Threat intelligence detection method and system, storage medium and electronic equipment
Through the graph neural network, multi-source threat intelligence data is processed, information silos and detection accuracy problems are solved, efficient threat intelligence analysis and real-time detection are achieved, and network security adaptability and defense capabilities are improved.
Patent Information
- Application Number
- CN202411240623.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-29
Smart Images

Figure CN119051955A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of information security, and in particular to a threat intelligence detection method, system, storage medium and electronic device. Background Art
[0002] At present, network security faces the major challenge of how to efficiently process and analyze a large amount of threat intelligence data. These data usually come from multiple sources (such as network traffic, log files, third-party threat intelligence platforms, open source intelligence, etc.), and the data formats are diverse and massive. Traditional methods rely on manual analysis and simple feature matching, and it is difficult to fuse the heterogeneous data they process. That is, the formats of multi-source data are different and difficult to process uniformly, resulting in information islands. In addition, traditional feature engineering is difficult to effectively capture complex network entity relationships, resulting in reduced detection accuracy. In addition, new threats emerge in an endless stream, and model updates and adaptability are poor. Summary of the invention
[0003] The purpose of this application is to provide a threat intelligence detection method, system, storage medium and electronic device to improve the efficiency and accuracy of intelligence analysis by applying a graph training model.
[0004] In order to solve the above technical problems, this application provides a threat intelligence detection method, and the specific technical solution is as follows:
[0005] Obtain intelligence data including threat intelligence;
[0006] The intelligence data is represented as graph structure data; the nodes in the graph structure data represent network entities, and the edges represent the relationships between the network entities;
[0007] Using the graph structure data as training data, training a preset graph neural network to obtain an intelligence feature recognition model;
[0008] Applying the intelligence feature recognition model to extract characteristic attributes of the threat intelligence, and outputting the threat level of each intelligence node corresponding to the threat intelligence;
[0009] The priority of each threat intelligence is determined based on the threat level.
[0010] Optionally, after obtaining intelligence data containing threat intelligence, it also includes:
[0011] Standardizing the format of the threat intelligence;
[0012] The threat intelligence is compared using a text similarity algorithm, and duplicate intelligence exceeding a similarity threshold is deleted.
[0013] Optionally, before comparing the threat intelligence using a text similarity algorithm and deleting duplicate intelligence exceeding a similarity threshold, the method further includes:
[0014] A clustering algorithm is applied to group or merge similar intelligence; wherein the similar intelligence is at least two threat intelligences whose text similarity is less than the similarity threshold and greater than the similarity threshold value.
[0015] Optionally, the using the graph structure data as training data to train a preset graph neural network to obtain an intelligence feature recognition model includes:
[0016] Determine the network structure of the preset graph neural network;
[0017] Determine the nonlinear features in the graph structure data using a nonlinear activation function, and propagate through the network structure hierarchy to expand the reception range of the nonlinear features;
[0018] Aggregating the nonlinear features to generate a comprehensive feature representation; the comprehensive feature representation is used to reflect the reference relationship between nodes and the context information of the relationship network;
[0019] Each node updates its own node status according to the comprehensive feature representation;
[0020] After model training and optimization, the intelligence feature recognition model is obtained.
[0021] Optionally, determining the priority of each threat intelligence based on the threat level includes:
[0022] Determine a characteristic score of the intelligence node corresponding to each threat intelligence;
[0023] The priority of each threat intelligence is determined according to the feature score.
[0024] Optionally, determining the priority of each threat intelligence according to the feature score includes:
[0025] Calculate the global influence of each intelligence node using graph theory algorithms;
[0026] The feature score is used as the local influence and combined with the global influence to determine the priority of each threat intelligence.
[0027] Optionally, after applying the intelligence feature recognition model to extract the characteristic attributes of the threat intelligence and outputting the threat level of each intelligence node corresponding to the threat intelligence, the method further includes:
[0028] The intelligence nodes whose global influence and local influence both meet corresponding thresholds are marked as emergency threat intelligence; the emergency threat intelligence is used as a real-time warning source and intelligence response source.
[0029] The present application also provides a threat intelligence detection system, including:
[0030] A data acquisition module, used to acquire intelligence data including threat intelligence;
[0031] A structure conversion module, used to represent the intelligence data as graph structure data; the nodes in the graph structure data represent network entities, and the edges represent the relationships between the network entities;
[0032] A model training module is used to use the graph structure data as training data to train a preset graph neural network to obtain an intelligence feature recognition model;
[0033] A feature detection module, used to apply the intelligence feature recognition model to extract the feature attributes of the threat intelligence and output the threat level of each intelligence node corresponding to the threat intelligence;
[0034] The intelligence sorting module is used to determine the priority of each threat intelligence based on the threat level.
[0035] The present application also provides a computer-readable storage medium having a computer program stored thereon, and the computer program implements the steps of the above-mentioned method when executed by a processor.
[0036] The present application also provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps of the above-mentioned method when calling the computer program in the memory.
[0037] The present application provides a method for detecting threat intelligence, including: obtaining intelligence data containing threat intelligence; representing the intelligence data as graph structure data; the nodes in the graph structure data represent network entities, and the edges represent the relationships between the network entities; using the graph structure data as training data, training a preset graph neural network, and obtaining an intelligence feature recognition model; applying the intelligence feature recognition model to extract feature attributes of the threat intelligence, and outputting the threat level of each intelligence node corresponding to the threat intelligence; and determining the priority of each threat intelligence based on the threat level.
[0038] By introducing graph neural networks, this application can efficiently integrate different types of data, break through information islands, and achieve unified processing of multi-source data. At the same time, by constructing graph structure data, different network entities and their relationships are represented as graph nodes and edges, so that the graph neural network can accurately identify complex network relationships, so that hidden threat patterns and attack paths can be identified in a timely manner, and the accuracy and precision of threat detection can be improved. Utilizing the parallel computing characteristics of graph neural networks, the present invention can significantly improve the data processing speed, realize real-time intelligence detection, ensure the timeliness and effectiveness of network security intelligence, and be able to quickly respond to sudden network threats. In addition, because the intelligence feature recognition model based on graph neural networks has a continuous learning and updating mechanism, it can dynamically adjust the model parameters to adapt to new threat patterns, improve the adaptability and flexibility of the system, and can respond to changing network attacks and enhance overall defense capabilities.
[0039] The present application also provides a threat intelligence detection system, a computer-readable storage medium, and an electronic device, which have the above-mentioned beneficial effects and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] 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, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.
[0041] Figure 1 A flowchart of a method for detecting threat intelligence provided by an embodiment of the present application;
[0042] Figure 2 A schematic diagram of the structure of the diagram provided in the embodiment of the present application;
[0043] Figure 3 A schematic diagram of the structure of a threat intelligence detection system provided in an embodiment of the present application;
[0044] Figure 4 A structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0045] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0046] The object information involved in this application, including but not limited to the object device information, object personal information, etc., and data, including but not limited to data used for analysis, stored data, displayed data, etc., are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the laws, regulations and standards of relevant countries and regions.
[0047] Currently, existing methods have difficulty in uniformly processing data from different sources. Although these methods use big data technology to collect and preprocess data, due to the strong heterogeneity of data sources and diverse formats, information cannot be efficiently integrated, forming "information islands". This approach limits the comprehensive utilization and efficient analysis of data. Information silos refer to the phenomenon that information and data are isolated and stored in different systems or databases within an organization or between different organizations, resulting in limited information sharing and circulation. Information silos may hinder effective decision-making and business processes.
[0048] Secondly, existing methods rely on traditional feature engineering techniques to extract useful features, but it is difficult to accurately identify complex network relationships and often difficult to capture the deep and complex relationships between network entities, which seriously affects the accuracy and precision of detection.
[0049] Compared with the existing technology, this application can efficiently integrate different types of data, break through information silos, and realize unified processing of multi-source heterogeneous data by introducing graph neural network (GNN) technology. Multi-source heterogeneous data refers to data from multiple different sources with different characteristics and formats. These data may include structured data (such as tabular data in databases), semi-structured data (such as XML and JSON), and unstructured data (such as text, images, and videos).
[0050] At the same time, the graph neural network GNN can accurately identify complex network relationships and improve the accuracy and precision of threat detection. By utilizing the parallel computing characteristics of GNN, the present invention can significantly improve the data processing speed and achieve real-time detection. More importantly, through continuous learning and model updates, GNN can quickly identify and respond to new threats and provide more flexible response measures.
[0051] See also Figure 1 , Figure 1 A flowchart of a method for detecting threat intelligence provided by an embodiment of the present application, the method comprising:
[0052] S101: Obtain intelligence data including threat intelligence;
[0053] S102: Representing the intelligence data as graph structure data; the nodes in the graph structure data represent network entities, and the edges represent the relationships between the network entities;
[0054] S103: Using the graph structure data as training data, training a preset graph neural network to obtain an intelligence feature recognition model;
[0055] S104: Applying the intelligence feature recognition model to extract characteristic attributes of the threat intelligence, and outputting the threat level of each intelligence node corresponding to the threat intelligence;
[0056] S105: Determine the priority of each threat intelligence based on the threat level.
[0057] Threat intelligence as described in this application refers to providing forward-looking predictions and early warnings of network attacks and their behaviors by collecting, processing, analyzing and evaluating various information. The sources of threats may include attackers, hacker organizations, etc. Attack methods are not limited to phishing attacks, denial of service attacks, etc.
[0058] First, it is necessary to obtain intelligence data. There is no limitation on how to obtain intelligence data. Intelligence data containing threat intelligence can be collected from a variety of data sources such as log files, network traffic, sensor data, etc. In a feasible implementation, unified threat intelligence data can be collected from several sources such as network traffic, log files, third-party threat intelligence platforms, and data collected by local security products. By integrating the first-line data collected by the security products themselves into a large database, a holistic data aggregation is made for attack types and other types of attacks. In a more preferred implementation, it can be ensured that the intelligence data comes from a variety of heterogeneous sources to cover a wide range of network security intelligence. Different intelligence data belongs to different sources, which can provide more accurate detection rates and coverage.
[0059] In a feasible implementation, after obtaining intelligence data, the threat intelligence can be formatted and standardized. There is no limitation on how to perform format standardization, for example, format conversion, data cleaning, and feature extraction can be performed. Data cleaning aims to remove redundant and noisy information in the data, and can also check and repair missing values or outliers in the data. When standardizing data, data in different formats can be converted into a unified structured format. For example, the text information of a log file is converted into structured tabular data for subsequent processing.
[0060] In addition, the threat intelligence can also be compared using a text similarity algorithm, and duplicate intelligence that exceeds a similarity threshold can be deleted. For example, a large model (such as BERT or GPT) can be used to extract important feature information from unstructured data (such as text logs), convert it into structured data, and convert the extracted features into numerical vectors through standardization processing, so that clustering algorithms can be applied to group or merge similar intelligence. Similar intelligence refers to at least two threat intelligences whose text similarity is less than the similarity threshold and greater than the similarity threshold value. This allows for further processing and analysis, so that data from different data sources can be effectively integrated.
[0061] After that, the processed data is represented as a graph structure. In the graph, nodes can represent network entities, which are not specifically limited here, including but not limited to IP addresses, domain names, file hash values, C2 relationships, botnet relationships, intranet addresses, etc. Edges represent relationships between entities, which may include communication relationships, dependency relationships, etc. This representation method can more intuitively display the complex associations between multi-source data.
[0062] Specifically, when executing step S102, nodes may be defined first, that is, nodes of the graph may be defined according to entities, so that the nodes represent various network entities, such as IP addresses, domain names, file hash values, etc. Then edges of the graph may be defined, and edges may be used to represent relationships between network entities, such as communication relationships, dependency relationships, etc. A graph example may be constructed to show different types of nodes and the relationships between them.
[0063] For example, see Figure 2 , Figure 2 A schematic diagram of the structure of the diagram provided in the embodiment of the present application is shown in FIG. Figure 2 There is a communication relationship between the node A IP address and the node B domain name, a command and control relationship between the node C file hash and the node D intranet address, and a C2 relationship between the node B domain name and the node D intranet address (the C2 relationship is a special form of network communication, usually involving intranet penetration technology), and a dependency relationship between the node A IP address and the node C file hash.
[0064] After that, the graph structure data is used as training data to train the preset graph neural network to obtain an intelligence feature recognition model. At this time, all network entities and relationships in the graph structure data used as training data are fully displayed.
[0065] In a feasible implementation, the intelligence feature recognition model may include the following steps:
[0066] The first step is to determine the network structure of the preset graph neural network;
[0067] Step 2: using a nonlinear activation function to determine the nonlinear features in the graph structure data, and propagating them through the network structure hierarchy to expand the receiving range of the nonlinear features;
[0068] Step 3: Aggregate the nonlinear features to generate a comprehensive feature representation; the comprehensive feature representation is used to reflect the reference relationship between nodes and the context information of the relationship network;
[0069] Step 4: Each node updates its own node status according to the comprehensive feature representation;
[0070] Step 5: After model training and optimization, the intelligence feature recognition model is obtained.
[0071] Before training the model, you can choose a suitable graph neural network model, such as GCN (Graph Convolutional Network) or GAT (Graph Attention Network).
[0072] After that, the network structure of the model can be further designed, including the number of layers, activation functions, etc. In intelligence processing, by designing a graph neural network model with a multi-layer structure, the number of layers is used to deepen the depth of feature propagation, so that the model can stack and learn more complex feature representations layer by layer, such as citation relationships and technical classifications. At the same time, the use of nonlinear activation functions such as ReLU enables the model to better capture and express the nonlinear features in patent citations and classification relationships, thereby improving the graph neural network model's ability to understand and process complex patent intelligence relationships.
[0073] In the second step, through multi-layer propagation, the graph neural network can learn the high-level feature representation of each node from the neighboring nodes. In the network threat intelligence processing, by performing feature propagation and aggregation on the graph, each node passes its features to the neighboring nodes and receives the features of the neighboring nodes. The feature reception range is expanded through layer-by-layer propagation, and these features are aggregated to form a new comprehensive feature representation. This process enables the reference feature to reflect the reference relationship between the node and others. At the same time, the features of different technical classifications are aggregated to form a comprehensive feature representation in the entire technical field. Through multi-level propagation and aggregation, the graph neural network can learn high-level feature representations from the neighboring nodes, which not only reflects the information of the node itself, but also contains the contextual information of its relationship network, improving the overall understanding and processing capabilities of intelligence.
[0074] Feature propagation and aggregation are performed on graph structure data. Each node transmits its features to its neighbor nodes, expanding the feature reception range through layer-by-layer propagation, and aggregating the received neighbor node features to form a new feature representation. In other words, the reference relationship features are propagated layer by layer, so that each network threat intelligence node can reflect its reference relationship with others; and the features of different technical classifications are aggregated, so that each node forms a comprehensive feature representation in the entire technical field, thereby enhancing the understanding and processing capabilities of intelligence data.
[0075] The model parameters can then be optimized to improve the accuracy of classification and prediction.
[0076] Through the above training steps, the generated intelligence feature recognition model will be able to accurately predict the threat level of network threat intelligence and can perform threat scoring on each node based on the learned features. At the same time, network security threat intelligence items can be classified and sorted to determine which network security threats are high-risk intelligence and need to be dealt with immediately. Therefore, the intelligence feature recognition model can be used to extract the characteristic attributes of threat intelligence and output the threat level of the intelligence node corresponding to each threat intelligence, and determine the priority of each threat intelligence based on the threat level.
[0077] It should be noted that threat intelligence of different priorities may belong to the same threat level. Therefore, the threat level is a classification of several threat intelligence, while the priority indicates the degree of threat between different intelligence, so as to determine the threat intelligence that needs to be handled first and the priority order according to the priority.
[0078] Specifically, the feature score of the intelligence node corresponding to each threat intelligence may be determined, thereby determining the priority of each threat intelligence according to the feature score.
[0079] After that, the threat intelligence can be prioritized based on the priority. The purpose of prioritization of threat intelligence is to ensure that the most urgent and important threats can be responded to and handled quickly. A feasible prioritization process can be as follows:
[0080] A1. Learning the features of graph structure data to determine the local influence of nodes: GNN is used to learn the features of nodes from the constructed graph structure data, and the relevant attributes of each intelligence node (such as association relationship, node influence, etc.) are extracted to determine the local influence of intelligence nodes;
[0081] A2. Feature priority scoring for intelligence nodes: score each intelligence node based on the learned features. The higher the score, the higher the threat level. Graph theory algorithms (such as PageRank) can be used to further adjust and optimize the score in combination with the association between nodes.
[0082] A3. Use the PageRank algorithm to mark the global influence of nodes: The PageRank algorithm can evaluate the global influence of nodes. By calculating the PageRank value of each node, its importance in the graph can be determined, which in turn affects the priority score determined in step A2.
[0083] A4. Comprehensive evaluation: Combine the feature learning results and the global influence calculated by the PageRank algorithm to generate the final priority score for each intelligence node.
[0084] Through this sorting mechanism, the system can automatically identify and sort the most threatening intelligence, ensuring that high-risk threats are dealt with first. For example, intelligence nodes whose global and local influences meet corresponding thresholds can be marked as emergency threat intelligence. Emergency threat intelligence can be used as a real-time warning source and intelligence response source.
[0085] Identification of urgent threats is a crucial part of threat intelligence processing, as it requires discovering and responding to high-risk threats in the shortest possible time to prevent large-scale cyber attacks and data breaches.
[0086] When identifying an emergency threat, you can evaluate the global and local influence of a node, and combine it with the contextual information of the intelligence to comprehensively judge its threat level. High-impact and strongly associated nodes are marked and prioritized for response as emergency threats. These nodes are sorted, and the higher the priority, the more urgent the threat, and the more immediate protective measures need to be taken.
[0087] For example, if a ransomware attack occurs: a node is identified as the host IP of the ransomware and is urgently marked through its high correlation with other nodes in the graph (such as a large number of visits to other nodes and rapid spread).
[0088] If a phishing attack is detected: A domain name node frequently appears in phishing attack intelligence and is accessed through multiple IP addresses. The node's importance score is used to quickly identify it as an urgent threat.
[0089] By using an intelligence feature recognition model built based on graph neural networks and graph theory algorithms, the system can intelligently identify and sort high-risk threats, thereby achieving real-time early warning and rapid response, and maximizing network security.
[0090] By introducing graph neural networks, this application can efficiently integrate different types of data, break through information islands, and achieve unified processing of multi-source data. At the same time, by constructing graph structure data, different network entities and their relationships are represented as graph nodes and edges, so that the graph neural network can accurately identify complex network relationships, so that hidden threat patterns and attack paths can be identified in a timely manner, and the accuracy and precision of threat detection can be improved. Utilizing the parallel computing characteristics of graph neural networks, the present invention can significantly improve the data processing speed, realize real-time intelligence detection, ensure the timeliness and effectiveness of network security intelligence, and be able to quickly respond to sudden network threats. In addition, because the intelligence feature recognition model based on graph neural networks has a continuous learning and updating mechanism, it can dynamically adjust the model parameters to adapt to new threat patterns, improve the adaptability and flexibility of the system, and can respond to changing network attacks and enhance overall defense capabilities.
[0091] In other embodiments of the present application, due to the wide and diverse sources of intelligence in the processing of network security threat intelligence, repeated or similar intelligence entries may appear. By executing an intelligence aggregation strategy when performing data preprocessing, these intelligences can be efficiently organized. The intelligence aggregation strategy may include the following steps:
[0092] Step 1: Data extraction and standardization: Extract threat intelligence from multi-source data and standardize data in different formats to ensure that they can be processed under the same framework.
[0093] Step 2: Similarity calculation: Use text similarity algorithms (such as TF-IDF, Jaccard similarity, etc.) or behavior pattern similarity to compare intelligence items and evaluate their repeatability.
[0094] Step 3: Cluster analysis: Use clustering algorithms (such as K-means, DBSCAN, etc.) to group similar intelligence to determine which intelligence items can be merged.
[0095] Step 4: Merge and remove duplicates: Merge similar or duplicate intelligence items to generate comprehensive intelligence items and remove redundant information.
[0096] It can be seen that through the intelligence aggregation strategy, the interference of duplicate information can be reduced, the quality of the intelligence database can be optimized, and the threat detection system can process and utilize this information more efficiently.
[0097] See also Figure 3 , Figure 3 A schematic diagram of the structure of a threat intelligence detection system provided in an embodiment of the present application, the system includes:
[0098] A data acquisition module, used to acquire intelligence data including threat intelligence;
[0099] A structure conversion module, used to represent the intelligence data as graph structure data; the nodes in the graph structure data represent network entities, and the edges represent the relationships between the network entities;
[0100] A model training module is used to use the graph structure data as training data to train a preset graph neural network to obtain an intelligence feature recognition model;
[0101] A feature detection module, used to apply the intelligence feature recognition model to extract the feature attributes of the threat intelligence and output the threat level of each intelligence node corresponding to the threat intelligence;
[0102] The intelligence sorting module is used to determine the priority of each threat intelligence based on the threat level.
[0103] Based on the above embodiments, as a preferred embodiment, it also includes:
[0104] The data processing module is used to perform format standardization processing on the threat intelligence; compare the threat intelligence using a text similarity algorithm, and delete duplicate intelligence that exceeds a similarity threshold.
[0105] Based on the above embodiments, as a preferred embodiment, it also includes:
[0106] The intelligence clustering module is used to apply a clustering algorithm to group or merge similar intelligence; wherein the similar intelligence is at least two threat intelligences whose text similarity is less than the similarity threshold and greater than the similarity threshold value.
[0107] Based on the above embodiment, as a preferred embodiment, the model training module is a module for performing the following steps:
[0108] Determine the network structure of the preset graph neural network;
[0109] Determine the nonlinear features in the graph structure data using a nonlinear activation function, and propagate through the network structure hierarchy to expand the reception range of the nonlinear features;
[0110] Aggregating the nonlinear features to generate a comprehensive feature representation; the comprehensive feature representation is used to reflect the reference relationship between nodes and the context information of the relationship network;
[0111] Each node updates its own node status according to the comprehensive feature representation;
[0112] After model training and optimization, the intelligence feature recognition model is obtained.
[0113] Based on the above embodiment, as a preferred embodiment, the intelligence sorting module includes:
[0114] A scoring unit, used to determine a characteristic score of each intelligence node corresponding to each threat intelligence;
[0115] A priority determination unit is used to determine the priority of each threat intelligence according to the feature score.
[0116] Based on the above embodiments, as a preferred embodiment, the priority determination unit is a unit for calculating the global influence of each of the intelligence nodes using a graph theory algorithm; taking the feature score as the local influence and combining the global influence to determine the priority of each of the threat intelligence.
[0117] Based on the above embodiments, as a preferred embodiment, it also includes:
[0118] The emergency intelligence determination module is used to mark the intelligence nodes whose global influence and local influence both meet the corresponding thresholds as emergency threat intelligence; the emergency threat intelligence is used as a real-time warning source and intelligence response source.
[0119] The present application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed, the steps provided in the above embodiment can be implemented. The storage medium may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.
[0120] The present application also provides an electronic device, see Figure 4 , a structural diagram of an electronic device provided in an embodiment of the present application, such as Figure 4 As shown, a processor 1410 and a memory 1420 may be included.
[0121] Among them, the processor 1410 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 1410 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 1410 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 1410 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 1410 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.
[0122] The memory 1420 may include one or more computer-readable storage media, which may be non-transitory. The memory 1420 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In this embodiment, the memory 1420 is at least used to store the following computer program 1421, wherein, after the computer program is loaded and executed by the processor 1410, it can implement the relevant steps in the method performed by the electronic device side disclosed in any of the aforementioned embodiments. In addition, the resources stored in the memory 1420 may also include an operating system 1422 and data 1423, etc., and the storage method may be temporary storage or permanent storage. Among them, the operating system 1422 may include Windows, Linux, Android, etc.
[0123] In some embodiments, the electronic device may further include a display screen 1430 , an input / output interface 1440 , a communication interface 1450 , a sensor 1460 , a power source 1470 , and a communication bus 1480 .
[0124] certainly, Figure 4 The structure of the electronic device shown does not constitute a limitation on the electronic device in the embodiments of the present application. In actual applications, the electronic device may include Figure 4 More or fewer components than shown, or combinations of certain components.
[0125] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other. For the system provided in the embodiment, since it corresponds to the method provided in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.
[0126] Specific examples are used herein to illustrate the principles and implementation methods of the present application, and the description of the above embodiments is only used to help understand the method and core ideas of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of the present application.
[0127] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.
Claims
1. A method for detecting threat intelligence, characterized in that: include: Obtain intelligence data including threat intelligence; Representing the intelligence data as graph structured data; The nodes in the graph structure data represent network entities, and the edges represent the relationships between the network entities; Using the graph structure data as training data, training a preset graph neural network to obtain an intelligence feature recognition model; Applying the intelligence feature recognition model to extract characteristic attributes of the threat intelligence, and outputting the threat level of each intelligence node corresponding to the threat intelligence; The priority of each threat intelligence is determined based on the threat level.
2. The method for detecting threat intelligence according to claim 1, characterized in that: After obtaining intelligence data containing threat intelligence, it also includes: Standardizing the format of the threat intelligence; The threat intelligence is compared using a text similarity algorithm, and duplicate intelligence exceeding a similarity threshold is deleted.
3. The method for detecting threat intelligence according to claim 2, characterized in that: Before comparing the threat intelligence using a text similarity algorithm and deleting duplicate intelligence exceeding a similarity threshold, the following steps are also included: A clustering algorithm is applied to group or merge similar intelligence; wherein the similar intelligence is at least two threat intelligences whose text similarity is less than the similarity threshold and greater than the similarity threshold value.
4. The method for detecting threat intelligence according to claim 1, characterized in that: The method of using the graph structure data as training data to train the preset graph neural network to obtain the intelligence feature recognition model includes: Determine the network structure of the preset graph neural network; Determine the nonlinear features in the graph structure data using a nonlinear activation function, and propagate through the network structure hierarchy to expand the reception range of the nonlinear features; Aggregating the nonlinear features to generate a comprehensive feature representation; the comprehensive feature representation is used to reflect the reference relationship between nodes and the context information of the relationship network; Each node updates its own node status according to the comprehensive feature representation; After model training and optimization, the intelligence feature recognition model is obtained.
5. The method for detecting threat intelligence according to claim 1, characterized in that: Determining the priority of each threat intelligence based on the threat level includes: Determine a characteristic score of the intelligence node corresponding to each threat intelligence; The priority of each threat intelligence is determined according to the feature score.
6. The method for detecting threat intelligence according to claim 5, characterized in that: Determining the priority of each threat intelligence according to the feature score includes: Calculate the global influence of each intelligence node using graph theory algorithms; The feature score is used as the local influence and combined with the global influence to determine the priority of each threat intelligence.
7. The method for detecting threat intelligence according to claim 6, characterized in that: After applying the intelligence feature recognition model to extract the characteristic attributes of the threat intelligence and outputting the threat level of each intelligence node corresponding to the threat intelligence, the method further includes: The intelligence nodes whose global influence and local influence both meet corresponding thresholds are marked as emergency threat intelligence; the emergency threat intelligence is used as a real-time warning source and intelligence response source.
8. A threat intelligence detection system, characterized in that: include: A data acquisition module, used to acquire intelligence data including threat intelligence; A structure conversion module, used for representing the intelligence data as graph structure data; The nodes in the graph structure data represent network entities, and the edges represent the relationships between the network entities; A model training module is used to use the graph structure data as training data to train a preset graph neural network to obtain an intelligence feature recognition model; A feature detection module, used to apply the intelligence feature recognition model to extract the feature attributes of the threat intelligence and output the threat level of each intelligence node corresponding to the threat intelligence; The intelligence sorting module is used to determine the priority of each threat intelligence based on the threat level.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the threat intelligence detection method as described in any one of claims 1 to 7 are implemented.
10. An electronic device, characterized in that: It includes a memory and a processor, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, the steps of the threat intelligence detection method according to any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Network security threat monitoring method and system based on artificial intelligence
CN119628963A
Industrial control network security advanced threat detection system fused with artificial intelligence
CN120474777A
Multi-source security intelligence collaborative analysis method and system fused with AI intelligent agent
CN120768691A
Terminal threat event detection method and device, computer equipment and storage medium
CN120811787A
Terminal threat event detection method and apparatus, computer device, and storage medium
CN120811787B
