Project compliance evaluation method and system based on atlas analysis

Abstract

The invention discloses a project compliance assessment method and system based on atlas analysis, and the method comprises the steps: collecting and processing the source feature data of a project, risk graph network construction parameters and a project detection correlation coefficient, and obtaining the compliance assessment features of a detection project; and constructing a cross-modal project compliance evaluation model by using the feature and the corresponding project compliance detection level, receiving a to-be-detected project compliance evaluation feature obtained by processing a to-be-evaluated project, and processing the to-be-detected project compliance evaluation feature by the constructed model to generate a real-time project compliance detection level for evaluation. According to the method, an optimized atlas analysis method is adopted, a project objective equipment parameter improvement coefficient is detected, and compliance evaluation level classification is carried out according to a big data neural network model constructed by project compliance keywords and constructed atlas network parameters, so that personalized feature processing and acquisition of data are realized; project compliance assessment is more automatic, resources are saved, project risks are reduced, and the reasonability of project compliance decision making is enhanced.

Description

Technical Field

[0001] The present invention belongs to the field of compliance assessment, and in particular relates to a project compliance assessment method and system based on graph analysis. Background Art

[0002] Evaluating the feasibility of a testing project lays a solid foundation for the research and development of subsequent testing projects. The testing project evaluation system can compare existing planned projects with those in the database and evaluate them in accordance with existing industry standards.

[0003] Some documents, such as the Chinese Patent Gazette CN112364174A (Patent Document 1), describe a method and system for evaluating similarity of patient medical records based on a knowledge graph, which describes prior art documents on entity recognition. Information extraction uses a bidirectional LSTM network with a conditional random field layer based on an attention mechanism to extract entity relationships.

[0004] In addition, for pharmaceutical projects based on knowledge graphs, statistical models and deep learning technologies are integrated for named entity recognition, and the application of bidirectional long short-term memory neural networks further improves the accuracy of information extraction.

[0005] Existing technologies use deep learning methods to extract information from databases. By repeatedly using convolutional neural networks and recurrent neural networks to mine and extract important knowledge in projects, the concept of subject knowledge graphs is expanded to professional fields. Entity recognition and relationship extraction are performed under the current situation where entity types are relatively limited, thereby improving the accuracy of project evaluation.

[0006] However, compared with the compliance assessment of testing projects, the above methods are not applicable, and the applied technology is very complex. If the knowledge graph is directly applied to the assessment of testing projects, there are obvious deficiencies, and more detailed compliance assessments for projects are even more lacking. The assessment of the compliance of testing projects has not yet been invented. In order to obtain more automated project compliance assessment results for testing project assessment scenarios, how to integrate cross-modal objective source parameters, risk feature parameters for constructing graphs, and testing project correlation coefficients, adopt optimized graph analysis methods and testing project objective equipment parameter improvement coefficients, and classify compliance assessment levels using a big data neural network model constructed with project compliance keywords and constructed graph network parameters, realize personalized feature processing and collection of project compliance big data, make project compliance assessments more automated, provide more accurate and comprehensive project compliance assessment information, help save resources and reduce project risks, and enhance the rationality of project compliance decisions, are the difficulties faced by existing technologies.

[0007] This invention proposes a project compliance assessment method and system based on graph analysis. The system collects project source feature data (national, provincial, municipal, company, and financial data), risk graph network construction parameters (a relationship graph constructed between project vocabulary and compliance-related vocabulary, calculated using a graph network relationship calculation formula (an improved similarity method to calculate various network feature values)), and project detection correlation coefficients (a calculation coefficient that combines the project's planned improvement feature parameters with the actual feature function of the improved detection equipment). These features are then matrixed to generate feature vectors to obtain compliance assessment features for the detection projects. This feature and existing corresponding project compliance detection levels are then used to construct a cross-modal project compliance assessment model. The model then receives the compliance assessment features of the detected projects, which are then processed and improved based on application scenarios. This model then processes the compliance assessment features of the detected projects to generate real-time project compliance detection levels for evaluation. Project evaluators then make compliance judgments based on the real-time project compliance detection levels, thereby making corresponding compliance decisions. Summary of the Invention

[0008] In order to solve the above technical problems, the present invention proposes a project compliance assessment method and system based on graph analysis.

[0009] In a first aspect of the present invention, a project compliance assessment method based on graph analysis is provided, the method comprising the following steps:

[0010] T1. Obtain source feature data of the inspection items and use graph network relationship processing to obtain risk graph network construction parameters. Calculate the item inspection correlation coefficient based on the inspection items and the features of the improved inspection equipment.

[0011] T2. Processing the source feature data, the risk graph network construction parameters, and the project detection correlation coefficient to form a detection project compliance assessment feature vector;

[0012] T3. Collecting project compliance detection levels obtained by evaluating the detection items, and constructing a cross-modal project compliance assessment model using the detection item compliance assessment feature vectors and the project compliance detection levels;

[0013] T4. Use the test data to evaluate and confirm the cross-modal project compliance assessment model, select the cross-modal project compliance assessment model with a higher coefficient than the test data as the optimal cross-modal project compliance assessment model, and use the optimal cross-modal project compliance assessment model to conduct compliance assessment of the project to be tested.

[0014] Furthermore, the source characteristic data includes project level coefficient, project source level and financial situation data.

[0015] Furthermore, before obtaining the risk graph network construction parameters by using graph network relationship processing, a relationship graph between project vocabulary and compliance vocabulary is constructed, and a risk graph network is constructed based on the top H compliance-related vocabulary appearing in the project and the top K most utilized compliance vocabulary. The risk graph network is constructed by quantifying the vocabulary feature vectors of the top H vocabulary and the top K vocabulary and representing them as a graph network in the form of node values ​​and edge degrees.

[0016] Furthermore, the risk graph network construction parameters are obtained by calculating and processing the graph network relationship.

[0017] Furthermore, the project detection correlation coefficient is obtained by comparing the practical features of the planned improvement of the detection equipment in the detection project with the improvement features of the actual detection equipment.

[0018] Furthermore, the cross-modal project compliance assessment model or the optimal cross-modal project compliance assessment model adopts an improved neural network model based on graph analysis.

[0019] Furthermore, the test data coefficient is obtained by testing projects that have undergone compliance assessment, and the test project compliance detection level obtained by outputting the cross-modal project compliance assessment model based on the projects that have undergone compliance assessment is compared with the actual project compliance detection level. The test data coefficient threshold is 0.8, that is, the cross-modal project compliance assessment model with a test data coefficient greater than or equal to 0.8 is selected as the optimal cross-modal project compliance assessment model.

[0020] Furthermore, the source characteristic data, the risk map network construction parameters and the project detection correlation coefficient are processed to form a detection project compliance assessment characteristic vector, which is obtained by adopting a characteristic vector reconstruction method, specifically by performing horizontal matrix splicing of the characteristic vector values.

[0021] A project compliance assessment system based on graph analysis is also provided. The system includes an Internet feature acquisition module, a graph network construction module, a risk graph network construction parameter processing module, a project detection correlation coefficient acquisition module, a cross-modal project compliance assessment model construction module, and a detection project assessment analysis module. The system is characterized by:

[0022] The Internet feature acquisition module is used to acquire source feature data of the detection items;

[0023] The graph network construction module: constructs a graph network using compliance vocabulary and project-wide compliance vocabulary;

[0024] The risk graph network construction parameter processing module is used to obtain risk graph network construction parameters by processing graph network relationships;

[0025] The project detection correlation coefficient acquisition module is used to obtain the project detection correlation coefficient using feature processing for the detection items and the improved detection equipment;

[0026] The cross-modal project compliance assessment model construction module processes the source feature data, the risk graph network construction parameters, and the project detection correlation coefficient to form a detection project compliance assessment feature vector; collects the project compliance detection levels obtained by evaluating the detection projects, and uses the detection project compliance assessment feature vector and the project compliance detection levels to construct a cross-modal project compliance assessment model; uses test data to evaluate and confirm the cross-modal project compliance assessment model, and selects a cross-modal project compliance assessment model with a coefficient higher than the test data as the optimal cross-modal project compliance assessment model;

[0027] The detection project evaluation and analysis module is connected to the cross-modal project compliance assessment model construction module, receives and uses the optimal cross-modal project compliance assessment model to perform compliance assessment on the project to be detected.

[0028] Furthermore, before obtaining risk graph network construction parameters through graph network relationship processing, a relationship graph between project-appearing words and compliance words is constructed, and a risk graph network is constructed based on the top H compliance-related words appearing in the project and the top K most utilized compliance words. In constructing the risk graph network, the top H words and the top K words are quantified into vocabulary feature vectors and represented as a graph network in the form of node values ​​and edge degrees.

[0029] The risk graph network construction parameters are calculated based on the graph network relationship calculation formula, and are specifically calculated using the vocabulary feature vector values ​​of compliance-related vocabulary in the detection project, the feature vector values ​​of compliance vocabulary, and the degree of network edges.

[0030] This invention proposes a project compliance assessment method and system based on graph analysis. The system collects project source feature data (national, provincial, municipal, company, and financial data), risk graph network construction parameters (a relationship graph constructed between project vocabulary and compliance-related vocabulary, calculated using a graph network relationship calculation formula (an improved similarity method to calculate various network feature values)), and project detection correlation coefficients (a calculation coefficient that combines the project's planned improvement feature parameters with the actual feature function of the improved detection equipment). These features are then matrixed to generate feature vectors to obtain compliance assessment features for the detection projects. This feature and existing corresponding project compliance detection levels are then used to construct a cross-modal project compliance assessment model. The model then receives the compliance assessment features of the detected projects, which are then processed and improved based on application scenarios. This model then processes the compliance assessment features of the detected projects to generate real-time project compliance detection levels for evaluation. Project evaluators then make compliance judgments based on the real-time project compliance detection levels, thereby making corresponding compliance decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a flow chart of a project compliance assessment method based on graph analysis of the present invention;

[0032] Figure 2 This is a schematic diagram of the structure of a project compliance assessment system based on graph analysis of the present invention;

[0033] Figure 3 It is a schematic diagram of the graph network after the graph network in the present invention is constructed;

[0034] Figure 4 Schematic diagram of the calculation principle of similarity in the present invention;

[0035] Figure 5 The figure is a schematic diagram of the structure of an electronic device for implementing the method of the present invention according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The invention is further described below with reference to the accompanying drawings and specific implementation methods.

[0037] The first embodiment of the present invention:

[0038] The present invention proposes a project compliance assessment method and system based on graph analysis.

[0039] In a first aspect of the present invention, a project compliance assessment method based on graph analysis is provided, the method comprising the following steps:

[0040] T1. Obtain source feature data of the inspection items and use graph network relationship processing to obtain risk graph network construction parameters. Calculate the item inspection correlation coefficient based on the inspection items and the features of the improved inspection equipment.

[0041] T2. Processing the source feature data, the risk graph network construction parameters, and the project detection correlation coefficient to form a detection project compliance assessment feature vector;

[0042] T3. Collecting project compliance detection levels obtained by evaluating the detection items, and constructing a cross-modal project compliance assessment model using the detection item compliance assessment feature vectors and the project compliance detection levels;

[0043] T4. Use the test data to evaluate and confirm the cross-modal project compliance assessment model, select the cross-modal project compliance assessment model with a higher coefficient than the test data as the optimal cross-modal project compliance assessment model, and use the optimal cross-modal project compliance assessment model to conduct compliance assessment of the project to be tested.

[0044] Furthermore, the source characteristic data includes the project level coefficient, the project source level and financial situation data, and the calculation formula is: ;

[0045] Since project compliance and project sources have a certain impact, most project compliance issues arise from problems at the project level and financial situation. Therefore, this invention fully considers the impact of this type of data on project compliance and optimizes the calculation of source feature data suitable for subsequent model classification processing to classify project compliance levels. is the source characteristic data, i represents the source of the i-th project, m represents the total number of project sources, L represents the project level coefficient, and its value is obtained according to the project level. The project level is divided into national, provincial, municipal, and company levels, corresponding to 1, 2, 3, and 4 respectively. is the level of the source of the i-th project, which is obtained by taking values ​​according to the unit level. The unit level is divided into ministry level, department level, bureau level and town level, and the values ​​are 1, 2, 3 and 4 respectively. j represents the j-th project provider of the project, and n represents the total project providers. represents the average annual expenditure of the j-th project provider, It represents the average annual income of the j-th project provider. Generally speaking, the value of m is equal to n. However, some projects have more project sources than project providers. This is because some project sources may not provide financial support.

[0046] Furthermore, before obtaining the risk graph network construction parameters by using graph network relationship processing, a relationship graph between project vocabulary and compliance vocabulary is constructed, and a risk graph network is constructed based on the top H compliance-related vocabulary appearing in the project and the top K most utilized compliance vocabulary. The risk graph network is constructed by quantifying the vocabulary feature vectors of the top H vocabulary and the top K vocabulary and representing them as a graph network in the form of node values ​​and edge degrees.

[0047] The simple form of some networks constructed by the graph network is shown in Figure 2 As shown, the compliance vocabulary and the compliance vocabulary within the project are nodes, the node values ​​are the vocabulary feature vector values ​​converted from the compliance vocabulary and the compliance vocabulary within the project, and the association between the vocabulary is constructed by connecting the edges and the weight values ​​of the edges.

[0048] Furthermore, the risk graph network construction parameters are calculated according to the graph network relationship calculation formula, specifically using the vocabulary feature vector values ​​of compliance-related words in the detection project, the feature vector values ​​of compliance words, and the degree of the network edge: ;

[0049] Since the more compliance-related words appear in a project, the greater the compliance risk of the project, the corresponding graph analysis and calculation are performed using the risk graph network construction parameters after the graph relationship between the compliance-related words and the compliance-related words in the project is constructed. Specifically, the corresponding similarity function is used in combination with the word feature vector value to perform similarity calculation to obtain the risk graph network construction parameters to represent the compliance of the detection project. In the formula, Constructing parameters for the risk graph network, The number of top K types of words that are used most frequently for compliant words, The number of the top H types of compliance-related words that appear in the test items, represents the compliance word that is used most frequently in the ath test item, and b represents the compliance-related words that appear in the bth test item. represents the vocabulary feature vector value of the ath most compliant vocabulary, Represents the vocabulary feature vector value of compliance-related words appearing in the b-th detection item, Represents the similarity calculation benchmark value. Since some words in a project will simultaneously generate graph construction associations with multiple qualified words, the maximum degree and the value of the graph relationship are used to correct the mean similarity between each word. is the maximum number of edges in the graph network, is the maximum weight of the edge, is the minimum weight of the edge.

[0050] Furthermore, the project detection correlation coefficient is calculated using the ratio of the practical features of the planned improved detection equipment in the detection project to the actual detection equipment improvement features: ;

[0051] Inspection projects are generally accepted based on the improvement of the functional parameters of the inspection equipment. Therefore, the ratio of the improvement of the inspection equipment function according to the project plan and the improvement of the inspection equipment function obtained in the actual project is used as the calculation of the project inspection correlation coefficient to evaluate the compliance of the inspection project. This is also an indispensable part of project evaluation. The correlation coefficient of the project is detected. In this invention, the main focus is on the accuracy, sensitivity and consistency of the detection equipment. 、 、 They respectively indicate plans to improve the accuracy, sensitivity and consistency of testing equipment in the testing project. 、 、 They respectively indicate that the testing items actually improve the accuracy, sensitivity and detection consistency of the testing equipment.

[0052] Furthermore, the cross-modal project compliance assessment model or the optimal cross-modal project compliance assessment model adopts an improved neural network model based on graph analysis, and its activation function calculation formula is as follows: ;

[0053] is the activation function value, X is the input test item compliance assessment feature vector, Constructing parameters for the risk graph network. According to the graph analysis in the present invention, the activation function can introduce nonlinear transformations into the neural network, enabling it to fit any complex function, limit the neuron output, and adapt to the scene requirements. The graph network in the present invention has a significant impact on the construction of the neural network. By using the risk graph network to construct parameters, the gradient propagation adjustment of the neural network with scene adaptation can be performed, so that the model training effect is more optimized. In order to adapt to the changes with the basic activation function as a variant, Taking the corresponding logarithmic value is more consistent with preventing the impact of gradient explosion and providing strong support for model construction.

[0054] Furthermore, the test data coefficient is obtained by testing projects that have undergone compliance assessment, and the test project compliance detection level obtained by outputting the cross-modal project compliance assessment model based on the projects that have undergone compliance assessment is compared with the actual project compliance detection level. The test data coefficient threshold is 0.8, that is, the cross-modal project compliance assessment model with a test data coefficient greater than or equal to 0.8 is selected as the optimal cross-modal project compliance assessment model.

[0055] Furthermore, the source characteristic data, the risk map network construction parameters and the project detection correlation coefficient are processed to form a detection project compliance assessment characteristic vector, which is obtained by adopting a characteristic vector reconstruction method, specifically by matrix splicing of the characteristic vector values.

[0056] A project compliance assessment system based on graph analysis is also provided. The system includes an Internet feature acquisition module, a graph network construction module, a risk graph network construction parameter processing module, a project detection correlation coefficient acquisition module, a cross-modal project compliance assessment model construction module, and a detection project assessment analysis module. The system is characterized by:

[0057] The Internet feature acquisition module is used to acquire source feature data of the detection items;

[0058] The graph network construction module: constructs a graph network using compliance vocabulary and project-wide compliance vocabulary;

[0059] The risk graph network construction parameter processing module is used to obtain risk graph network construction parameters by processing graph network relationships;

[0060] The project detection correlation coefficient acquisition module is used to obtain the project detection correlation coefficient using feature processing for the detection items and the improved detection equipment;

[0061] The cross-modal project compliance assessment model construction module processes the source feature data, the risk graph network construction parameters, and the project detection correlation coefficient to form a detection project compliance assessment feature vector; collects the project compliance detection levels obtained by evaluating the detection projects, and uses the detection project compliance assessment feature vector and the project compliance detection levels to construct a cross-modal project compliance assessment model; uses test data to evaluate and confirm the cross-modal project compliance assessment model, and selects a cross-modal project compliance assessment model with a coefficient higher than the test data as the optimal cross-modal project compliance assessment model;

[0062] The detection project evaluation and analysis module is connected to the cross-modal project compliance assessment model construction module, receives and uses the optimal cross-modal project compliance assessment model to perform compliance assessment on the project to be detected.

[0063] Furthermore, before obtaining risk graph network construction parameters through graph network relationship processing, a relationship graph between project-appearing words and compliance words is constructed, and a risk graph network is constructed based on the top H compliance-related words appearing in the project and the top K most utilized compliance words. In constructing the risk graph network, the top H words and the top K words are quantified into vocabulary feature vectors and represented as a graph network in the form of node values ​​and edge degrees.

[0064] The risk graph network construction parameters are calculated based on the graph network relationship calculation formula, specifically using the vocabulary feature vector values ​​of compliance-related words in the test project, the feature vector values ​​of compliance words, and the degree of the network edge: ;

[0065] Since the more compliance-related words appear in a project, the greater the compliance risk of the project, the corresponding graph analysis and calculation are performed using the risk graph network construction parameters after the graph relationship between the compliance-related words and the compliance-related words in the project is constructed. Specifically, the corresponding similarity function is used in combination with the word feature vector value to perform similarity calculation to obtain the risk graph network construction parameters to represent the compliance of the detection project. In the formula, Constructing parameters for the risk graph network, The number of top K types of words that are used most frequently for compliant words, The number of the top H types of compliance-related words that appear in the test items, represents the compliance word that is used most frequently in the ath test item, and b represents the compliance-related words that appear in the bth test item. represents the vocabulary feature vector value of the ath most compliant vocabulary, Represents the vocabulary feature vector value of compliance-related words appearing in the b-th detection item, Represents the similarity calculation benchmark value. Since some words in a project will simultaneously generate graph construction associations with multiple qualified words, the maximum degree and the value of the graph relationship are used to correct the mean similarity between each word. is the maximum number of edges in the graph network, is the maximum weight of the edge, is the minimum weight of the edge.

[0066] This invention proposes a project compliance assessment method and system based on graph analysis. The system collects project source feature data (national, provincial, municipal, company, and financial data), risk graph network construction parameters (a relationship graph constructed between project vocabulary and compliance-related vocabulary, calculated using a graph network relationship calculation formula (an improved similarity method to calculate various network feature values)), and project detection correlation coefficients (a calculation coefficient that combines the project's planned improvement feature parameters with the actual feature function of the improved detection equipment). These features are then matrixed to generate feature vectors to obtain compliance assessment features for the detection projects. This feature and existing corresponding project compliance detection levels are then used to construct a cross-modal project compliance assessment model. The model then receives the compliance assessment features of the detected projects, which are then processed and improved based on application scenarios. This model then processes the compliance assessment features of the detected projects to generate real-time project compliance detection levels for evaluation. Project evaluators then make compliance judgments based on the real-time project compliance detection levels, thereby making corresponding compliance decisions.

[0067] The combination of multiple embodiments of the present invention can achieve all of the above-mentioned effects, but it is not required that every embodiment of the present invention achieve all of the above-mentioned advantages and effects, because each embodiment of the present invention can constitute an independent technical solution and make one or more contributions to the prior art.

[0068] For any module structure not specifically defined in this invention, the prior art shall prevail. The prior art mentioned in the aforementioned background and specific embodiments of this invention may be considered as part of this invention and used to understand the meaning of certain technical features or parameters. The scope of protection of this invention shall be based on the actual content of the claims.

Claims

1. A project compliance assessment method based on graph analysis, characterized in that: The method comprises the following steps: T1. Obtain source feature data of the inspection items and use graph network relationship processing to obtain risk graph network construction parameters. Calculate the item inspection correlation coefficient based on the inspection items and the features of the improved inspection equipment. T2. Processing the source feature data, the risk graph network construction parameters, and the project detection correlation coefficient to form a detection project compliance assessment feature vector; T3. Collecting project compliance detection levels obtained by evaluating the detection items, and constructing a cross-modal project compliance assessment model using the detection item compliance assessment feature vectors and the project compliance detection levels; T4. Use the test data to evaluate and confirm the cross-modal project compliance assessment model, select the cross-modal project compliance assessment model with a higher coefficient than the test data as the optimal cross-modal project compliance assessment model, and use the optimal cross-modal project compliance assessment model to conduct compliance assessment of the project to be tested.

2. The project compliance assessment method based on graph analysis according to claim 1, characterized in that: The source characteristic data includes project level coefficient, project source level and financial situation data.

3. The project compliance assessment method based on graph analysis according to claim 1, characterized in that: Before obtaining the risk graph network construction parameters by using graph network relationship processing, a relationship graph between project vocabulary and compliance vocabulary is constructed, and a risk graph network is constructed based on the top H compliance-related vocabulary appearing in the project and the top K most utilized compliance vocabulary. The risk graph network is constructed by quantizing the vocabulary feature vectors of the top H vocabulary and the top K vocabulary and representing them as a graph network in the form of node values ​​and edge degrees.

4. The project compliance assessment method based on graph analysis according to claim 3, characterized in that: The risk graph network construction parameters are obtained by calculating and processing the graph network relationship.

5. The project compliance assessment method based on graph analysis according to claim 1, characterized in that: The project detection correlation coefficient is obtained by comparing the practical features of the planned improvement of the detection equipment in the detection project with the improvement features of the actual detection equipment.

6. The project compliance assessment method based on graph analysis according to claim 2, 4 or 5, characterized in that: The cross-modal project compliance assessment model or the optimal cross-modal project compliance assessment model adopts an improved neural network model based on graph analysis.

7. The project compliance assessment method based on graph analysis according to claim 6, characterized in that: The test data coefficient is obtained by testing projects that have undergone compliance assessments, and the test project compliance detection level obtained by outputting the cross-modal project compliance assessment model based on the projects that have undergone compliance assessments is compared with the actual project compliance detection level. The test data coefficient threshold is 0.8, that is, the cross-modal project compliance assessment model with a test data coefficient greater than or equal to 0.8 is selected as the optimal cross-modal project compliance assessment model.

8. The project compliance assessment method based on graph analysis according to claim 7, characterized in that: The source characteristic data, the risk map network construction parameters and the project detection correlation coefficient are processed to form a detection project compliance assessment characteristic vector, which is obtained by adopting a characteristic vector reconstruction method, and specifically by performing horizontal matrix splicing of the characteristic vector values.

9. A project compliance assessment system based on graph analysis, comprising an internet feature acquisition module, a graph network construction module, a risk graph network construction parameter processing module, a project detection correlation coefficient acquisition module, a cross-modal project compliance assessment model construction module, and a detection project assessment analysis module, characterized by: The Internet feature acquisition module is used to acquire source feature data of the detection items; The graph network construction module: constructs a graph network using compliance vocabulary and project-wide compliance vocabulary; The risk graph network construction parameter processing module is used to obtain risk graph network construction parameters by processing graph network relationships; The project detection correlation coefficient acquisition module is used to obtain the project detection correlation coefficient using feature processing for the detection items and the improved detection equipment; The cross-modal project compliance assessment model construction module processes the source feature data, the risk graph network construction parameters, and the project detection correlation coefficient to form a detection project compliance assessment feature vector; collects the project compliance detection levels obtained by evaluating the detection projects, and uses the detection project compliance assessment feature vector and the project compliance detection levels to construct a cross-modal project compliance assessment model; uses test data to evaluate and confirm the cross-modal project compliance assessment model, and selects a cross-modal project compliance assessment model with a coefficient higher than the test data as the optimal cross-modal project compliance assessment model; The detection project evaluation and analysis module is connected to the cross-modal project compliance assessment model construction module, receives and uses the optimal cross-modal project compliance assessment model to perform compliance assessment on the project to be detected.

10. The project compliance assessment system based on graph analysis according to claim 9, characterized in that: Before obtaining the risk graph network construction parameters through graph network relationship processing, a relationship graph between project-appearing words and compliance words is constructed, and a risk graph network is constructed based on the top H compliance-related words appearing in the project and the top K most utilized compliance words. In constructing the risk graph network, the top H words and the top K words are quantified into vocabulary feature vectors and represented as a graph network in the form of node values ​​and edge degrees; The risk graph network construction parameters are calculated based on the graph network relationship calculation formula, and are specifically calculated using the vocabulary feature vector values ​​of compliance-related vocabulary in the detection project, the feature vector values ​​of compliance vocabulary, and the degree of network edges.

Images