Vulnerability analysis method and device, equipment and storage medium

By acquiring and preprocessing vulnerability report data, a vulnerability analysis model for iterative training is built, which solves the problem of difficult to identify new vulnerabilities in the existing technology, realizes the ability to quickly identify new vulnerabilities, and improves the network security protection effect.

CN118944960APending Publication Date: 2024-11-12E SURFING IOT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411212175.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

Existing vulnerability scanning tools are poor in identifying new vulnerabilities, making it difficult to quickly adapt to new vulnerabilities, making it difficult to identify network security threats.

Method used

By obtaining vulnerability report data output by multiple vulnerability scanning tools, performing feature extraction and data preprocessing, building a trained vulnerability analysis model, and using iterative training technology to make the model self-learning ability and quickly identify new vulnerabilities.

Benefits of technology

It improves the accuracy and efficiency of vulnerability scanning, can quickly adapt to new vulnerabilities, reduces the difficulty of identifying network security threats, and avoids the defects that are difficult to identify new vulnerabilities in the existing technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118944960A_ABST
    Figure CN118944960A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of network security, in particular to a vulnerability analysis method, device and equipment and a storage medium. Vulnerability report data output by multiple vulnerability scanning tools are obtained, feature extraction is conducted on the vulnerability report data, and therefore key element features of vulnerabilities are extracted; the historical vulnerability analysis model is iteratively trained based on the key element features of the vulnerabilities, so that the model can be iteratively trained according to the element features of the new vulnerabilities when facing the new vulnerabilities, the ability of self-learning vulnerability scanning is obtained, the vulnerability scanning effect is improved, and the problem that in the prior art, novel vulnerabilities are difficult to recognize is solved. And network security threats are formed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of network security technology, and in particular to a vulnerability analysis method, device, equipment and storage medium. Background Art

[0002] With the development of network technology, the problem of network security protection has become increasingly prominent, especially the scanning direction for network security vulnerabilities, which has become the top priority of enterprise security protection. Existing vulnerability scanning is generally performed by scanning several vulnerability scanning tools, but the scanning effect of existing vulnerability scanning tools is poor and the accuracy of vulnerability identification is low. Therefore, in order to better perform vulnerability scanning, technical personnel in this field perform model training based on a large amount of data to improve the accuracy of vulnerability scanning, but this method is difficult to achieve good results when facing new vulnerabilities.

[0003] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0004] The main purpose of the present invention is to provide a vulnerability analysis method, device, equipment and storage medium, aiming to solve the technical problem that it is difficult to identify new vulnerabilities in the prior art and form network security threats.

[0005] To achieve the above object, the present invention provides a vulnerability analysis method, which comprises the following steps:

[0006] Obtain vulnerability report data output by multiple vulnerability scanning tools;

[0007] Perform feature extraction on each vulnerability report data to obtain a meta-feature dataset;

[0008] Input the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model;

[0009] Vulnerability analysis is performed using the trained vulnerability analysis model.

[0010] In an optional embodiment, the feature extraction is performed on each vulnerability report data to obtain a meta-feature data set, including:

[0011] Preprocess each vulnerability report data to obtain target vulnerability report data;

[0012] Feature extraction is performed on the target vulnerability report data to obtain meta-feature data sets of multiple dimensions.

[0013] In an optional embodiment, the preprocessing of each vulnerability report data includes at least one of the following steps:

[0014] Performing data cleaning on the vulnerability report data;

[0015] Tabulating the vulnerability report data based on vulnerability attribute information of the vulnerability report data;

[0016] Determine the vulnerability scanning tool corresponding to the vulnerability report data, and classify the vulnerability scanning tool according to the vulnerability scanning tool.

[0017] In an optional embodiment, the target vulnerability data includes at least: a text attribute, a numerical attribute, and a classification attribute;

[0018] The feature extraction of the vulnerability report data includes:

[0019] Extracting keyword features of the text attributes;

[0020] Performing numerical preprocessing on the numerical attribute, wherein the numerical preprocessing at least includes numerical normalization and numerical standardization;

[0021] Feature extraction is performed on the categorical attributes through one-hot encoding.

[0022] In an optional embodiment, before inputting the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain the trained vulnerability analysis model, the method further includes:

[0023] Performing vector conversion on the meta-feature data to obtain a feature data vector;

[0024] Determining the category of the meta-feature data and the input channel corresponding to the category;

[0025] The feature data vector is input into the historical vulnerability analysis model through an input channel corresponding to the category of the meta feature data for iterative training to obtain a trained vulnerability analysis model.

[0026] In an optional embodiment, the step of inputting the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model includes:

[0027] Obtain model parameters of the historical vulnerability analysis model;

[0028] Extracting a target training set and a target test set based on the meta-feature data sets of each dimension, wherein the target training set and the target test set contain the same data in at least one dimension;

[0029] Iterative training is performed based on the historical vulnerability analysis model of the target training set and the target test set to update the model parameters and obtain a trained vulnerability analysis model.

[0030] In an optional embodiment, the model parameters include feature extraction parameters and network parameters;

[0031] The iterative training of the historical vulnerability analysis model based on the target training set and the target test set to update the model parameters and obtain a trained vulnerability analysis model includes:

[0032] Based on the target training set, input into the historical vulnerability analysis model to obtain a vulnerability prediction result;

[0033] Determine a loss function based on the vulnerability prediction result and the manual analysis result;

[0034] Globally updating the feature extraction parameters and the network parameters according to the loss function and the target test set;

[0035] When the loss function satisfies a preset convergence condition, determining current feature extraction parameters and current network parameters;

[0036] A trained vulnerability analysis model is generated according to the current feature extraction parameters and the current network parameters.

[0037] In addition, to achieve the above-mentioned purpose, the present invention also proposes a vulnerability analysis device, which includes:

[0038] An acquisition module is used to obtain vulnerability report data output by multiple vulnerability scanning tools;

[0039] An extraction module is used to extract features from each vulnerability report data to obtain a meta-feature data set;

[0040] The training module is used to input the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model;

[0041] The analysis module is used to perform vulnerability analysis using the trained vulnerability analysis model.

[0042] In addition, to achieve the above-mentioned purpose, the present invention also proposes a vulnerability analysis device, which includes: a memory, a processor, and a vulnerability analysis program stored in the memory and executable on the processor, and the vulnerability analysis program is configured to implement the steps of the vulnerability analysis method described above.

[0043] In addition, to achieve the above-mentioned purpose, the present invention also proposes a storage medium, on which a vulnerability analysis program is stored, and when the vulnerability analysis program is executed by a processor, the steps of the vulnerability analysis method described above are implemented.

[0044] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the vulnerability analysis method described above are implemented.

[0045] One or more technical solutions proposed in the present application have at least the following technical effects: The present application discloses a vulnerability analysis method, which includes: obtaining vulnerability report data output by multiple vulnerability scanning tools; performing feature extraction on each vulnerability report data to obtain a meta-feature data set; inputting the meta-feature data sets of each dimension into a historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model; performing vulnerability analysis through the trained vulnerability analysis model. Compared with the prior art, the present application obtains vulnerability report data output by multiple vulnerability scanning tools, and performs feature extraction on each acquired vulnerability report data, thereby extracting key meta-features of vulnerabilities in multiple dimensions, and iteratively training the historical vulnerability analysis model based on the key meta-features of multiple dimensions, so that when the model faces new vulnerabilities, iterative training can be performed according to the meta-features of the new vulnerability report to obtain the ability of self-learning vulnerability scanning. In the absence of a large number of vulnerability reports output by new vulnerability scanning tools, a vulnerability scanning tool that adapts to new vulnerabilities can be quickly obtained, thereby quickly identifying new vulnerabilities and improving the effect of vulnerability scanning, thereby avoiding the technical problem of difficulty in identifying new vulnerabilities and forming network security threats in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0048] Figure 1 This is a flow chart of a first embodiment of a vulnerability analysis method of the present invention;

[0049] Figure 2 A schematic diagram of a vulnerability analysis model architecture of an embodiment of a vulnerability analysis method of the present invention;

[0050] Figure 3 This is a flow chart of a second embodiment of the vulnerability analysis method of the present invention;

[0051] Figure 4 This is a flow chart of a third embodiment of the vulnerability analysis method of the present invention;

[0052] Figure 5 A schematic diagram of a model training iteration process of an embodiment of a vulnerability analysis method of the present invention;

[0053] Figure 6 This is a structural block diagram of the first embodiment of the vulnerability analysis device of the present invention.

[0054] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0055] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0056] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0057] Based on this, the embodiment of the present invention provides a vulnerability analysis method, referring to Figure 1 , Figure 1 The figure is a flow chart of a first embodiment of a vulnerability analysis method of the present invention.

[0058] In this embodiment, the vulnerability analysis method includes the steps of:

[0059] Step S10: Acquire vulnerability report data output by multiple vulnerability scanning tools.

[0060] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device, a terminal device, etc. that can realize the above functions. The following terminal device is taken as an example to illustrate this embodiment and the following embodiments.

[0061] It is understandable that with the rapid development of information technology, network security issues have become increasingly prominent and have become a major challenge faced by enterprises, governments and individuals. Vulnerability scanning, as a key link in network security protection, aims to identify potential vulnerabilities in network systems and take corresponding security measures to prevent them. However, existing vulnerability scanning tools generally face the problem of high false alarm rates when generating scanning reports. These false alarms not only consume a lot of time and energy of security personnel for manual analysis, but may also cover up real security threats and bring potential risks to network security. In addition, with the passage of time, various vulnerability scanning tools emerge in an endless stream, and new vulnerabilities continue to emerge, which brings huge challenges to the work of security personnel.

[0062] Faced with the above challenges, technicians in this field use machine learning technology to improve the accuracy and efficiency of vulnerability scanning. However, traditional machine learning methods usually require a large amount of labeled data to train the model, and when faced with new vulnerabilities and new vulnerability scanning tools, the model often needs to be retrained, which not only increases the training cost of the model, but may also lead to a decrease in the model's detection ability for new vulnerabilities.

[0063] It should be understood that the vulnerability scanning tool in this embodiment can be at least one vulnerability scanning tool, so as to obtain multiple vulnerability report data, improve data diversity, and further improve the effect of model training, wherein the vulnerability report data includes but is not limited to vulnerability type, vulnerability identification results and vulnerability solution methods.

[0064] The vulnerability report data output by multiple vulnerability scanning tools may be obtained from a networked database or from other data storage devices that can provide vulnerability report data of vulnerability scanning tools, and this embodiment does not impose any specific restrictions on this.

[0065] Step S20: extract features from each vulnerability report data to obtain a meta-feature data set.

[0066] It can be understood that after feature extraction of vulnerability report data from various vulnerability scanning tools, the meta-feature data set obtained can be used for subsequent model iteration training. Specifically, the feature extraction process can be to extract features from vulnerability report data from multiple dimensions, such as text attributes, numerical attributes, and classification attributes, etc., to improve the generalization of the meta-feature data set.

[0067] Step S30: Input the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model.

[0068] In this embodiment, the historical vulnerability analysis model refers to a model previously used for vulnerability analysis. Since the training samples of the historical vulnerability analysis model lack identification data of new vulnerabilities, when faced with new vulnerabilities, it is impossible to identify new vulnerabilities and provide targeted solutions, which reduces the reliability of network security.

[0069] Iterative training of the historical vulnerability analysis model may be iterative modification of parameters of the model, such as feature extraction parameters and network parameters, etc., which is not specifically limited in this embodiment.

[0070] Furthermore, before inputting the meta-feature data set into the initial vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model, the method further includes:

[0071] Performing vector conversion on the meta-feature data to obtain a feature data vector;

[0072] Determining the category of the meta-feature data and the input channel corresponding to the category;

[0073] The feature data vector is input into the initial vulnerability analysis model through the input channel corresponding to the category of the meta feature data for iterative training to obtain a trained vulnerability analysis model.

[0074] In the specific implementation, refer to Figure 2 , Figure 2 The figure is a schematic diagram of the architecture of the vulnerability analysis model. By using the output of the meta-feature extraction module as the input of the convolutional neural network, each type of feature can be used as an independent input channel for feature extraction and input into the convolutional neural network with ReLU as the activation function for linear transformation. Dropout is used for regularization to improve the generalization ability of the model and reduce the risk of overfitting. Finally, a fully connected layer is used to learn a more advanced feature representation, and the final output is the result of predicting user preferences.

[0075] Step S40: performing vulnerability analysis using the trained vulnerability analysis model.

[0076] During the training process, the trained vulnerability analysis model performs feature analysis of specific dimensions on the vulnerability report data of the new vulnerability scanning tool. Even if the training sample is small in magnitude, the previously pre-trained historical model has strong analysis capabilities, allowing the model to scan and self-learn, thereby quickly adapting to the new vulnerability scanning tool and realizing rapid identification of new vulnerabilities. At the same time, when faced with the problem that new vulnerabilities are difficult to identify in existing technologies, the trained model will also make its own judgments based on the learning experience in various other vulnerability tools, solve potential new network security threats, and improve the effectiveness of network security protection.

[0077] This embodiment obtains vulnerability report data output by multiple vulnerability scanning tools and performs feature extraction on the vulnerability report data, thereby extracting key meta-features of vulnerabilities in multiple dimensions, and iteratively trains the historical vulnerability analysis model based on the key meta-features of the vulnerabilities in each dimension. When the model faces new vulnerabilities, it can perform iterative training based on the meta-features of the new vulnerabilities, thereby obtaining the ability of self-learning vulnerability scanning, improving the effect of vulnerability scanning, and avoiding the technical problem in the prior art that it is difficult to identify new vulnerabilities and form network security threats.

[0078] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 3 , step S20, comprising:

[0079] Step S201: pre-process each vulnerability report data to obtain target vulnerability report data.

[0080] Step S202: extracting features from the target vulnerability report data to obtain meta-feature data sets of multiple dimensions.

[0081] It should be noted that preprocessing the vulnerability report data can make the data format of the vulnerability report data more consistent with the format of model iterative training, avoiding invalid data that cannot be recognized by the model and affecting the effect of model training.

[0082] Further, the preprocessing of the vulnerability report data comprises at least one of the following steps:

[0083] Performing data cleaning on the vulnerability report data;

[0084] Tabulating the vulnerability report data based on vulnerability attribute information of the vulnerability report data;

[0085] Determine the vulnerability scanning tool corresponding to the vulnerability report data, and classify the vulnerability scanning tool according to the vulnerability scanning tool.

[0086] It is understandable that vulnerability attribute information includes but is not limited to vulnerability number, vulnerability level, vulnerability category, detailed scoring information of the Common Vulnerability Scoring System (CVSS), vulnerability affected components, vulnerability affected versions, repair suggestions, unique identifier (CVE Numbering Authority, CNA), vulnerability discovery time, vulnerability verification method, vulnerability description, risk source code information and other information.

[0087] In the specific implementation, data cleaning can remove invalid data in the vulnerability report data and reduce the invalid training workload of subsequent model training; tabular processing and classification of the vulnerability report data can standardize the vulnerability report data and improve the efficiency of model training. At the same time, when the user needs to check the vulnerability report data, it can also improve the user's viewing experience and make the visualization stronger.

[0088] Furthermore, the feature extraction of the vulnerability report data includes:

[0089] Extracting keyword features of the text attributes;

[0090] Performing numerical preprocessing on the numerical attribute, wherein the numerical preprocessing at least includes numerical normalization and numerical standardization;

[0091] Feature extraction is performed on the categorical attributes through one-hot encoding.

[0092] Specifically, in the feature extraction process, text attributes such as vulnerability description, repair suggestions, vulnerability verification methods and risk source code information in the vulnerability report data are processed, and keyword features are extracted using technologies such as bag-of-words model, TF-IDF, word vector model (such as Word2Vec, GloVe); numerical attributes such as vulnerability number, CVSS detailed scoring information, vulnerability discovery time, etc. can be directly used as features, or normalized and standardized; for classification attributes such as vulnerability level, vulnerability category, vulnerability affected component, vulnerability affected version, CVE number, etc., unique hot encoding is used for feature extraction. Finally, the above features are extracted and connected together as input data for model iterative training, so that the model can be optimized to obtain a trained vulnerability analysis model.

[0093] It should be understood that when facing the problem of identifying new vulnerabilities, since the data level of the relevant vulnerability report data of the new vulnerabilities is relatively small, even after model iterative training, the trained vulnerability analysis model only has the ability to preliminarily identify vulnerabilities, and the recognition effect is not good. In order to improve the recognition efficiency of the trained vulnerability analysis model in identifying new vulnerabilities, in this embodiment, after the model iteration, the recognition data of each vulnerability scanning tool facing the new vulnerability can be counted, and then these recognition data can be used as training sets to perform targeted iterative training on the trained vulnerability analysis model to increase the recognition ability of the trained vulnerability analysis model for new vulnerabilities.

[0094] This embodiment obtains target vulnerability report data by preprocessing the vulnerability report data, and then extracts features from the target vulnerability report data to obtain a meta-feature data set, simplifies the vulnerability report data, and normalizes the vulnerability report data to improve the efficiency of subsequent model iterative training and reduce workload.

[0095] Based on the second embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to the above description, and will not be described in detail later. Figure 4 , step S30, comprising:

[0096] Step S301: Obtain model parameters of an initial vulnerability analysis model.

[0097] Step S302: extracting a target training set and a target test set based on the meta-feature data sets of each dimension, wherein the target training set and the target test set contain the same data of at least one dimension.

[0098] Step S303: performing iterative training on the initial vulnerability analysis model based on the target training set and the target test set to update the model parameters and obtain a trained vulnerability analysis model.

[0099] It should be noted that the model parameters include but are not limited to feature extraction parameters θ1 and network parameters θ2. The meta-feature data set is divided into a target training set and a target test set. The corresponding vulnerability analysis results are then predicted through the target training set. Therefore, the network parameters can be reversely corrected through the predicted vulnerability analysis results and the actual vulnerability analysis results to improve the convergence of the model.

[0100] Furthermore, the iterative training of the initial vulnerability analysis model based on the target training set and the target test set to update the model parameters and obtain a trained vulnerability analysis model includes:

[0101] Based on the target training set, input into the initial vulnerability analysis model to obtain a vulnerability prediction result;

[0102] Determine a loss function based on the vulnerability prediction result and the manual analysis result;

[0103] Globally updating the feature extraction parameters and the network parameters according to the loss function and the target test set;

[0104] When the loss function satisfies a preset convergence condition, determining current feature extraction parameters and current network parameters;

[0105] A trained vulnerability analysis model is generated according to the current feature extraction parameters and the current network parameters.

[0106] Specifically, refer to Figure 5 In this embodiment, the data of a vulnerability scanning tool is selected from the training set as a task, and the data of the task is divided into a target support set and a target query set. The model is trained through the target support set and the results of manual analysis (whether it is a false alarm) for each vulnerability. The model locally updates the network parameter θ2 through back propagation, and can be trained and updated through multiple groups of tasks. This step trains to obtain F(W), and based on the loss function, the feature extraction parameter θ1 and the network parameter θ2 are globally updated through the query set until the model converges. When there are new vulnerabilities or new scanning tools, the model can continue to be trained to enable the model to learn new knowledge and improve its vulnerability analysis capabilities. The fω obtained by this step has good generalization ability. When it is necessary to analyze the vulnerability report of a vulnerability scanning tool, only a small amount of the data of the vulnerability scanning tool is needed to train fω and obtain fθ. In this way, only a specific vulnerability scanning tool needs to be trained to obtain a suitable model. When a new vulnerability appears, the model is directly trained based on fθ to obtain a model fθ' that can solve the new vulnerability. This allows the model to have a better recognition effect when facing new vulnerabilities. As the number of recognitions increases, the corresponding recognition effect will also be better, which greatly reduces the workload of manual recognition by users.

[0107] This embodiment obtains model parameters of an initial vulnerability analysis model; extracts a target training set and a target test set based on meta-feature data sets of each dimension, wherein the target training set and the target test set contain at least one dimension of identical data; iteratively trains the initial vulnerability analysis model based on the target training set and the target test set to update the model parameters to obtain a trained vulnerability analysis model, iteratively updates the vulnerability analysis model based on vulnerability scanning data of at least one vulnerability scanning tool to obtain updated model parameters, and obtains a vulnerability recognition model that has a better recognition effect on new vulnerabilities, thereby improving the reliability of vulnerability scanning.

[0108] This application also provides a vulnerability analysis device, please refer to Figure 6 , the vulnerability analysis device comprises:

[0109] The acquisition module 10 is used to acquire vulnerability report data output by multiple vulnerability scanning tools.

[0110] The extraction module 20 is used to extract features from each vulnerability report data to obtain a meta-feature data set.

[0111] The training module 30 is used to input the meta-feature data set into the initial vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model.

[0112] The analysis module 40 is used to perform vulnerability analysis using the trained vulnerability analysis model.

[0113] This embodiment obtains vulnerability report data output by multiple vulnerability scanning tools, and performs feature extraction on the vulnerability report data, thereby extracting key meta-features of the vulnerabilities, and iteratively trains the historical vulnerability analysis model based on the key meta-features of the vulnerabilities. When the model faces new vulnerabilities, it can perform iterative training according to the meta-features of the new vulnerabilities, thereby obtaining the ability of self-learning vulnerability scanning, improving the effect of vulnerability scanning, and avoiding the technical problem in the prior art that it is difficult to identify new vulnerabilities and form network security threats.

[0114] In one embodiment, the extraction module 20 is further used to preprocess the vulnerability report data to obtain target vulnerability report data; perform feature extraction on the target vulnerability report data to obtain a meta-feature data set of multiple dimensions.

[0115] In one embodiment, the extraction module 20 is also used to perform data cleaning on the vulnerability report data; tabulate the vulnerability report data based on vulnerability attribute information of the vulnerability report data; determine the vulnerability scanning tool corresponding to the vulnerability report data, and classify the vulnerability scanning tool according to the vulnerability scanning tool.

[0116] In one embodiment, the extraction module 20 is further used to extract keyword features of the text attributes; perform numerical preprocessing on the numerical attributes, and the numerical preprocessing includes at least numerical normalization and numerical standardization; and perform feature extraction on the categorical attributes through one-hot encoding.

[0117] In one embodiment, the training module 30 is also used to perform vector conversion on the meta-feature data to obtain a feature data vector; determine the category of the meta-feature data and the input channel corresponding to the category; input the feature data vector into the initial vulnerability analysis model through the input channel corresponding to the category of the meta-feature data for iterative training to obtain a trained vulnerability analysis model.

[0118] In one embodiment, the training module 30 is also used to obtain model parameters of the initial vulnerability analysis model; extract a target training set and a target test set based on the meta-feature data sets of each dimension, and the target training set and the target test set contain at least one dimension of the same data; iteratively train the initial vulnerability analysis model based on the target training set and the target test set to update the model parameters and obtain a trained vulnerability analysis model.

[0119] In one embodiment, the training module 30 is also used to input the target training set into the initial vulnerability analysis model to obtain a vulnerability prediction result; determine a loss function based on the vulnerability prediction result and the manual analysis result; globally update the feature extraction parameters and the network parameters based on the loss function and the target test set; determine the current feature extraction parameters and the current network parameters when the loss function meets the preset convergence conditions; and generate a trained vulnerability analysis model based on the current feature extraction parameters and the current network parameters.

[0120] The present application provides a vulnerability analysis device, which includes: at least one processor; and a memory that is communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the vulnerability analysis method in the above-mentioned embodiment one.

[0121] The vulnerability analysis device provided by the present application adopts the vulnerability analysis method in the above embodiment to solve the technical problem of vulnerability analysis. Compared with the prior art, the beneficial effects of the vulnerability analysis device provided by the present application are the same as the beneficial effects of the vulnerability analysis method provided by the above embodiment, and the other technical features in the vulnerability analysis device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0122] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0123] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0124] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, wherein the computer-readable program instructions are used to execute the vulnerability analysis method in the above-mentioned embodiment.

[0125] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0126] The computer-readable storage medium may be included in the vulnerability analysis device; or may exist independently without being assembled into the vulnerability analysis device.

[0127] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the vulnerability analysis device, the vulnerability analysis device performs vulnerability analysis.

[0128] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0129] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0130] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0131] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned vulnerability analysis method, and can solve the technical problems of vulnerability analysis. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the vulnerability analysis method provided in the above-mentioned embodiment, and will not be elaborated here.

[0132] The present application also provides a computer program product, including a computer program, which implements the steps of the vulnerability analysis method as described above when executed by a processor.

[0133] The computer program product provided by this application can solve the technical problem of vulnerability analysis. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the vulnerability analysis method provided by the above embodiment, which will not be repeated here.

[0134] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A vulnerability analysis method, characterized in that: The vulnerability analysis method comprises: Obtain vulnerability report data output by multiple vulnerability scanning tools; Perform feature extraction on each vulnerability report data to obtain meta-feature datasets in multiple dimensions; Input the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model; Vulnerability analysis is performed using the trained vulnerability analysis model.

2. The method according to claim 1, characterized in that The feature extraction is performed on each vulnerability report data to obtain a meta-feature data set, including: Preprocess each vulnerability report data to obtain target vulnerability report data; Feature extraction is performed on the target vulnerability report data to obtain meta-feature data sets of multiple dimensions.

3. The method according to claim 2, characterized in that The preprocessing of each vulnerability report data includes at least one of the following steps: Performing data cleaning on the vulnerability report data; Tabulating the vulnerability report data based on vulnerability attribute information of the vulnerability report data; Determine the vulnerability scanning tool corresponding to the vulnerability report data, and classify the vulnerability scanning tool according to the vulnerability scanning tool.

4. The method according to claim 2, characterized in that The target vulnerability data at least includes: text attributes, numerical attributes and classification attributes; The feature extraction of the vulnerability report data includes: Extracting keyword features of the text attributes; Performing numerical preprocessing on the numerical attribute, wherein the numerical preprocessing at least includes numerical normalization and numerical standardization; Feature extraction is performed on the categorical attributes through one-hot encoding.

5. The method according to claim 1, characterized in that Before inputting the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain the trained vulnerability analysis model, the method further includes: Performing vector conversion on the meta-feature data to obtain a feature data vector; Determining the category of the meta-feature data and the input channel corresponding to the category; The feature data vector is input into the historical vulnerability analysis model through an input channel corresponding to the category of the meta feature data for iterative training to obtain a trained vulnerability analysis model.

6. The method according to claim 5, characterized in that The meta-feature data sets of each dimension are input into the historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model, including: Obtain model parameters of the historical vulnerability analysis model; Extracting a target training set and a target test set based on the meta-feature data sets of each dimension, wherein the target training set and the target test set contain the same data in at least one dimension; Iterative training is performed based on the historical vulnerability analysis model of the target training set and the target test set to update the model parameters and obtain a trained vulnerability analysis model.

7. The method according to claim 6, characterized in that The model parameters include feature extraction parameters and network parameters; The iterative training of the historical vulnerability analysis model based on the target training set and the target test set to update the model parameters and obtain a trained vulnerability analysis model includes: Based on the target training set, input into the historical vulnerability analysis model to obtain a vulnerability prediction result; Determine a loss function based on the vulnerability prediction result and the manual analysis result; Globally updating the feature extraction parameters and the network parameters according to the loss function and the target test set; When the loss function satisfies a preset convergence condition, determining current feature extraction parameters and current network parameters; A trained vulnerability analysis model is generated according to the current feature extraction parameters and the current network parameters.

8. A vulnerability analysis device, characterized in that: The vulnerability analysis device comprises: An acquisition module is used to obtain vulnerability report data output by multiple vulnerability scanning tools; An extraction module is used to extract features from each vulnerability report data to obtain a meta-feature data set; The training module is used to input the meta-feature data sets of each dimension into the historical vulnerability analysis model for iterative training to obtain a trained vulnerability analysis model; The analysis module is used to perform vulnerability analysis using the trained vulnerability analysis model.

9. A vulnerability analysis device, characterized in that: The vulnerability analysis device comprises: a memory, a processor, and a vulnerability analysis program stored in the memory and executable on the processor, wherein the vulnerability analysis program is configured to implement the steps of the vulnerability analysis method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium stores a vulnerability analysis program, which, when executed by a processor, implements the steps of the vulnerability analysis method according to any one of claims 1 to 7.