unifies the collection of industry-education related data scattered in different information systems and forms an original data set. Subsequently, in S12, the collected initial industry-education data is processed... The data undergoes cleaning, format conversion, denoising, and standardization. Data cleaning algorithms remove invalid, duplicate, and missing data; data format conversion tools unify data from different systems into a standard format; data denoising algorithms filter irrelevant noise; and standardization algorithms convert data with different dimensions and units into dimensionless standardized data. This generates industry-education integration data with a unified structure, enabling analysis and processing of data from different sources and with different structures within the same data system. This implementation method completes the unified collection and data quality processing of multi-source industry-education integration data, ensuring that industry-education integration-related data has a unified structure and computability, providing a consistent data foundation for subsequent dimensional classification, weight evaluation, and measurement calculations.

[0027] Example 3

[0028] Please refer to Figure 2 Specifically: S2 includes S21 and S22; S21. Randomly initialize all dimensionless standardized data points in the industry-education integration data into several cluster centers. Use the K-means clustering algorithm to calculate the Euclidean distance from each data point to each cluster center and assign the data points to the nearest cluster center. After each round of clustering assignment, recalculate the mean coordinates of the data points in each cluster based on the coordinates of all data points in the current cluster as the new cluster center. Compare the cluster labels of each data point in this round with the cluster labels of each data point in the previous round one by one. When the cluster labels of all data points in this round are completely consistent with the corresponding labels of the clusters of the data points in the previous round, it is determined that the clustering result of this round is consistent with the clustering result of the previous round, and the assignment is stopped. The data points assigned to the same cluster center are classified into the dimension of similar patterns. S22. Perform pairwise comparisons on any two dimensions in all dimensions and construct a judgment matrix. Preset dimension evaluation rules and assign scale values ​​to each dimension in the judgment matrix based on the comparison results of each dimension with the indicator evaluation rules. Score the comparison results of each pair of dimensions in the judgment matrix using a scale from 1 to 9, where 1 indicates that the data features corresponding to the two dimensions are at the same evaluation level, and 9 indicates that the data features corresponding to the current dimension are at a higher evaluation level than the other dimension. The preset dimension evaluation rules read the historical industry-education integration data corresponding to each dimension in the historical industry-education integration database, count the number of samples in the historical industry-education integration data of each dimension and divide them into intervals to generate several continuous evaluation intervals. Then, according to the order of the number of samples in each evaluation interval, a level identifier is written for each evaluation interval to form a dimension evaluation level sequence. The dimension name, corresponding evaluation interval and level identifier are written into the industry-education integration indicator library as dimension evaluation rules.

[0029] In this embodiment, S21 reads the preprocessed industry-education integration data, inputs each individual data point as a data point into the K-means clustering calculation process, randomly initializes several cluster centers for all data points, calculates the Euclidean distance from each data point to each cluster center, assigns the data points to the corresponding cluster centers according to the principle of minimum distance, and continues to perform the data point assignment operation after recalculating the cluster center position in each round until the cluster center reaches the preset stopping condition and terminates the iteration process, thereby grouping data points with similar data distribution characteristics into the same cluster center, and determining the data set corresponding to the cluster center as the dimensional structure of similar patterns; S22 After completing the dimensional classification, based on the dimensions formed by the clustering results, performs pairwise comparison operations on any two dimensions, and scores the importance relationship between dimensions according to the scale from 1 to 9 to construct a judgment matrix, where a value of 1 indicates that the two dimensions are equally important, and a value of 9 indicates that one dimension has a higher importance priority than the other dimension, thereby transforming the dimensional relationship formed by clustering into a comparative structure that can be quantitatively calculated. Through the above processing flow, the original industry-education integration data, after completing the classification of similar patterns, further establishes comparative relationships between dimensions. The dimensional structure is composed of data distribution relationships and dimensional comparison relationships, providing structured input data for subsequent calculation of the maximum eigenvalue, consistency ratio assessment, and comprehensive weight calculation. This transforms the industry-education integration evaluation process from a raw data set into an analytical data system with clear dimensional relationships and comparative structures.

[0030] Example 4

[0031] Please refer to Figure 2 Specifically: S3 includes S31; S31. Read the number of dimensions in the judgment matrix, and call the matrix eigenvalue solving algorithm to solve the characteristic equation of the judgment matrix to obtain the eigenvalue set of the judgment matrix. Then, extract the largest eigenvalue from the eigenvalue set, and perform consistency index calculation based on the largest eigenvalue and the number of dimensions of the judgment matrix. Specifically: Then, the consistency index obtained in real time is divided by the average historical consistency index to obtain the consistency ratio. The judgment matrix is ​​then preliminarily weighted based on the consistency ratio, as follows. Within a historically stable operating range, a sequence of consistency ratios is statistically analyzed, and the median value is set as the consistency threshold. When the consistency ratio is less than or equal to the consistency threshold, it means that the current judgment matrix is ​​consistent and compliant, and the current ratio scale is retained. When the consistency ratio is greater than the consistency threshold, it indicates that the current judgment matrix is ​​not compliant. At this time, the current ratio scale is increased by 1, and the dimensions in the judgment matrix are adjusted for iterative analysis until the current judgment matrix is ​​compliant.

[0032] S3 further includes S32 and S33; S32. Extract the data column corresponding to each dimension in the judgment matrix, perform probability distribution calculation on the current data column to generate a dimension probability distribution sequence, perform information entropy calculation based on the probability distribution sequence to obtain the entropy value of the corresponding dimension, and form an entropy value sequence by combining the entropy values ​​of each dimension. Then, calculate the mean and standard deviation of the entropy value sequence using statistical methods, set the sum of the mean and standard deviation as the entropy value critical threshold, and compare it with the entropy values ​​of each dimension. Generate a secondary weight evaluation based on the comparison results, as follows. When the entropy value is less than the entropy threshold, it means that the data points in the current dimension are concentrated and the information content of the current data point in the current dimension does not meet the current weight standard. At this time, the current scale is increased by 1 and the dimensions in the judgment matrix are adjusted for iterative analysis until the dimension meets the current weight standard. When the entropy value is greater than or equal to the entropy threshold, it means that the data points in the current dimension are evenly distributed and the information content of the current dimension meets the requirements for weight calculation. At this time, the current scale is retained. S33. After the preliminary weight assessment and the secondary weight assessment, when the preliminary weight assessment shows that the judgment matrix is ​​consistent and compliant and the secondary weight assessment shows that it meets the current weight standard, the adjusted judgment matrix is ​​read, and the weight vector solving algorithm is called to perform a multiplication calculation on each column element of the judgment matrix to generate a row product value. The dimension root operation is performed on each row product value to obtain the row feature value, and then all row feature values ​​are normalized to generate a dimension weight vector. Subsequently, the entropy value corresponding to each dimension is read and the entropy weight algorithm is called to divide the entropy value of each dimension by the sum of the entropy values ​​to obtain the objective weight sequence of the dimension. Then, the weight fusion algorithm is called to perform dimension-by-dimensional summation calculation on the dimension weight vector and the objective weight sequence of the dimension in the order of the dimensions, and the summation result is normalized to generate the comprehensive weight sequence of each dimension.

[0033] In this embodiment, S31 calculates the eigenvalues ​​of each dimension in the judgment matrix using regression analysis and extracts the maximum eigenvalue. The eigenvalue is a numerical parameter obtained by solving the characteristic equation of the judgment matrix, representing the proportion of a dimension that remains unchanged but undergoes only a proportional change when the judgment matrix is ​​applied to it. In the judgment matrix, it represents the pairwise comparison relationship between dimensions, and the matrix eigenvalue reflects the proportional change characteristics of this comparison relationship structure under matrix operations. A consistency index is calculated by subtracting the maximum eigenvalue from the number of dimensions in the judgment matrix and comparing it with the number of dimensions minus 1. The consistency index is then compared with a random consistency index to obtain a consistency ratio. A consistency ratio sequence is statistically analyzed within a historical stable operating range. A consistency threshold is set by the sum of the median and quartile values, and a preliminary weight evaluation is performed on the judgment matrix. When the consistency ratio is less than or equal to the consistency threshold, the current proportional scale is retained; when the consistency ratio is greater than the consistency threshold, the proportional scale is incremented, and the dimensional relationships in the judgment matrix are iteratively analyzed. The process continues until consistency and compliance are achieved. S32, after completing the initial weight assessment, calculates the entropy value of each dimension using the Shannon entropy formula and extracts the entropy distribution characteristics. A second weight assessment is performed by setting a critical threshold for the entropy value by summing the mean and standard deviation of all dimension entropy values. When the entropy value is less than the critical threshold, the scale is incremented and the dimensional structure of the judgment matrix is ​​readjusted until the current weight standard is met. When the entropy value is greater than or equal to the critical threshold, the current scale is retained. S33 then jointly judges the results of the initial and second weight assessments. When both the consistency and compliance of the judgment matrix and the entropy value meet the current weight standard are met, the comprehensive weight of each dimension is calculated using a weighted average method. Through this process, the consistency check of the judgment matrix structure and the analysis of the dimensional information distribution simultaneously participate in the dimensional weight construction process. This ensures that the comprehensive weight satisfies the consistency constraints of the dimensional comparison relationship and reflects the data distribution characteristics of each dimension, thus forming a dimensional weight structure for subsequent collaborative relationship analysis and industry-education integration measurement calculations.

[0034] Example 5

[0035] Please refer to Figure 2 Specifically: S4 includes S41; S41. After obtaining the comprehensive weights, read the scale values ​​between each dimension in the judgment matrix, construct a dimension relationship matrix based on the scale values, and then perform relationship measurement calculations based on the scale values ​​and the comprehensive weights of each dimension to generate the dimension relationship value Wg, specifically: Among them, Wg ij W represents the dimensional relationship value between the i-th and j-th dimensions. i and W j Let a represent the combined weights of the i-th and j-th dimensions, respectively. ij This represents the scale value between the i-th and j-th dimensions in the judgment matrix.

[0036] S4 further includes S42 and S43; S42. Select one dimension as the dependent variable and the other dimensions as independent variables. Use the least squares algorithm to calculate the regression coefficient of each independent variable on the dependent variable. After obtaining the regression coefficients, calculate the p-value of each regression coefficient. If the p-value is less than 0.05, it indicates that there is a statistical causal relationship and a causal relationship report is generated. If the p-value is greater than or equal to 0.05, it indicates that there is no statistical causal relationship and a no-causal relationship report is generated. S43. Calculate the collaborative metric value by performing a nonlinear function transformation based on the dimensional relationship values ​​of each pair of dimensions in the correlation matrix, specifically: , where C ij r represents the collaborative metric between dimension i and dimension j. ij For the dimensional relationship values ​​in the correlation matrix, when calculating the synergistic metric, if the causal relationship is statistically causal, a causal relationship correction factor is introduced. Adjust the synergy metric, i.e., C. ij =C ij ×w ij , where w ij β is a causal relationship correction factor. ij Let C be the regression coefficient of dimension i on dimension j, k be the index of all independent variable dimensions, and C be the regression coefficient of dimension i on dimension j. ij ′ represents the adjusted collaborative metric value.

[0037] In this embodiment, after obtaining the comprehensive weights of each dimension, S41 first reads the scale values ​​between each dimension in the judgment matrix, constructs a dimension relationship matrix using the scale values, and calculates the strength of the relationship between the dimensions in conjunction with the comprehensive weights of each dimension, thus obtaining the dimension relationship value Wg between each dimension, thereby forming a relationship description result reflecting the degree of structural correlation between different dimensions; S42, after completing the construction of the correlation matrix, selects one dimension as the dependent variable and the remaining dimensions as independent variables, and uses the least squares algorithm to calculate the regression relationship between each independent variable and the dependent variable, obtaining the regression coefficient β. ijFurthermore, p-values ​​are calculated for each regression coefficient to determine statistical causality. A report indicating the existence of statistical causality is generated when the p-value is less than 0.05, and a report indicating no causality is generated when the p-value is greater than or equal to 0.05. In statistics, the significance level is typically set at 0.05, meaning that a 5% probability of random misjudgment is allowed. The reason for commonly using 0.05 is twofold: first, it has long been widely used as the standard significance level in statistical inference, exhibiting high versatility in social sciences, economic analysis, and educational evaluation; second, 0.05 represents a trade-off between the risk of misjudgment and the ability to identify causality. If the threshold is too small, many genuine relationships may not be identified; if the threshold is too large, random fluctuations are easily misjudged as genuine relationships. Therefore, this method uses a p-value of 0.05 as the statistical relationship determination boundary, which can identify causal relationships while controlling the probability of random misjudgment. The statistically significant correlations between dimensions provide a basis for subsequent calculation of collaborative metrics and weight adjustments. After determining the correlation and statistical causal relationships, a nonlinear function transformation is performed based on the dimensional relationship values ​​in the correlation matrix to calculate the collaborative metrics. When statistical causal relationships exist, a causal relationship correction factor is introduced to correct the collaborative metrics, resulting in adjusted collaborative metrics that simultaneously reflect the degree of correlation and regression structure between dimensions. Through the above processing, the statistical correlations and regression relationships between dimensions can be uniformly quantified into collaborative metrics, establishing a foundation for subsequent weight adjustments and industry-education integration measurement calculations. This ensures that dimensional relationships no longer rely solely on a single correlation judgment but are comprehensively characterized by combining the correlation matrix and regression relationships, forming a collaborative relationship measurement result that reflects the multidimensional data correlation structure.

[0038] Example 6

[0039] Please refer to Figure 2 Specifically: S5 includes S51; S51. Adjust the overall weight using the synergy metric as an adjustment factor. The specific adjustment scheme is as follows: Statistical analysis of historical collaborative metrics is performed using statistical methods to calculate the mean and standard deviation of collaborative metrics for each dimension in historical data, and the sum of the mean and standard deviation is preset as the interactive judgment threshold. When the collaborative metric value is less than or equal to the interaction judgment threshold, it indicates that the interaction effect of the current dimension is significant. At this time, the comprehensive weight is increased. The specific weight increase amount is calculated as (collaborative metric value - interaction judgment threshold) × the maximum weight increase value allowed within the standard range. When the collaborative metric value is greater than the interaction judgment threshold, it indicates that the interaction effect of the current dimension is not significant. At this time, the overall weight is reduced. The specific weight reduction amount is calculated as (threshold - collaborative metric value) × the maximum weight reduction value allowed within the standard range.

[0040] S5 also includes S52; S52. Read the feature values ​​corresponding to each dimension, and call the vector weighting algorithm to perform a dimension-by-dimensional product operation on the feature values ​​of each dimension and the adjusted dimension weights to generate a sequence of score values ​​for each dimension. Then, perform interval normalization processing on the score value sequence of the same dimension to obtain the dimension evaluation value of each dimension. Perform cumulative aggregation operation on the dimension evaluation values ​​according to the dimension order to generate the industry-education integration measurement value. The system reads the sample sequence of industry-education integration metrics from the historical evaluation database, sorts them according to their numerical values ​​to generate an ordered sample sequence, calls the quantile interval partitioning algorithm, and divides the ordered sample sequence into several continuous intervals according to the preset number of intervals. The starting sample value of each interval is read as the lower bound value, and the ending sample value of each interval is read as the upper bound value. Finally, a weighted matching status evaluation is performed with the industry-education integration metrics, as detailed below: When the industry-education integration metric value is greater than or equal to the lower bound of the interval and less than the upper bound of the interval, it is determined as an interval hit, indicating that the current dimension weight matches, and the status level identifier corresponding to the current interval is determined as the current industry-education integration status level. When the industry-education integration metric value equals a certain boundary value, it indicates that the current dimension weight is matched. Based on the boundary inclusion rule, its belonging interval is selected, and the status level identifier corresponding to the belonging interval is determined as the current industry-education integration status level. When the industry-education integration metric does not meet any interval inclusion rule, it is determined as an interval miss. An interval miss indicates that the comprehensive index has exceeded the limit, which means that the current dimension weights are mismatched, and an out-of-bounds direction determination is performed, as follows: When the industry-education integration metric value is less than the lower bound of the minimum interval, the out-of-bounds direction is marked as a lower out-of-bounds, and the lower out-of-bounds handling process is triggered. The status level identifier is set to the out-of-bounds status code, and the weight is reduced through S4 iterative analysis. The weight matching status is re-evaluated. When the industry-education integration metric value is greater than or equal to the upper limit of the maximum interval, the out-of-bounds direction is marked as an upper out-of-bounds event, and the upper out-of-bounds handling process is triggered. The status level identifier is set to the out-of-bounds status code, and the weight is increased through S4 iterative analysis, and the weight matching status assessment is performed again.

[0041] In this embodiment, according to S51, the collaborative measurement value is used as an adjustment coefficient to adjust the comprehensive weight. Statistical analysis is performed on the historical collaborative measurement values, and the average and standard deviation of the collaborative measurement values ​​of each dimension in the historical data are calculated. The sum of the average and standard deviation is set as the interaction judgment threshold. Then, the current collaborative measurement value is compared with the interaction judgment threshold. When the collaborative measurement value is less than or equal to the interaction judgment threshold, the comprehensive weight is adjusted; when the collaborative measurement value is greater than the interaction judgment threshold, the comprehensive weight is adjusted again to form an adjusted dimension weight that reflects the change in the interaction relationship between dimensions. Subsequently, in S52, the feature values ​​corresponding to each dimension are read, and the feature values ​​of each dimension are multiplied with the adjusted dimension weights one dimension at a time using a vector weighting algorithm to generate a scoring value sequence. The scoring value sequence is then subjected to interval normalization to obtain the dimension evaluation value of each dimension. Then, according to the dimension order, accumulation and aggregation are performed to generate the industry-education integration measurement value. At the same time, the industry-education integration measurement value sample sequence in the historical evaluation database is read and sorted according to the numerical size to form an ordered sample sequence. The sequence is then divided into quantile intervals. The algorithm divides the ordered sample sequence into several continuous intervals and extracts the lower and upper bounds of each interval. Then, it performs matching analysis between the industry-education integration metric and the boundaries of each interval. When the interval inclusion rule is met, it is determined that the interval is hit and the corresponding industry-education integration status level is determined. When any interval inclusion rule is not met, it is determined that the interval is not hit and an out-of-bounds direction judgment is executed. When the industry-education integration metric is less than the minimum lower bound of the interval, a lower out-of-bounds flag is written and the lower out-of-bounds handling process is triggered. When the industry-education integration metric is greater than or equal to the maximum upper bound of the interval, an upper out-of-bounds flag is written and the upper out-of-bounds handling process is triggered. At the same time, iterative analysis is performed through S4 to readjust the dimension weights and perform weight matching status evaluation again. Through the above implementation process, the collaborative metric participates in the dynamic adjustment of the comprehensive weight, and a matching relationship between the industry-education integration metric and the historical sample interval is constructed based on the weight calculation results. This is used to identify the dimension weight matching status and the industry-education integration status level, so that the industry-education integration metric results reflect the collaborative correlation between multi-dimensional features and the changes in weight matching status.

[0042] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. A method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping, characterized by: Includes the following steps: S1. After obtaining legal access authorization through the data source, collect initial industry-education integration data from multiple data sources through the API interface, and obtain industry-education integration data after preprocessing. S2. Classify the industry-education integration data by similarity pattern dimension, and construct a judgment matrix to score the dimension comparison results; S3. Extract the largest eigenvalue of the judgment matrix and construct a consistency ratio for preliminary weight evaluation. At the same time, construct the entropy values ​​of each dimension for secondary weight evaluation. When the preliminary weight evaluation meets the consistency compliance and the secondary weight evaluation meets the current weight standard, construct the comprehensive weight of each dimension through the proportional scale. S4. Construct a correlation matrix and select the dependent variable to calculate the regression coefficient and its p-value, determine the statistical causal relationship, and then perform a nonlinear function transformation on the dimensional relationship value to obtain the synergistic measure value and make corrections. S5. Adjust the overall weights by using the collaborative metric as an adjustment coefficient, fit the adjusted dimensional weights with the corresponding feature values ​​of each dimension, and construct the industry-education integration metric for weight matching status evaluation.

2. The method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping according to claim 1, characterized in that: S1 includes S11; S11. After obtaining legal access authorization from the data source, collect initial industry-education data from multiple data sources through the API interface; The data sources include school academic affairs management systems, enterprise ERP systems, and government public data platforms; The initial industry-education data includes the depth of cooperation, resource matching degree, output results, and industry adaptability; The depth of cooperation includes the quantity, form, and stability of school-enterprise cooperation; the degree of resource matching includes the funds and resources invested by enterprises; the output results include the technology transfer of cooperation results; and the industry fit includes the employment quality and job placement rate of students.

3. The method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping according to claim 2, characterized in that: S1 further includes S12; S12. Process the acquired initial industry-education integration data to obtain industry-education integration data; The data processing includes data cleaning, format conversion, noise reduction, and standardization. The data cleaning is used to remove invalid data, duplicate data, and missing values ​​from the initial data of the industry-education integration program based on the data cleaning algorithm. The format conversion is used to convert different data in the initial data of industry-education integration into a standard format using a data format conversion tool. The denoising is used to filter irrelevant noise in the initial data of the production and education program based on a data denoising algorithm. The standardization process is used to convert raw data with different dimensions and units into dimensionless standardized data points according to the standardization algorithm, and to organize the dimensionless standardized data points to obtain industry-education integration data.

4. The method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping according to claim 3, characterized in that: S2 includes S21 and S22; S21. Randomly initialize all dimensionless standardized data points in the industry-education integration data into several cluster centers. Use the K-means clustering algorithm to calculate the Euclidean distance from each data point to each cluster center and assign the data points to the nearest cluster center. After each round of clustering assignment, recalculate the mean coordinates of the data points in each cluster based on the coordinates of all data points in the current cluster as the new cluster center. Compare the cluster labels of each data point in this round with the cluster labels of each data point in the previous round one by one. When the cluster labels of all data points in this round are completely consistent with the corresponding labels of the clusters of the data points in the previous round, it is determined that the clustering result of this round is consistent with the clustering result of the previous round, and the assignment is stopped. The data points assigned to the same cluster center are classified into the dimension of similar patterns. S22. Perform pairwise comparisons on any two dimensions in all dimensions and construct a judgment matrix. Preset dimension evaluation rules and assign scale values ​​to each dimension in the judgment matrix based on the comparison results of each dimension with the indicator evaluation rules. Score the comparison results of each pair of dimensions in the judgment matrix using a scale from 1 to 9, where 1 indicates that the data features corresponding to the two dimensions are at the same evaluation level, and 9 indicates that the data features corresponding to the current dimension are at a higher evaluation level than the other dimension. The preset dimension evaluation rules read the historical industry-education integration data corresponding to each dimension in the historical industry-education integration database, count the number of samples in the historical industry-education integration data of each dimension and divide them into intervals to generate several continuous evaluation intervals. Then, according to the order of the number of samples in each evaluation interval, a level identifier is written for each evaluation interval to form a dimension evaluation level sequence. The dimension name, corresponding evaluation interval and level identifier are written into the industry-education integration indicator library as dimension evaluation rules.

5. The method of claim 4, wherein the method is characterized in that: S3 includes S31; S31. Read the number of dimensions in the judgment matrix, and call the matrix eigenvalue solving algorithm to solve the characteristic equation of the judgment matrix to obtain the eigenvalue set of the judgment matrix. Then, extract the largest eigenvalue from the eigenvalue set, and perform consistency index calculation based on the largest eigenvalue and the number of dimensions of the judgment matrix. Specifically: Then, the consistency index obtained in real time is divided by the average historical consistency index to obtain the consistency ratio. The judgment matrix is ​​then preliminarily weighted based on the consistency ratio, as follows. Within a historically stable operating range, a sequence of consistency ratios is statistically analyzed, and the median value is set as the consistency threshold. When the consistency ratio is less than or equal to the consistency threshold, it means that the current judgment matrix is ​​consistent and compliant, and the current ratio scale is retained. When the consistency ratio is greater than the consistency threshold, it indicates that the current judgment matrix is ​​not compliant. At this time, the current ratio scale is increased by 1, and the dimensions in the judgment matrix are adjusted for iterative analysis until the current judgment matrix is ​​compliant.

6. The method of claim 5, wherein the method is characterized in that: S3 further includes S32 and S33; S32. Extract the data column corresponding to each dimension in the judgment matrix, perform probability distribution calculation on the current data column to generate a dimension probability distribution sequence, perform information entropy calculation based on the probability distribution sequence to obtain the entropy value of the corresponding dimension, and form an entropy value sequence by combining the entropy values ​​of each dimension. Then, calculate the mean and standard deviation of the entropy value sequence using statistical methods, set the sum of the mean and standard deviation as the entropy value critical threshold, and compare it with the entropy values ​​of each dimension. Generate a secondary weight evaluation based on the comparison results, as follows. When the entropy value is less than the entropy threshold, it means that the data points in the current dimension are concentrated and the information content of the current data point in the current dimension does not meet the current weight standard. At this time, the current scale is increased by 1 and the dimensions in the judgment matrix are adjusted for iterative analysis until the dimension meets the current weight standard. When the entropy value is greater than or equal to the entropy threshold, it means that the data points in the current dimension are evenly distributed and the information content of the current dimension meets the requirements for weight calculation. At this time, the current scale is retained. S33. After the preliminary weight assessment and the secondary weight assessment, when the preliminary weight assessment shows that the judgment matrix is ​​consistent and compliant and the secondary weight assessment shows that it meets the current weight standard, the adjusted judgment matrix is ​​read, and the weight vector solving algorithm is called to perform a multiplication calculation on each column element of the judgment matrix to generate a row product value. The dimension root operation is performed on each row product value to obtain the row feature value, and then all row feature values ​​are normalized to generate a dimension weight vector. Subsequently, the entropy value corresponding to each dimension is read and the entropy weight algorithm is called to divide the entropy value of each dimension by the sum of the entropy values ​​to obtain the objective weight sequence of the dimension. Then, the weight fusion algorithm is called to perform dimension-by-dimensional summation calculation on the dimension weight vector and the objective weight sequence of the dimension in the order of the dimensions, and the summation result is normalized to generate the comprehensive weight sequence of each dimension.

7. The method of claim 6, wherein the method is characterized in that: S4 includes S41; S41. After obtaining the comprehensive weights, read the scale of proportions between each dimension in the judgment matrix, and perform relationship measurement calculation based on the scale and the comprehensive weights of each dimension to generate the dimensional relationship value Wg, specifically: , where C ij W represents the dimensional relationship value between the i-th and j-th dimensions. i and W j Let a represent the combined weights of the i-th and j-th dimensions, respectively. ij This represents the scale between the i-th and j-th dimensions in the judgment matrix. A correlation matrix is ​​generated by calculating the dimensional relationship values ​​between all dimensions.

8. The method for measuring the collaborative relationship between industry and education based on multidimensional feature mapping according to claim 7, characterized in that: S4 further includes S42 and S43; S42. Select one dimension as the dependent variable and the other dimensions as independent variables. Use the least squares algorithm to calculate the regression coefficient of each independent variable on the dependent variable. After obtaining the regression coefficients, calculate the p-value of each regression coefficient. If the p-value is less than 0.05, it indicates that there is a statistical causal relationship and a causal relationship report is generated. If the p-value is greater than or equal to 0.05, it indicates that there is no statistical causal relationship and a no-causal relationship report is generated. S43. Calculate the synergistic metric value by performing a nonlinear function transformation based on the dimensional relationship values ​​of each pair of dimensions in the correlation matrix. When calculating the synergistic metric value, if the causal relationship is statistically causal, then introduce a causal relationship correction factor to adjust the synergistic metric value.

9. The method of claim 8, wherein the method is characterized in that: S5 includes S51; S51. Adjust the overall weight using the synergy metric as an adjustment factor. The specific adjustment scheme is as follows: Statistical analysis of historical collaborative metrics is performed using statistical methods to calculate the mean and standard deviation of collaborative metrics for each dimension in historical data, and the sum of the mean and standard deviation is preset as the interactive judgment threshold. When the collaborative metric value is less than or equal to the interaction judgment threshold, it indicates that the interaction effect of the current dimension is significant, and the overall weight is increased accordingly. When the collaborative metric value is greater than the interaction judgment threshold, it indicates that the interaction effect of the current dimension is not significant, and the overall weight is reduced accordingly.

10. The method of claim 9, wherein the method is based on a multi-dimensional feature mapping of the production and education integration. S5 also includes S52; S52. Read the feature values ​​corresponding to each dimension, and call the vector weighting algorithm to perform a dimension-by-dimensional product operation on the feature values ​​of each dimension and the adjusted dimension weights to generate a sequence of score values ​​for each dimension. Then, perform interval normalization processing on the score value sequence of the same dimension to obtain the dimension evaluation value of each dimension. Perform cumulative aggregation operation on the dimension evaluation values ​​according to the dimension order to generate the industry-education integration measurement value. The system reads the sample sequence of industry-education integration metrics from the historical evaluation database, sorts them according to their numerical values ​​to generate an ordered sample sequence, calls the quantile interval partitioning algorithm, and divides the ordered sample sequence into several continuous intervals according to the preset number of intervals. The starting sample value of each interval is read as the lower bound value, and the ending sample value of each interval is read as the upper bound value. Finally, a weighted matching status evaluation is performed with the industry-education integration metrics, as detailed below: When the industry-education integration metric value is greater than or equal to the lower bound of the interval and less than the upper bound of the interval, it is determined as an interval hit, indicating that the current dimension weight matches, and the status level identifier corresponding to the current interval is determined as the current industry-education integration status level. When the industry-education integration metric value equals a certain boundary value, it indicates that the current dimension weight is matched. Based on the boundary inclusion rule, its belonging interval is selected, and the status level identifier corresponding to the belonging interval is determined as the current industry-education integration status level. When the industry-education integration metric does not meet any interval inclusion rule, it is determined as an interval miss. An interval miss indicates that the comprehensive index has exceeded the limit, which means that the current dimension weights are mismatched, and an out-of-bounds direction determination is performed, as follows: When the industry-education integration metric value is less than the lower bound of the minimum interval, the out-of-bounds direction is marked as a lower out-of-bounds, and the lower out-of-bounds handling process is triggered. The status level identifier is set to the out-of-bounds status code, and the weight is reduced through S4 iterative analysis. The weight matching status is re-evaluated. When the industry-education integration metric value is greater than or equal to the upper limit of the maximum interval, the out-of-bounds direction is marked as an upper out-of-bounds event, and the upper out-of-bounds handling process is triggered. The status level identifier is set to the out-of-bounds status code, and the weight is increased through S4 iterative analysis, and the weight matching status assessment is performed again.

Images