Database security monitoring method and system based on privacy protection

Abstract

The invention provides a database security monitoring method and system based on privacy protection, database security real-time monitoring is realized by using an operation and maintenance center, an inspection center, a privacy protection center and platform management, and the privacy protection center carries out data desensitization on monitoring data by constructing a desensitization strategy decision model and establishing a desensitization parameter optimizer. According to the method, business requirements and data analysis requirements are comprehensively considered, the self-adaptive data desensitization can better balance the relationship between data security and data availability, and a more comprehensive solution is provided for database operation and maintenance and security management; the distance calculation mode in the desensitization strategy decision model is improved, so that different features have the same scale, the distribution difference between samples is reflected more accurately, and the classification effect is further improved. Besides, according to the data quality function calculation method provided by the invention, the information loss in the data desensitization process is measured more accurately by considering the uncertainty of the original data after the desensitization data is given and the dependency relationship between the data attributes.

Description

Technical Field

[0001] The invention relates to a database security monitoring method and system based on privacy protection, and belongs to the field of database security. Background Art

[0002] Database monitoring systems play a key role in today's information security environment, especially in the face of increasingly complex security threats and strict privacy compliance requirements. Database monitoring platforms based on third parties protect the security of enterprise databases while posing certain risks to the privacy security of internal data. In the prior art, data desensitization technology is an important technical means in the field of data security and privacy protection. It processes sensitive data to reduce the risk level of data, thereby reducing the possibility of data leakage and abuse. In order to adapt to the complex and changing data environment and user needs, the adaptive desensitization strategy decision model can dynamically select appropriate desensitization algorithms based on data characteristics and user needs. However, in the process of constructing the desensitization strategy decision model, the existing desensitization strategy decision model has poor effects in high-dimensional space, which is specifically manifested in the following aspects:

[0003] First, the complexity of high-dimensional data makes it difficult for traditional distance calculation methods such as Euclidean distance to accurately measure the similarities and differences between data. In high-dimensional space, data is sparsely distributed, and traditional distance measurement methods are no longer applicable, resulting in the selection of desensitization strategies deviating from actual needs and failing to accurately meet the requirements of data security and privacy protection.

[0004] Secondly, in business data application scenarios, the dependencies between data features are relatively complex. Existing models ignore the complex dependencies between data features and have deviations in their understanding of the intrinsic structure and patterns of the data, resulting in the desensitization results being unable to accurately reflect the true situation of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] Figure 1 Diagram of a database security monitoring system based on privacy protection. Summary of the invention

[0006] Aiming at the potential threat of third-party database monitoring platform to enterprise database data during deployment, especially how to better provide security services for enterprise database without infringing enterprise data privacy, the present invention proposes a database security monitoring method based on privacy protection, which uses an adaptive data desensitization strategy decision model to achieve intelligent desensitization of sensitive monitoring data, and solves the problems of insufficient accuracy and unsatisfactory classification effect in high-dimensional data space. Specifically, the following steps are included:

[0007] (1) Implementation of operation and maintenance, which provides system monitoring, fault handling, backup and recovery, and security management functions;

[0008] (2) Perform inspections, conduct regular inspections of the system, and provide inspection plans, security scans, and generate inspection reports;

[0009] (3) Construct a desensitization strategy decision model to desensitize the monitoring data by constructing a desensitization strategy decision model and establishing a desensitization parameter optimizer;

[0010] (4) Management platform: provides the overall management and configuration interface of the system and is responsible for user management, permission control and system configuration.

[0011] Furthermore, in step (3), when constructing the desensitization strategy decision model, the objective function is calculated Among them, Max is the value of maximizing the objective function, Acc i is the classification accuracy of the i-th sample, Dist i is the quotient of the distance within the i-th sample class and the distance between classes, C is the total number of categories; M is the number of samples in each category; z c,m is the center point of the feature vector of the mth test sample in the cth class; u c,m is the center point of the feature vector of the mth training sample in the cth class; (z c,m -u c,m ) T , (z j,m -u j,m ) T are the transpose of the center point difference vector between the test sample feature vector and its corresponding training sample feature vector; are the covariance matrices Σ c and Σ j The inverse matrix of .

[0012] Furthermore, in step (3), when establishing the desensitization parameter optimizer, the data quality function is calculated Where X d = {R1, R2, ..., R n} is the original data, Y d =(R1′,R′2,...,R′ n ) is the anonymized data, r is the data set X d The number of attributes of data set X d The number of records.

[0013] Based on the above method, the present invention further proposes a database security monitoring system based on privacy protection, which includes:

[0014] (1) Operation and maintenance center, which provides system monitoring, fault handling, backup and recovery, and security management functions;

[0015] (2) Inspection center, which performs regular inspections on the system and provides inspection plans, security scans, and inspection report generation functions;

[0016] (3) Privacy Protection Center, which desensitizes monitoring data by building a desensitization strategy decision model and establishing a desensitization parameter optimizer;

[0017] (4) Platform management: provides the overall management and configuration interface of the system and is responsible for user management, permission control, and system configuration.

[0018] Furthermore, in the privacy protection center, when constructing the desensitization strategy decision model, the objective function is calculated Among them, Max is the value of maximizing the objective function, Acc i is the classification accuracy of the i-th sample, Dist i is the quotient of the distance within the i-th sample class and the distance between classes, C is the total number of categories; M is the number of samples in each category; z c,m is the center point of the feature vector of the mth test sample in the cth class; u c,m is the center point of the feature vector of the mth training sample in the cth class; (z c,m -u c,m ) T , (z j,m -u j,m ) T are the transpose of the center point difference vector between the test sample feature vector and its corresponding training sample feature vector; are the covariance matrices Σ c and Σ j The inverse matrix of .

[0019] Furthermore, in the privacy protection center, when establishing the desensitization parameter optimizer, the data quality function is calculated Where X d = {R1, R2, ..., R n} is the original data, Y d =(R1′,R′2,...,R′ n ) is the anonymized data, r is the data set X d The number of attributes of data set X d The number of records.

[0020] Compared with the prior art, the technical advantages of the present invention are:

[0021] (1) In the database security monitoring system, by comprehensively considering business needs and data analysis requirements, adaptive data desensitization can better balance the relationship between data security and data availability, and provide a more comprehensive solution for database operation and maintenance and security management.

[0022] (2) Improve the distance calculation method in the desensitization strategy decision model so that different features have the same scale, more accurately reflect the distribution differences between samples, and further improve the classification effect. In addition, the data quality function calculation method proposed in the present invention more accurately measures the information loss in the data desensitization process by considering the uncertainty of the original data after the given desensitized data and the dependency between data attributes. DETAILED DESCRIPTION

[0023] The system of the present invention comprises the following modules:

[0024] 1. Operation and Maintenance Center

[0025] The operation and maintenance center mainly provides daily operation and maintenance support for the system to ensure system stability and availability. Its core functions include system monitoring, fault handling, backup and recovery, and security management.

[0026] (1) System monitoring

[0027] The system monitoring module is responsible for real-time monitoring of the usage of system resources (such as CPU, memory, network, etc.), and monitoring database performance (such as query speed, number of connections, response time, etc.), formulating monitoring indicators, using dynamic threshold algorithms to monitor system anomalies, and setting up an alarm system to promptly notify operation and maintenance personnel when an abnormal situation is detected.

[0028] (2) Troubleshooting

[0029] The fault handling module ensures the normal operation of the system through fault detection and automatic repair functions, while recording and reporting faults, generating detailed fault reports for analysis and improvement, and supporting manual fault handling processes.

[0030] (3) Backup and recovery

[0031] Backup and recovery is responsible for regular and automatic backing up of databases and system configurations, and encrypting and storing backup data to ensure its security; it also provides a fast data recovery mechanism in the event of data loss or system failure.

[0032] (4) Safety management

[0033] Security management is responsible for regularly updating and maintaining system security patches, managing user permissions and access control, and implementing logging and auditing to ensure the traceability of system operations.

[0034] 2. Inspection Center

[0035] The inspection center conducts regular inspections of the system to ensure the security of the system and data and detect potential security vulnerabilities and risks.

[0036] (1) Inspection plan

[0037] The inspection plan module allows you to customize the inspection plan, set the frequency, time and scope of the inspection, and automate the scheduling of inspection tasks.

[0038] (2) Security Scan

[0039] The security scanning module is responsible for regularly scanning databases and systems to detect potential security vulnerabilities and configuration errors, while using security tools to scan for malware, viruses, and other threats.

[0040] (3) Inspection report

[0041] The inspection report module generates inspection reports, records the problems found and treatment suggestions, and automatically sends the reports to relevant personnel for processing and decision-making.

[0042] 3. Privacy Protection Center

[0043] In order to protect the privacy data security of platform users, the Privacy Protection Center is responsible for desensitizing the monitoring data. It performs intelligent desensitization on sensitive data involved in user database monitoring. Its main functions include building a desensitization strategy decision model and establishing a desensitization parameter optimizer.

[0044] (1) Constructing a desensitization strategy decision model

[0045] When selecting a desensitization algorithm, the system will establish a set of decision-making factors to design a desensitization strategy based on user needs and data analysis results. This system covers different levels and types of data, including hierarchical classification of data, specific classification of personal data, overall feature analysis of the entire data set, and in-depth understanding of data attribute characteristics. Based on these factors, the system uses a supervised learning model for training, creates a preliminary desensitization strategy model, and determines preliminary desensitization parameters and strategies.

[0046] In order to improve the accuracy of the desensitization strategy decision model, the present invention designs an improved algorithm based on the small sample multi-classification problem. The objective function based on the small sample multi-classification model is defined as follows:

[0047]

[0048] Among them, Max aims to maximize the value of the objective function, Acc i Indicates the classification accuracy of the i-th class sample, Dist iRepresents the quotient of the intra-class distance and the inter-class distance of the i-th sample. For small sample learning problems, K samples are used for K-fold cross validation. However, the existing distance calculation formula is prone to overfitting in small sample problems, and the measurement of intra-class and inter-class distances is not accurate enough. The traditional Euclidean distance has poor effect in high-dimensional space, and does not consider the covariance relationship between features, resulting in unsatisfactory classification effect. In order to solve this problem, the present invention introduces the Mahalanobis distance to enhance the measurement of intra-class and inter-class distances. The Mahalanobis distance takes into account the covariance structure of the data, so that different features have the same scale, and more accurately reflects the distribution differences between samples. Specifically, the distance calculation formula is as follows:

[0049]

[0050] Where C represents the total number of categories; M represents the number of samples in each category; z c,m represents the center point of the feature vector of the mth test sample in the cth class. This vector contains the values ​​of the sample in all feature dimensions; u c,m The center point of the feature vector of the mth training sample in the cth class can be regarded as the mean of the feature vectors of all training samples in the cth class; (z c,m -u c,m ) T and (z j,m -u j,m ) T They represent the transpose of the difference vector between the test sample feature vector and the center point of its corresponding training sample feature vector. This difference vector represents the degree of deviation between the test sample and the center point of the training sample; and Denote the covariance matrix Σ c and Σ j The inverse matrix of is used to standardize the feature vectors so that different features have the same scale.

[0051] (2) Establish a desensitization parameter optimizer

[0052] The optimizer determines the final parameters based on the privacy risk function and the data quality function. In order to more accurately measure the information loss in the data desensitization process, the uncertainty of the original data after the desensitized data is given and the dependency between data attributes is considered. This paper proposes a new data quality function calculation method:

[0053]

[0054] Among them, the original data X d = {R1, R2, ..., R n}, desensitized data Y d =(R1′,R′2,...,R′ n), r is the dataset X d The number of attributes in the dataset X d The number of records.

[0055] 4. Platform management

[0056] Platform management provides the overall management and configuration interface of the system and is responsible for user management, permission control and system configuration.

[0057] (1) User Management

[0058] The user management module is used to create, delete and manage user accounts, assign user roles and permissions, and ensure that each user can only access the functions and data for which they are authorized. In addition, different user roles and permissions use different desensitization strategies in the privacy protection center.

[0059] (2) Permission Control

[0060] The permission control workload defines and manages roles and permissions in the system, implements fine-grained permission control strategies, and ensures the security of data and functions.

[0061] (3) System Configuration

[0062] The system configuration module provides a system configuration interface that allows administrators to configure system parameters and policies, and supports configuration export and import functions to facilitate system migration and recovery.

[0063] A more specific embodiment:

[0064] The system of the present invention comprises the following parts:

[0065] 1. Operation and Maintenance Center

[0066] The operation and maintenance center mainly handles database resources, performance monitoring, and database anomalies. It uses Prometheus to monitor CPU, memory, network and other resources in real time, and stores the monitoring data in a time series database. It uses the database's built-in monitoring tools to monitor key performance indicators such as query speed, number of connections, and response time, and creates a visual dashboard to display the real-time status of each indicator.

[0067] In addition, the system contains a deep learning-based dynamic threshold algorithm to monitor the abnormal conditions of various indicators. When the indicator exceeds the dynamic threshold, an alarm is triggered and the operation and maintenance personnel are notified. In the fault handling stage, the system has a built-in knowledge base and a large language model to assist the operation and maintenance personnel in troubleshooting the root cause of the fault and completing the fault repair. At the same time, the ELK stack is used to record all system operation logs and regularly generate audit reports to ensure the traceability of system operations.

[0068] 2. Inspection Center

[0069] The inspection center allows administrators to set inspection frequency, time and scope, and set automatic scheduling of inspection tasks and other security scanning tasks. After the inspection is completed, the system automatically generates an inspection report, records the problems found and treatment suggestions, and sends it to relevant personnel.

[0070] 3. Privacy Protection Center

[0071] The Privacy Protection Center intelligently desensitizes sensitive data involved in monitoring the interest-free user database. First, the system identifies sensitive information and uses it as input, including personal identity information, business sensitive information, etc. Based on the data usage scenario and the different users, all sensitive data are graded and classified. For personal identity information, the data needs to be anonymized or pseudonymized, and masking or password-based pseudonymization technology can be used. For business sensitive information, a desensitization algorithm based on differential privacy is used.

[0072] Assume that there is a data set with 3 categories (C=3), each category has 5 samples (M=5), each sample contains 3 feature dimensions (d=3), and its training data and test data are used as input respectively, and the desensitization strategy and data leakage risk assessment are output.

[0073] Specifically, the input sensitive data is divided into personal identity information and business sensitive data, among which business sensitive data is further divided into business data 1 and business data 2. The data set of the desensitization strategy decision model includes training set and test data.

[0074] The specific steps are:

[0075] Step 1: Identify sensitive information, that is, the system identifies personal identity information and business sensitive data, and classifies and rates all sensitive information.

[0076] Step 2: Train the model

[0077] (1) Input the training data set, train the model based on data of different categories, and calculate the distance and classification accuracy of each category. Assume that the feature vector of the training data is as follows:

[0078] Business data 1: u 1,1 =(1,2,1),u 1,2 =(1.1,2.1,1.2),u 1,3 =(0.9,1.9,0.8),u 1,4 =(1.2,2.2,1.1),u 1,5 =(1,2,1);

[0079] Business data 2: u 2,1 =(3,3,3),u 2,2 =(3.1,3.1,3.1),u 2,3=(2.9,2.9,2.8),u 2,4 =(3,3.2,3.1),u 1,5 =(3,3,3);

[0080] Personally Identifiable Information: 3,1 =(5,5,5),u 3,2 =(4.9,4.9,5.1),u 3,3 =(5.1,5.1,5),u 3,4 =(5,5.2,5.2),u 3,5 =(5,5,5);

[0081] Assuming that the covariance matrix of each category is not the identity matrix I, calculate the business data 1

[0082]

[0083] Similarly, the Dist of business data 2 and personal identity information is calculated, Dist2 = 0.345, Dist3 = 0.158. Assume that the classification accuracy Acc of each category is Acc1 = 0.9, Acc2 = 0.85, Acc3 = 0.8.

[0084] Step 3: Determine the desensitization algorithm and calculate

[0085] Step 4: Test data input, assuming the test data is:

[0086] z 1,1 =(1,2,1.1), z 2,1 =(3,3.1,3), z 3,1 =(5.1,5.1,5).

[0087] Step 5: Calculate your risk of a data breach

[0088] Assume that in the generalization / noise case, let Pr(R e (l),R e ′)=0.1,Pr(R e (l) | R e ')=0.8. DQ(X d ,Y d )=-5×3×0.1log20.8=0.48285.

[0089] Output: According to the calculation results, the optimization parameters are the trade-off between privacy risk and data quality. The parameters that minimize privacy risk and maximize data quality are selected as output. Through the above calculation and decision-making process, the desensitization strategy is determined, and finally the intelligent desensitization strategy model selects the appropriate desensitization algorithm to desensitize sensitive data and conducts risk assessment on the calculation results.

[0090] 4. Platform management

[0091] There are 6 roles in platform management: system administrator, database administrator, security administrator, data analyst, developer and ordinary user. System administrator has full system configuration authority and is sensitive to all sensitive data in the user database, such as user ID, password, etc. Database administrator is also sensitive to user personal identity information, but is allowed to view anonymized data related to performance. Security administrators are forced to desensitize all users' personal sensitive data and are only allowed to view desensitized security logs and alarm information; data analysts use partial desensitization strategies to blur personal identity information, such as using general labels instead of specific data; ordinary users are only allowed to access desensitized data related to themselves.

[0092] By integrating the above functional modules into the database monitoring platform, comprehensive monitoring and protection of sensitive data in the database can be achieved, ensuring that user personal privacy and business data are fully protected when using and sharing data, and being able to dynamically adjust and optimize desensitization strategies as needed to improve the overall security and compliance of the database.

[0093] The units, devices or modules described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described separately by functions divided into various modules. Of course, when implementing this application, the functions of each module can be implemented in the same or more software and / or hardware, or the modules that implement the same function can be implemented by a combination of multiple sub-modules or sub-units. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0094] Those skilled in the art also know that, in addition to implementing the controller in a purely computer-readable program code, the controller can be made to implement the same function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, such a controller can be considered as a hardware component, and the devices for implementing various functions included therein can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software modules for implementing the method and structures within the hardware component.

[0095] The present application may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc. that perform specific tasks or implement specific abstract data types. The present application may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0096] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly contributed to the prior art in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0097] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.

[0098] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A database security monitoring method based on privacy protection, characterized in that: The following steps are involved: (1) Implementation of operation and maintenance, which provides system monitoring, fault handling, backup and recovery, and security management functions; (2) Perform inspections, conduct regular inspections of the system, and provide inspection plans, security scans, and generate inspection reports; (3) Construct a desensitization strategy decision model to desensitize the monitoring data by constructing a desensitization strategy decision model and establishing a desensitization parameter optimizer; (4) Management platform: provides the overall management and configuration interface of the system and is responsible for user management, permission control and system configuration.

2. A database security monitoring method based on privacy protection as claimed in claim 1, characterized in that: In step (3), when constructing the desensitization strategy decision model, the objective function is calculated Among them, Max is the value of maximizing the objective function, Acc i is the classification accuracy of the i-th sample, Dist i is the quotient of the distance within the i-th sample class and the distance between classes, C is the total number of categories; M is the number of samples in each category; z c,m is the center point of the feature vector of the mth test sample in the cth class; u c,m is the center point of the feature vector of the mth training sample in the cth class; (z c,m -u c,m ) T , (z j,m -u j,m ) T are the transpose of the center point difference vector between the test sample feature vector and its corresponding training sample feature vector; are the covariance matrices Σ c and Σ j The inverse matrix of .

3. A database security monitoring method based on privacy protection as claimed in claim 1, characterized in that: In step (3), when establishing the desensitization parameter optimizer, the data quality function is calculated Where X d = {R1, R2, ..., R n } is the original data, Y d =(R′1,R′2,...,R′ n ) is the anonymized data, r is the data set X d The number of attributes of the dataset X d The number of records.

4. A database security monitoring system based on privacy protection, characterized in that: The system includes: (1) Operation and maintenance center, which provides system monitoring, fault handling, backup and recovery, and security management functions; (2) Inspection center, which performs regular inspections on the system and provides inspection plans, security scans, and inspection report generation functions; (3) Privacy Protection Center, which desensitizes monitoring data by building a desensitization strategy decision model and establishing a desensitization parameter optimizer; (4) Platform management: provides the overall management and configuration interface of the system and is responsible for user management, permission control, and system configuration.

5. A database security monitoring system based on privacy protection as claimed in claim 4, characterized in that: In the Privacy Protection Center, when building a desensitization strategy decision model, calculate the objective function Among them, Max is the value of maximizing the objective function, Acc i is the classification accuracy of the i-th sample, Dist i is the quotient of the distance within the i-th sample class and the distance between classes, C is the total number of categories; M is the number of samples in each category; z c,m is the center point of the feature vector of the mth test sample in the cth class; u c,m is the center point of the feature vector of the mth training sample in the cth class; (z c,m -u c,m ) T , (z j,m -u j,m ) T are the transpose of the center point difference vector between the test sample feature vector and its corresponding training sample feature vector; are the covariance matrices Σ c and Σ j The inverse matrix of .

6. A database security monitoring system based on privacy protection as claimed in claim 4, characterized in that: In the Privacy Protection Center, when establishing a desensitization parameter optimizer, calculate the data quality function Where X d = {R1, R2, ..., R n } is the original data, Y d =(R′1,R′2,...,R′ n ) is the anonymized data, r is the data set X d The number of attributes of data set X d The number of records.

Images