A network information security analysis method and system based on big data

The network information security analysis method enhances the accuracy and sensitivity of threat detection by analyzing firewall logs, performing traffic capture and behavior logic analysis, and deploying a big data attack sniffing engine on a cloud platform, addressing the limitations of traditional methods.

CN118337485BActive Publication Date: 2024-12-03GUANGZHOU BAOLU INTELLIGENT TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410549703.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-06
Publication Date
2024-12-03
Estimated Expiration
2044-05-06

AI Technical Summary

Technical Problem

Traditional network information security analysis methods based on big data suffer from inaccurate perception of threatening behaviors and low sensitivity to early sniffing of attack behaviors.

Method used

A network information security analysis method and system that includes obtaining firewall log data, extracting abnormal activity connection addresses, performing abnormal behavior monitoring, traffic capture processing, threat behavior situation awareness, attack behavior structure logic analysis, and designing a big data attack sniffing engine with automated firmware for real-time monitoring and analysis on a cloud platform.

Benefits of technology

Improves the accuracy of perceiving threatening behaviors and sensitivity to early sniffing of attack behaviors, enabling timely detection and response to various types of network attacks, including known and unknown patterns, with reduced operation and maintenance costs and increased scalability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118337485B_ABST
    Figure CN118337485B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of network security analysis, and in particular to a network information security analysis method and system based on big data. The method comprises the following steps: extracting abnormal activity connection addresses from firewall log data and monitoring abnormal behavior data to obtain abnormal behavior monitoring data; capturing abnormal behavior traffic from abnormal behavior monitoring data and performing threat behavior situation awareness to obtain threat behavior situation awareness data; performing attack behavior structure logic analysis based on threat behavior situation awareness data to obtain attack behavior structure logic data; performing big data attack sniffing engine design and automated firmware design based on attack behavior structure logic data to obtain big data attack sniffing automated firmware; and sending the big data attack sniffing automated firmware to a cloud platform to perform network information security analysis. The present invention makes security analysis more accurate by optimizing the network information security analysis technology.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of network security analysis, and in particular to a network information security analysis method and system based on big data. Background Art

[0002] In today's digital age, network information security has become a global focus. With the popularization of the Internet and the development of information technology, the forms and scale of network attacks are increasing. The previous security defense methods can no longer meet the rapidly changing threats. For this reason, the network information security analysis method based on big data has emerged. This method uses big data technology to collect, store and analyze massive amounts of network data to mine potential security threats and signs of attacks. Compared with previous security defense methods, the security analysis method based on big data has stronger real-time, comprehensive and predictive performance, and can detect and respond to various network attack behaviors in a timely manner, providing strong support for network security. The emergence of this method marks that network security defense has entered a new stage, which is of great significance for protecting personal privacy and corporate interests. However, the traditional network information security analysis method based on big data has the problem of inaccurate perception of threatening behaviors and low sensitivity to early sniffing of attack behaviors. Summary of the invention

[0003] Based on this, it is necessary to provide a network information security analysis method and system based on big data to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a network information security analysis method based on big data is provided, the method comprising the following steps:

[0005] Step S1: Obtain firewall log data; extract abnormal activity connection addresses from the firewall log data to obtain abnormal activity connection addresses; monitor abnormal behavior data based on the abnormal activity connection addresses to obtain abnormal behavior monitoring data;

[0006] Step S2: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data to obtain abnormal behavior traffic capture data; performing threat behavior situation awareness based on the abnormal behavior traffic capture data to obtain threat behavior situation awareness data; performing attack behavior structure logic analysis based on the threat behavior situation awareness data to obtain attack behavior structure logic data;

[0007] Step S3: performing attack behavior data association mining based on the attack behavior structure logic data to obtain attack behavior association mining data; performing big data attack sniffing engine design based on the attack behavior association mining data to obtain a big data attack sniffing engine;

[0008] Step S4: Automated firmware design is performed on the big data attack sniffing engine to obtain the big data attack sniffing automated firmware; the big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis.

[0009] The present invention can timely discover abnormal activities in the network, such as potential intrusion attempts, malicious traffic or abnormal data transmission, by obtaining firewall log data and extracting abnormal activity connection addresses. This helps to identify potential security threats early and take corresponding countermeasures. By performing traffic capture processing on abnormal behavior monitoring data, the specific characteristics and traffic patterns of abnormal activities can be deeply understood. This helps to improve the perception of threatening behaviors, including the detection and identification of known and unknown threats. According to the abnormal behavior traffic capture data, threat behavior situation awareness can be performed, that is, the trend and pattern of threat behavior in the network can be analyzed. This helps to form a comprehensive understanding of the network security situation, and timely adjust security policies and measures to cope with the ever-changing threat environment. By performing attack behavior structure logical analysis on threat behavior situation awareness data, the behavior patterns, means and strategies of attackers can be revealed. This helps to understand the attacker's way of thinking and attack chain, and provides a reference for formulating more effective security defense strategies. By performing attack behavior data association mining based on attack behavior structure logical data, the correlation and pattern between different attack behaviors can be identified and analyzed. This helps to deeply understand the attacker's behavior chain and attack strategy, thereby improving the ability to identify and respond to complex attacks. Based on the attack behavior association mining data, the design of a big data attack sniffing engine can realize real-time monitoring and analysis of network traffic and activities. This engine can use big data technology and algorithms to quickly and effectively detect and respond to various types of attack behaviors, including known and unknown attack patterns. The big data attack sniffing engine is designed with automated firmware, and its functions and algorithms can be encapsulated into deployable firmware to achieve automated deployment and management. This can improve the scalability and universality of security solutions, and reduce operation and maintenance costs and complexity. The big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis, which can achieve comprehensive monitoring and management of distributed network environments. The cloud platform provides the advantages of elastic resources and high availability, and can effectively cope with the processing requirements of large-scale network traffic and security events. Therefore, the present invention provides an optimization process made by a network information security analysis method based on big data, which solves the problems of inaccurate perception of threatening behaviors and low sensitivity to early sniffing of attack behaviors in traditional network information security analysis methods based on big data, improves the accuracy of perception of threatening behaviors, and improves the sensitivity to early sniffing of attack behaviors.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: Obtain firewall log data;

[0012] Step S12: Analyze abnormal behavior activities according to the firewall log data to obtain abnormal behavior activity data;

[0013] Step S13: extracting abnormal activity connection addresses from firewall log data according to abnormal behavior activity data to obtain abnormal activity connection addresses;

[0014] Step S14: abnormal behavior data monitoring is performed based on the abnormal activity connection address and the abnormal behavior activity data to obtain abnormal behavior monitoring data.

[0015] By acquiring firewall log data, the present invention can collect various activities and connection information in the network, provide a data basis for subsequent security analysis and monitoring, and perform abnormal behavior activity analysis on firewall log data, which helps to identify and analyze abnormal activities, including potential attack behaviors, abnormal traffic patterns, or unusual data transmission. This helps to detect potential security threats early, and extracts abnormal activity connection addresses from firewall log data based on abnormal behavior activity data, and can screen out specific connection addresses related to abnormal activities. This helps to focus analysis and monitoring, and increase attention to attack targets. Abnormal behavior data monitoring based on abnormal activity connection addresses and abnormal behavior activity data can monitor and analyze the development trends and patterns of abnormal behaviors in real time. This helps to respond quickly to security incidents and take necessary defensive measures to mitigate possible security risks.

[0016] Preferably, step S2 comprises the following steps:

[0017] Step S21: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data according to the abnormal activity connection address to obtain abnormal behavior traffic capture data;

[0018] Step S22: performing attack program packet capture processing on the abnormal behavior traffic capture data to obtain attack program packet capture data; performing deep packet inspection on the attack program packet capture data to obtain deep packet inspection data;

[0019] Step S23: Perform threat behavior situation awareness based on deep packet inspection data and abnormal behavior traffic capture data to obtain threat behavior situation awareness data;

[0020] Step S24: Performing an attack behavior structure logic analysis on the attack program packet capture data according to the threat behavior situation awareness data to obtain attack behavior structure logic data.

[0021] The present invention performs abnormal behavior traffic capture processing on abnormal behavior monitoring data according to the abnormal activity connection address, and can obtain network traffic data related to abnormal activities. This helps to gain a deep understanding of the nature and characteristics of abnormal behavior, so as to more effectively perform subsequent analysis and processing. The attack program packet capture processing is performed on the abnormal behavior traffic capture data, and potential attack programs and malicious codes can be captured. Through deep packet detection, attack behaviors and malicious behaviors in network traffic can be further analyzed and identified, including the detection of protocols, data packet structures and contents. Threat behavior situation awareness is performed based on deep packet detection data and abnormal behavior traffic capture data, and threat behavior trends and patterns in the network can be monitored and analyzed in real time. This helps to quickly discover and respond to various types of network attacks, and take corresponding security measures. According to the threat behavior situation awareness data, the attack behavior structure logic analysis is performed on the attack program packet capture data, and the attacker's behavior pattern and attack chain can be deeply understood. This helps to discover the attacker's intentions and purposes, and provides a reference for formulating targeted security defense strategies.

[0022] Preferably, step S23 includes the following steps:

[0023] Step S231: performing threat code analysis on the deep packet inspection data to obtain threat code data; performing threat code data structure analysis on the threat code data to obtain threat code structure data; performing code reverse engineering analysis on the threat code structure data to obtain code threat reverse analysis data;

[0024] Step S232: performing dependency association analysis on the threatening code structure data according to the code threat reverse analysis data to obtain threatening code structure dependency association data;

[0025] Step S233: performing process hijacking analysis according to the threatening code structure dependency association data and the code threat reverse analysis data to obtain threatening process hijacking analysis data; performing process hijacking execution command deduction according to the threatening process hijacking analysis data and the code threat reverse analysis data to obtain process hijacking execution command deduction data;

[0026] Step S234: performing a variance balance measurement analysis on the process hijacking execution command deduction data to obtain command deduction variance balance measurement data;

[0027] Step S235: constructing a balanced measurement policy gradient learning model for the command deduction variance balanced measurement data based on the Policy Gradient algorithm to obtain a command deduction balanced measurement policy model;

[0028] Step S236: performing a threat situation gradient analysis on the process hijacking execution command deduction data according to the command deduction balance measurement strategy model to obtain the hijacking command threat situation gradient data;

[0029] Step S237: Perform threat behavior situation awareness according to the hijacking command threat situation gradient data and the abnormal behavior traffic capture data to obtain threat behavior situation awareness data.

[0030] The present invention can identify and analyze malicious code and attack payloads present in network traffic by performing threatening code analysis and reverse engineering on deep packet detection data. This helps to understand the attacker's means and attack methods, and provides a basis for further security response. Dependent association analysis is performed based on code threatening reverse analysis data, and the correlation between the structure and function of the malicious code can be revealed. This helps to understand the execution logic and possible threatening behaviors of malicious code, analyzes threatening process hijacking analysis data and execution command deduction data, and can identify and understand the attacker's hijacking behavior on the system process, and deduce possible attack commands and operations. This helps to predict the attacker's next move and take corresponding defensive measures. The deduction variance balance measurement data is analyzed, and the degree of difference and risk level between different deduction schemes can be evaluated. This helps to determine the most threatening attack paths and behaviors, and optimize security defense strategies. Based on the Policy Gradient algorithm, a balanced measurement policy learning model is constructed, which can realize dynamic identification and response to threatening behaviors. This helps to establish an adaptive security defense mechanism and improve the ability to respond to new threats. Combining the hijacking command threat situation gradient data and abnormal behavior traffic capture data for threat behavior situation awareness can fully understand the threat behavior trends and patterns in the network. This helps to timely discover and respond to evolving security threats.

[0031] Preferably, performing variance balance measurement analysis on the process hijacking execution command deduction data includes the following steps:

[0032] The execution command is repeatedly deduced on the process hijacking execution command deduction data using the preset execution command deduction count threshold to obtain a hijacking command deduction data set;

[0033] Perform hijacking command difference analysis on the hijacking command deduction data set to obtain hijacking command difference data; perform random two-to-one matching chi-square test on the hijacking command deduction data set based on the hijacking command difference data to obtain deduction difference deviation degree data;

[0034] According to the deduced difference deviation degree data, the confidence interval between the two random hijacking command difference data is calculated to obtain the hijacking command difference confidence interval;

[0035] Based on the deduction difference deviation degree data and the hijacking command difference confidence interval, the deduction variance balance measurement analysis is performed to obtain the command deduction variance balance measurement data.

[0036] The present invention repeatedly deduces the process hijacking execution command deduction data by utilizing the preset execution command deduction count threshold value to obtain the hijacking command deduction data set. This helps to collect enough deduction data for subsequent analysis and comparison, and performs hijacking command difference analysis on the hijacking command deduction data set to obtain hijacking command difference data. This helps to understand the differences and characteristics between different deduction data. According to the hijacking command difference data, the hijacking command deduction data set is subjected to a random two-to-one matching chi-square test to obtain deduction difference deviation degree data. This helps to determine whether the difference between the deduction data is significant, and quantify the degree of difference. According to the deduction difference deviation degree data, the hijacking command difference data is subjected to a random confidence interval calculation between the two to obtain the hijacking command difference confidence interval. This helps to determine the confidence level and statistical significance of the difference data, and combines the deduction difference deviation degree data and the hijacking command difference confidence interval to perform deduction variance balance measurement analysis to obtain command deduction variance balance measurement data. This helps to evaluate the differences between different deduction schemes and determine the most threatening deduction path.

[0037] Preferably, step S24 includes the following steps:

[0038] Step S241: performing attack behavior context correlation analysis on the attack program packet capture data according to the threat behavior situation awareness data to obtain attack behavior context correlation data;

[0039] Step S242: performing attack behavior topology analysis on the attack behavior contextual correlation data to obtain attack behavior topology data; identifying attack key target nodes according to the attack behavior contextual correlation data and the attack behavior topology data to obtain attack behavior key target node data;

[0040] Step S243: Performing attack behavior structure logic analysis on key target node data according to attack behavior context-related data to obtain attack behavior structure logic data.

[0041] The present invention can identify the contextual relationship and mutual influence between attack behaviors by analyzing the threat behavior situation awareness data. This helps to understand the attacker's strategy and means, so as to better deal with and prevent potential security threats. By analyzing the upper and lower associated data of the attack behavior, the topological structure of the attack behavior can be established to reveal the hierarchical relationship and dependency relationship between the attack behaviors. This helps to identify the key nodes and attack paths in the attack chain, and provides an important reference for the formulation of defense strategies. Based on the attack behavior topological data, the key target nodes in the attack chain, i.e. the attacker's main target or the key link on the attack path, can be identified. The identification of these nodes helps to take defensive measures in time to protect the security of key systems and data resources. By performing a logical analysis of the attack behavior structure on the key target node data, the execution logic and internal structure of the attack behavior can be deeply understood. This helps to identify the attacker's behavior pattern and attack strategy, and provides an important reference for formulating targeted security defense strategies.

[0042] Preferably, step S3 comprises the following steps:

[0043] Step S31: performing attack behavior data association mining according to the attack behavior structure logic data and the threat behavior situation awareness data to obtain attack behavior association mining data;

[0044] Step S32: performing attack behavior pattern analysis on the attack behavior association mining data to obtain attack behavior pattern data;

[0045] Step S33: Design a big data attack sniffing engine based on the attack behavior association mining data and the attack behavior pattern data to obtain a big data attack sniffing engine.

[0046] The present invention can determine the correlation and dependency between attack behaviors by performing attack behavior contextual correlation analysis on the attack program packet capture data according to the threat behavior situation awareness data. This helps to understand the attacker's attack path and strategy, and discover the hidden relationship between attack behaviors; by performing attack behavior topological analysis on the attack behavior contextual correlation data, the topological structure of the attack behavior can be constructed, and the hierarchical relationship and connection mode between the attack behaviors can be clearly displayed. According to the topological analysis results, key target nodes are identified, that is, the most important nodes in the attack and the nodes that are most concerned by the attacker. This helps to focus analysis and defense forces, improve the level of protection for key target nodes, and perform attack behavior structure logical analysis on key target node data according to the attack behavior contextual correlation data, which can deeply understand the structure and logic of the attack behavior. This helps to reveal the attacker's attack ideas and methods, and provide a reference for formulating corresponding defense strategies.

[0047] Preferably, step S33 includes the following steps:

[0048] Step S331: performing attack load analysis on the attack behavior association mining data to obtain attack behavior load data; performing attack flow fluctuation analysis on the attack behavior association mining data according to the attack behavior load data to obtain attack flow fluctuation data;

[0049] Step S332: analyzing the instantaneous traffic fluctuation growth rate of the attack traffic fluctuation data to obtain instantaneous traffic growth rate data; when the instantaneous traffic growth rate data is less than a preset instantaneous traffic growth threshold, marking the attack traffic fluctuation data in the period as regular access traffic data;

[0050] Step S333: When the instantaneous traffic growth rate data is greater than or equal to the preset instantaneous traffic growth threshold, the attack traffic fluctuation data in the period is marked as risky access traffic data;

[0051] Step S334: performing risk access behavior pattern mapping processing on the attack behavior pattern data according to the risk access traffic data to obtain risk access behavior pattern data;

[0052] Step S335: Perform unauthorized access request frequency analysis on the risk access behavior pattern data to obtain unauthorized access request frequency data;

[0053] Step S336: using a gradient boosting tree algorithm to perform risk behavior feature memory learning on the risk access behavior pattern data and the unauthorized access request frequency data to obtain risk behavior feature memory data;

[0054] Step S337: Design a big data attack sniffing engine based on the risk behavior characteristic memory data to obtain a big data attack sniffing engine.

[0055] The present invention can understand the intensity and frequency of the attack by performing attack load analysis on the attack behavior associated mining data. Performing attack flow fluctuation analysis on the attack behavior associated mining data according to the attack behavior load data can identify the changing trend and pattern of the attack flow, which is helpful to predict the development trend and law of the attack behavior. By analyzing the instantaneous growth rate of the attack flow, the sudden increase of the attack flow can be quickly identified. By comparing the instantaneous flow growth rate with the preset instantaneous flow growth threshold, the conventional access flow and the risk access flow can be effectively distinguished. The attack behavior pattern data is mapped and processed according to the risk access flow data, and the characteristic pattern and law of the attack behavior can be identified. This helps to establish a pattern recognition system for attack behavior, improve the detection and identification capabilities of potential threats, and analyze the frequency data of unauthorized access requests to understand the frequency of unauthorized access by attackers to the system. This helps to discover potential security vulnerabilities and attack risks, and take corresponding security measures in time for defense. The gradient boosting tree algorithm is a powerful machine learning technology that can effectively process complex nonlinear relationships and high-dimensional data. In this step, the gradient boosting tree algorithm is used to learn the risk access behavior pattern data and the frequency data of unauthorized access requests, which can dig out the characteristics and rules hidden behind the data. Through the gradient boosting tree algorithm learning, the risk behavior feature memory data can be obtained, which contains the memory of the key features and patterns of risk access behavior and unauthorized access requests. These data are crucial for subsequent attack detection and analysis, and can help the system identify potential security threats. According to the risk behavior feature memory data, a big data attack sniffing engine can be designed, which can monitor and analyze large-scale network traffic data in real time. The big data attack sniffing engine can identify and intercept potential attack behaviors, including malicious access, unauthorized operations, etc., thereby improving the system's detection and response capabilities to security threats. The designed sniffing engine can analyze network traffic in real time, detect abnormal behaviors in time and take corresponding defense measures. At the same time, based on the risk behavior feature memory data, the engine has a high degree of accuracy and can accurately identify attack behaviors to avoid false positives and false negatives.

[0056] Preferably, performing risk behavior feature memory learning on risk access behavior pattern data and unauthorized access request frequency data comprises the following steps:

[0057] The risk access behavior pattern data and the unauthorized access request frequency data are divided into time series to obtain the risk access pattern training set, the risk access pattern test set, and the unauthorized access request training set and the unauthorized access request test set respectively;

[0058] Perform linear correlation training set feature analysis on the risk access pattern training set and the unauthorized access request training set to obtain the risk-unauthorized linear feature training set;

[0059] Perform random feature sampling on the risk access pattern test set and the unauthorized access request test set to obtain a risk-unauthorized random feature sampling test set;

[0060] The gradient boosting tree algorithm is used to construct the risk-overreaching behavior feature memory model for the risk-overreaching linear feature training set, and the risk behavior feature memory model is obtained;

[0061] The risk-overriding random feature sampling test set is input into the risk behavior feature memory model for risk behavior feature memory learning to obtain risk behavior feature memory data.

[0062] The present invention divides the original data into training set and test set according to the time series by time series division, which can ensure the generalization ability of the model on future data and avoid the problem of overfitting or underfitting of the model. The risk access pattern training set and the unauthorized access request training set can be analyzed to identify the linear features related to risk behavior and unauthorized request. These features include request frequency, access path, user identity, etc., which are helpful to establish a basic risk behavior feature memory model. Random feature sampling on the test set can reduce the computational burden during the test process, and improve the efficiency and speed of model evaluation while maintaining data distribution. The gradient boosting tree algorithm is used to construct a model for the risk-unauthorized linear feature training set, which can establish a model memory of the complex nonlinear relationship between risk behavior and unauthorized request, improve the model's fitting ability and generalization ability for data, and input the risk-unauthorized random feature sampling test set into the risk behavior feature memory model to perform risk behavior feature memory learning. The purpose of this step is to allow the model to learn on the test data to obtain more comprehensive data patterns and features, and further improve the performance and accuracy of the model.

[0063] Preferably, the present invention also provides a network information security analysis system based on big data, which is used to execute the network information security analysis method based on big data as described above, and the network information security analysis system based on big data includes:

[0064] The abnormal behavior monitoring module is used to obtain firewall log data; extract abnormal activity connection addresses from the firewall log data to obtain abnormal activity connection addresses; monitor abnormal behavior data based on the abnormal activity connection addresses to obtain abnormal behavior monitoring data;

[0065] The threat behavior situation awareness module is used to perform abnormal behavior traffic capture processing on abnormal behavior monitoring data to obtain abnormal behavior traffic capture data; perform threat behavior situation awareness based on abnormal behavior traffic capture data to obtain threat behavior situation awareness data; perform attack behavior structure logic analysis based on threat behavior situation awareness data to obtain attack behavior structure logic data;

[0066] A big data attack sniffing engine design module is used to perform attack behavior data association mining based on attack behavior structure logic data to obtain attack behavior association mining data; a big data attack sniffing engine is designed based on attack behavior association mining data to obtain a big data attack sniffing engine;

[0067] The network information security analysis module is used to perform automated firmware design on the big data attack sniffing engine to obtain the big data attack sniffing automated firmware; and send the big data attack sniffing automated firmware to the cloud platform to perform network information security analysis.

[0068] The beneficial effect of the present invention is that by obtaining firewall log data and extracting abnormal activity connection addresses, abnormal activities in the network, such as potential intrusion attempts, malicious traffic or abnormal data transmission, can be discovered in time. This helps to identify potential security threats early and take corresponding countermeasures. By performing traffic capture processing on abnormal behavior monitoring data, the specific characteristics and traffic patterns of abnormal activities can be deeply understood. This helps to improve the perception of threat behaviors, including the detection and identification of known and unknown threats. According to the abnormal behavior traffic capture data, threat behavior situation awareness can be performed, that is, the threat behavior trends and patterns in the network are analyzed. This helps to form a comprehensive understanding of the network security situation and adjust security policies and measures in time to cope with the ever-changing threat environment. By performing attack behavior structure logical analysis on threat behavior situation awareness data, the attacker's behavior patterns, means and strategies can be revealed. This helps to understand the attacker's way of thinking and attack chain, and provides a reference for formulating more effective security defense strategies. By performing attack behavior data association mining based on attack behavior structure logical data, the correlation and pattern between different attack behaviors can be identified and analyzed. This helps to deeply understand the attacker's behavior chain and attack strategy, thereby improving the ability to identify and respond to complex attacks. Based on the attack behavior association mining data, the design of a big data attack sniffing engine can realize real-time monitoring and analysis of network traffic and activities. This engine can use big data technology and algorithms to quickly and effectively detect and respond to various types of attack behaviors, including known and unknown attack patterns. The big data attack sniffing engine is designed with automated firmware, and its functions and algorithms can be encapsulated into deployable firmware to achieve automated deployment and management. This can improve the scalability and universality of security solutions, and reduce operation and maintenance costs and complexity. The big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis, which can achieve comprehensive monitoring and management of distributed network environments. The cloud platform provides the advantages of elastic resources and high availability, and can effectively cope with the processing requirements of large-scale network traffic and security events. Therefore, the present invention provides an optimization process made by a network information security analysis method based on big data, which solves the problems of inaccurate perception of threatening behaviors and low sensitivity to early sniffing of attack behaviors in traditional network information security analysis methods based on big data, improves the accuracy of perception of threatening behaviors, and improves the sensitivity to early sniffing of attack behaviors. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a schematic diagram of the steps of a network information security analysis method based on big data;

[0070] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0071] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0072] 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

[0073] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0074] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0075] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0076] To achieve this, please refer to Figures 1 to 3 , a network information security analysis method based on big data, the method comprising the following steps:

[0077] Step S1: Obtain firewall log data; extract abnormal activity connection addresses from the firewall log data to obtain abnormal activity connection addresses; monitor abnormal behavior data based on the abnormal activity connection addresses to obtain abnormal behavior monitoring data;

[0078] Step S2: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data to obtain abnormal behavior traffic capture data; performing threat behavior situation awareness based on the abnormal behavior traffic capture data to obtain threat behavior situation awareness data; performing attack behavior structure logic analysis based on the threat behavior situation awareness data to obtain attack behavior structure logic data;

[0079] Step S3: performing attack behavior data association mining based on the attack behavior structure logic data to obtain attack behavior association mining data; performing big data attack sniffing engine design based on the attack behavior association mining data to obtain a big data attack sniffing engine;

[0080] Step S4: Automated firmware design is performed on the big data attack sniffing engine to obtain the big data attack sniffing automated firmware; the big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis.

[0081] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of a network information security analysis method based on big data of the present invention. In this example, the network information security analysis method based on big data includes the following steps:

[0082] Step S1: Obtain firewall log data; extract abnormal activity connection addresses from the firewall log data to obtain abnormal activity connection addresses; monitor abnormal behavior data based on the abnormal activity connection addresses to obtain abnormal behavior monitoring data;

[0083] In the embodiment of the present invention, by configuring the firewall device, setting the log recording function to capture network traffic data, using the log analysis tool, extracting abnormal connection addresses that are inconsistent with normal network traffic, for example, detecting a large number of connection requests from the same IP address, monitoring the activities of the abnormal connection addresses, including recording the source, destination, and protocol used of the connection. For example, the firewall log shows that a certain IP address has sent a large number of connection requests to the internal server in a short period of time, and the IP address is not a common visitor to the internal network, so it is identified as an abnormal activity connection address.

[0084] Step S2: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data to obtain abnormal behavior traffic capture data; performing threat behavior situation awareness based on the abnormal behavior traffic capture data to obtain threat behavior situation awareness data; performing attack behavior structure logic analysis based on the threat behavior situation awareness data to obtain attack behavior structure logic data;

[0085] In the embodiment of the present invention, a traffic capture tool (such as Wireshark) is used to capture and analyze the traffic involved in the abnormal behavior monitoring data, and abnormal traffic data that does not match the normal traffic is screened out, such as abnormal frequent port scanning and a large number of abnormal data packet transmissions. According to the captured abnormal traffic data, the type, source, target and other key points of the threat behavior are analyzed, and the threat intelligence data and security analysis tools are used to match the abnormal behavior with the known threat characteristics to determine the possible threat situation. Based on the threat behavior situation awareness data, the structure and logic of the attack behavior are analyzed, the patterns and laws in the abnormal traffic are explored, and the possible motivations and strategies of the attacker are analyzed. Through behavior analysis, the possible goals, means and potential threats of the attacker are identified, and the basis for further countermeasures is provided. For example, after the abnormal behavior monitoring data is processed by traffic capture, it is found that a certain IP address frequently sends a large number of data packets to a certain port of the internal server in a short period of time, and the content of the data packet has obvious malicious characteristics (such as containing malicious code or instructions). After analysis, it is confirmed that the IP address is in the process of launching a remote command execution (Remote Code Execution) attack.

[0086] Step S3: performing attack behavior data association mining based on the attack behavior structure logic data to obtain attack behavior association mining data; performing big data attack sniffing engine design based on the attack behavior association mining data to obtain a big data attack sniffing engine;

[0087] In the embodiment of the present invention, key information in the logical data of the attack behavior structure, such as the attacker's IP address, attack target, attack type, etc., is used to perform data association mining, and an association mining algorithm (such as association rule mining, frequent item set mining, etc.) is used to find the association relationship and pattern between the attack behaviors. According to the mining results, the possible association rules between the attack behaviors and the attacker's potential attack strategy are determined. Based on the attack behavior association mining data, a big data attack sniffing engine is designed for real-time monitoring and detection of attack behaviors in the network. The architecture and components of the sniffing engine are determined, including a data acquisition module, a data processing module, and a rule matching engine. The algorithm and logic of the sniffing engine are developed and implemented, including real-time data stream processing, rule matching, anomaly detection and other functions.

[0088] Step S4: Automated firmware design is performed on the big data attack sniffing engine to obtain the big data attack sniffing automated firmware; the big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis.

[0089] In an embodiment of the present invention, according to the functions and design requirements of the big data attack sniffing engine, an automated firmware design is performed, the architecture and components of the firmware are determined, including the firmware's operating system, driver, and application program, and the functional modules of the firmware are developed and implemented to ensure that it can meet the needs of the sniffing engine and has automated monitoring, detection, and response capabilities. The designed automated firmware is uploaded to the storage space of the cloud platform, and a corresponding deployment and execution environment is established on the cloud platform to perform network information security analysis. The automated task scheduler on the cloud platform is configured to execute the big data attack sniffing automated firmware periodically or in real time, and the results are fed back to the security analysis system.

[0090] The present invention can timely discover abnormal activities in the network, such as potential intrusion attempts, malicious traffic or abnormal data transmission, by obtaining firewall log data and extracting abnormal activity connection addresses. This helps to identify potential security threats early and take corresponding countermeasures. By performing traffic capture processing on abnormal behavior monitoring data, the specific characteristics and traffic patterns of abnormal activities can be deeply understood. This helps to improve the perception of threatening behaviors, including the detection and identification of known and unknown threats. According to the abnormal behavior traffic capture data, threat behavior situation awareness can be performed, that is, the trend and pattern of threat behavior in the network can be analyzed. This helps to form a comprehensive understanding of the network security situation, and timely adjust security policies and measures to cope with the ever-changing threat environment. By performing attack behavior structure logical analysis on threat behavior situation awareness data, the behavior patterns, means and strategies of attackers can be revealed. This helps to understand the attacker's way of thinking and attack chain, and provides a reference for formulating more effective security defense strategies. By performing attack behavior data association mining based on attack behavior structure logical data, the correlation and pattern between different attack behaviors can be identified and analyzed. This helps to deeply understand the attacker's behavior chain and attack strategy, thereby improving the ability to identify and respond to complex attacks. Based on the attack behavior association mining data, the design of a big data attack sniffing engine can realize real-time monitoring and analysis of network traffic and activities. This engine can use big data technology and algorithms to quickly and effectively detect and respond to various types of attack behaviors, including known and unknown attack patterns. The big data attack sniffing engine is designed with automated firmware, and its functions and algorithms can be encapsulated into deployable firmware to achieve automated deployment and management. This can improve the scalability and universality of security solutions, and reduce operation and maintenance costs and complexity. The big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis, which can achieve comprehensive monitoring and management of distributed network environments. The cloud platform provides the advantages of elastic resources and high availability, and can effectively cope with the processing requirements of large-scale network traffic and security events. Therefore, the present invention provides an optimization process made by a network information security analysis method based on big data, which solves the problems of inaccurate perception of threatening behaviors and low sensitivity to early sniffing of attack behaviors in traditional network information security analysis methods based on big data, improves the accuracy of perception of threatening behaviors, and improves the sensitivity to early sniffing of attack behaviors.

[0091] Preferably, step S1 comprises the following steps:

[0092] Step S11: Obtain firewall log data;

[0093] Step S12: Analyze abnormal behavior activities according to the firewall log data to obtain abnormal behavior activity data;

[0094] Step S13: extracting abnormal activity connection addresses from firewall log data according to abnormal behavior activity data to obtain abnormal activity connection addresses;

[0095] Step S14: abnormal behavior data monitoring is performed based on the abnormal activity connection address and the abnormal behavior activity data to obtain abnormal behavior monitoring data.

[0096] In an embodiment of the present invention, a firewall device is set to enable a logging function to capture network traffic data, determine the level and format of the log record, including selecting the type of information to be recorded (such as connection request, data packet information), the storage location and format of the log (such as text file, database), and the log rotation strategy, use a log analysis tool to parse and analyze the firewall log data to identify abnormal behavior activities, analyze key indicators in the log data, such as connection frequency, source IP address, target port, etc., to determine the pattern and characteristics of the abnormal behavior, generate abnormal behavior activity data according to the characteristics of the abnormal behavior, including the time, source IP address, and target port of the abnormal connection request, filter out connection request records related to the abnormal behavior based on the abnormal behavior activity data, extract the source IP address and target IP address in the abnormal connection request as the abnormal activity connection address, deduplicate and count the abnormal activity connection address, so as to facilitate subsequent abnormal behavior data monitoring; monitor the behavior of the abnormal activity connection address, including the frequency of connection, target port, and communication protocol, analyze the difference between the abnormal behavior activity data and the normal behavior, so as to identify potential security threats, and generate abnormal behavior monitoring data according to the monitoring results, including detailed information of the abnormal connection, trend analysis of the abnormal behavior, etc., to provide a basis for subsequent security protection.

[0097] By acquiring firewall log data, the present invention can collect various activities and connection information in the network, provide a data basis for subsequent security analysis and monitoring, and perform abnormal behavior activity analysis on firewall log data, which helps to identify and analyze abnormal activities, including potential attack behaviors, abnormal traffic patterns, or unusual data transmission. This helps to detect potential security threats early, and extracts abnormal activity connection addresses from firewall log data based on abnormal behavior activity data, and can screen out specific connection addresses related to abnormal activities. This helps to focus analysis and monitoring, and increase attention to attack targets. Abnormal behavior data monitoring based on abnormal activity connection addresses and abnormal behavior activity data can monitor and analyze the development trends and patterns of abnormal behaviors in real time. This helps to respond quickly to security incidents and take necessary defensive measures to mitigate possible security risks.

[0098] Preferably, step S2 comprises the following steps:

[0099] Step S21: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data according to the abnormal activity connection address to obtain abnormal behavior traffic capture data;

[0100] Step S22: performing attack program packet capture processing on the abnormal behavior traffic capture data to obtain attack program packet capture data; performing deep packet inspection on the attack program packet capture data to obtain deep packet inspection data;

[0101] Step S23: Perform threat behavior situation awareness based on deep packet inspection data and abnormal behavior traffic capture data to obtain threat behavior situation awareness data;

[0102] Step S24: Performing an attack behavior structure logic analysis on the attack program packet capture data according to the threat behavior situation awareness data to obtain attack behavior structure logic data.

[0103] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0104] Step S21: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data according to the abnormal activity connection address to obtain abnormal behavior traffic capture data;

[0105] In an embodiment of the present invention, a network traffic capture tool (such as tcpdump, Wireshark, etc.) is used to capture the network traffic involved in the abnormal active connection address, and a capture filter is set to capture only the traffic data related to the abnormal active connection address to reduce the amount of data and improve processing efficiency. The captured traffic data may include TCP / IP data packets, UDP data packets, and ICMP data packets to obtain abnormal behavior traffic capture data.

[0106] Step S22: performing attack program packet capture processing on the abnormal behavior traffic capture data to obtain attack program packet capture data; performing deep packet inspection on the attack program packet capture data to obtain deep packet inspection data;

[0107] In an embodiment of the present invention, a packet capture tool (such as Snort, Suricata, etc.) is used to capture data of abnormal behavior traffic and perform packet capture processing. By setting packet capture rules, data packets that may contain attack programs, such as data packets with malicious characteristics and data packets of abnormal size, are screened out, and the attack program data obtained by packet capture is saved as a packet capture file for further analysis and processing. A deep packet inspection tool (such as Snort, Suricata, etc.) is used to perform in-depth analysis on the attack program packet capture data, and the content (code), protocol, header information, etc. of the data packet are analyzed to identify possible attack behaviors or malicious activities. According to the results of the deep packet inspection, deep packet inspection data is generated, including the attack type, attack target, and attacker IP address.

[0108] Step S23: Perform threat behavior situation awareness based on deep packet inspection data and abnormal behavior traffic capture data to obtain threat behavior situation awareness data;

[0109] In an embodiment of the present invention, deep packet inspection data and abnormal behavior traffic capture data are integrated to form comprehensive threat behavior situation awareness data, and abnormal behavior patterns, attack types, attack targets and other information in the data are analyzed to perceive the situation and trend of network threats, conduct timeliness analysis of situation awareness, and identify the current threat status and development trends.

[0110] Step S24: Performing an attack behavior structure logic analysis on the attack program packet capture data according to the threat behavior situation awareness data to obtain attack behavior structure logic data.

[0111] In an embodiment of the present invention, based on threat behavior situation awareness data, a structural logical analysis is performed on the attack program packet capture data, and the logical associations and structural features between the attack behaviors are analyzed, such as the attacker's behavior pattern and the way the attack link is constructed. Data mining and association analysis techniques are used to mine the association rules and patterns between the attack behaviors to form attack behavior structural logical data.

[0112] The present invention performs abnormal behavior traffic capture processing on abnormal behavior monitoring data according to the abnormal activity connection address, and can obtain network traffic data related to abnormal activities. This helps to gain a deep understanding of the nature and characteristics of abnormal behavior, so as to more effectively perform subsequent analysis and processing. The attack program packet capture processing is performed on the abnormal behavior traffic capture data, and potential attack programs and malicious codes can be captured. Through deep packet detection, attack behaviors and malicious behaviors in network traffic can be further analyzed and identified, including the detection of protocols, data packet structures and contents. Threat behavior situation awareness is performed based on deep packet detection data and abnormal behavior traffic capture data, and threat behavior trends and patterns in the network can be monitored and analyzed in real time. This helps to quickly discover and respond to various types of network attacks, and take corresponding security measures. According to the threat behavior situation awareness data, the attack behavior structure logic analysis is performed on the attack program packet capture data, and the attacker's behavior pattern and attack chain can be deeply understood. This helps to discover the attacker's intentions and purposes, and provides a reference for formulating targeted security defense strategies.

[0113] Preferably, step S23 includes the following steps:

[0114] Step S231: performing threat code analysis on the deep packet inspection data to obtain threat code data; performing threat code data structure analysis on the threat code data to obtain threat code structure data; performing code reverse engineering analysis on the threat code structure data to obtain code threat reverse analysis data;

[0115] Step S232: performing dependency association analysis on the threatening code structure data according to the code threat reverse analysis data to obtain threatening code structure dependency association data;

[0116] Step S233: performing process hijacking analysis according to the threatening code structure dependency association data and the code threat reverse analysis data to obtain threatening process hijacking analysis data; performing process hijacking execution command deduction according to the threatening process hijacking analysis data and the code threat reverse analysis data to obtain process hijacking execution command deduction data;

[0117] Step S234: performing a variance balance measurement analysis on the process hijacking execution command deduction data to obtain command deduction variance balance measurement data;

[0118] Step S235: constructing a balanced measurement policy gradient learning model for the command deduction variance balanced measurement data based on the Policy Gradient algorithm to obtain a command deduction balanced measurement policy model;

[0119] Step S236: performing a threat situation gradient analysis on the process hijacking execution command deduction data according to the command deduction balance measurement strategy model to obtain the hijacking command threat situation gradient data;

[0120] Step S237: Perform threat behavior situation awareness according to the hijacking command threat situation gradient data and the abnormal behavior traffic capture data to obtain threat behavior situation awareness data.

[0121] In an embodiment of the present invention, a potentially threatening code sample, such as malware or attack script, is extracted from deep packet inspection data. A malware analysis tool (such as IDAPro, Ghidra, etc.) is used to perform static and dynamic analysis on the extracted code sample to identify the threatening behavior therein, analyze the structural features of the threatening code, including function call relationships, data structures, and control flows, extract key features in the threatening code, such as malicious function modules and abnormal behavior triggering conditions, use reverse engineering technology to decompile and disassemble the threatening code to restore its source code or high-level language representation, analyze the logical structure, algorithm, data processing method, etc. of the code, so as to deeply understand the working principle and behavior of the threatening code, and identify the dependency association relationship in the threatening code based on the results of the code threat reverse analysis, including function call dependency, data dependency, and classification. Analyze the characteristics and patterns of dependency associations to determine the key dependencies and associations in the code; based on the threat code structure dependency association data, identify the key dependencies that lead to process hijacking, such as dependencies on system calls, process control, etc., and combine the code threat reverse analysis data to analyze the hijacking method and implementation logic of the process by the threat code. Based on the threat process hijacking analysis data, deduce the hijacking execution process of the process by the threat code. Use the code threat reverse analysis data to restore the process hijacking execution commands that the threat code may use, such as system calls and process injections. Analyze the deduced process hijacking execution commands, count their execution status and impact in different environments, and evaluate the execution variance of different commands, that is, the difference caused by executing the same command in different systems or environments. According to the variance balance measurement, evaluate the stability and reliability of the command deduction results;Based on the Policy Gradient algorithm, a balanced measurement policy gradient learning model is constructed for the command deduction variance balanced measurement data. The command deduction variance balanced measurement data is collected as a training data set to construct a learning model of the Policy Gradient algorithm. The model will learn how to output a balanced measurement strategy based on the input command deduction variance balanced measurement data. The model is trained using the training data set to optimize the model parameters so that it can more accurately learn the balanced measurement strategy of command deduction. The constructed command deduction balanced measurement strategy model is used to analyze the process hijacking execution command deduction data. According to the balanced measurement strategy output by the model, the threat situation gradient of the hijacking command is evaluated, that is, the degree of influence of the hijacking command on the system security situation. The hijacking command threat situation gradient data is integrated and analyzed with the abnormal behavior traffic capture data. The threat behavior pattern, situation change trend and other information in the data are analyzed to perceive the situation and evolution of network threats. According to the analysis results, comprehensive threat behavior situation awareness data is formed to provide decision support for network security protection. The balanced measurement strategy learning model constructed by the Gradient algorithm learns and optimizes the command deduction variance balanced measurement data to obtain the command deduction balanced measurement strategy model. The model is used to analyze the process hijacking execution command deduction data to obtain the threat situation gradient data of the hijacking command. These data are integrated and analyzed with the abnormal behavior traffic capture data to form comprehensive threat behavior situation awareness data to help the network security team identify and respond to potential network threats in a timely manner. ;

[0122] The present invention can identify and analyze malicious code and attack payloads present in network traffic by performing threatening code analysis and reverse engineering on deep packet detection data. This helps to understand the attacker's means and attack methods, and provides a basis for further security response. Dependent association analysis is performed based on code threatening reverse analysis data, and the correlation between the structure and function of the malicious code can be revealed. This helps to understand the execution logic and possible threatening behaviors of malicious code, analyzes threatening process hijacking analysis data and execution command deduction data, and can identify and understand the attacker's hijacking behavior on the system process, and deduce possible attack commands and operations. This helps to predict the attacker's next move and take corresponding defensive measures. The deduction variance balance measurement data is analyzed, and the degree of difference and risk level between different deduction schemes can be evaluated. This helps to determine the most threatening attack paths and behaviors, and optimize security defense strategies. Based on the Policy Gradient algorithm, a balanced measurement policy learning model is constructed, which can realize dynamic identification and response to threatening behaviors. This helps to establish an adaptive security defense mechanism and improve the ability to respond to new threats. Combining the hijacking command threat situation gradient data and abnormal behavior traffic capture data for threat behavior situation awareness can fully understand the threat behavior trends and patterns in the network. This helps to timely discover and respond to evolving security threats.

[0123] Preferably, performing variance balance measurement analysis on the process hijacking execution command deduction data includes the following steps:

[0124] The execution command is repeatedly deduced on the process hijacking execution command deduction data using the preset execution command deduction count threshold to obtain a hijacking command deduction data set;

[0125] Perform hijacking command difference analysis on the hijacking command deduction data set to obtain hijacking command difference data; perform random two-to-one matching chi-square test on the hijacking command deduction data set based on the hijacking command difference data to obtain deduction difference deviation degree data;

[0126] According to the deduced difference deviation degree data, the confidence interval between the two random hijacking command difference data is calculated to obtain the hijacking command difference confidence interval;

[0127] Based on the deduction difference deviation degree data and the hijacking command difference confidence interval, the deduction variance balance measurement analysis is performed to obtain the command deduction variance balance measurement data.

[0128] In an embodiment of the present invention, a preset execution command deduction count threshold is set. For example, when the count of a certain command in the deduction exceeds the threshold, repeated deduction is performed, the process hijacking execution command deduction data is analyzed, the commands exceeding the threshold are identified, and repeated deduction operations are performed on these commands to obtain more deduction data, analyze the command sequence in the hijacking command deduction data set, identify the differences and change patterns therein, compare the differences between different commands, such as differences in command parameters, execution order, etc., randomly group the commands in the hijacking command deduction data set, match them into pairs one by one, perform a chi-square test on each pair of commands, evaluate the degree of deviation of the deduction difference, that is, judge whether the deduction difference between them is significant, specifically set the execution command deduction count threshold to 10 times, analyze the process hijacking execution command deduction data, and find that the count of a certain command in the deduction exceeds the threshold, so repeated deduction operations are performed, and then, the commands in the deduction data set are analyzed for differences, and the differences between different commands are compared. Then, the commands are randomly grouped, matched into pairs one by one, and a chi-square test is performed on each pair of commands to evaluate the degree of deviation of the deduction difference between them. These steps help to deeply understand the change pattern of hijacking commands and the differences in the deduction process, and provide data support for further threat analysis. The deduction difference deviation degree data is used to calculate the mean and standard deviation of the hijacking command difference data. Based on the calculated mean and standard deviation, the confidence interval between the two random values ​​is calculated using statistical methods (such as Z distribution or T distribution). The deduction difference deviation degree data and the hijacking command difference confidence interval are combined to comprehensively evaluate the variance of command deduction and analyze the variance balance of command deduction, that is, the stability and consistency of command differences in the deduction process. According to the analysis results, the command deduction variance balance measurement data is obtained to evaluate the reliability and consistency level of the deduction process. For example, according to the deduction difference deviation degree data, the mean of the hijacking command difference data is calculated to be 0.75 and the standard deviation is 0.1. The confidence interval is calculated using statistical methods, and the confidence interval at the 95% confidence level is (0.65, 0.85). Next, the confidence interval is combined with the deduction difference deviation degree data to analyze the variance balance of command deduction. If the variance balance measurement data of the inference falls within the confidence interval, it means that the variance of the inference process is relatively stable and reliable. Otherwise, the inference process needs to be further adjusted to improve its consistency level.

[0129] The present invention repeatedly deduces the process hijacking execution command deduction data by utilizing the preset execution command deduction count threshold value to obtain the hijacking command deduction data set. This helps to collect enough deduction data for subsequent analysis and comparison, and performs hijacking command difference analysis on the hijacking command deduction data set to obtain hijacking command difference data. This helps to understand the differences and characteristics between different deduction data. According to the hijacking command difference data, the hijacking command deduction data set is subjected to a random two-to-one matching chi-square test to obtain deduction difference deviation degree data. This helps to determine whether the difference between the deduction data is significant, and quantify the degree of difference. According to the deduction difference deviation degree data, the hijacking command difference data is subjected to a random confidence interval calculation between the two to obtain the hijacking command difference confidence interval. This helps to determine the confidence level and statistical significance of the difference data, and combines the deduction difference deviation degree data and the hijacking command difference confidence interval to perform deduction variance balance measurement analysis to obtain command deduction variance balance measurement data. This helps to evaluate the differences between different deduction schemes and determine the most threatening deduction path.

[0130] Preferably, step S24 includes the following steps:

[0131] Step S241: performing attack behavior context correlation analysis on the attack program packet capture data according to the threat behavior situation awareness data to obtain attack behavior context correlation data;

[0132] Step S242: performing attack behavior topology analysis on the attack behavior contextual correlation data to obtain attack behavior topology data; identifying attack key target nodes according to the attack behavior contextual correlation data and the attack behavior topology data to obtain attack behavior key target node data;

[0133] Step S243: Performing attack behavior structure logic analysis on key target node data according to attack behavior context-related data to obtain attack behavior structure logic data.

[0134] In the embodiment of the present invention, threat behavior situation awareness data and attack program packet capture data are used to analyze the relevant characteristics and behavior patterns of attack behaviors, perform contextual analysis of attack behaviors, identify the association relationship between attack behaviors, such as communication and data transmission between attack programs, build the topological structure of attack behaviors based on contextual data of attack behaviors, display the association and dependency between attack behaviors, analyze the topology of attack behaviors, identify the network structure and topological characteristics between attack behaviors, identify the key target nodes in attack behaviors, that is, important targets or key systems in attacks, analyze contextual data of attack behaviors, evaluate the logical structure and behavior characteristics of key target nodes in attack behaviors, identify the structural logic of attack behaviors, that is, the logical relationship and execution path between each node in attack behaviors. Specifically, the attack program packet capture data is analyzed according to the threat behavior situation awareness data, the characteristics and patterns of attack behaviors are identified, and then contextual analysis of attack behaviors is performed to find the communication relationship between attack programs. Then, based on the above analysis results, the topological structure of attack behaviors is constructed, the key target nodes of attack behaviors are identified, and their logical structure and characteristics in attacks are analyzed. Finally, through the logical analysis of the attack behavior structure, we can understand the logical relationship between the various nodes in the attack behavior and provide guidance for further security protection and response measures.

[0135] The present invention can identify the contextual relationship and mutual influence between attack behaviors by analyzing the threat behavior situation awareness data. This helps to understand the attacker's strategy and means, so as to better deal with and prevent potential security threats. By analyzing the upper and lower associated data of the attack behavior, the topological structure of the attack behavior can be established to reveal the hierarchical relationship and dependency relationship between the attack behaviors. This helps to identify the key nodes and attack paths in the attack chain, and provides an important reference for the formulation of defense strategies. Based on the attack behavior topological data, the key target nodes in the attack chain, i.e. the attacker's main target or the key link on the attack path, can be identified. The identification of these nodes helps to take defensive measures in time to protect the security of key systems and data resources. By performing a logical analysis of the attack behavior structure on the key target node data, the execution logic and internal structure of the attack behavior can be deeply understood. This helps to identify the attacker's behavior pattern and attack strategy, and provides an important reference for formulating targeted security defense strategies.

[0136] Preferably, step S3 comprises the following steps:

[0137] Step S31: performing attack behavior data association mining according to the attack behavior structure logic data and the threat behavior situation awareness data to obtain attack behavior association mining data;

[0138] Step S32: performing attack behavior pattern analysis on the attack behavior association mining data to obtain attack behavior pattern data;

[0139] Step S33: Design a big data attack sniffing engine based on the attack behavior association mining data and the attack behavior pattern data to obtain a big data attack sniffing engine.

[0140] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0141] Step S31: performing attack behavior data association mining according to the attack behavior structure logic data and the threat behavior situation awareness data to obtain attack behavior association mining data;

[0142] In the embodiment of the present invention, the attack behavior structure logic data and the threat behavior situation awareness data are combined to mine the association relationship and pattern between the attack behaviors, and use data mining technology, such as association rule mining or sequence pattern mining, to discover the potential association between the attack behaviors.

[0143] Step S32: performing attack behavior pattern analysis on the attack behavior association mining data to obtain attack behavior pattern data;

[0144] In the embodiment of the present invention, the attack behavior association mining data is analyzed to identify the attack behavior patterns and laws therein, and to explore the association patterns between the attack behaviors, such as common attack combinations or sequences.

[0145] Step S33: Design a big data attack sniffing engine based on the attack behavior association mining data and the attack behavior pattern data to obtain a big data attack sniffing engine.

[0146] In the embodiment of the present invention, the structure and algorithm of a big data attack sniffing engine are designed based on attack behavior association mining data and attack behavior pattern data, the real-time, accuracy and scalability of the engine are considered, an attack detection scheme suitable for a large-scale network environment is designed, the big data attack sniffing engine is implemented and deployed, and performance testing and optimization are performed to ensure that it can effectively identify and prevent various types of attack behaviors.

[0147] The present invention can determine the correlation and dependency between attack behaviors by performing attack behavior contextual correlation analysis on the attack program packet capture data according to the threat behavior situation awareness data. This helps to understand the attacker's attack path and strategy, and discover the hidden relationship between attack behaviors; by performing attack behavior topological analysis on the attack behavior contextual correlation data, the topological structure of the attack behavior can be constructed, and the hierarchical relationship and connection mode between the attack behaviors can be clearly displayed. According to the topological analysis results, key target nodes are identified, that is, the most important nodes in the attack and the nodes that are most concerned by the attacker. This helps to focus analysis and defense forces, improve the level of protection for key target nodes, and perform attack behavior structure logical analysis on key target node data according to the attack behavior contextual correlation data, which can deeply understand the structure and logic of the attack behavior. This helps to reveal the attacker's attack ideas and methods, and provide a reference for formulating corresponding defense strategies.

[0148] Preferably, step S33 includes the following steps:

[0149] Step S331: performing attack load analysis on the attack behavior association mining data to obtain attack behavior load data; performing attack flow fluctuation analysis on the attack behavior association mining data according to the attack behavior load data to obtain attack flow fluctuation data;

[0150] Step S332: analyzing the instantaneous traffic fluctuation growth rate of the attack traffic fluctuation data to obtain instantaneous traffic growth rate data; when the instantaneous traffic growth rate data is less than a preset instantaneous traffic growth threshold, marking the attack traffic fluctuation data in the period as regular access traffic data;

[0151] Step S333: When the instantaneous traffic growth rate data is greater than or equal to the preset instantaneous traffic growth threshold, the attack traffic fluctuation data in the period is marked as risky access traffic data;

[0152] Step S334: performing risk access behavior pattern mapping processing on the attack behavior pattern data according to the risk access traffic data to obtain risk access behavior pattern data;

[0153] Step S335: Perform unauthorized access request frequency analysis on the risk access behavior pattern data to obtain unauthorized access request frequency data;

[0154] Step S336: using a gradient boosting tree algorithm to perform risk behavior feature memory learning on the risk access behavior pattern data and the unauthorized access request frequency data to obtain risk behavior feature memory data;

[0155] Step S337: Design a big data attack sniffing engine based on the risk behavior characteristic memory data to obtain a big data attack sniffing engine.

[0156] In an embodiment of the present invention, the attack behavior load in the attack behavior association mining data is analyzed, that is, the degree of occupation of network or system resources by the attack, and the attack behavior load data, such as the number, size, frequency, etc. of attack data packets, is calculated to evaluate the intensity and impact of the attack. Based on the attack behavior load data, the fluctuation of the attack traffic is analyzed, that is, the changing trend and fluctuation degree of the attack traffic, and the attack traffic fluctuation data is extracted, including the peak value, valley value, and fluctuation period information of the attack traffic. The instantaneous growth rate of the attack traffic fluctuation data is calculated, that is, the changing speed of the attack traffic per unit time, and the changing trend of the instantaneous traffic growth rate is analyzed. The growth of the attack traffic is evaluated, and the attack traffic fluctuation data is compared with a preset instantaneous traffic growth threshold. If it is less than the threshold, it is considered that the attack traffic is at a regular access level and is not risky. If the instantaneous traffic growth rate data reaches or exceeds the preset threshold, it is considered that the attack traffic fluctuation is risky and there is an attack activity. The attack behavior pattern in the risky access traffic data is analyzed to identify Identify the risky access behavior patterns, map the risky access behavior patterns with the original attack behavior pattern data, obtain the behavior pattern data for risky access, analyze the frequency of unauthorized access requests in the risky access behavior pattern data, that is, the frequency of malicious access requests, and count the frequency of different types of unauthorized access requests in a specific time period to evaluate the activity and threat level of the attack. Use the gradient boosting tree algorithm to train and learn the risky access behavior pattern data and the unauthorized access request frequency data. The model learns the characteristics of risky behavior, such as attack mode, frequency, target, etc., to identify and predict future risky behavior. Based on the risk behavior feature memory data learned by the gradient boosting tree model, design the detection algorithm and rules of the big data attack sniffing engine. Considering the real-time and accuracy of the engine, design a detection engine that can effectively identify and prevent various types of attack behaviors, implement and deploy the big data attack sniffing engine, and perform performance testing and optimization to ensure its efficient operation in a large-scale network environment. Specifically, by analyzing the risky access traffic data, it is found that some access behaviors have unauthorized access patterns, and their request frequencies are counted. The gradient boosting tree algorithm is used to learn these data to obtain the feature memory data of risky behavior. Based on these data, a big data attack sniffing engine was designed, which can promptly identify and prevent similar attack behaviors and improve network security.

[0157] The present invention can understand the intensity and frequency of the attack by performing attack load analysis on the attack behavior associated mining data. Performing attack flow fluctuation analysis on the attack behavior associated mining data according to the attack behavior load data can identify the changing trend and pattern of the attack flow, which is helpful to predict the development trend and law of the attack behavior. By analyzing the instantaneous growth rate of the attack flow, the sudden increase of the attack flow can be quickly identified. By comparing the instantaneous flow growth rate with the preset instantaneous flow growth threshold, the conventional access flow and the risk access flow can be effectively distinguished. The attack behavior pattern data is mapped and processed according to the risk access flow data, and the characteristic pattern and law of the attack behavior can be identified. This helps to establish a pattern recognition system for attack behavior, improve the detection and identification capabilities of potential threats, and analyze the frequency data of unauthorized access requests to understand the frequency of unauthorized access by attackers to the system. This helps to discover potential security vulnerabilities and attack risks, and take corresponding security measures in time for defense. The gradient boosting tree algorithm is a powerful machine learning technology that can effectively process complex nonlinear relationships and high-dimensional data. In this step, the gradient boosting tree algorithm is used to learn the risk access behavior pattern data and the frequency data of unauthorized access requests, which can dig out the characteristics and rules hidden behind the data. Through the gradient boosting tree algorithm learning, the risk behavior feature memory data can be obtained, which contains the memory of the key features and patterns of risk access behavior and unauthorized access requests. These data are crucial for subsequent attack detection and analysis, and can help the system identify potential security threats. According to the risk behavior feature memory data, a big data attack sniffing engine can be designed, which can monitor and analyze large-scale network traffic data in real time. The big data attack sniffing engine can identify and intercept potential attack behaviors, including malicious access, unauthorized operations, etc., thereby improving the system's detection and response capabilities to security threats. The designed sniffing engine can analyze network traffic in real time, detect abnormal behaviors in time and take corresponding defense measures. At the same time, based on the risk behavior feature memory data, the engine has a high degree of accuracy and can accurately identify attack behaviors to avoid false positives and false negatives.

[0158] Preferably, performing risk behavior feature memory learning on risk access behavior pattern data and unauthorized access request frequency data comprises the following steps:

[0159] The risk access behavior pattern data and the unauthorized access request frequency data are divided into time series to obtain the risk access pattern training set, the risk access pattern test set, and the unauthorized access request training set and the unauthorized access request test set respectively;

[0160] Perform linear correlation training set feature analysis on the risk access pattern training set and the unauthorized access request training set to obtain the risk-unauthorized linear feature training set;

[0161] Perform random feature sampling on the risk access pattern test set and the unauthorized access request test set to obtain a risk-unauthorized random feature sampling test set;

[0162] The gradient boosting tree algorithm is used to construct the risk-overreaching behavior feature memory model for the risk-overreaching linear feature training set, and the risk behavior feature memory model is obtained;

[0163] The risk-overriding random feature sampling test set is input into the risk behavior feature memory model for risk behavior feature memory learning to obtain risk behavior feature memory data.

[0164] In an embodiment of the present invention, the risk access behavior pattern data and the unauthorized access request frequency data are divided into a training set and a test set according to the time sequence, and the risk access pattern training set and the unauthorized access request training set are subjected to linear correlation feature analysis to identify linear correlation features, and random feature sampling is performed on the risk access pattern test set and the unauthorized access request test set to generate a risk-unauthorized random feature sampling test set, and a gradient boosting tree algorithm is used to construct a model for the risk-unauthorized linear feature training set to generate a risk behavior feature memory model, and the risk-unauthorized random feature sampling test set is input into the risk behavior feature memory model to perform risk behavior feature memory learning to obtain risk behavior feature memory data. Specifically, there are risk access behavior pattern data and unauthorized access request frequency data within a period of time. These data are divided into a training set and a test set according to time. On the training set, the linear correlation features between the risk access behavior pattern data and the unauthorized access request data are analyzed, and feature analysis is performed. Then, random feature sampling is performed on the test set to generate a random feature sampling test set for testing. Next, a risk behavior feature memory model is constructed using the gradient boosting tree algorithm, and the model can identify the features of risk behavior. Finally, the random feature sampling test set was input into the risk behavior feature memory model, and the risk behavior feature memory learning was performed to obtain the risk behavior feature memory data.

[0165] The present invention divides the original data into training set and test set according to the time series by time series division, which can ensure the generalization ability of the model on future data and avoid the problem of overfitting or underfitting of the model. The risk access pattern training set and the unauthorized access request training set can be analyzed to identify the linear features related to risk behavior and unauthorized request. These features include request frequency, access path, user identity, etc., which are helpful to establish a basic risk behavior feature memory model. Random feature sampling on the test set can reduce the computational burden during the test process, and improve the efficiency and speed of model evaluation while maintaining data distribution. The gradient boosting tree algorithm is used to construct a model for the risk-unauthorized linear feature training set, which can establish a model memory of the complex nonlinear relationship between risk behavior and unauthorized request, improve the model's fitting ability and generalization ability for data, and input the risk-unauthorized random feature sampling test set into the risk behavior feature memory model to perform risk behavior feature memory learning. The purpose of this step is to allow the model to learn on the test data to obtain more comprehensive data patterns and features, and further improve the performance and accuracy of the model.

[0166] Preferably, the present invention also provides a network information security analysis system based on big data, which is used to execute the network information security analysis method based on big data as described above, and the network information security analysis system based on big data includes:

[0167] The abnormal behavior monitoring module is used to obtain firewall log data; extract abnormal activity connection addresses from the firewall log data to obtain abnormal activity connection addresses; monitor abnormal behavior data based on the abnormal activity connection addresses to obtain abnormal behavior monitoring data;

[0168] The threat behavior situation awareness module is used to perform abnormal behavior traffic capture processing on abnormal behavior monitoring data to obtain abnormal behavior traffic capture data; perform threat behavior situation awareness based on abnormal behavior traffic capture data to obtain threat behavior situation awareness data; perform attack behavior structure logic analysis based on threat behavior situation awareness data to obtain attack behavior structure logic data;

[0169] A big data attack sniffing engine design module is used to perform attack behavior data association mining based on attack behavior structure logic data to obtain attack behavior association mining data; a big data attack sniffing engine is designed based on attack behavior association mining data to obtain a big data attack sniffing engine;

[0170] The network information security analysis module is used to perform automated firmware design on the big data attack sniffing engine to obtain the big data attack sniffing automated firmware; and send the big data attack sniffing automated firmware to the cloud platform to perform network information security analysis.

[0171] The present invention can timely discover abnormal activities in the network, such as potential intrusion attempts, malicious traffic or abnormal data transmission, by obtaining firewall log data and extracting abnormal activity connection addresses. This helps to identify potential security threats early and take corresponding countermeasures. By performing traffic capture processing on abnormal behavior monitoring data, the specific characteristics and traffic patterns of abnormal activities can be deeply understood. This helps to improve the perception of threatening behaviors, including the detection and identification of known and unknown threats. According to the abnormal behavior traffic capture data, threat behavior situation awareness can be performed, that is, the trend and pattern of threat behavior in the network can be analyzed. This helps to form a comprehensive understanding of the network security situation, and timely adjust security policies and measures to cope with the ever-changing threat environment. By performing attack behavior structure logical analysis on threat behavior situation awareness data, the behavior patterns, means and strategies of attackers can be revealed. This helps to understand the attacker's way of thinking and attack chain, and provides a reference for formulating more effective security defense strategies. By performing attack behavior data association mining based on attack behavior structure logical data, the correlation and pattern between different attack behaviors can be identified and analyzed. This helps to deeply understand the attacker's behavior chain and attack strategy, thereby improving the ability to identify and respond to complex attacks. Based on the attack behavior association mining data, the design of a big data attack sniffing engine can realize real-time monitoring and analysis of network traffic and activities. This engine can use big data technology and algorithms to quickly and effectively detect and respond to various types of attack behaviors, including known and unknown attack patterns. The big data attack sniffing engine is designed with automated firmware, and its functions and algorithms can be encapsulated into deployable firmware to achieve automated deployment and management. This can improve the scalability and universality of security solutions, and reduce operation and maintenance costs and complexity. The big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis, which can achieve comprehensive monitoring and management of distributed network environments. The cloud platform provides the advantages of elastic resources and high availability, and can effectively cope with the processing requirements of large-scale network traffic and security events. Therefore, the present invention provides an optimization process made by a network information security analysis method based on big data, which solves the problems of inaccurate perception of threatening behaviors and low sensitivity to early sniffing of attack behaviors in traditional network information security analysis methods based on big data, improves the accuracy of perception of threatening behaviors, and improves the sensitivity to early sniffing of attack behaviors.

[0172] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0173] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A network information security analysis method based on big data, characterized in that: The following steps are involved: Step S1: Obtain firewall log data; extract abnormal activity connection addresses from the firewall log data to obtain abnormal activity connection addresses; monitor abnormal behavior data based on the abnormal activity connection addresses to obtain abnormal behavior monitoring data; Step S2: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data to obtain abnormal behavior traffic capture data; Perform threat behavior situation awareness based on abnormal behavior traffic capture data to obtain threat behavior situation awareness data; Performing an attack behavior structure logic analysis based on the threat behavior situation awareness data to obtain attack behavior structure logic data; wherein step S2 includes: Step S21: performing abnormal behavior traffic capture processing on the abnormal behavior monitoring data according to the abnormal activity connection address to obtain abnormal behavior traffic capture data; Step S22: performing attack program packet capture processing on the abnormal behavior traffic capture data to obtain attack program packet capture data; performing deep packet inspection on the attack program packet capture data to obtain deep packet inspection data; Step S23: Perform threat behavior situation awareness based on deep packet inspection data and abnormal behavior traffic capture data to obtain threat behavior situation awareness data; wherein step S23 includes: Step S231: performing threat code analysis on the deep packet inspection data to obtain threat code data; performing threat code data structure analysis on the threat code data to obtain threat code structure data; performing code reverse engineering analysis on the threat code structure data to obtain code threat reverse analysis data; Step S232: performing dependency association analysis on the threatening code structure data according to the code threat reverse analysis data to obtain threatening code structure dependency association data; Step S233: performing process hijacking analysis according to the threatening code structure dependency association data and the code threat reverse analysis data to obtain threatening process hijacking analysis data; performing process hijacking execution command deduction according to the threatening process hijacking analysis data and the code threat reverse analysis data to obtain process hijacking execution command deduction data; Step S234: performing variance balance measurement analysis on the process hijacking execution command deduction data to obtain command deduction variance balance measurement data; wherein, performing variance balance measurement analysis on the process hijacking execution command deduction data includes: The execution command is repeatedly deduced on the process hijacking execution command deduction data using the preset execution command deduction count threshold to obtain a hijacking command deduction data set; Perform hijacking command difference analysis on the hijacking command deduction data set to obtain hijacking command difference data; perform random two-to-one matching chi-square test on the hijacking command deduction data set based on the hijacking command difference data to obtain deduction difference deviation degree data; According to the deduced difference deviation degree data, the confidence interval between the two random hijacking command difference data is calculated to obtain the hijacking command difference confidence interval; Based on the deduction difference deviation degree data and the hijacking command difference confidence interval, the deduction variance balance measurement analysis is performed to obtain the command deduction variance balance measurement data; Step S235: constructing a balanced measurement policy gradient learning model for the command deduction variance balanced measurement data based on the Policy Gradient algorithm to obtain a command deduction balanced measurement policy model; Step S236: performing a threat situation gradient analysis on the process hijacking execution command deduction data according to the command deduction balance measurement strategy model to obtain the hijacking command threat situation gradient data; Step S237: Perform threat behavior situation awareness according to the hijacking command threat situation gradient data and the abnormal behavior traffic capture data to obtain threat behavior situation awareness data; Step S24: performing an attack behavior structure logic analysis on the attack program packet capture data according to the threat behavior situation awareness data to obtain attack behavior structure logic data; wherein step S24 includes: Step S241: performing attack behavior context correlation analysis on the attack program packet capture data according to the threat behavior situation awareness data to obtain attack behavior context correlation data; Step S242: performing attack behavior topology analysis on the attack behavior contextual correlation data to obtain attack behavior topology data; identifying attack key target nodes according to the attack behavior contextual correlation data and the attack behavior topology data to obtain attack behavior key target node data; Step S243: performing an attack behavior structure logic analysis on key target node data according to the attack behavior context association data to obtain attack behavior structure logic data; Step S3: performing attack behavior data association mining according to the attack behavior structure logic data to obtain attack behavior association mining data; performing big data attack sniffing engine design according to the attack behavior association mining data to obtain a big data attack sniffing engine; wherein step S3 includes: Step S31: performing attack behavior data association mining according to the attack behavior structure logic data and the threat behavior situation awareness data to obtain attack behavior association mining data; Step S32: performing attack behavior pattern analysis on the attack behavior association mining data to obtain attack behavior pattern data; Step S33: Designing a big data attack sniffing engine based on the attack behavior association mining data and the attack behavior pattern data to obtain a big data attack sniffing engine; wherein step S33 includes: Step S331: performing attack load analysis on the attack behavior association mining data to obtain attack behavior load data; performing attack flow fluctuation analysis on the attack behavior association mining data according to the attack behavior load data to obtain attack flow fluctuation data; Step S332: analyzing the instantaneous traffic fluctuation growth rate of the attack traffic fluctuation data to obtain instantaneous traffic growth rate data; when the instantaneous traffic growth rate data is less than a preset instantaneous traffic growth threshold, marking the attack traffic fluctuation data in the period as regular access traffic data; Step S333: When the instantaneous traffic growth rate data is greater than or equal to the preset instantaneous traffic growth threshold, the attack traffic fluctuation data in the period is marked as risky access traffic data; Step S334: performing risk access behavior pattern mapping processing on the attack behavior pattern data according to the risk access traffic data to obtain risk access behavior pattern data; Step S335: Perform unauthorized access request frequency analysis on the risk access behavior pattern data to obtain unauthorized access request frequency data; Step S336: using a gradient boosting tree algorithm to perform risk behavior feature memory learning on the risk access behavior pattern data and the unauthorized access request frequency data to obtain risk behavior feature memory data; Step S337: Designing a big data attack sniffing engine according to the risk behavior characteristic memory data to obtain a big data attack sniffing engine; Step S4: Automated firmware design is performed on the big data attack sniffing engine to obtain the big data attack sniffing automated firmware; the big data attack sniffing automated firmware is sent to the cloud platform to perform network information security analysis.

2. The network information security analysis method based on big data according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: Obtain firewall log data; Step S12: Analyze abnormal behavior activities according to the firewall log data to obtain abnormal behavior activity data; Step S13: extracting abnormal activity connection addresses from firewall log data according to abnormal behavior activity data to obtain abnormal activity connection addresses; Step S14: abnormal behavior data monitoring is performed based on the abnormal activity connection address and the abnormal behavior activity data to obtain abnormal behavior monitoring data.

3. The network information security analysis method based on big data according to claim 1 is characterized in that: The risk behavior feature memory learning of the risk access behavior pattern data and the unauthorized access request frequency data includes the following steps: The risk access behavior pattern data and the unauthorized access request frequency data are divided into time series to obtain the risk access pattern training set, the risk access pattern test set, and the unauthorized access request training set and the unauthorized access request test set respectively; Perform linear correlation training set feature analysis on the risk access pattern training set and the unauthorized access request training set to obtain the risk-unauthorized linear feature training set; Perform random feature sampling on the risk access pattern test set and the unauthorized access request test set to obtain a risk-unauthorized random feature sampling test set; The gradient boosting tree algorithm is used to construct the risk-overreaching behavior feature memory model for the risk-overreaching linear feature training set, and the risk behavior feature memory model is obtained; The risk-overriding random feature sampling test set is input into the risk behavior feature memory model for risk behavior feature memory learning to obtain risk behavior feature memory data.

4. A network information security analysis system based on big data, characterized in that: Used to execute the network information security analysis method based on big data as claimed in claim 1, the network information security analysis system based on big data includes: The abnormal behavior monitoring module is used to obtain firewall log data; extract abnormal activity connection addresses from the firewall log data to obtain abnormal activity connection addresses; monitor abnormal behavior data based on the abnormal activity connection addresses to obtain abnormal behavior monitoring data; The threat behavior situation awareness module is used to perform abnormal behavior traffic capture processing on abnormal behavior monitoring data to obtain abnormal behavior traffic capture data; perform threat behavior situation awareness based on abnormal behavior traffic capture data to obtain threat behavior situation awareness data; perform attack behavior structure logic analysis based on threat behavior situation awareness data to obtain attack behavior structure logic data; A big data attack sniffing engine design module is used to perform attack behavior data association mining based on attack behavior structure logic data to obtain attack behavior association mining data; a big data attack sniffing engine is designed based on attack behavior association mining data to obtain a big data attack sniffing engine; The network information security analysis module is used to perform automated firmware design on the big data attack sniffing engine to obtain the big data attack sniffing automated firmware; and send the big data attack sniffing automated firmware to the cloud platform to perform network information security analysis.