Project data processing method and system applied to scientific research management system, and medium

Abstract

The invention discloses a project data processing method and system applied to a scientific research management system and a medium, and relates to the technical field of data processing, and the method comprises the steps: obtaining a feature set corresponding to each piece of project data; calculating the feature weight of each feature vector in the feature set, which is linearly represented by other feature vectors, and obtaining a feature weight set formed by the feature weights of the feature vectors; constructing and forming a corresponding target set through feature weights which are not less than a preset value in the feature weight set; performing mapping on a target data matrix established by the target set by adopting PCA to obtain a construction data matrix formed by the feature construction set of each item data; calculating an abnormal score of each piece of data in the item data, and determining the data with the abnormal score not less than a threshold value as abnormal data; the data processing difficulty is reduced, residual data, abnormal data, repeated data and the like can be removed, related personnel can clearly know the scientific research process, and meanwhile the data processing aspect is more efficient.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically, to a project data processing method, system and medium applied to a scientific research management system. Background Art

[0002] In order to improve the efficiency and quality of scientific research management, promote the transformation of scientific research results, and support the career development of scientific researchers, so as to enhance the company's competitiveness in the field of scientific research, many companies have designed scientific research management systems. Scientific research management systems improve the efficiency and quality of scientific research work through automation and intelligent means, and can achieve full-process management of scientific research projects, including application, review, progress tracking, and funding management, thereby improving the overall level of scientific research management; through standardized process management, ensure the smooth progress of scientific research projects. From project application, project establishment, execution to conclusion, each step has clear processes and standards, which reduces the arbitrariness of management and improves the standardization of project management. It provides companies with a platform for the evaluation and transformation of scientific and technological achievements, helps scientific researchers evaluate and commercialize scientific and technological achievements, and improves the transformation effect and economic value of scientific research achievements.

[0003] The current scientific research management system has simple data processing, complicated scientific research processes and irregular management. In addition, data is not interoperable during the data entry management process, which will lead to a lot of residual, abnormal and duplicate data between data of different project types, making the system complex in data processing and consuming more manpower costs. Summary of the invention

[0004] The purpose of the present invention is to provide a project data processing method, system and medium applied to a scientific research management system to solve the problems existing in the above-mentioned background technology.

[0005] The above technical objectives of the present invention are achieved through the following technical solutions: In the first aspect, the present application provides a project data processing method applied to a scientific research management system, comprising the following specific steps: According to the project data of each imported scientific research project, the feature vector corresponding to each data in each of the project data is obtained, and the feature set corresponding to each of the project data is obtained; For each of the feature sets, calculate the feature weight of each feature vector in the feature set linearly represented by other feature vectors to obtain a feature weight set composed of the feature weights of each feature vector; For each of the feature weight sets, a corresponding target set is formed by constructing a feature weight in the feature weight set that is not less than a preset value; Establish a target data matrix through each of the target sets, map the target data matrix using PCA to obtain a constructed data matrix, and obtain a feature construction set of each project data through the constructed data matrix; The abnormality score of each data in the project data is calculated by using the feature construction set and the feature set, and the data with an abnormality score not less than a threshold is determined as abnormal data.

[0006] The beneficial effects of the present invention are as follows: in the present solution, firstly, for the project data of each scientific research project imported into the scientific research management system, different methods can be used to obtain the feature vectors of each data, such as the text data can use the bag-of-words model to obtain the feature vectors, the image data can use feature recognition or deep learning methods to extract, the numerical data can use machine learning methods to extract features, etc., so as to obtain the feature set corresponding to each project data, and for each feature set, the feature weight of the feature vector can be obtained by linearly representing one of the feature vectors in each feature set by other feature vectors, and a feature weight set corresponding to the feature set and composed of the feature weights of each feature vector can be obtained; secondly, a certain number of feature weights are selected from each feature weight set according to a preset value to re-establish the target set, and the target set of each project data is established. The target data matrix is ​​constructed, and the constructed data matrix formed by the feature construction set of each project data is obtained by PCA mapping; finally, the abnormal score of the data is calculated by constructing the feature vectors of each feature in the data matrix and the corresponding data, and the data with an abnormal score not less than the threshold is determined as abnormal data. For the original data determined as abnormal data, it will be marked in the scientific research management system and the relevant staff will be reminded to verify it manually. After verification, the original data that is abnormal data can be retained or deleted according to the actual situation; in this way, when processing the data of each project in the system, the difficulty of data processing is reduced. Selective retention or deletion after abnormal data identification can make the data presentation clearer, and residual, abnormal and repeated data can be eliminated, so that relevant personnel can clearly understand the scientific research process, and at the same time, it is more efficient in data processing.

[0007] In this scheme, when identifying abnormal data, feature construction can be used to convert the original data into a more expressive form. By combining and transforming the original features, more discriminative new features can be obtained. By displaying one of the feature vectors in a linear representation through other feature vectors, the quantitative correlation between the data features in each category can be obtained, thereby better mining the deeper feature relationships between the sample data and forming a mesh topological structure, which avoids the loss of the original sample features and obtains more information from the sample, ultimately improving the final anomaly detection efficiency.

[0008] Based on the above technical solution, the present invention can also be improved as follows.

[0009] Furthermore, before acquiring the feature vectors of the above-mentioned project data, the data of each project is cleaned, specifically: Discretize the attribute items of each original data in the normalized project data, and transform each attribute value obtained after discretization into a preset integer range according to the size; The information gain rate of each attribute item of the original data is calculated based on the transformed attribute value, and the attribute set is constructed through each information gain rate that is not less than a preset value; Insert the corresponding data in the attribute set into the preset prefix tree, traverse each leaf node in the prefix tree to delete duplicate data, and obtain the project data after data cleaning.

[0010] The beneficial effect of adopting the above further solution is that the data cleaning process performed by calculating the information gain rate and inserting the prefix tree can reduce the time complexity of the detection process while ensuring the accuracy of the project data.

[0011] Furthermore, the above method also includes: After the abnormal data is processed, the random forest algorithm is used to establish a decision tree containing some random data in each project data, and the establishment of the decision tree is stopped when each data in each project data is selected at least twice; Based on each established decision tree, the first classification error of each decision tree is calculated, and by adding noise to each project data and re-establishing each new decision tree, the second classification error of each new decision tree is calculated; Based on the first classification error and the second classification error, the importance of each data is calculated using the information gain rate of each data in each project data.

[0012] The beneficial effect of adopting the above further scheme is: since the project data information is very complex and there are some redundant attribute items or attribute items that are irrelevant to data analysis, the existence of these features brings unnecessary influence to data analysis and reduces the accuracy of data analysis. Therefore, when calculating the importance of the original project data, the data with high importance can be highlighted and the attributes with low importance can be reduced, thereby improving the efficiency of data analysis.

[0013] Furthermore, the feature weight of each of the above feature vectors is linearly represented by other feature vectors, specifically: ; In the formula, Indicates the feature set i The feature weights of the feature vectors, Indicates that the feature set is i Among the eigenvectors other than the eigenvector j feature vectors, Indicates the coefficient matrix i The eigenvector is relative to the j The transformation coefficients of the eigenvectors, Indicates i The residual coefficients of the eigenvectors.

[0014] Furthermore, the information gain rate of the attribute items of the above-mentioned original data is specifically: ,in: ; In the formula, Represents project data D Medium attribute item A The information gain rate is Represents project data D Medium attribute item A The information gain of Represents an attribute item A The value is i The number of samples, Represents project data D The total number of samples in n Represents an attribute item A The number of values ​​of Represents an intermediate variable.

[0015] Furthermore, the importance of each of the above data is as follows: ; In the formula, Represents project data D Medium attribute item A The information gain rate is Represents project data D Medium attribute item A The importance of the corresponding data, represents the second classification error, represents the first classification error, N Represents the total number of decision trees in the random forest algorithm; where the first classification error or the second classification error is obtained by: ; In the formula, represents the first classification error or the second classification error, A Represents the number of samples correctly classified in the random forest algorithm, N Represents the total number of samples in the random forest algorithm.

[0016] Furthermore, the above calculation of the abnormality score of each data in the project data is specifically: ; In the formula, Indicates the feature structure set X The feature vector of the data is constructed. Indicates the feature set X The feature vector of the data, Indicates the feature set X Anomaly score of data.

[0017] In a second aspect, the present application provides a project data processing system applied to a scientific research management system, and a project data processing method applied to a scientific research management system applied to any one of the first aspects, including: The first module is used to obtain the feature vector corresponding to each data in each project data according to the project data of each imported scientific research project, and obtain the feature set corresponding to each project data; The second module is used to calculate, for each feature set, the feature weight of each feature vector in the feature set linearly represented by other feature vectors, and obtain a feature weight set composed of the feature weights of each feature vector; The third module is used to construct a corresponding target set for each feature weight set through feature weights in the feature weight set that are not less than a preset value; The fourth module is used to establish a target data matrix through each target set, and use PCA to map the target data matrix to obtain a construction data matrix formed by the feature construction set of each project data; The fifth module is used to calculate the anomaly score of each data in the project data through the feature construction set and the feature set, and determine the data with an anomaly score not less than a threshold as abnormal data.

[0018] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any one of the methods in the first aspect when executing the computer program.

[0019] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any one of the methods in the first aspect.

[0020] Compared with the prior art, the present invention has at least the following beneficial effects: In the present application, firstly, for the project data of each scientific research project imported into the scientific research management system, different methods can be used to obtain the feature vectors of each data, such as text data can use the bag-of-words model to obtain feature vectors, image data can use feature recognition or deep learning methods to extract, numerical data can use machine learning methods to extract features, etc., so as to obtain the feature set corresponding to each project data, and for each feature set, the feature weight of the feature vector can be obtained by linearly representing one of the feature vectors in each feature set by other feature vectors, and a feature weight set corresponding to the feature set and composed of the feature weights of each feature vector is obtained; secondly, a certain number of feature weights are selected from each feature weight set according to preset values ​​to re-establish the target set, and the target data matrix is ​​established through the target set of each project data. Matrix, and obtain the constructed data matrix formed by the feature construction set of each project data by PCA mapping; finally, the abnormal score of the data is calculated by constructing the feature vectors of each feature in the data matrix and the corresponding data, and the data with an abnormal score not less than the threshold is determined as abnormal data. For the original data determined as abnormal data, it will be marked in the scientific research management system and the relevant staff will be reminded to verify it manually. After verification, the original data that is abnormal data can be retained or deleted according to the actual situation; in this way, when processing the data of each project in the system, the difficulty of data processing is reduced. Selective retention or deletion after abnormal data identification can make the data presentation clearer, and residual, abnormal and repeated data can be eliminated, so that relevant personnel can clearly understand the scientific research process, and at the same time, it is more efficient in data processing.

[0021] In the present application, when identifying abnormal data, feature construction can be used to convert the original data into a more expressive form. By combining and transforming the original features, more discriminative new features can be obtained. By displaying one of the feature vectors in a linear representation through other feature vectors, the quantitative correlation between the various data features in each category can be obtained, thereby better mining the deeper feature relationships between the sample data, forming a mesh topological structure, avoiding the loss of the original sample features and obtaining more information from the sample, ultimately improving the final anomaly detection efficiency.

[0022] In the present application, the data cleaning process performed by calculating the information gain rate and inserting a prefix tree can reduce the time complexity of the detection process while ensuring the accuracy of the project data; since the project data information is very complex and there are some redundant attribute items or attribute items that are irrelevant to data analysis, the existence of these features brings unnecessary impacts to data analysis and reduces the accuracy of data analysis. Therefore, when calculating the importance of the original project data, the data with high importance can be highlighted and the attributes with low importance can be reduced, thereby improving the efficiency of data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings: Figure 1 A method flow chart of a processing method in an embodiment of the present invention; Figure 2 A connection diagram of a processing system in an embodiment of the present invention; Figure 3 Schematic diagram of the connection of electronic equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0025] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0027] In the description of the embodiments of the present invention, "plurality" means at least 2.

[0028] Embodiment 1: In order to solve the problems that the data processing in the current scientific research management system is simple, the processed data cannot intuitively display the data attributes and meanings, and there are many residual, abnormal and duplicate data between the data of different project types, which makes the system complex in data processing and consumes more manpower costs, this embodiment provides a project data processing method applied to the scientific research management system, such as Figure 1 As shown, the specific steps include: S1, according to the project data of each scientific research project imported, obtain the feature vector corresponding to each data in each project data, and obtain the feature set corresponding to each project data.

[0029] Among them, the feature vector corresponding to each data in the project data can be obtained in different ways; for example, the feature vector of text data can be obtained using the bag-of-words model, image data can be extracted using feature recognition or deep learning methods, and numerical data can be extracted using machine learning methods.

[0030] Specifically, before acquiring the feature vectors of the above-mentioned project data, the data of each project is cleaned, specifically: S11, discretizing the attribute items of each original data in the normalized project data, and transforming each attribute value obtained after discretization into a preset integer range according to the size.

[0031] S12, calculating the information gain rate of each attribute item of the original data based on the transformed attribute value, and constructing an attribute set through each information gain rate that is not less than a preset value.

[0032] Among them, the information gain rate of the attribute items of each of the above original data is specifically: ,in: ; In the formula, Represents project data D Medium attribute item A The information gain rate is Represents project data D Medium attribute item A The information gain of Represents an attribute item A The value is i The number of samples, Represents project data D The total number of samples in n Represents an attribute item A The number of values ​​of Represents an intermediate variable.

[0033] S13, inserting each corresponding data in the attribute set into a preset prefix tree, traversing each leaf node in the prefix tree to delete duplicate data, and obtaining project data after data cleaning.

[0034] Specifically, the improvements in the characteristics and structure of the preset prefix tree are as follows: 1) Non-leaf nodes are used as index split items and do not store data information.

[0035] 2) Only leaf nodes store data, and can store multiple pieces of data.

[0036] 3) There are n attributes as index items of data, with n+2 layers, where the first layer is the root node and the last layer stores sample data information.

[0037] Among them, when detecting duplicate data in the prefix tree, first, traverse the data in each leaf node in turn; secondly, compare each sample point with the data in the leaf node where it is located. If the similarity between the two data is greater than the given threshold, it will be output to the similar data set. After comparing with other data, the data will be marked as compared data and will not be compared again during the next traversal. For example, if A and B are in the same leaf node, A will be compared with B. If their similarity is greater than the given threshold, A will be output to the similar data set and deleted from the leaf node where it is located. There is only one piece of data in the leaf nodes where C, D, and G are located, so no comparison is required, that is, these single existing data are not duplicate data.

[0038] Specifically, after the prefix tree is improved, similar data can be quickly gathered in the same leaf node, reducing the calculation process, and then traversing the leaf nodes, calculating the similarity between data within the leaf nodes, completing the detection of duplicate data, and improving the efficiency of duplicate data detection.

[0039] S2, for each feature set, calculate the feature weight of each feature vector in the feature set linearly represented by other feature vectors, and obtain a feature weight set composed of the feature weights of each feature vector.

[0040] Among them, since the project data information is very complex and there are some redundant attribute items or attribute items that are irrelevant to data analysis, the existence of these features has brought unnecessary influence on data analysis and reduced the accuracy of data analysis. Therefore, when calculating the importance of the original project data, the data with high importance can be highlighted and the attributes with low importance can be reduced, thereby improving the efficiency of data analysis.

[0041] Optionally, the feature weight of each of the above feature vectors linearly represented by other feature vectors is: ; In the formula, Indicates the feature set i The feature weights of the feature vectors, Indicates that the feature set is i Among the eigenvectors other than the eigenvector j feature vectors, Indicates the coefficient matrix i The eigenvector is relative to j The transformation coefficients of the eigenvectors, Indicates i The residual coefficients of the eigenvectors.

[0042] S3, for each feature weight set, a corresponding target set is formed by constructing feature weights in the feature weight set that are not less than a preset value.

[0043] Specifically, for each feature weight set, the feature weights in the feature weight set are arranged in order of size. The target set can be established by feature weights that are not less than a preset value, or by a preset ratio, such as establishing the target set by the first 50% of the feature weights in the feature weight set arranged in order of size.

[0044] S4, establish a target data matrix through each target set, and use PCA to map the target data matrix to obtain a constructed data matrix, and obtain the feature construction set of each project data through the constructed data matrix.

[0045] Among them, using PCA for mapping can reduce the dimension of data while retaining the most important features of the data. PCA mapping is currently a more mature method, and its steps can be simplified as follows: 1. Data standardization: Standardize the original data (target data matrix) to ensure that the mean of each feature is 0 and the variance is 1 (or the standard deviation is 1). This step is to eliminate the dimensional differences between different features so that each feature contributes relatively equally to PCA; 2. Calculate the covariance matrix: Use the standardized data to calculate the covariance matrix between features. The covariance matrix reflects the degree of correlation between features and is the basis of PCA. 3. Calculate eigenvalues ​​and eigenvectors; perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors. The eigenvector represents the direction of the new coordinate axis after the data transformation, while the eigenvalue represents the variance in this direction; 4. Select the principal components; arrange the eigenvalues ​​from large to small, and select the eigenvectors corresponding to the largest k eigenvalues ​​as the principal components. Here k is the number of dimensions you want to retain, that is, the target dimension after dimensionality reduction; 5. Data projection: Project the original data onto the selected principal components to obtain the reduced-dimensional data. This step achieves data dimensionality reduction while retaining the main information of the original data as much as possible.

[0046] S5, calculating the anomaly score of each data in the project data through the feature construction set and the feature set, and determining the data with an anomaly score not less than the threshold as abnormal data.

[0047] Optionally, the above calculation of the anomaly score of each data in the project data is specifically: ; In the formula, Indicates the feature structure set X The feature vector of the data is constructed. Indicates the feature set X The feature vector of the data, Indicates the feature set X Anomaly score of data.

[0048] Among them, the original data determined to be abnormal data will be marked in the scientific research management system and the relevant staff will be reminded to manually verify it. After verification, the original data that is abnormal data can be retained or deleted according to the actual situation; in this way, when processing data for each project in the system, the difficulty of data processing is reduced. Selective retention or deletion after abnormal data identification can make the data presentation clearer, and residual, abnormal and duplicate data can be eliminated, so that relevant personnel can clearly understand the scientific research process, and at the same time, it is more efficient in data processing.

[0049] Optionally, the above method further includes: S51, using the random forest algorithm to build a decision tree containing some random data in each project data for each project data after the abnormal data is processed, and stopping the building of the decision tree when each data in each project data is selected at least twice; the building of the decision tree in the random forest algorithm is a relatively mature method and will not be repeated here, and stopping the building of the decision tree when each data in each project data is selected at least twice can ensure that after all decision trees are built, all attribute items participate in the building of the random forest.

[0050] S52, based on each established decision tree, a first classification error of each decision tree is calculated, and a second classification error of each new decision tree is calculated by adding noise to each project data and re-establishing each new decision tree.

[0051] S53, based on the first classification error and the second classification error, the importance of each data is calculated using the information gain rate of each data in each project data.

[0052] Specifically, since the project data information is very complex and there are some redundant attribute items or attribute items that are irrelevant to data analysis, the existence of these features has brought unnecessary influence on data analysis and reduced the accuracy of data analysis. Therefore, when calculating the importance of the original project data, the data with high importance can be highlighted and the attributes with low importance can be reduced, thereby improving the efficiency of data analysis.

[0053] Optionally, the importance of each of the above data is as follows: ; In the formula, Represents project data D Medium attribute item A The information gain rate is Represents project data D Medium attribute item A The importance of the corresponding data, represents the second classification error, represents the first classification error, N Represents the total number of decision trees in the random forest algorithm; where the first classification error or the second classification error is obtained by: ; In the formula, represents the first classification error or the second classification error, A Represents the number of samples correctly classified in the random forest algorithm, N Represents the total number of samples in the random forest algorithm.

[0054] Embodiment 2: This embodiment of the application provides a project data processing system applied to a scientific research management system, and a project data processing method applied to a scientific research management system in any one of Embodiment 1, such as Figure 2 As shown, including: The first module is used to obtain the feature vector corresponding to each data in each project data according to the project data of each imported scientific research project, and obtain the feature set corresponding to each project data; The second module is used to calculate, for each feature set, the feature weight of each feature vector in the feature set linearly represented by other feature vectors, and obtain a feature weight set composed of the feature weights of each feature vector; The third module is used to construct a corresponding target set for each feature weight set through feature weights in the feature weight set that are not less than a preset value; The fourth module is used to establish a target data matrix through each target set, and use PCA to map the target data matrix to obtain a construction data matrix formed by the feature construction set of each project data; The fifth module is used to calculate the anomaly score of each data in the project data through the feature construction set and the feature set, and determine the data with an anomaly score not less than a threshold as abnormal data.

[0055] Embodiment 3: This embodiment of the present application provides an electronic device, such as Figure 3 As shown, it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any method in Embodiment 1 is implemented.

[0056] Embodiment 4: The embodiment of the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any method in Embodiment 1.

[0057] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A project data processing method applied to a scientific research management system, characterized in that: The specific steps include: According to the project data of each imported scientific research project, the feature vector corresponding to each data in each of the project data is obtained, and the feature set corresponding to each of the project data is obtained; For each of the feature sets, calculate the feature weight of each feature vector in the feature set linearly represented by other feature vectors to obtain a feature weight set composed of the feature weights of each feature vector; For each of the feature weight sets, a corresponding target set is formed by constructing a feature weight in the feature weight set that is not less than a preset value; Establish a target data matrix through each of the target sets, map the target data matrix using PCA to obtain a constructed data matrix, and obtain a feature construction set of each project data through the constructed data matrix; The abnormality score of each data in the project data is calculated by using the feature construction set and the feature set, and the data with an abnormality score not less than a threshold is determined as abnormal data.

2. The project data processing method applied to a scientific research management system according to claim 1 is characterized in that: Before acquiring the feature vector of each of the project data, each of the project data is cleaned, specifically: Discretize the attribute items of each original data in the normalized project data, and transform each attribute value obtained after discretization into a preset integer range according to the size; The information gain rate of each attribute item of the original data is calculated based on the transformed attribute value, and the attribute set is constructed through each information gain rate that is not less than a preset value; Insert the corresponding data in the attribute set into the preset prefix tree, traverse each leaf node in the prefix tree to delete duplicate data, and obtain the project data after data cleaning.

3. The project data processing method applied to a scientific research management system according to claim 1 is characterized in that: The method further comprises: Using the random forest algorithm to build a decision tree containing some random data in each project data after the abnormal data is processed, and stop building the decision tree when each data in each project data is selected at least twice; Based on each established decision tree, the first classification error of each decision tree is calculated, and by adding noise to each project data and re-establishing each new decision tree, the second classification error of each new decision tree is calculated; Based on the first classification error and the second classification error, the importance of each data is calculated using the information gain rate of each data in the project data.

4. The project data processing method applied to a scientific research management system according to claim 1 is characterized in that: The feature weight of each feature vector linearly represented by other feature vectors is specifically: ； In the formula, Indicates the feature set i The feature weights of the feature vectors, Indicates that the feature set is i Among the eigenvectors other than the eigenvector j feature vectors, Indicates the coefficient matrix i The eigenvector is relative to the j The transformation coefficients of the eigenvectors, Indicates i The residual coefficients of the eigenvectors.

5. The project data processing method applied to a scientific research management system according to claim 2 is characterized in that: The information gain rate of the attribute items of each original data is as follows: ,in: ; In the formula, Represents project data D Medium attribute item A The information gain rate, Represents project data D Medium attribute item A The information gain of Represents an attribute item A The value is i The number of samples, Represents project data D The total number of samples in n Represents an attribute item A The number of values ​​of Represents an intermediate variable.

6. The project data processing method applied to a scientific research management system according to claim 3 is characterized in that: The importance of each data is as follows: ； In the formula, Represents project data D Medium attribute item A The information gain rate, Represents project data D Medium attribute item A The importance of the corresponding data, represents the second classification error, represents the first classification error, N Represents the total number of decision trees in the random forest algorithm; where the first classification error or the second classification error is obtained by: ; In the formula, represents the first classification error or the second classification error, A Represents the number of samples correctly classified in the random forest algorithm, N Represents the total number of samples in the random forest algorithm.

7. The project data processing method applied to a scientific research management system according to claim 1 is characterized in that: Calculate the anomaly score of each data in the project data, specifically: ； In the formula, Indicates the feature structure set X The feature vector of the data is constructed. Indicates the feature set X The feature vector of the data, Indicates the feature set X Anomaly score of data.

8. A project data processing system applied to a scientific research management system, applied to a project data processing method applied to a scientific research management system as claimed in any one of claims 1 to 7, characterized in that: include: The first module is used to obtain the feature vector corresponding to each data in each project data according to the project data of each imported scientific research project, and obtain the feature set corresponding to each project data; The second module is used to calculate, for each of the feature sets, a feature weight of each feature vector in the feature set linearly represented by other feature vectors, to obtain a feature weight set composed of the feature weights of each feature vector; A third module is used to construct a corresponding target set for each of the feature weight sets by using the feature weight in the feature weight set that is not less than a preset value; The fourth module is used to establish a target data matrix through each of the target sets, and map the target data matrix using PCA to obtain a construction data matrix formed by the feature construction sets of each project data; The fifth module is used to calculate the abnormality score of each data in the project data through the feature construction set and the feature set, and determine the data with an abnormality score not less than a threshold as abnormal data.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the method according to any one of claims 1 to 7 is implemented when the processor executes the computer program.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which enable a computer to execute the method of any one of claims 1-7.

Images