Dynamic data masking method and system, medium, and processor

Abstract

The present application provides a dynamic data masking method, comprising the following steps: intercepting a data request sent by a client; forwarding the data request of the client to a database server at a back end; capturing response data sent by the database server, the response data being a result returned after the database server processes the data request of the client; analyzing the returned response data, using a machine learning algorithm to establish a dynamic sensitive data identification rule, and identifying sensitive data in the response data; performing real-time dynamic masking processing on the identified sensitive data on the basis of a configured masking strategy; and sending the masked data back to the client. By establishing a dynamic sensitive data identification rule, the level of intelligence and specialization of masking systems is improved. Dynamic data masking technology allows immediate masking during data access, reducing the processing and storage requirements for original data.

Description

Dynamic data desensitization method, system, medium and processor TECHNICAL FIELD

[0001] The present application relates to the technical field of data desensitization, in particular to a dynamic data desensitization method, system, medium and processor. BACKGROUND

[0002] With the development of the Internet, the widespread application of cloud computing, big data and artificial intelligence technology, communication operator production systems have massive sensitive data. For example, enterprise business sensitive data such as financial reports, strategic documents and product pricing policies, as well as personal privacy data such as user name, location, home address and identity card number. If the operator leaks, tampers or other security problems occur during the development test and business operation of these sensitive data, it will not only have a serious impact on the company and the industry, but also cause irreparable loss to the country. In particular, with the gradual development of 5G business, the security problem of sensitive data is more prominent. There are many solutions to solve the security problem of sensitive data, and the most commonly used is data desensitization technology.

[0003] Traditional data desensitization technology focuses on the final result, affecting the user's usage habits, and lacks scene diversification prediction in the process. Artificial identification of sensitive data is required for massive data, and the desensitization system is low in intelligence and professionalism.

[0004] Therefore, a dynamic data desensitization method, system, medium and processor are needed.

[0005] SUMMARY

[0006] In view of the low intelligence and professionalism of the desensitization system in the prior art, the present application provides a dynamic data desensitization method, system, medium and processor, which can establish dynamic sensitive data identification rules, improve the intelligence and professionalism of the desensitization system, and allow instant desensitization when data is accessed through real-time dynamic desensitization processing, reducing the processing and storage requirements of the original data. The specific technical solutions are as follows:

[0007] A dynamic data desensitization method applied to a proxy server, the proxy server being located on the data link between a client and a database server; comprising the following steps:

[0008] Intercepting a data request sent by the client;

[0009] Forwarding the data request of the client to the database server in the back end;

[0010] Capturing response data sent by the database server, the response data being the result returned by the database server after processing the data request of the client;

[0011] analyzing the returned response data, using a machine learning algorithm to establish a dynamic sensitive data identification rule, and identifying sensitive data in the response data;

[0012] According to the configured desensitization strategy, the identified sensitive data is dynamically desensitized in real time;

[0013] The desensitized data is sent back to the client.

[0014] Further, the use of machine learning algorithm to establish dynamic sensitive data identification rule includes the following steps:

[0015] Sampling the content in the big data cluster and synchronizing the sampling data to the sample library;

[0016] Using clustering or classification analysis techniques to convert sample data into ordered samples with features;

[0017] Using semantic similarity analysis, rule analysis or dictionary analysis techniques to divide the obtained ordered samples into several data models according to content and meaning;

[0018] Classify the classified data model and finally form a dynamic sensitive data identification rule and classification standard.

[0019] Further, when the identified sensitive data is desensitized in real time according to the configured desensitization strategy, different desensitization rules are applied according to data features, context information or visitor identity and permission, so that users can only access data that meets their permission level.

[0020] Further, the different desensitization rules according to the different identities and permissions of the visitors include the following steps:

[0021] First, define different identity features according to different roles and responsibilities of users;

[0022] According to the identity features, the sensitive data is hidden, shielded, encrypted or audited desensitized to different degrees, so that different levels of users must access the sensitive data according to their different identity features, and cannot modify any sensitive data.

[0023] Further, the data features include name, ID number, bank account, address, phone number, enterprise name or organization code.

[0024] Further, the dynamic desensitization process includes partial masking, reversible desensitization, mixed desensitization, synonymous replacement or deterministic desensitization.

[0025] Further, the dynamic data desensitization method further includes the following steps:

[0026] On the basis of identifying sensitive data, the adjustment of the sensitive data relationship is completed to ensure the correlation of the data.

[0027] A dynamic data desensitization system applied to the dynamic data desensitization method described above, comprising:

[0028] An interception module for intercepting a data request issued by a client;

[0029] A forwarding module for forwarding the data request of the client to a database server of the backend;

[0030] A capture module for capturing response data issued by the database server, the response data being a result returned after the database server processes the data request of the client;

[0031] An analysis module for analyzing the returned response data, using a machine learning algorithm to establish a dynamic sensitive data identification rule, and identifying sensitive data in the response data;

[0032] A desensitization module for performing real-time dynamic desensitization processing on the identified sensitive data according to a configured desensitization strategy;

[0033] A sending module for sending the desensitized data back to the client.

[0034] A computer-readable storage medium comprising a stored program, wherein the computer-readable storage medium controls the device where the computer-readable storage medium is located to perform the dynamic data desensitization method as described above when the program is running.

[0035] A processor for running a program, wherein the program performs the dynamic data desensitization method as described above when the program is running.

[0036] Compared with the prior art, the beneficial effects of the present application are:

[0037] 1. The dynamic sensitive data identification rule can be established to improve the intelligent and professional level of the desensitization system.

[0038] 2. The dynamic data desensitization technology allows instant desensitization when data is accessed, reducing the processing and storage requirements of the original data.

[0039] 3. Through the desensitization technology, sensitive data is converted into a non-sensitive form during storage and transmission, reducing the risk of data leakage. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the specific embodiments of the present application or the prior art, the drawings needed to be used in the specific embodiments or prior art description will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference signs. In the drawings, the elements or parts are not necessarily drawn according to the actual proportions.

[0041] Fig. 1 is a flow diagram of a dynamic data de-sensitization method;

[0042] Fig. 2 is a flow diagram of data de-sensitization according to visitor identity and permission;

[0043] Fig. 3 is a flow diagram of a Storm-based stream data de-sensitization processing method;

[0044] Fig. 4 is a flow diagram of a batch data de-sensitization processing method. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of the present application.

[0046] It should be understood that the terms "comprise" and "include" as used in the specification and the appended claims indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0047] It should also be understood that the terms used in the present application specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0048] It should be further understood that the term "and / or" as used in the present application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0049] Embodiment One

[0050] In the era of big data, all data needs to be efficiently fused to maximize the benefits of data. In the process of efficient data fusion (i.e. development, testing, production, application, etc.), how to ensure the security of data will become an important task for information security management, operation and maintenance departments. With the rapid development of informationization of State Grid Corporation of China, various application systems are widely used. In order to provide services to the outside world and fully play its use value, it is necessary to effectively interface with external systems. However, the interface will bring potential security risks to data, and sensitive data needs to be desensitized. The specific data desensitization requirements include:

[0051] (1) Prevent sensitive data in the production library from being leaked. The important sensitive information such as user card number, address, identity, mobile phone number in the production library is disturbed and confused, and then the related data is provided to the third party, so as to prevent the sensitive data in the production library from being leaked.

[0052] (2) Ensure the relevance of data in the development, testing, application and other stages. By using data desensitization strategy and data desensitization algorithm, the integrity (i.e. the meaning of data does not appear to be lost, the length of data does not change), effectiveness (the business format and data type of data do not change), and relationship (the correlation between the data in the table and the data between the tables do not change) of the desensitized data are ensured, so as to improve the usability and authenticity of data in the development, testing, application and other stages.

[0053] (3) Ensure the security of data sharing and data maintenance. By formulating and implementing different access strategies and access permissions for different database accessers, and monitoring the usernames, total access time, tool types and IP addresses used by these accessers throughout the process, differential management and control are realized to ensure the security of the accessed data.

[0054] (4) Ensure the compliance of sensitive data management policy. The desensitization of sensitive data, the processing of sensitive data and the desensitization rules must comply with the requirements of data management of State Grid Corporation of China, and must be carried out under the conditions allowed by the relevant policies and regulations of State Grid Corporation of China.

[0055] Data desensitization can be divided into static data desensitization and dynamic data desensitization.

[0056] Static data desensitization refers to desensitizing and de-privatizing data files while ensuring the correlation between data; then sending the data to a third-party company for development testing or data analysis, and after obtaining the analysis results, the analyzed data can be traced back. This desensitization method is suitable for project development units that need to obtain complete data to ensure the smooth completion of data analysis work, and for data providers who do not want sensitive data to be leaked out. In this case, a traceable desensitization method is needed to desensitize the data to ensure that the data sent out does not contain sensitive information. When the project development is completed, the analysis system or result data is traced back to the real source data, which ensures data sharing and result consistency during development, and ensures that real data will not be leaked during development.

[0057] Dynamic data desensitization refers to desensitizing sensitive data in the background database server by the user in the front-end client application, and then presenting the desensitized data on the front-end client. At the same time, the sensitive data of each business system can be desensitized in a transparent and real-time manner by deploying a proxy on the communication layer, that is, a computer data processing system including a client, a database server and a proxy server, the proxy server is located on the data link between the client and the database server, and is connected with the client and the database server respectively; the proxy server is used to desensitize the data when the client accesses the data in the database server and then send it to the client, so as to avoid the security risk of data leakage caused by the client accessing sensitive information.

[0058] As shown in FIG. 1, it is a flowchart of a dynamic data desensitization method, which is applied to the proxy server mentioned above, and the proxy server is located on the data link between the client and the database server; including the following steps:

[0059] Deploying a proxy server on the data transmission path, usually between the client and the database server;

[0060] Intercepting the data request sent by the client, including the query request;

[0061] Forwarding the data request of the client to the database server in the back end;

[0062] Capturing the response data sent by the database server, the response data being the result returned by the database server after processing the data request of the client;

[0063] Analyzing the returned response data, using a machine learning algorithm to establish a dynamic sensitive data identification rule, and identifying the sensitive data in the response data;

[0064] According to the configured desensitization strategy such as masking, replacement, encryption, etc., the identified sensitive data is dynamically desensitized in real time;

[0065] The desensitized data is sent back to the client, and the data received by the client has been processed and does not contain sensitive information.

[0066] The application scheme can establish dynamic sensitive data identification rules, and improve the intelligentization and specialization level of the desensitization system. At the same time, the dynamic data desensitization technology allows instant desensitization during data access, reducing the processing and storage requirements of the original data. Using the application scheme, the work of the proxy server is transparent to the client and the database server, and will not change the existing data access mode and application logic. The proxy server can effectively protect sensitive data without changing the existing database and application architecture, while ensuring data availability and compliance. The proxy server needs to efficiently process data to complete the desensitization task with minimal delay, ensuring that the user experience is not affected. The design of the proxy server needs to consider security to prevent becoming a target of attack, and needs to comply with data protection regulations. The proxy server records logs of all data access and desensitization operations to facilitate auditing and monitoring of data access behavior.

[0067] In specific implementation, the use of machine learning algorithm to establish dynamic sensitive data identification rules includes the following steps:

[0068] Sampling the content in the big data cluster and synchronizing the sampled data to the sample library;

[0069] Using clustering or classification analysis technology to convert sample data into ordered samples with features;

[0070] Using semantic similarity analysis, rule analysis or dictionary analysis technology to divide the obtained ordered samples into several data models according to content and meaning;

[0071] Classifying the classified data models and finally forming a dynamic and targeted sensitive data identification rule and classification standard.

[0072] In specific implementation, when the identified sensitive data is desensitized in real time according to the configured desensitization strategy, different desensitization rules are applied according to data features, context information, or different access identities and permissions, to ensure that users can only access data that meets their permission level.

[0073] Further, as shown in FIG. 2, the proxy server can apply different desensitization rules according to different user permissions and roles to ensure that users can only access data that meets their permission level. The different desensitization rules according to different access identities and permissions include the following steps:

[0074] For each type of dynamic data returned in the production database (a database server), first define different identity characteristics according to the different roles and responsibilities of the users;

[0075] According to the identity characteristics, sensitive data is hidden, shielded, encrypted or audited desensitized to different degrees, so that users of different levels must access sensitive data according to their different identity characteristics, and cannot modify any type of sensitive data.

[0076] In specific implementation, the dynamic desensitization processing includes partial masking, reversible desensitization, hybrid desensitization, synonym replacement or deterministic desensitization. Generally, different user identity characteristics can be desensitized by using corresponding data dynamic desensitization algorithms. Different dynamic desensitization technologies have different characteristics and are suitable for different scenes and needs. The following is a brief description of partial masking, reversible desensitization, hybrid desensitization, synonym replacement and deterministic desensitization:

[0077] (1) Partial Masking (Partial Masking):

[0078] This method only masks part of the data, for example, for a credit card number, only the last four digits are displayed, and the rest are replaced with stars or similar placeholders, such as "************1234".

[0079] (2) Reversible Data Masking (Reversible Data Masking):

[0080] Reversible desensitization technology allows sensitive data to be converted into a reversible format, which means that the original data can be recovered through some algorithm or key. This is often used in scenarios where the original format of the data needs to be preserved for analysis but also needs to be protected.

[0081] (3) Hybrid Masking (Hybrid Masking):

[0082] Hybrid desensitization combines multiple desensitization technologies, such as partial masking and data replacement, to provide more comprehensive protection. This method can flexibly select the most suitable desensitization strategy according to the sensitivity of the data and the needs of the access.

[0083] (4) Synonym Replacement (Synonym Replacement):

[0084] Synonym replacement is a method of replacing sensitive data with synonyms or similar terms, but this method may not change the actual meaning of the data, so it may not be suitable for scenarios that require strict protection of personal privacy.

[0085] (5) Deterministic Data Masking:

[0086] Deterministic data masking means that for the same original data, each time the data masking process will produce the same result. This helps in testing and development in environments where data consistency needs to be maintained, as the same input will always produce the same output.

[0087] Each data masking technique has its applicable scenarios and advantages and disadvantages, and when choosing a data masking strategy, the sensitivity of the data, business needs, compliance requirements, and the usability of the data after data masking need to be considered. For example, reversible data masking can be very useful in scenarios where data needs to be analyzed but privacy needs to be protected, while partial masking is suitable for scenarios where data needs to be quickly identified but the complete content of the data is not required.

[0088] In specific implementations, the data features include name, certificate number, bank account, address, phone number, enterprise name, or organization code.

[0089] In specific implementations, data masking can also be divided into streaming data masking and batch data masking.

[0090] Streaming data refers to data that is continuously generated, calculated in real time, dynamically increased, and requires timely response. It has the characteristics of mass and real-time, and the real-time or quasi-real-time data processing technology is generally classified as streaming data processing technology, including Apache Storm, Spark Streaming, etc. As shown in FIG. 3, Storm is a distributed, reliable, fault-tolerant data stream processing system. The input stream of the cluster is managed by the spout component, that is, after the spout transmits data to the bolt, the bolt transmits data to other bolts or saves data to some storage, and a Storm cluster is a series of bolts that convert data transmitted by the spout. Since the data processing method of Storm is incremental real-time processing, the data desensitization module should have the function of incremental data desensitization. When the data has not been completely transmitted, the desensitization module can be used to read historical data (historical data needs to be read to restore the previous state to ensure the continuity and consistency of data processing) and combine the corresponding algorithm to desensitize the data, remove sensitive words, and generalize the data according to the desensitization rules. The advantage of streaming data desensitization is that data processing starts from the moment of data transmission, and the disadvantage is that it cannot use full data for complex correlation processing. For micro-batch processing, state calculation is required, and only one delivery is allowed, but high latency is not considered. SparkStreaming can be used for streaming data desensitization. If the data user performs machine learning, graph operation or accesses the SQL database, the stack in Apache Spark will combine the data stream with some library to provide the data user with a convenient and integrated programming model.

[0091] As shown in FIG. 4, batch data access refers to that the data source comes from a stable and basically unchanged storage medium, and the data is collected to the big data platform by data scanning at one time. The data is mainly historical data, and the data source generally comes from files, relational databases, NoSQL databases, etc. The processing technology includes Flume, Sqoop, etc. Batch data desensitization can be desensitized during data import, or desensitized after data enters the big data platform by calling the desensitization program module. Batch data desensitization can combine the correlation of the data and use complex desensitization algorithms to achieve better desensitization effect. Based on the data collection method of Flume, the interceptor can be written to call the data desensitization program in the interceptor, and output the desensitized data. Sqoop is suitable for data collection of relational databases, and can be used to establish an intermediate table, write a user-defined function (User-Defined Function, UDF) and program, and finally perform batch data desensitization through a task scheduling program.

[0092] In the embodiment, the dynamic data desensitization method further comprises the following steps: on the basis of identifying the sensitive data, adjusting the sensitive data relationship to ensure the correlation of the data.

[0093] Therefore, the process of implementing data desensitization in the proxy server generally includes four steps: sensitive data discovery, sensitive data analysis, desensitization scheme development, and desensitization task execution. In combination with the rules, algorithms, and environment of data desensitization, the best data desensitization effect is obtained. The detailed introduction of each process is as follows:

[0094] Sensitive data discovery: sensitive data discovery includes manual discovery and automatic discovery. For relatively fixed business data of the power grid company, manual screening can be used to specify which columns and which data in the library need to be desensitized. Generally, the structure and length of such data will not change, and most of them are numerical and fixed-length characters. For example, unit code, account number, account name, and electricity address identification column. For these data, manual desensitization rules and different data access strategies can be used to ensure that sensitive information is not leaked. In addition, with the help of sensitive data information library and word segmentation system, sensitive information contained in the database can be automatically identified, which can reduce the workload and prevent omissions compared with manual identification.

[0095] Sensitive data analysis: on the basis of discovering sensitive data, the relationship of sensitive data is adjusted to ensure the correlation of the data. Through various data desensitization algorithms such as deformation, shielding, replacement, randomization, strong encryption, format retention encryption, etc., data masking is performed for different data types.

[0096] Desensitization scheme development: for different data desensitization needs, on the basis of the basic desensitization algorithm, a special desensitization strategy can be configured. The development of the desensitization scheme mainly relies on the reuse of the desensitization strategy and the desensitization algorithm, and the optimal scheme is developed by configuring and expanding the desensitization algorithm.

[0097] Desensitization task execution: the start, stop, and pause operations of the desensitization task support the interruption or continuation of the desensitization task, and the parallel processing of the desensitization task.

[0098] Selection of data desensitization algorithm: different data desensitization algorithms are selected according to different data characteristics. For example, when performing data desensitization on names, ID numbers, bank accounts, addresses, phone numbers, enterprise names, and organization codes, the data desensitization algorithms generally include deformation, shielding, randomization, replacement, strong encryption, and format retention encryption.

[0099] Formulation of data desensitization rules: Data desensitization rules can be generally divided into two categories: recoverable rules and non-recoverable rules. Recoverable rules are rules that can restore sensitive data after desensitization to the original sensitive data through certain means, such as various encryption and decryption algorithms. Non-recoverable rules are rules that cannot restore sensitive data after desensitization to the original sensitive data using any means, such as the two data desensitization algorithms of generation algorithm and replacement algorithm. Common desensitization rules are shown in Table 1.

[0100] Table 1

[0101] Selection of data desensitization environment: Data desensitization environments can be divided into two categories: production environment data desensitization and non-production environment data desensitization. If classified by application scenarios, it can be divided into dynamic data desensitization and static data desensitization. The biggest difference between the two is whether data desensitization is being performed when sensitive data is being used.

[0102] The application of the above-mentioned dynamic data desensitization method in power system enterprises can bring many benefits:

[0103] (1) Enhance data security:

[0104] Through desensitization technology, sensitive data is converted into non-sensitive form during storage and transmission, reducing the risk of data leakage.

[0105] (2) Protect personal privacy:

[0106] Personal sensitive information (such as name, address, ID number, etc.) is processed through desensitization to ensure that personal privacy is not accessed by unauthorized persons.

[0107] (3) Improve customer trust:

[0108] By implementing data desensitization measures, enterprises show respect and protection for customer privacy, thereby enhancing customer trust and loyalty.

[0109] (4) Reduce the cost of violations:

[0110] By reducing the risk of data leakage, enterprises can avoid high fines and compensation for violating data protection regulations.

[0111] (5) Promote data sharing:

[0112] Desensitized data can be shared without exposing sensitive information, supporting cross-department or cross-organizational data cooperation.

[0113] (6) Support secondary use of data:

[0114] Data de-sensitization allows data to be used for other purposes such as testing, development, analysis, etc., without worrying about leaking sensitive information.

[0115] (7) Improve data processing efficiency:

[0116] Dynamic data de-sensitization technology allows real-time de-sensitization when data is accessed, reducing the processing and storage requirements of original data.

[0117] (8) Optimize data backup and recovery:

[0118] Using de-sensitization technology during data backup and recovery can ensure the security of backup data and reduce the risk during the recovery process.

[0119] (9) Enhance the auditability of data:

[0120] The recording and logging of de-sensitization operations can improve the transparency of data operations and facilitate security audits and monitoring.

[0121] (10) Promote data-driven decision-making:

[0122] De-sensitized data can be used for analysis and reporting to help businesses make data-based decisions while protecting sensitive information.

[0123] Embodiment two

[0124] A dynamic data de-sensitization system applied to the dynamic data de-sensitization method described above, comprising:

[0125] An interception module for intercepting data requests sent by a client;

[0126] A forwarding module for forwarding the client's data request to a database server in the backend;

[0127] A capture module for capturing response data sent by the database server, which is the result returned after the database server processes the client's data request;

[0128] An analysis module for analyzing the returned response data, using machine learning algorithms to establish dynamic sensitive data identification rules and identifying sensitive data in the response data;

[0129] A de-sensitization module for real-time dynamic de-sensitization processing of identified sensitive data according to the configured de-sensitization strategy;

[0130] A sending module for sending the de-sensitized data back to the client.

[0131] Through the construction of the data desensitization system, the establishment of the desensitization algorithm library based on the big data platform is realized, and the sensitive data privacy can be desensitized efficiently and in parallel according to the data desensitization rules; a measurement model is established for the user privacy data according to the data desensitization theory, so that the risk of possible leakage of sensitive data can be quantitatively, qualitatively and accurately measured. The access mechanism based on approval for the privacy data is established by combining the permission management system, the user authentication system and the permission management system of the privacy data under different protection levels of the big data platform, and the value of data mining and analysis is maximized under the premise of minimizing the risk of sensitive data leakage and protecting the user privacy data in combination with various rules and regulations, laws and regulations of the power grid company.

[0132] Embodiment three

[0133] A computer readable storage medium comprising a stored program, wherein the computer readable storage medium controls the device where the computer readable storage medium is located to execute the dynamic data desensitization method as described above when the program is running.

[0134] Embodiment four

[0135] A processor for running a program, wherein the processor executes the dynamic data desensitization method as described above when the program is running.

[0136] The application provides a dynamic data desensitization method applied to a proxy server located on a data link between a client and a database server, which comprises the following steps: intercepting a data request sent by the client; forwarding the data request of the client to a database server in the back end; capturing response data sent by the database server, wherein the response data is the result returned by the database server after processing the data request of the client; analyzing the returned response data, using a machine learning algorithm to establish dynamic sensitive data identification rules, and identifying sensitive data in the response data; performing real-time dynamic desensitization processing on the identified sensitive data according to a configured desensitization strategy; and sending the desensitized data back to the client. By establishing dynamic sensitive data identification rules, the intelligent and professional level of the desensitization system is improved. The dynamic data desensitization technology allows instant desensitization when data is accessed, reducing the processing and storage requirements for original data. Through the desensitization technology, sensitive data is converted into a non-sensitive form during storage and transmission, reducing the risk of data leakage.

[0137] Those skilled in the art can understand that the units of the examples described in combination with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components of the examples have been described in general terms in the above description. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0138] In the embodiments provided by the present application, it should be understood that the division of units is only a logical functional division, and there can be another division manner in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0139] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0140] When the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application. The aforementioned storage medium includes: a U disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0141] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the claims and the description of the present application.

Claims

1. A dynamic data anonymization method, characterized in that, Applied in a proxy server, the proxy server is located on the data link between the client and the database server; including the following steps: Intercept data requests sent by the client; Forward the client's data requests to the backend database server; Capture response data sent by the database server, which is the result returned by the database server after processing the client's data request; Analyze the returned response data, use machine learning algorithms to establish dynamic sensitive data identification rules, and identify sensitive data in the response data; The identified sensitive data is dynamically de-identified in real time according to the configured de-identification strategy; The anonymized data is sent back to the client.

2. The dynamic data desensitization method according to claim 1, characterized in that, The method of establishing dynamic sensitive data identification rules using machine learning algorithms includes the following steps: The content in the big data cluster is sampled, and the sampled data is synchronized to the sample database; Use clustering or classification analysis techniques to transform sample data into ordered samples with features; The obtained ordered samples are divided into several data models according to their content and meaning using semantic similarity analysis, rule analysis, or dictionary analysis techniques. The categorized data models are then classified and tiered to ultimately form a dynamic and targeted set of rules and classification standards for identifying sensitive data.

3. The dynamic data desensitization method according to claim 1, characterized in that, When performing real-time desensitization processing on the identified sensitive data according to the configured desensitization strategy, different desensitization rules are applied based on data characteristics, context information, or the identity and permissions of the visitor, to ensure that users can only access data that meets their permission level.

4. The dynamic data desensitization method according to claim 3, characterized in that, The method of applying different de-identification rules based on the visitor's identity and permissions includes the following steps: First, different identity characteristics are defined based on the different roles and responsibilities of users; Sensitive data is hidden, masked, encrypted, or audit-desensitized to varying degrees based on identity characteristics, thereby ensuring that users at different levels can only access sensitive data according to their different identity characteristics and cannot modify any type of sensitive data.

5. The dynamic data desensitization method according to claim 3, characterized in that, The data features include name, ID number, bank account, address, telephone number, company name or organization code.

6. The dynamic data desensitization method according to claim 1, characterized in that, The dynamic desensitization process includes partial masking, reversible desensitization, hybrid desensitization, synonym substitution, or deterministic desensitization.

7. The dynamic data desensitization method according to claim 1, characterized in that, It also includes the following steps: Based on the identification of sensitive data, the relationships between the sensitive data are adjusted to ensure the correlation between the data.

8. A dynamic data anonymization system, characterized in that, The dynamic data desensitization method applied to any one of claims 1 to 7 includes: The interception module is used to intercept data requests sent by the client. The forwarding module is used to forward client data requests to the backend database server; The capture module is used to capture response data sent by the database server, which is the result returned by the database server after processing the client's data request; The analysis module is used to analyze the returned response data, use machine learning algorithms to establish dynamic sensitive data identification rules, and identify sensitive data in the response data. The desensitization module is used to perform real-time dynamic desensitization processing on the identified sensitive data according to the configured desensitization strategy. The sending module is used to send the anonymized data back to the client.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the dynamic data desensitization method according to any one of claims 1 to 7.

10. A processor, characterized in that, The processor is used to run a program, wherein the program executes the dynamic data desensitization method according to any one of claims 1 to 7 when it runs.

Images