Network security level protection management system and method based on artificial intelligence

Through natural language processing technology based on deep learning, semantic understanding of the basic requirements for network security level protection and equipment evaluation items evidence, and using fine-grained point-by-point one-way attention interaction matching model for evaluation, solving the low efficiency of manual evaluation in the existing technology and the problem of poor accuracy has realized the automation and intelligent security level protection evaluation of network equipment, improving the evaluation efficiency and accuracy.

CN118709925BActive Publication Date: 2024-11-19FUJIAN ZHONGXIN NET SAFETY INFORMATION TECHNOLOGY CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411204409.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-19
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

The existing network security level assessment mainly relies on manual review, which is inefficient and difficult to guarantee the accuracy and consistency of the evaluation results, resulting in difficulty in improving the efficiency and accuracy of network security level protection.

Method used

Using natural language processing technology based on deep learning, semantic understanding of the basic requirements for network security level protection and network device unit evaluation evidence, semantic feature matching is performed through fine-grained point-by-point one-way attention interaction matching model, and intelligently judge network device units The level of network security protection.

Benefits of technology

The automated and intelligent security level protection evaluation of network equipment has been realized, which reduces the workload and subjectivity of manual evaluation, and improves the efficiency and accuracy of evaluation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118709925B_ABST
    Figure CN118709925B_ABST
Patent Text Reader

Abstract

The present application relates to the field of network security technology, and specifically discloses a network security level protection management system and method based on artificial intelligence, which uses natural language processing technology based on deep learning to semantically understand the basic requirements of network security level protection and network equipment unit evaluation item evidence, and mines out the normative knowledge and evaluation standard information related to network security level protection, as well as the evaluation information related to network equipment security, and intelligently judges the network security protection level of the network equipment unit based on the semantic feature matching analysis results of the network equipment unit evaluation item evidence relative to the basic requirements of network security level protection. In this way, the automated and intelligent security level protection evaluation of network equipment is realized, which effectively reduces the workload and subjectivity of manual evaluation and improves the efficiency and accuracy of evaluation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of network security technology, and more specifically, to an artificial intelligence-based network security level protection management system and method. Background Art

[0002] With the rapid development of information technology, the Internet has become an indispensable infrastructure in modern society, playing an important role in promoting economic growth, improving public service efficiency, and enriching people's lives.

[0003] The cybersecurity graded protection system is a basic system for information security protection. It improves the overall security level of information systems by providing graded protection for information systems. According to the Cybersecurity Law and the Information Security Graded Protection Management Measures and other relevant regulations, each unit must carry out graded protection assessments for information systems involving important confidential information, social stability, economic order and public interests in accordance with the graded protection implementation guidelines and relevant standards to ensure that the information systems have the corresponding security protection capabilities.

[0004] However, the current network security level assessment mainly relies on manual review, testing and interviews. However, this assessment method is inefficient, and due to the large differences in the technical level and professional experience of the assessors, the accuracy and consistency of the assessment results are difficult to fully guarantee. Therefore, in order to improve the efficiency and accuracy of network security level protection, a network security level protection management system and method based on artificial intelligence is expected. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiments of this application provide a network security level protection management system and method based on artificial intelligence.

[0006] Accordingly, according to one aspect of the present application, there is provided a network security level protection management system based on artificial intelligence, which includes:

[0007] The network security information acquisition module is used to obtain a set of basic requirements for network security level protection and a set of evidence for network equipment unit evaluation items;

[0008] A network security information semantic coding module, used to semantically code the set of network equipment unit evaluation item evidence and the set of network security level protection basic requirements respectively to obtain a set of network equipment unit evaluation item evidence semantic coding feature vectors and a set of network security level protection basic requirements semantic coding feature vectors;

[0009] A one-way matching analysis module, used to calculate the one-way attention matching features of the set of semantic coding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic coding feature vectors of the network security level protection basic requirements to obtain a global matching result aggregation representation vector;

[0010] The network security protection level analysis module is used to determine the network security protection level label based on the aggregation representation vector of the global matching result.

[0011] In the above-mentioned artificial intelligence-based network security level protection management system, the network security information semantic coding module includes: an evaluation item evidence semantic coding unit, which is used to pass each network equipment unit evaluation item evidence in the set of network equipment unit evaluation item evidence through a semantic encoder based on the Bert model to obtain a set of semantic coding feature vectors of the network equipment unit evaluation item evidence; a basic requirement semantic coding unit, which is used to use the semantic encoder based on the Bert model to semantically encode each network security level protection basic requirement in the set of network security level protection basic requirements to obtain a set of semantic coding feature vectors of the network security level protection basic requirements.

[0012] In the above-mentioned artificial intelligence-based network security level protection management system, the one-way matching analysis module includes: a point-by-point one-way matching unit, which is used to use each network equipment unit evaluation item evidence semantic coding feature vector in the set of the network equipment unit evaluation item evidence semantic coding feature vector as a query feature vector, and input the query feature vector and the set of the network security level protection basic requirements semantic coding feature vector into a fine-grained point-by-point one-way attention interactive matching model to obtain a set of one-way matching result representation vectors; a global aggregation unit, which is used to cascade the set of the one-way matching result representation vectors to obtain the global matching result aggregation representation vector.

[0013] In the above-mentioned artificial intelligence-based network security level protection management system, the point-by-point one-way matching unit includes: a semantic measurement subunit, which is used to calculate the semantic measurement coefficient between the query feature vector and each network security level protection basic requirement semantic coding feature vector in the set of network security level protection basic requirement semantic coding feature vectors to obtain a set of one-way matching semantic measurement coefficients; a normalization subunit, which is used to normalize the set of one-way matching semantic measurement coefficients through a softmax function to obtain a set of one-way matching semantic measurement weights; an attention weighted subunit, which is used to calculate the weighted sum of the set of network security level protection basic requirement semantic coding feature vectors using the set of one-way matching semantic measurement weights as the weight to obtain a one-way matching interactive reshaping feature vector; a feature difference calculation subunit, which is used to calculate the differential feature vector between the query feature vector and the one-way matching interactive reshaping feature vector as the one-way matching result representation vector.

[0014] In the above-mentioned artificial intelligence-based network security level protection management system, the semantic measurement subunit is used to: calculate the square of each eigenvalue of each semantic coding feature vector of the network security level protection basic requirements in the set of semantic coding feature vectors of the network security level protection basic requirements to obtain a set of non-linear transformation semantic coding feature vectors of the network security level protection basic requirements; cascade the query feature vector with each non-linear transformation semantic coding feature vector of the network security level protection basic requirements in the set of non-linear transformation semantic coding feature vectors of the network security level protection basic requirements to obtain a set of semantic association feature vectors; use a predetermined weight vector to multiply each semantic association feature vector in the set of semantic association feature vectors to obtain a set of semantic similarities; add a predetermined bias parameter to each semantic similarity in the set of semantic similarities and then pass the sigmoid activation function to obtain the set of one-way matching semantic measurement coefficients.

[0015] In the above-mentioned artificial intelligence-based network security level protection management system, the network security protection level analysis module is used to: input the global matching result aggregation representation vector into the classifier-based security level protection evaluation result generator to obtain the evaluation result, and the evaluation result is used to represent the network security protection level label.

[0016] In the above-mentioned artificial intelligence-based network security level protection management system, it also includes a training module for training the Bert model-based semantic encoder, the fine-grained point-by-point one-way attention interaction matching model and the classifier-based security level protection evaluation result generator.

[0017] In the above-mentioned artificial intelligence-based network security level protection management system, the training module includes: a training data acquisition unit, used to acquire training data, the training data including a set of training network security level protection basic requirements, a set of training network equipment unit evaluation item evidence, and the true value of the network security protection level label; a training data evaluation item evidence semantic encoding unit, used to pass each training network equipment unit evaluation item evidence in the set of training network equipment unit evaluation item evidence through the semantic encoder based on the Bert model to obtain a set of semantic encoding feature vectors of training network equipment unit evaluation item evidence; a training data basic requirement semantic encoding unit, used to use the semantic encoder based on the Bert model to semantically encode each training network security level protection basic requirement in the set of training network security level protection basic requirements to obtain a set of semantic encoding feature vectors of training network security level protection basic requirements; a training data one-way matching unit, used to use the training network equipment unit evaluation item evidence to obtain a set of semantic encoding feature vectors of training network security level protection basic requirements. Each training network device unit evaluation item evidence semantic encoding feature vector in the set of item evidence semantic encoding feature vectors is used as a training query feature vector, and the training query feature vector and the set of training network security level protection basic requirements semantic encoding feature vectors are input into the fine-grained point-by-point one-way attention interactive matching model to obtain a set of training one-way matching result representation vectors; a training data matching result global aggregation unit is used to cascade the set of training one-way matching result representation vectors to obtain a training global matching result aggregation representation vector; a classification loss unit is used to input the training global matching result aggregation representation vector into the classifier-based security level protection evaluation result generator to obtain a classification loss function value; a model training unit is used to train the gated response-based feature vector dynamic interaction fusion module and the classifier-based flushing status monitor with the classification loss function value, wherein in each round of iteration of the training, the training global matching result aggregation representation vector is iteratively optimized.

[0018] According to another aspect of the present application, there is provided a network security level protection management method based on artificial intelligence, which includes:

[0019] Obtain a collection of basic requirements for cybersecurity level protection and a collection of evidence for network equipment unit evaluation items;

[0020] Semantically encoding the set of network equipment unit evaluation item evidence and the set of network security level protection basic requirements respectively to obtain a set of network equipment unit evaluation item evidence semantic encoding feature vectors and a set of network security level protection basic requirements semantic encoding feature vectors;

[0021] Calculate the one-way attention matching features of the set of semantic coding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic coding feature vectors of the network security level protection basic requirements to obtain a global matching result aggregate representation vector;

[0022] Determine a network security protection level label based on the global matching result aggregation representation vector;

[0023] Among them, calculating the one-way attention matching features of the set of semantic coding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic coding feature vectors of the network security level protection basic requirements to obtain a global matching result aggregate representation vector includes:

[0024] Each of the network device unit evaluation item evidence semantic coding feature vectors in the set of the network device unit evaluation item evidence semantic coding feature vectors is used as a query feature vector, and the query feature vector and the set of the network security level protection basic requirements semantic coding feature vectors are input into a fine-grained point-by-point one-way attention interactive matching model to obtain a set of one-way matching result representation vectors;

[0025] The sets of the one-way matching result representation vectors are cascaded to obtain the global matching result aggregate representation vector.

[0026] Compared with the prior art, the network security level protection management system and method based on artificial intelligence provided by this application uses natural language processing technology based on deep learning to semantically understand the basic requirements of network security level protection and network equipment unit evaluation item evidence, dig out the normative knowledge and evaluation standard information related to network security level protection, and the evaluation information related to network equipment security, and intelligently judge the network security protection level of the network equipment unit based on the semantic feature matching analysis results of the network equipment unit evaluation item evidence relative to the basic requirements of network security level protection. In this way, the automated and intelligent security level protection evaluation of network equipment is realized, which effectively reduces the workload and subjectivity of manual evaluation and improves the efficiency and accuracy of evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0028] Figure 1 It is a block diagram of an artificial intelligence-based network security level protection management system according to an embodiment of the present application.

[0029] Figure 2 This is a schematic diagram of the architecture of an artificial intelligence-based network security level protection management system according to an embodiment of the present application.

[0030] Figure 3 It is a block diagram of a network security information semantic encoding module in an artificial intelligence-based network security level protection management system according to an embodiment of the present application.

[0031] Figure 4 It is a block diagram of a one-way matching analysis module in an artificial intelligence-based network security level protection management system according to an embodiment of the present application.

[0032] Figure 5 It is a block diagram of a point-by-point one-way matching unit in an artificial intelligence-based network security level protection management system according to an embodiment of the present application.

[0033] Figure 6 It is a block diagram of a training module in an artificial intelligence-based network security level protection management system according to an embodiment of the present application.

[0034] Figure 7 This is a flowchart of an artificial intelligence-based network security level protection management method according to an embodiment of the present application. DETAILED DESCRIPTION

[0035] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above and other purposes, features and advantages of the present application will become more apparent. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0036] As mentioned in the above background technology, the current network security level assessment mainly relies on manual review, testing and interviews. However, this assessment method is inefficient, and due to the large differences in the technical level and professional experience of the assessors, the accuracy and consistency of the assessment results are difficult to be fully guaranteed. In response to the above technical problems, the technical concept of this application is to use natural language processing technology based on deep learning to semantically understand the basic requirements of network security level protection and network equipment unit evaluation item evidence, to mine the normative knowledge and evaluation standard information related to network security level protection, as well as the evaluation information related to network equipment security, and based on the semantic feature matching analysis results of the network equipment unit evaluation item evidence relative to the basic requirements of network security level protection, intelligently judge the network security protection level of the network equipment unit. In this way, the automated and intelligent security level protection assessment of network equipment is realized, which effectively reduces the workload and subjectivity of manual assessment and improves the efficiency and accuracy of the assessment.

[0037] Figure 1 It is a block diagram of an artificial intelligence-based network security level protection management system according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of a network security level protection management system based on artificial intelligence according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the artificial intelligence-based network security level protection management system 100 includes: a network security information acquisition module 110, which is used to obtain a set of basic requirements for network security level protection, and a set of network equipment unit evaluation item evidence; a network security information semantic encoding module 120, which is used to semantically encode the set of network equipment unit evaluation item evidence and the set of basic requirements for network security level protection respectively to obtain a set of semantic encoding feature vectors of network equipment unit evaluation item evidence and a set of semantic encoding feature vectors of basic requirements for network security level protection; a one-way matching analysis module 130, which is used to calculate the one-way attention matching features of the set of semantic encoding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic encoding feature vectors of the basic requirements for network security level protection to obtain a global matching result aggregate representation vector; a network security protection level analysis module 140, which is used to determine the network security protection level label based on the global matching result aggregate representation vector.

[0038] In the above-mentioned artificial intelligence-based network security level protection management system 100, the network security information acquisition module 110 is used to obtain a set of basic requirements for network security level protection and a set of network equipment unit evaluation item evidence. It should be understood that the basic requirements for network security level protection are the benchmark for network security management and evaluation. They define the security requirements and control measures that should be met under different security levels. They are usually formulated by relevant industry standard organizations and provide clear standards for the security level protection of network equipment. The network equipment unit evaluation item evidence is a specific manifestation of the actual network equipment security, including relevant evidence such as the security status, configuration information, and security events of the network equipment in actual operation. By obtaining the evaluation item evidence of the network equipment unit, it can be compared and analyzed with the basic requirements for network security level protection to evaluate whether the network equipment meets the corresponding security level requirements, thereby providing targeted security reinforcement and risk control for the security protection of the network equipment.

[0039] In the above-mentioned network security level protection management system 100 based on artificial intelligence, the network security information semantic coding module 120 is used to semantically code the set of network equipment unit evaluation item evidence and the set of network security level protection basic requirements to obtain a set of network equipment unit evaluation item evidence semantic coding feature vectors and a set of network security level protection basic requirements semantic coding feature vectors. Among them, Figure 3 FIG. 1 is a block diagram of a network security information semantic encoding module in a network security level protection management system based on artificial intelligence according to an embodiment of the present application. Figure 3 As shown, the network security information semantic encoding module 120 includes: an evaluation item evidence semantic encoding unit 121, which is used to pass each network equipment unit evaluation item evidence in the set of the network equipment unit evaluation item evidence through a semantic encoder based on the Bert model to obtain a set of semantic encoding feature vectors of the network equipment unit evaluation item evidence; a basic requirement semantic encoding unit 122, which is used to use the semantic encoder based on the Bert model to semantically encode each network security level protection basic requirement in the set of network security level protection basic requirements to obtain a set of semantic encoding feature vectors of the network security level protection basic requirements.

[0040] Specifically, the evaluation item evidence semantic encoding unit 121 is used to pass each network device unit evaluation item evidence in the set of network device unit evaluation item evidence through a semantic encoder based on the Bert model to obtain a set of semantic encoding feature vectors of the network device unit evaluation item evidence. It should be understood that considering that the network device unit evaluation item evidence usually contains a large amount of natural language text, there are problems such as diverse expressions and complex structures, which makes it difficult to directly apply traditional text processing methods (such as keyword matching, TF-IDF, etc.). Therefore, in order to accurately understand and express the semantic information in each evaluation item evidence, in the technical solution of the present application, a semantic encoder based on the Bert model is further used to perform deep semantic understanding and feature representation on each network device unit evaluation item evidence. Specifically, the Bert model is a bidirectional encoder based on Transformer. By pre-training on a large amount of unlabeled text, it has a strong semantic understanding ability, can fully capture the rich language knowledge and contextual semantic information in each evaluation item evidence, accurately understand the security meaning behind it, and convert it into a feature vector form that can be understood and processed by a computer to obtain a set of semantic encoding feature vectors of network device unit evaluation item evidence.

[0041] Specifically, the basic requirement semantic encoding unit 122 is used to use the semantic encoder based on the Bert model to semantically encode each basic requirement of the network security level protection in the set of basic requirements for network security level protection to obtain a set of semantic encoding feature vectors of the basic requirements for network security level protection. Similarly, the basic requirements for network security level protection are also usually text descriptions of a series of security standards and measures, which may have different expressions, wording habits and sentence structures. In the technical solution of the present application, the semantic encoder based on the Bert model is also used to semantically encode each basic requirement for network security level protection, so as to utilize the powerful semantic understanding ability of the Bert model to deeply explore the deep meaning and logical relationship behind each basic requirement for network security level protection, and convert the diverse text descriptions into a unified and standardized feature vector form, thereby obtaining a set of semantic encoding feature vectors of the basic requirements for network security level protection, providing a standardized data basis for subsequent matching analysis.

[0042] In the above-mentioned network security level protection management system 100 based on artificial intelligence, the one-way matching analysis module 130 is used to calculate the one-way attention matching features of the set of semantic encoding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic encoding feature vectors of the network security level protection basic requirements to obtain a global matching result aggregation representation vector. Among them, Figure 4 FIG. 1 is a block diagram of a one-way matching analysis module in a network security level protection management system based on artificial intelligence according to an embodiment of the present application. Figure 4 As shown, the one-way matching analysis module 130 includes: a point-by-point one-way matching unit 131, which is used to use each network device unit evaluation item evidence semantic encoding feature vector in the set of the network device unit evaluation item evidence semantic encoding feature vector as a query feature vector, and input the query feature vector and the set of the network security level protection basic requirements semantic encoding feature vectors into a fine-grained point-by-point one-way attention interactive matching model to obtain a set of one-way matching result representation vectors; a global aggregation unit 132, which is used to cascade the set of the one-way matching result representation vectors to obtain the global matching result aggregation representation vector.

[0043] Specifically, the point-by-point one-way matching unit 131 is used to use each semantic encoding feature vector of the network device unit evaluation item evidence in the set of semantic encoding feature vectors of the network device unit evaluation item evidence as a query feature vector, and input the query feature vector and the set of semantic encoding feature vectors of the basic requirements for network security level protection into a fine-grained point-by-point one-way attention interactive matching model to obtain a set of one-way matching result representation vectors. It should be understood that since the semantic association between the network device unit evaluation item evidence and the basic requirements for network security level protection is relatively complex, it is difficult to accurately measure the degree of matching and correlation between the two by simple text matching or distance calculation between feature vectors. Therefore, in the technical solution of the present application, a fine-grained point-by-point one-way attention interactive matching model is further introduced to achieve accurate mining and representation of the deep semantic relationship between the network device unit evaluation item evidence and the basic requirements for network security level protection. Specifically, the fine-grained point-by-point one-way attention interaction matching model is based on the attention mechanism, which generates attention weights by calculating the semantic association of the semantic encoding feature vector of the network equipment unit evaluation item evidence relative to the set of semantic encoding feature vectors of the network security level protection basic requirements, and performs attention weighted fusion on the set of semantic encoding feature vectors of the network security level protection basic requirements to comprehensively represent the part of all network security level protection basic requirements information related to the evaluation item evidence. Furthermore, by calculating the feature difference between the semantic encoding feature vector of the network equipment unit evaluation item evidence and the feature vector after attention fusion, the mismatch or difference between the evaluation item evidence and the network security level protection basic requirements at the semantic level is further revealed, thereby obtaining the one-way matching result representation vector between each network equipment unit evaluation item evidence and the network security level protection basic requirements, effectively capturing the subtle semantic differences and associations between each evaluation item evidence and the network security level protection basic requirements, and providing accurate and detailed judgment basis for subsequent network security level protection level judgment.

[0044] Figure 5 FIG. 1 is a block diagram of a point-by-point one-way matching unit in an artificial intelligence-based network security level protection management system according to an embodiment of the present application. Figure 5As shown, the point-by-point one-way matching unit 131 includes: a semantic measurement subunit 1311, which is used to calculate the semantic measurement coefficient between the query feature vector and each semantic coding feature vector of the basic requirements for network security level protection in the set of semantic coding feature vectors of the basic requirements for network security level protection to obtain a set of one-way matching semantic measurement coefficients; a normalization subunit 1312, which is used to normalize the set of one-way matching semantic measurement coefficients through a softmax function to obtain a set of one-way matching semantic measurement weights; an attention weighting subunit 1313, which is used to calculate the weighted sum of the set of semantic coding feature vectors of the basic requirements for network security level protection using the set of one-way matching semantic measurement weights as weights to obtain a one-way matching interactive reshaping feature vector; a feature difference calculation subunit 1314, which is used to calculate the differential feature vector between the query feature vector and the one-way matching interactive reshaping feature vector as the one-way matching result representation vector.

[0045] Specifically, the semantic measurement subunit 1311 is used to: calculate the square of each eigenvalue of each semantic coding feature vector of the basic requirements for network security level protection in the set of semantic coding feature vectors of the basic requirements for network security level protection to obtain a set of non-linear transformation semantic coding feature vectors of the basic requirements for network security level protection; cascade the query feature vector with each non-linear transformation semantic coding feature vector of the basic requirements for network security level protection in the set of non-linear transformation semantic coding feature vectors of the basic requirements for network security level protection to obtain a set of semantic association feature vectors; use a predetermined weight vector to multiply each semantic association feature vector in the set of semantic association feature vectors to obtain a set of semantic similarities; add a predetermined bias parameter to each semantic similarity in the set of semantic similarities and then pass the sigmoid activation function to obtain the set of one-way matching semantic measurement coefficients.

[0046] That is, the point-by-point one-way matching unit 131 is used to process the query feature vector and the set of semantic coding feature vectors of the basic requirements for network security level protection using the following one-way interactive matching formula to obtain the one-way matching result representation vector, wherein the one-way interactive matching formula is:

[0047]

[0048] in, represents the query feature vector, The first one in the set of semantic coding feature vectors of the basic requirements for network security level protection A semantic coding feature vector of basic requirements for network security level protection, Indicates cascade operation, is the predetermined weight vector, is the predetermined bias parameter, express Activation function, Represents the query feature vector and the The semantic similarity between the semantic encoding feature vectors of the basic requirements for network security level protection, is the number of semantically encoded feature vectors of the basic requirements for network security level protection, represents the exponential function operation with e as the base, Represents the query feature vector and the The one-way matching semantic metric coefficient between the semantic encoding feature vectors of the basic requirements for network security level protection is represents the one-way matching interaction reshaping feature vector, It means subtracting by position. Indicates that the one-way matching result represents a vector.

[0049] Specifically, the global aggregation unit 132 is used to cascade the set of one-way matching result representation vectors to obtain the global matching result aggregation representation vector. That is, the one-way matching result representation vectors between the evaluation item evidence of each network device unit and the basic requirements of network security level protection are integrated to form a global assessment of the overall security of the network device unit. In the technical solution of the present application, a cascading method is adopted to splice the one-way matching result representation vectors to reflect the overall performance of the network device unit in multiple security evaluation items, thereby providing a more comprehensive and integrated evaluation perspective.

[0050] In the above-mentioned network security level protection management system 100 based on artificial intelligence, the network security protection level analysis module 140 is used to determine the network security protection level label based on the global matching result aggregation representation vector. In a specific example of the present application, the implementation method of determining the network security protection level label based on the global matching result aggregation representation vector is to input the global matching result aggregation representation vector into the security level protection evaluation result generator based on the classifier to obtain the evaluation result, and the evaluation result is used to represent the network security protection level label. It should be understood that in the technical solution of the present application, the security level protection evaluation result generator is based on the classification algorithm, and after training and optimization, it can automatically learn and identify the characteristic patterns under different security levels, and accordingly map the global matching result aggregation representation vector to the corresponding network security protection level label, make accurate classification decisions, and determine the network security protection level of the network device unit, so that corresponding security policies and measures can be taken according to the evaluation results. Through this automated evaluation method based on machine learning, not only the evaluation efficiency is improved, manual participation is reduced, but also the accuracy and objectivity of the evaluation are improved.

[0051] It should be understood that before using the above neural network model, it needs to be trained. That is to say, in the network security level protection management system based on artificial intelligence of the present application, it also includes a training module for training the semantic encoder based on the Bert model, the fine-grained point-by-point one-way attention interaction matching model and the security level protection evaluation result generator based on the classifier.

[0052] Figure 6 FIG. 1 is a block diagram of a training module in a network security level protection management system based on artificial intelligence according to an embodiment of the present application. Figure 6As shown, the training module 200 includes: a training data acquisition unit 210, which is used to acquire training data, and the training data includes a set of training network security level protection basic requirements, a set of training network equipment unit evaluation item evidence, and a true value of the network security protection level label; a training data evaluation item evidence semantic encoding unit 220, which is used to pass each training network equipment unit evaluation item evidence in the set of training network equipment unit evaluation item evidence through the semantic encoder based on the Bert model to obtain a set of semantic encoding feature vectors of training network equipment unit evaluation item evidence; a training data basic requirement semantic encoding unit 230, which is used to use the semantic encoder based on the Bert model to semantically encode each training network security level protection basic requirement in the set of training network security level protection basic requirements to obtain a set of semantic encoding feature vectors of training network security level protection basic requirements; a training data one-way matching unit 240, which is used to semantically encode the training network equipment unit evaluation item evidence with the semantic encoder based on the Bert model Each training network device unit evaluation item evidence semantic encoding feature vector in the set of feature vectors is used as a training query feature vector, and the training query feature vector and the set of training network security level protection basic requirements semantic encoding feature vectors are input into the fine-grained point-by-point one-way attention interactive matching model to obtain a set of training one-way matching result representation vectors; a training data matching result global aggregation unit 250 is used to cascade the set of training one-way matching result representation vectors to obtain a training global matching result aggregation representation vector; a classification loss unit 260 is used to input the training global matching result aggregation representation vector into the classifier-based security level protection evaluation result generator to obtain a classification loss function value; a model training unit 270 is used to train the gated response-based feature vector dynamic interaction fusion module and the classifier-based flushing status monitor with the classification loss function value, wherein, in each round of iteration of the training, the training global matching result aggregation representation vector is iteratively optimized.

[0053] Here, the set of semantically encoded feature vectors of training network equipment unit evaluation item evidence and the set of semantically encoded feature vectors of training network security level protection basic requirements respectively express the encoded semantic features of the set of training network equipment unit evaluation item evidence and the set of training network security level protection basic requirements. When the set of semantically encoded feature vectors of training network equipment unit evaluation item evidence and the set of semantically encoded feature vectors of training network security level protection basic requirements are matched with each other using fine-grained point-by-point one-way attention interaction, the aggregated representation vector of the training global matching result may also cause the lack of causal correlation of the posterior class probability relative to the source semantic distribution representation due to the difference in fine-grained point-by-point one-way attention interaction weights caused by the feature misalignment of the encoded semantic features based on the source semantic distribution, thereby affecting the accuracy of the classification results.

[0054] Based on this, in a preferred embodiment, when the model is trained with the classification loss function value, in each iteration of the training, the training global matching result aggregate representation vector is optimized, specifically comprising the following steps: obtaining the training global matching result aggregate representation vector input into the security level protection evaluation result generator based on the classifier and obtaining the representation vector respectively; Network security protection level labels correspond to label probability values; The label probability values ​​are subtracted from After dividing by one, calculate the square root of the sum of the squares and add One-half is added to obtain a level probability value; each eigenvalue of the training global matching result aggregate representation vector is probabilistically processed based on the sigmoid probability activation function to obtain a probabilistic global matching result aggregate representation vector; a global matching result aggregate cognitive symbol value is determined based on a comparison between each eigenvalue of the probabilistic global matching result aggregate representation vector and the level probability value, wherein the global matching result aggregate cognitive symbol value is equal to one, zero and negative one respectively in response to the eigenvalue of the probabilistic global matching result aggregate representation vector being greater than, equal to and less than the level probability value; the mean of all eigenvalues ​​of the probabilistic global matching result aggregate representation vector is calculated to obtain a global matching result aggregate integral. body phase shift value; multiply each eigenvalue of the probabilistic global matching result aggregate representation vector by the global matching result aggregate cognitive symbol value and the global matching result aggregate overall phase shift value to obtain a probabilistic global matching result aggregate cognitive feature vector and a probabilistic global matching result aggregate phase shift feature vector; multiply the probabilistic global matching result aggregate cognitive feature vector and the probabilistic global matching result aggregate phase shift feature vector by the first weight and the second weight as hyperparameters respectively, and then calculate their difference to obtain a probabilistic global matching result aggregate differential representation vector; take the absolute value of each eigenvalue of the probabilistic global matching result aggregate differential representation vector to obtain the optimized trained global matching result aggregate representation vector.

[0055] That is, the training global matching result aggregate representation vector is optimized by the following optimization formula to obtain an optimized training global matching result aggregate representation vector, wherein the optimization formula is:

[0056]

[0057] in, is the probabilistic global matching result aggregate representation vector obtained by passing the training global matching result aggregate representation vector through the probability activation function. eigenvalues, Represents the label probability value corresponding to each network security protection level label, is the grade probability value, and denote the first weight and the second weight as hyperparameters, respectively. Aggregate cognitive symbol values ​​for global matching results, is the mean of all eigenvalues ​​in the probabilistic global matching result aggregation representation vector, represents multiplication, Indicates taking the absolute value, is the first vector of the aggregate representation of the global matching result after optimization training feature values.

[0058] Here, in the above-mentioned preferred embodiment, the level cognitive phase transformation response of the training global matching result aggregate representation vector is obtained by comparing the probabilistic amplitude of the eigenvalue of the training global matching result aggregate representation vector with the level probability representation, and the phase shift response of the eigenvalue of the training global matching result aggregate representation vector relative to the level probability representation of the feature as a whole is subjected to a feature distribution sequence invariance transformation based on the level probability differential distribution expansion, so as to realize the causal constraint of the posterior class probability of the training global matching result aggregate representation vector on its prior semantic feature distribution representation, so as to improve the accuracy of the evaluation result obtained by inputting the training global matching result aggregate representation vector into the security level protection evaluation result generator based on the classifier.

[0059] In summary, the network security level protection management system based on artificial intelligence according to the embodiment of the present application is explained, which uses natural language processing technology based on deep learning to semantically understand the basic requirements of network security level protection and network equipment unit evaluation item evidence, and mines out the normative knowledge and evaluation standard information related to network security level protection, as well as the evaluation information related to network equipment security, and intelligently judges the network security protection level of the network equipment unit based on the semantic feature matching analysis results of the network equipment unit evaluation item evidence relative to the basic requirements of network security level protection. In this way, the automated and intelligent security level protection evaluation of network equipment is realized, which effectively reduces the workload and subjectivity of manual evaluation and improves the efficiency and accuracy of evaluation.

[0060] Figure 7 Flowchart of the network security level protection management method based on artificial intelligence according to the embodiment of the present application. Figure 7 As shown, according to the embodiment of the present application, the network security level protection management method based on artificial intelligence includes the steps of: S1, obtaining a set of basic requirements for network security level protection, and a set of evidence of network equipment unit evaluation items; S2, respectively performing semantic encoding on the set of network equipment unit evaluation item evidence and the set of basic requirements for network security level protection to obtain a set of semantic encoding feature vectors of network equipment unit evaluation item evidence and a set of semantic encoding feature vectors of basic requirements for network security level protection; S3, calculating the unidirectional attention matching features of the set of semantic encoding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic encoding feature vectors of the basic requirements for network security level protection to obtain a global matching result aggregate representation vector; S4, determining the network security protection level label based on the global matching result aggregate representation vector.

[0061] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned network security level protection management method based on artificial intelligence have been referred to above. Figures 1 to 6The invention has been introduced in detail in the description of the artificial intelligence-based network security level protection management system, and therefore, its repeated description will be omitted.

[0062] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0063] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical sub-unit, that is, it may be located in one place, or it may be distributed on multiple network sub-units. Some or all of the units can be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0064] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0065] In addition, it is obvious that the word "comprising" does not exclude other sub-units or steps, and the singular does not exclude the plural. Multiple sub-units stated in the system claim can also be implemented by one sub-unit through software or hardware.

[0066] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A network security level protection management system based on artificial intelligence, characterized in that: include: The network security information acquisition module is used to obtain a set of basic requirements for network security level protection and a set of evidence for network equipment unit evaluation items; A network security information semantic coding module, used to semantically code the set of network equipment unit evaluation item evidence and the set of network security level protection basic requirements respectively to obtain a set of network equipment unit evaluation item evidence semantic coding feature vectors and a set of network security level protection basic requirements semantic coding feature vectors; A one-way matching analysis module, used to calculate the one-way attention matching features of the set of semantic coding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic coding feature vectors of the network security level protection basic requirements to obtain a global matching result aggregation representation vector; A network security protection level analysis module, used to determine a network security protection level label based on the global matching result aggregation representation vector; Wherein, the one-way matching analysis module includes: A point-by-point one-way matching unit, used to use each of the network device unit evaluation item evidence semantic coding feature vectors in the set of the network device unit evaluation item evidence semantic coding feature vectors as a query feature vector, and input the query feature vector and the set of the network security level protection basic requirements semantic coding feature vectors into a fine-grained point-by-point one-way attention interactive matching model to obtain a set of one-way matching result representation vectors; A global aggregation unit, used for cascading the set of one-way matching result representation vectors to obtain the global matching result aggregation representation vector; The network security information semantic encoding module includes: An evaluation item evidence semantic encoding unit, used for respectively passing each network device unit evaluation item evidence in the set of network device unit evaluation item evidence through a semantic encoder based on a Bert model to obtain a set of semantic encoding feature vectors of the network device unit evaluation item evidence; A basic requirement semantic encoding unit, used to use the semantic encoder based on the Bert model to semantically encode each network security level protection basic requirement in the set of network security level protection basic requirements to obtain a set of semantic encoding feature vectors of the network security level protection basic requirements; Wherein, the point-by-point one-way matching unit comprises: A semantic metric subunit, used for calculating the semantic metric coefficient between the query feature vector and each semantic metric vector of the network security level protection basic requirements in the set of semantic coded feature vectors of the network security level protection basic requirements to obtain a set of one-way matching semantic metric coefficients; A normalization subunit, used for normalizing the set of one-way matching semantic metric coefficients through a softmax function to obtain a set of one-way matching semantic metric weights; An attention weighting subunit, used for calculating the weighted sum of the set of semantic encoding feature vectors of the basic requirements for network security level protection using the set of one-way matching semantic measurement weights as weights to obtain a one-way matching interactive reshaping feature vector; The feature difference calculation subunit is used to calculate the difference feature vector between the query feature vector and the one-way matching interactive reshaping feature vector as the one-way matching result representation vector.

2. The network security level protection management system based on artificial intelligence according to claim 1 is characterized in that: The semantic metric subunit is used to: Calculate the square of each eigenvalue of each semantic coding feature vector of the basic requirements for cybersecurity level protection in the set of semantic coding feature vectors of the basic requirements for cybersecurity level protection to obtain a set of semantic coding feature vectors of the basic requirements for cybersecurity level protection; Cascading the query feature vector with each nonlinear transformation network security level protection basic requirements semantic coding feature vector in the set of nonlinear transformation network security level protection basic requirements semantic coding feature vectors to obtain a set of semantically associated feature vectors; Multiplying each semantically associated feature vector in the set of semantically associated feature vectors by a predetermined weight vector to obtain a set of semantic similarities; Each semantic similarity in the set of semantic similarities is added with a predetermined bias parameter and then passed through a sigmoid activation function to obtain a set of one-way matching semantic measurement coefficients.

3. The network security level protection management system based on artificial intelligence according to claim 2 is characterized in that: The network security protection level analysis module is used to: The global matching result aggregation representation vector is input into a classifier-based security level protection evaluation result generator to obtain an evaluation result, and the evaluation result is used to represent a network security protection level label.

4. The network security level protection management system based on artificial intelligence according to claim 3 is characterized in that: It also includes a training module for training the Bert model-based semantic encoder, the fine-grained point-by-point one-way attention interaction matching model and the classifier-based security level protection evaluation result generator.

5. The network security level protection management system based on artificial intelligence according to claim 4 is characterized in that: The training module comprises: A training data acquisition unit, used to acquire training data, wherein the training data includes a set of basic requirements for training network security level protection, a set of evidence of training network equipment unit evaluation items, and a true value of a network security protection level label; A training data evaluation item evidence semantic encoding unit, used for passing each training network device unit evaluation item evidence in the set of training network device unit evaluation item evidence through the semantic encoder based on the Bert model to obtain a set of training network device unit evaluation item evidence semantic encoding feature vectors; A training data basic requirement semantic encoding unit, used to use the semantic encoder based on the Bert model to semantically encode each training network security level protection basic requirement in the set of training network security level protection basic requirements to obtain a set of training network security level protection basic requirement semantic encoding feature vectors; A training data one-way matching unit, used to use each training network device unit evaluation item evidence semantic coding feature vector in the set of the training network device unit evaluation item evidence semantic coding feature vectors as a training query feature vector, and input the training query feature vector and the set of the training network security level protection basic requirements semantic coding feature vectors into the fine-grained point-by-point one-way attention interactive matching model to obtain a set of training one-way matching result representation vectors; A training data matching result global aggregation unit, used for cascading the set of training unidirectional matching result representation vectors to obtain a training global matching result aggregation representation vector; A classification loss unit, used for inputting the training global matching result aggregation representation vector into the classifier-based security level protection evaluation result generator to obtain a classification loss function value; A model training unit is used to train the gated response-based feature vector dynamic interaction fusion module and the classifier-based flushing state monitor with the classification loss function value, wherein in each round of iteration of the training, the training global matching result aggregate representation vector is iteratively optimized.

6. A network security level protection management method based on artificial intelligence, characterized in that: include: Obtain a collection of basic requirements for cybersecurity level protection and a collection of evidence for network equipment unit evaluation items; Semantically encoding the set of network equipment unit evaluation item evidence and the set of network security level protection basic requirements respectively to obtain a set of network equipment unit evaluation item evidence semantic encoding feature vectors and a set of network security level protection basic requirements semantic encoding feature vectors; Calculate the one-way attention matching features of the set of semantic coding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic coding feature vectors of the network security level protection basic requirements to obtain a global matching result aggregate representation vector; Determine a network security protection level label based on the global matching result aggregation representation vector; Among them, calculating the one-way attention matching features of the set of semantic coding feature vectors of the network equipment unit evaluation item evidence relative to the set of semantic coding feature vectors of the network security level protection basic requirements to obtain a global matching result aggregate representation vector includes: Each of the network device unit evaluation item evidence semantic coding feature vectors in the set of the network device unit evaluation item evidence semantic coding feature vectors is used as a query feature vector, and the query feature vector and the set of the network security level protection basic requirements semantic coding feature vectors are input into a fine-grained point-by-point one-way attention interactive matching model to obtain a set of one-way matching result representation vectors; Cascading the set of one-way matching result representation vectors to obtain the global matching result aggregate representation vector; The set of network equipment unit evaluation item evidence and the set of network security level protection basic requirements are semantically encoded to obtain a set of network equipment unit evaluation item evidence semantic encoding feature vectors and a set of network security level protection basic requirements semantic encoding feature vectors, including: Each piece of network device unit evaluation item evidence in the set of network device unit evaluation item evidence is passed through a semantic encoder based on a Bert model to obtain a set of semantic encoding feature vectors of the network device unit evaluation item evidence; Using the semantic encoder based on the Bert model, semantically encode each of the network security level protection basic requirements in the set of network security level protection basic requirements to obtain a set of semantic encoding feature vectors of the network security level protection basic requirements; Among them, each network device unit evaluation item evidence semantic coding feature vector in the set of the network device unit evaluation item evidence semantic coding feature vector is used as a query feature vector, and the query feature vector and the set of the network security level protection basic requirements semantic coding feature vector are input into the fine-grained point-by-point one-way attention interactive matching model to obtain a set of one-way matching result representation vectors, including: Calculating semantic metric coefficients between the query feature vector and each semantic coded feature vector of the network security level protection basic requirements in the set of semantic coded feature vectors of the network security level protection basic requirements to obtain a set of one-way matching semantic metric coefficients; Normalizing the set of one-way matching semantic metric coefficients through a softmax function to obtain a set of one-way matching semantic metric weights; The weighted sum of the set of semantic encoding feature vectors of the basic requirements for network security level protection is calculated using the set of one-way matching semantic measurement weights as weights to obtain a one-way matching interactive reshaping feature vector; A differential feature vector between the query feature vector and the one-way matching interactive reshaping feature vector is calculated as the one-way matching result representation vector.

Citation Information

Patent Citations

  • Public information security monitoring system and method thereof

    CN116866054A

  • Project data security information grading processing method

    CN117236334A

  • Data security monitoring method based on block chain

    CN117494228A

  • Computer network information security protection system and method

    CN117560201A

  • Network security level protection evaluation method, system and device based on artificial intelligence

    CN117768220A