Data sharing network security privacy hierarchical control trust access method and device

Abstract

The invention provides a data sharing network security privacy hierarchical control trust access method and device, and relates to the technical field of data access. The method comprises the following steps: acquiring a data identifier input by a target user at a target client, and performing primary verification; under the condition that the first-level verification is passed, the fingerprint of the target client is obtained, and second-level verification is carried out; under the condition that the second-level verification is passed, login information input by a target user at a target client is obtained, and first dynamic verification information and second dynamic verification information are sent to the target client; acquiring third dynamic verification information and fourth dynamic verification information input by the target user at the target client, and performing third-level verification; under the condition that the three-level verification is passed, determining an environment risk assessment value, and under the condition that the environment risk assessment value is smaller than a preset threshold value, passing the ultimate verification of the target user; a temporary token is generated for the target user. According to the method, the data sharing efficiency can be greatly improved while the compliance is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data access, and in particular, to a network security privacy hierarchical control trust access method and device for data sharing. Background Art

[0002] With the rapid development of big data and cross-domain collaboration, the access control mechanism of traditional data sharing systems faces severe challenges. Most existing technologies adopt static permission models based on roles (RBAC) or attributes (ABAC), and their defects in dynamic multi-domain environments are becoming increasingly prominent:

[0003] Trust evaluation lag: Traditional methods rely on preset rules and cannot fuse user behavior, device status, and environmental risks (such as abnormal geographical locations and insecure network access) in real time, resulting in response delays in scenarios such as APT attacks and internal threats. According to industry statistics, the misjudgment rate of such static models for new attacks is as high as 30%-40%;

[0004] Coarse-grained privacy classification: Most existing classification technologies are based on manual annotation or simple keyword matching, making it difficult to effectively process unstructured content such as medical images and genetic data, and lacking a timeliness decay mechanism (such as the sensitivity of medical records over five years not being dynamically degraded), resulting in both overprotection and leakage risks;

[0005] Absence of quantum security: Widely used encryption algorithms such as RSA and ECC face quantum computing threats. NIST tests show that the Shor algorithm can break a 2048-bit RSA key within a few minutes, and existing solutions lack a smooth transition mechanism for migrating to post-quantum cryptography;

[0006] Single behavior modeling: Traditional device fingerprints are easily forged (such as MAC address tampering), and biological behavior characteristics such as keystroke dynamics and mouse trajectories are not combined, resulting in a detection rate of less than 60% for impersonation attacks.

[0007] The above limitations make it difficult to balance security and availability in data sharing. Especially in scenarios such as intelligent healthcare and cross-border finance, business efficiency is hindered due to overly tightened permissions, or large-scale data leakage is caused by control oversights. Therefore, there is an urgent need for a new access control solution that integrates dynamic trust evaluation, quantum security encryption, and intelligent cross-domain collaboration to address the complex challenges of multi-domain data sharing. Summary of the Invention

[0008] The purpose of the present invention is to provide a network security privacy hierarchical control trust access method and device for data sharing, aiming to significantly improve data sharing efficiency while ensuring compliance, in view of the deficiencies in the above-mentioned existing technologies.

[0009] To achieve the above purpose, the technical solutions adopted in the embodiments of the present application are as follows:

[0010] In a first aspect, an embodiment of the present application provides a network security privacy hierarchical control trust access method for data sharing, which is applied to a server and includes: when receiving a data access request from a target user, obtaining a data identifier input by the target user on a target client, and performing a first-level verification on the data identifier using a data fingerprint; in the case of passing the first-level verification, obtaining the fingerprint of the target client, and performing a second-level verification on the target client according to the fingerprint of the target client; in the case of passing the second-level verification, obtaining the login information input by the target user on the target client, and sending a first dynamic verification information and a second dynamic verification information to the target client using a dynamic two-factor authentication mechanism; obtaining a third dynamic verification information and a fourth dynamic verification information input by the target user on the target client, and performing a third-level verification on the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information; in the case of passing the third-level verification, using an environmental risk index model to evaluate the operation environmental risk of the target user to obtain an environmental risk assessment value, and in the case that the environmental risk assessment value is less than a preset threshold, passing the ultimate verification of the target user; using an ABAC-RAdAC hybrid model to perform dynamic authorization on the target user, and generating a temporary token for the target user.

[0011] In one implementation, the data fingerprint is generated through the following process: the target client obtains original data; the original data includes structured data and unstructured data; the unstructured data includes text data, image data, and audio-video data; inputting the original data into a TensorFlow Lite model and a semantic classification engine to process the original data and output a privacy level label of the original data; the privacy level label includes L1 level, L3 level, and L5 level; encrypting and storing the L1-level original data using the national cryptography SM4 algorithm, encrypting and storing the L3-level original data using a homomorphic encryption combined with an attribute-based encryption algorithm, encrypting the L5-level original data using the post-quantum cryptography NTRU algorithm, and combining blockchain sharding storage; processing the original data using a hash algorithm to output a first hash value; processing the metadata and the privacy level label using a hash algorithm to output a second hash value; splicing the first hash value and the second hash value to obtain the data fingerprint and storing it.

[0012] In one embodiment, inputting the original data into the TensorFlow Lite model and the semantic grading engine to process the original data and output the privacy level label of the original data includes: inputting the original data into the TensorFlow Lite model, using the TensorFlow Lite model to extract features from the original data, and outputting a feature vector and the metadata; inputting the feature vector into the semantic grading engine, using the TF-IDF algorithm, the BiLSTM neural network, and the knowledge graph to perform semantic analysis on the feature vector, and grading the feature vector according to the semantic analysis result, and outputting the privacy level label of the original data.

[0013] In one embodiment, the fingerprint of the target client is generated through the following process: collecting the hardware layer information, software layer information, and network layer information of the target client, and using the SimHash algorithm to generate the unique device identifier of the target client.

[0014] In one embodiment, sending the first dynamic verification information and the second dynamic verification information to the target client by using the dynamic two-factor authentication mechanism includes: based on the national cryptography SM2 algorithm, sending the first dynamic verification information to the target client; the first dynamic verification information includes a dynamic verification code; using the Mel spectrogram combined with the ResNet-18 model to send the second dynamic verification information to the target client; the second dynamic verification information includes voiceprint verification information.

[0015] In one embodiment, verifying the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information includes: comparing the first dynamic verification information with the third dynamic verification information, and passing the verification if the first dynamic verification information is consistent with the third dynamic verification information; using to compare the second dynamic verification information with the fourth dynamic verification information, and if the second dynamic verification information is consistent with the fourth dynamic verification information, extracting the MFCC feature of the fourth dynamic verification information, and comparing the MFCC feature of the fourth dynamic verification information with the MFCC feature of the target user in the database, and passing the verification when the similarity between the MFCC feature of the fourth dynamic verification information and the MFCC feature of the target user in the database is greater than a preset threshold.

[0016] In one embodiment, before the triple verification is passed, the method further includes: obtaining the mouse movement trajectory and the keyboard keystroke interval sequence of the target user; preprocessing the mouse movement trajectory to generate a speed sequence; performing a discrete Fourier transform on the speed sequence and extracting the main frequency; discretizing the keyboard keystroke interval sequence, and generating corresponding state encodings for different intervals of the discretization result; constructing a state transition matrix based on the state encodings, inputting the state transition matrix into a Markov model, and processing the state transition matrix using the Markov model to obtain a prediction probability; determining whether there is an abnormality in the behavior of the target user according to the main frequency and the prediction probability; when there is no abnormality in the behavior of the target user, passing the triple verification.

[0017] In one embodiment, the operation environment risk of the target user is evaluated using an environmental risk index model to obtain an environmental risk assessment value, including: determining a network risk value according to the VPN usage, TOR exit node, and WiFi encryption level; determining a geographical risk value according to the deviation between the base station triangulation and the GPS coordinates; determining a time risk value according to the DTW distance between the access time period and the historical behavior; and determining the environmental risk assessment value according to the network risk value, the geographical risk value, and the time risk value.

[0018] In one embodiment, the ABAC-RAdAC hybrid model is used to perform dynamic authorization on the target user and generate a temporary token for the target user, including: using the ABAC-RAdAC hybrid model to determine the initial target access permission level of the target user according to the attributes of the target user, and adjusting the initial target access permission level according to the environmental risk assessment value to obtain the final target access permission level; determining the corresponding target access data according to the final target access permission level; using an encryption algorithm to generate a root key for accessing the target access data, deriving a session key according to the root key, and generating the temporary token according to the session key.

[0019] In a second aspect, an embodiment of the present application further provides a network security privacy hierarchical control trust access device for data sharing, which is applied to a server and includes: a primary verification module configured to, when receiving a data access request from a target user, obtain a data identifier input by the target user on a target client, and perform primary verification on the data identifier using a data fingerprint; a secondary verification module configured to, when the primary verification is passed, obtain the fingerprint of the target client, and perform secondary verification on the target client according to the fingerprint of the target client;

[0020] The three - level verification module is configured to, when the second - level verification is passed, obtain the login information input by the target user on the target client, and send the first dynamic verification information and the second dynamic verification information to the target client by using a dynamic two - factor authentication mechanism; obtain the third dynamic verification information and the fourth dynamic verification information input by the target user on the target client, and perform three - level verification on the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information.

[0021] The ultimate verification module is configured to, when the three - level verification is passed, use an environmental risk index model to evaluate the operation environmental risk of the target user, obtain an environmental risk assessment value, and pass the ultimate verification of the target user when the environmental risk assessment value is less than a preset threshold; The token generation module is configured to perform dynamic authorization on the target user by using an ABAC - RAdAC hybrid model and generate a temporary token for the target user.

[0022] In one implementation, the data fingerprint is generated through the following process: The target client obtains the original data; the original data includes structured data and unstructured data; the unstructured data includes text data, image data, and audio - video data; input the original data into the TensorFlow Lite model and the semantic classification engine to process the original data and output the privacy - level label of the original data; the privacy - level label includes L1 level, L3 level, and L5 level; encrypt and store the L1 - level original data by using the national cryptography SM4 algorithm, encrypt and store the L3 - level original data by using the homomorphic encryption combined with the attribute - based encryption algorithm, encrypt the L5 - level original data by using the post - quantum cryptography NTRU algorithm and store it in combination with blockchain sharding; process the original data by using a hash algorithm to output a first hash value; process the metadata and the privacy - level label by using a hash algorithm to output a second hash value; splice the first hash value and the second hash value to obtain the data fingerprint and store it.

[0023] In one embodiment, inputting the original data into the TensorFlow Lite model and the semantic grading engine to process the original data and output the privacy level label of the original data includes: inputting the original data into the TensorFlow Lite model, using the TensorFlow Lite model to extract features from the original data, and outputting the feature vector and the metadata; inputting the feature vector into the semantic grading engine, using the TF-IDF algorithm, the BiLSTM neural network, and the knowledge graph to perform semantic analysis on the feature vector, and grading the feature vector according to the semantic analysis result, and outputting the privacy level label of the original data.

[0024] In one embodiment, the fingerprint of the target client is generated through the following process: collecting the hardware layer information, software layer information, and network layer information of the target client, and using the SimHash algorithm to generate the unique device identifier of the target client.

[0025] In one embodiment, the three-level verification module is configured to: based on the national cryptography SM2 algorithm, send the first dynamic verification information to the target client; the first dynamic verification information includes a dynamic verification code; use the Mel spectrogram combined with the ResNet-18 model to send the second dynamic verification information to the target client; the second dynamic verification information includes voiceprint verification information.

[0026] In one embodiment, the three-level verification module is configured to: compare the first dynamic verification information with the third dynamic verification information, and pass the verification when the first dynamic verification information is consistent with the third dynamic verification information; use to compare the second dynamic verification information with the fourth dynamic verification information, and when the second dynamic verification information is consistent with the fourth dynamic verification information, extract the MFCC feature of the fourth dynamic verification information, and compare the MFCC feature of the fourth dynamic verification information with the MFCC feature of the target user in the database. When the similarity between the MFCC feature of the fourth dynamic verification information and the MFCC feature of the target user in the database is greater than a preset threshold, pass the verification.

[0027] In one embodiment, the three - level verification module is configured to: obtain the mouse movement trajectory and keyboard keystroke interval sequence of the target user; pre - process the mouse movement trajectory to generate a speed sequence; perform a discrete Fourier transform on the speed sequence and extract the main frequency; perform a discretization process on the keyboard keystroke interval sequence, and generate corresponding state encodings for different intervals of the discretization result; construct a state transition matrix based on the state encodings, and input the state transition matrix into a Markov model, and use the Markov model to process the state transition matrix to obtain a prediction probability; determine whether there is an abnormality in the behavior of the target user according to the main frequency and the prediction probability; when there is no abnormality in the behavior of the target user, pass the three - level verification.

[0028] In one embodiment, the ultimate verification module is configured to: determine a network risk value according to the VPN usage situation, TOR exit node, and WiFi encryption level; determine a geographical risk value according to the deviation between the base station triangulation and the GPS coordinates; determine a time risk value according to the DTW distance between the access time period and the historical behavior; determine the environmental risk assessment value according to the network risk value, the geographical risk value, and the time risk value.

[0029] In one embodiment, the token generation module is configured to: adopt the ABAC - RAdAC hybrid model, determine the initial target access permission level of the target user according to the attributes of the target user, and adjust the initial target access permission level according to the environmental risk assessment value to obtain the final target access permission level; determine the corresponding target access data according to the final target access permission level; adopt an encryption algorithm to generate a root key for accessing the target access data, derive a session key according to the root key, and generate the temporary token according to the session key.

[0030] In a third aspect, an embodiment of the present application provides a computer device, including: a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the computer device runs, the processor communicates with the storage medium through the bus, and the processor executes the program instructions to perform the steps of any of the above - mentioned methods.

[0031] In a fourth aspect, an embodiment of the present application provides a non - volatile computer - readable storage medium, on which a computer program is stored. When the computer program is run by a processor, it performs the steps of any of the above - mentioned methods.

[0032] The beneficial effects of this application are as follows: First, when receiving a data access request from a target user, obtain the data identifier input by the target user on the target client, and perform primary verification on the data identifier using a data fingerprint. Second, when the primary verification is passed, obtain the fingerprint of the target client, and perform secondary verification on the target client based on the fingerprint of the target client. Third, when the secondary verification is passed, obtain the login information input by the target user on the target client, and send the first dynamic verification information and the second dynamic verification information to the target client using a dynamic two-factor authentication mechanism. Then, obtain the third dynamic verification information and the fourth dynamic verification information input by the target user on the target client, and perform tertiary verification on the third dynamic verification information and the fourth dynamic verification information based on the first dynamic verification information and the second dynamic verification information. Then, when the tertiary verification is passed, use an environmental risk index model to evaluate the operation environment risk of the target user, obtain an environmental risk assessment value, and pass the ultimate verification of the target user when the environmental risk assessment value is less than a preset threshold. Finally, use an ABAC-RAdAC hybrid model to perform dynamic authorization on the target user and generate a temporary token for the target user. In this way, by constructing an intelligent hierarchical-evaluation-control closed-loop system, while ensuring compliance, the data sharing efficiency is greatly improved, which is especially suitable for scenarios that require complex data collaboration such as smart cities and fintech, and has significant market application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0034] Figure 1 It is a flowchart showing a method for network security privacy hierarchical control and trusted access for data sharing provided by an embodiment of this application;

[0035] Figure 2 It is a flowchart showing a method for network security privacy hierarchical control and trusted access for data sharing provided by an embodiment of this application;

[0036] Figure 3 It is a flowchart showing a method for network security privacy hierarchical control and trusted access for data sharing provided by an embodiment of this application;

[0037] Figure 4 It is a flowchart showing a method for network security privacy hierarchical control and trusted access for data sharing provided by an embodiment of this application;

[0038] Figure 5 Schematic flowchart of a network security privacy hierarchical control trust access method for data sharing provided by an embodiment of the present application;

[0039] Figure 6 Schematic flowchart of a network security privacy hierarchical control trust access method for data sharing provided by an embodiment of the present application;

[0040] Figure 7 Schematic flowchart of a network security privacy hierarchical control trust access method for data sharing provided by an embodiment of the present application;

[0041] Figure 8 Schematic flowchart of a network security privacy hierarchical control trust access method for data sharing provided by an embodiment of the present application;

[0042] Figure 9 Schematic structural diagram of a network security privacy hierarchical control trust access device for data sharing provided by an embodiment of the present application;

[0043] Figure 10 Schematic structural diagram of a computer device provided by an embodiment of the present application. Detailed implementation manners

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention.

[0045] Therefore, the detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application claimed, but merely represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the scope of protection of the present application.

[0046] In the description of the present application, it should be noted that if terms such as "upper", "lower", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship when the product of this application is normally placed, it is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present application.

[0047] In addition, the terms "first", "second", etc. in the description, claims and the above-mentioned drawings of the present invention are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0048] It should be noted that, without conflict, the features in the embodiments of the present application can be combined with each other.

[0049] Figure 1 It is a schematic flowchart of a network security privacy hierarchical control trust access method for data sharing provided by an embodiment of the present application; as Figure 1 shown, this method is applied to a server, and this method includes:

[0050] Step 110: When receiving a data access request from a target user, obtain the data identifier input by the target user on the target client, and perform a primary verification on the data identifier using a data fingerprint.

[0051] The data access request is a data identifier submitted by the target user (requestor) through the target client, such as: file name, database ID, etc.

[0052] The data fingerprint is data stored by the server in the database; specifically, as Figure 2 shown, the data fingerprint is generated through the following steps 210 to 260:

[0053] Step 210: The target client obtains the original data.

[0054] The original data includes structured data and unstructured data; the unstructured data includes text data, image data, audio and video data.

[0055] Structured data is specific data, including numbers, dates, texts, etc., which can be easily mapped to tables, relational databases or other structured data storage systems. Unstructured data includes irregular, non-fixed format and rule data such as pictures, audio, video, emails, social media posts, etc., which are difficult to store and process in tables or relational databases.

[0056] Original data, which is the initial form of the data that the user wants to access, such as: case records, hospitalization records, medical CT images, inspection reports, etc.

[0057] Step 220: Input the original data into the TensorFlow Lite model and the semantic grading engine, process the original data, and output the privacy level label of the original data.

[0058] Among them, the privacy level labels include L1 level, L3 level, and L5 level. Here, for different privacy levels, the encryption methods are different. The higher the privacy level, the more complex the encryption method and the higher the encryption degree.

[0059] The TensorFlow Lite model is a lightweight machine learning framework specifically used to deploy and run machine learning models in resource-constrained environments such as mobile devices, embedded devices, and Internet of Things devices. The main functions of TensorFlow Lite include model conversion, model optimization, and inference execution.

[0060] The TensorFlow Lite model is mainly used for feature extraction of unstructured data; the semantic grading engine is mainly used for dynamic privacy grading of the extracted features; specifically, as Figure 3 shown, the above step 220 may further include the following steps 310 and 320:

[0061] Step 310: Input the original data into the TensorFlow Lite model, use the TensorFlow Lite model to extract features from the original data, and output the feature vector and metadata.

[0062] Among them, for structured data, the TensorFlow Lite model can directly extract features and output the feature vector. For unstructured data, the TensorFlow Lite model can first perform a preliminary analysis (such as: OCR recognition of case text), then extract basic features (such as: detecting the coordinates of sensitive areas in the image), and generate metadata (such as: file format, data size, geographical location restrictions (such as: only within this hospital), time limit (such as: 3 years), etc.), and finally output the feature vector and metadata. Here, whether it is processing structured data or unstructured data, the output feature vector is a structured feature vector.

[0063] Step 320: Input the feature vector into the semantic grading engine, use the TF-IDF algorithm, BiLSTM neural network, and knowledge graph to perform semantic analysis on the feature vector, and classify the feature vector according to the semantic analysis result, and output the privacy level label of the original data.

[0064] Among them, the TF-IDF algorithm is used to extract keywords in the feature vector to analyze the feature vector at the syntax level; the BiLSTM neural network is used to analyze the context relevance (such as: the weight of "HIV positive" in the case) to analyze the feature vector at the context level; the knowledge graph can be a legal knowledge graph (such as: Article 9 of GDPR on special category data) to analyze the feature vector at the regulation level; furthermore, three-level semantic analysis is achieved.

[0065] Step 230: Encrypt and store the L1-level raw data using the national cipher SM4 algorithm, encrypt and store the L3-level raw data using the homomorphic encryption combined with the attribute-based encryption algorithm, and encrypt the L5-level raw data using the post-quantum cryptography NTRU algorithm, combined with blockchain sharding storage.

[0066] Among them, the national cipher algorithm, fully known as the cipher algorithm released by the State Cryptography Administration, is a series of cipher algorithms independently developed in China. These algorithms include: SM1, SM2, SM3, SM4, and SM9. These algorithms have important application values in the field of information security and are widely used in the fields of finance, government affairs, telecommunications, etc.

[0067] The national cipher SM4 algorithm is a symmetric encryption algorithm based on block ciphers that supports 128-bit keys. The SM4 algorithm has high security and computational efficiency and is suitable for application scenarios such as data encryption and decryption. The core of the SM4 symmetric encryption algorithm is the Feistel network structure, which encrypts the plaintext data through multiple iterations. In each iteration, the round function F is applied. The round function F consists of a linear transformation and a non-linear transformation. The linear transformation uses the XOR operation, and the non-linear transformation uses the modular addition operation.

[0068] Homomorphic encryption means that performing addition and multiplication operations on the plaintext in the ring and then encrypting is equivalent to performing the corresponding operations on the ciphertext after encryption. Due to this property, a third party can be entrusted to process the data without leaking information.

[0069] The attribute-based encryption algorithm (ABE) is an attribute-based encryption method that allows the data owner to control the access rights of the data according to the attributes of the user. The ABE algorithm is an extension of the identity-based encryption algorithm, representing the user identity through a series of descriptive attributes, allowing the data owner and users with specific attributes to share the encrypted data, realizing a one-to-many communication mode.

[0070] The core idea of the post - quantum cryptography NTRU algorithm is to convert plaintext or ciphertext into polynomials over a ring and achieve encryption and decryption through mathematical operations. Specifically, the post - quantum cryptography NTRU algorithm uses three basic data structures: finite fields, polynomial rings, and elliptic curves. Among them, finite fields are used to generate public and private keys, polynomial rings are used to implement encryption and decryption, and elliptic curves are used to verify signatures.

[0071] Step 240: Process the original data using a hash algorithm and output the first hash value.

[0072] Among them, the hash algorithm, also known as the hashing algorithm, is a technology that converts input data of any length (such as strings, files, etc.) into output data of a fixed length (usually shorter) (i.e., the hash value or hash value) through a specific algorithm.

[0073] Exemplarily, the first hash value is 0x8912...

[0074] Step 250: Process the metadata and privacy level label using a hash algorithm and output the second hash value.

[0075] Exemplarily, the second hash value is 0xab3f...

[0076] Step 260: Concatenate the first hash value and the second hash value to obtain a data fingerprint and store it.

[0077] Among them, concatenating the first hash value and the second hash value gives the data fingerprint, for example: 0x8912... (the first hash value)|0xab3f... (the second hash value).

[0078] Step 120: In the case of passing the first - level verification, obtain the fingerprint of the target client and perform second - level verification on the target client according to the fingerprint of the target client.

[0079] Among them, the fingerprint of the target client can be generated based on the hardware - layer features, software - layer features, and network - layer features of the target client; specifically, the fingerprint of the target client is generated through the following steps:

[0080] Collect the hardware - layer information, software - layer information, and network - layer information of the target client, and use the SimHash algorithm to generate a unique device identifier for the target client.

[0081] Among them, the hardware - layer information includes the CPU microcode version and the TPM chip ID; the software - layer information includes the operating system patch hash and the installed certificate fingerprint; the network - layer information includes the MAC address obfuscation value and the TCP timing information. Specifically, the SimHash algorithm can be used to process the hardware - layer information, software - layer information, and network - layer information to generate the fingerprint of the target client.

[0082] In actual operation, the fingerprint of the target client can be compared with the fingerprint of the target client stored in the database. When the fingerprint of the target client is consistent with the fingerprint of the target client stored in the database, it is determined that the secondary verification of the target client is passed; otherwise, the secondary verification fails.

[0083] Step 130: When the secondary verification is passed, obtain the login information input by the target user on the target client, and send the first dynamic verification information and the second dynamic verification information to the target client by using the dynamic two-factor authentication mechanism.

[0084] Among them, the first dynamic verification information can be a dynamic verification code; the second dynamic verification information can be voiceprint verification information; different verification information is generated in different ways. Specifically, as Figure 4 shown, the above step 130 can further include the following step 410 and step 420:

[0085] Step 410: Based on the national cryptography SM2 algorithm, send the first dynamic verification information to the target client.

[0086] Among them, the national cryptography SM2 algorithm is an asymmetric encryption algorithm based on elliptic curve cryptography, supporting a key length of 256 bits. The SM2 algorithm has high security and computational efficiency, and is applicable to application scenarios such as digital signatures and key exchanges.

[0087] In this step, after generating the first dynamic verification information by using the national cryptography SM2 algorithm, it can be directly sent to the target client.

[0088] Step 420: Use the Mel spectrogram combined with the ResNet-18 model to send the second dynamic verification information to the target client.

[0089] Among them, the Mel spectrogram is a spectral representation method of audio signals based on the Mel frequency scale. It better reflects the perceptual characteristics of the human auditory system by converting frequency to the Mel scale. The Mel scale is a non-linear frequency scale, which can be closer to the human subjective perception of pitch. Especially in the low-frequency region, the bandwidth is wider, while in the high-frequency region, the bandwidth is narrower.

[0090] The ResNet-18 model is a classic deep convolutional neural network model, mainly used for ImageNet image classification. The name ResNet-18 comes from the 18 convolutional layers it contains. The main feature is the introduction of the concept of residual blocks. Through skip connections, the network can learn residual information, thus better performing feature extraction and processing.

[0091] In this step, after generating the second dynamic verification information by combining the Mel spectrogram with the ResNet-18 model, it can be directly sent to the target client.

[0092] Step 140: Obtain the third dynamic verification information and the fourth dynamic verification information input by the target user on the target client, and perform three-level verification on the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information.

[0093] Among them, the three-level verification in this step is mainly achieved by comparison; specifically, as Figure 5 shown, the above step 140 may further include the following step 510 and step 520:

[0094] Step 510: Compare the first dynamic verification information with the third dynamic verification information. If the first dynamic verification information is consistent with the third dynamic verification information, the verification is passed.

[0095] Exemplarily, if the sent first dynamic verification information is 3327 and the received third verification information is 3327, then the first dynamic verification information is consistent with the third dynamic verification information, and the verification is passed; if the sent first dynamic verification information is 3327 and the received third verification information is 2273 or other numbers other than 3327, the verification fails.

[0096] Step 520: Compare the second dynamic verification information with the fourth dynamic verification information. If the second dynamic verification information is consistent with the fourth dynamic verification information, extract the MFCC features of the fourth dynamic verification information, and compare the MFCC features of the fourth dynamic verification information with the MFCC features of the target user in the database. When the similarity between the MFCC features of the fourth dynamic verification information and the MFCC features of the target user in the database is greater than the preset threshold, the verification is passed.

[0097] Among them, when comparing the second dynamic verification information with the fourth dynamic verification information, it is mainly to compare whether the pronunciations of the second dynamic verification information and the fourth dynamic verification information are the same.

[0098] MFCC (Mel Frequency Cepstral Coefficients) is a kind of speech feature parameter used for speech recognition and speaker recognition. It is designed based on the auditory characteristics of the human ear and can better reflect the spectral characteristics of speech. MFCC features are extracted from the speech signal through a series of processing steps (including pre-emphasis, framing, windowing, fast Fourier transform, discrete cosine transform). It is mainly used for speech recognition and speaker recognition and can provide information about the clarity and recognition of speech. MFCC features do not depend on the specific nature of the signal, so they can still maintain good recognition performance in the case of low signal-to-noise ratio.

[0099] In actual operation, the user behavior can also be detected; specifically, as Figure 6 shown, the network security privacy hierarchical control trust access method for data sharing provided by the embodiments of the present application may further include the following steps 610 to 660:

[0100] Step 610: Obtain the mouse movement trajectory and keyboard keystroke interval sequence of the target user.

[0101] Among them, the mouse movement trajectory can be collected after receiving the data access request of the target user and continuously collected until the user behavior detection is performed; or, during the period from receiving the data access request of the target user to performing the user behavior detection, the mouse movement trajectory within a preset duration is collected; there is no limitation here.

[0102] The keyboard keystroke interval sequence is the same as the above method for collecting the mouse movement trajectory, and will not be elaborated here.

[0103] Step 620: Preprocess the mouse movement trajectory to generate a speed sequence.

[0104] Among them, the preprocessing may include trajectory segmentation and coordinate normalization; specifically, the trajectory segmentation can be performed with continuous movement events as units (for example: segment when the interval > 500 ms); the coordinate normalization can convert the coordinates into relative displacement vectors, and thus the speed sequence can be obtained.

[0105] Step 630: Perform a discrete Fourier transform on the speed sequence and extract the main frequency.

[0106] Among them, the discrete Fourier transform (Discrete Fourier Transform, DFT) Fourier analysis method is the most basic method for signal analysis, and the Fourier transform is the core of Fourier analysis. Through it, the signal is transformed from the time domain to the frequency domain, and then the spectral structure and variation law of the signal are studied.

[0107] The main frequency can be obtained by retaining the frequency corresponding to the first n% of the energy; here, n% can be 10%, 15%, etc., and there is no limitation here.

[0108] Step 640: Discretize the keyboard keystroke interval sequence and generate corresponding state encodings for different intervals of the discretization result.

[0109] Among them, the state encoding can be obtained in the following way as shown in Table 1:

[0110] Interval Division (ms) Status Encoding Description [0，100) S0 Quick Keystroke [100，300) S1 Normal Speed [300，∞) S2 Long Interval / Pause

[0111] Table 1

[0112] Step 650: Construct a state transition matrix based on the state encoding, input the state transition matrix into a Markov model, and use the Markov model to process the state transition matrix to obtain the prediction probability.

[0113] Among them, the state transition matrix can include the state encoding and the events corresponding to the states. For example: S0: {S0: 0.2, S1: 0.7, S2: 0.1}, S1: {S0: 0.3, S1: 0.5, S2: 0.2}, S2: {S0: 0.6, S1: 0.3, S2: 0.1}.

[0114] The Markov model is a stochastic process model based on the Markov property assumption, used to describe the transfer behavior of the system among a series of states. It is widely applied in fields such as speech recognition, weather prediction, and gene sequence analysis.

[0115] The prediction probability, that is, the probability corresponding to each state encoding, can be understood as the probability of each state occurring. For example: output = {S0: 0.65, S1: 0.3, S2: 0.05}.

[0116] Step 660: Determine whether the behavior of the target user is abnormal according to the main frequency and the prediction probability; when the behavior of the target user is not abnormal, pass the three-level verification.

[0117] Among them, for the main frequency, the eigenvector of the main frequency can be determined through feature engineering, and then the Mahalanobis distance between the eigenvector and the nearest neighbor clustering center is calculated. When the Mahalanobis distance between the eigenvector and the nearest neighbor clustering center is greater than the preset threshold, it is determined that the behavior of the target user is abnormal.

[0118] For the prediction probability, it can also be determined whether the behavior of the target user is abnormal through a threshold; specifically, it can be determined whether the prediction probability is less than the preset threshold. When the prediction probability is less than the preset threshold, it is determined that the behavior of the target user is abnormal.

[0119] Step 150: In the case of passing the three-level verification, use the environmental risk index model to evaluate the operation environmental risk of the target user to obtain the environmental risk assessment value, and in the case where the environmental risk assessment value is less than the preset threshold, pass the ultimate verification of the target user.

[0120] Among them, the environmental risk index model can evaluate network risk, geographical risk, and time risk; specifically, as Figure 7 shown, the above Step 150 can further include the following Steps 710 to 740:

[0121] Step 710: Determine the network risk value according to the VPN usage situation, the TOR exit node, and the WiFi encryption level.

[0122] Among them, the network risk value can be determined through the following process: First, identify the VPN and TOR exit nodes, and determine the corresponding risk coefficients according to the WiFi encryption type; among them, there are corresponding weights for VPN usage, TOR exit nodes, and WiFi encryption levels; Second, determine the network risk value according to the network risk value calculation formula; Exemplarily, the network risk value calculation formula is shown as formula (1) below:

[0123] NetworkRisk = 0.6·VPN_Flag + 0.3·WiFi_Risk + 0.1·Proxy_Usage (1)

[0125] Step 720: Determine the geographical risk value according to the deviation between the base station triangulation and the GPS coordinates.

[0126] Among them, the geographical risk value can be determined through the following process: First, use the Mozilla Location Service API to parse the coordinates; Second, perform GPS anti-spoofing detection on the coordinates (including: speed mutation detection: acceleration > 1000m / s 2 judged as abnormal; elevation rationality: altitude exceeds the normal range of ±500m); Finally, determine the geographical risk value according to the geographical risk value calculation formula; Exemplarily, the geographical risk value calculation formula is shown as formula (2) below:

[0127] GeoRisk = 0.7·Distance_Score + 0.3·Velocity_Anomaly (2)

[0128] Among them, Distance_Score represents the Haversine distance between the base station location and the GPS coordinates (increasing 0.1 points per kilometer); Velocity_Anomaly represents the deviation degree based on the historical movement pattern.

[0129] Step 730: Determine the time risk value according to the DTW distance between the access time period and the historical behavior.

[0130] Among them, the time risk value can be determined through the following process: First, determine the DTW distance between the access time period and the historical behavior; Second, determine the corresponding time risk value according to the DTW distance between the access time period and the historical behavior.

[0131] Step 740: Determine the environmental risk assessment value according to the network risk value, geographical risk value, and time risk value.

[0132] Among them, different weights can be assigned to the network risk value, geographical risk value, and time risk value, and then the network risk value, geographical risk value, and time risk value are weighted to obtain the final environmental risk assessment value.

[0133] Exemplarily, the environmental risk index model is shown in the following formula (3):

[0134]

[0135] Among them, w i is the weight obtained by constructing a judgment matrix through expert scoring and calculating the consistency ratio (CR < 0.1); for example: network risk value (0.44), geographical risk value (0.36), time risk value (0.20); φ ij is statistically obtained based on historical attack data; for example: the combined risk of network anomalies and geographical location anomalies is increased by 30%; θ·ThreatIntel is to integrate external threat intelligence in real time, such as: IP blacklist, vulnerability warning, and dynamically adjust θ (default 0.1, 0.3 in case of emergency).

[0136] The above environmental risk index model uses the Analytic Hierarchy Process (AHP) to determine the weight and introduces a risk synergy term, which can quantify the interaction of multiple risk factors and improve the compound attack detection rate.

[0137] Step 160: Use the ABAC-RAdAC hybrid model to perform dynamic authorization on the target user and generate a temporary token for the target user.

[0138] Among them, the ABAC model is mainly used to generate access rules; RAdAC is mainly used to dynamically adjust access permissions; specifically, as Figure 8 shown, the above step 160 may further include the following steps 810 to 830:

[0139] Step 810: Use the ABAC-RAdAC hybrid model to determine the initial target access permission level of the target user according to the attributes of the target user, and adjust the initial target access permission level according to the environmental risk assessment value to obtain the final target access permission level.

[0140] Among them, the ABAC model (Attribute-Based Access Control, ABAC) is a dynamic and fine-grained access control model that determines access permissions by evaluating user attributes, resource attributes, environmental attributes, and action attributes. This model is more flexible than the traditional Role-Based Access Control (RBAC) because it allows permission decisions to be made based on multiple conditions and context information.

[0141] The RAdAC model can perform dynamic permission adjustment in the manner shown in Table 2 below:

[0142]

[0143]

[0144] Table 2

[0145] Step 820: Determine the corresponding target access data according to the final target access permission level.

[0146] Among them, the mapping relationship between the access permission level and the access data is stored in the database, and the corresponding target access data can be directly determined by referring to the mapping relationship table according to the final target access permission.

[0147] Step 830: Use an encryption algorithm to generate a root key for accessing the target access data, derive a session key based on the root key, and generate a temporary token based on the session key.

[0148] Among them, the encryption algorithm can adopt symmetric encryption algorithms, asymmetric encryption algorithms, hash functions, etc., which are not limited here.

[0149] Exemplarily, the generation of the temporary token can be performed using a dynamic grant model, and the dynamic grant model is shown in the following formula (4):

[0150]

[0151] Among them, GrantProb represents the permission matching degree; RoleMatch represents the role matching degree; RequestMatch represents the request matching degree; R max represents the maximum acceptable risk threshold.

[0152] The role matching degree RoleMatch is shown in the following formula (5):

[0153]

[0154] Among them, μ represents the user role; κ represents the required role.

[0155] The request matching degree RequestMatch is shown in the following formula (6):

[0156]

[0157] Among them, represents the requested permission; represents the minimum necessary permission.

[0158] The above dynamic grant model fuzzily matches the boundary conditions, such as: partial role matching; the probability model supports flexible policies, such as: 80% probability grant plus manual review.

[0159] The root key can be generated using any of the above encryption algorithms; the session key can be generated using any of the above encryption algorithms based on the root key; the temporary token can be generated using any of the above encryption algorithms based on the session key.

[0160] The data sharing network security privacy hierarchical control trust access method provided by the embodiments of this application. First, when receiving a data access request from a target user, obtain the data identifier input by the target user on the target client, and perform a primary verification on the data identifier using a data fingerprint. Second, when the primary verification is passed, obtain the fingerprint of the target client, and perform a secondary verification on the target client according to the fingerprint of the target client. Third, when the secondary verification is passed, obtain the login information input by the target user on the target client, and send the first dynamic verification information and the second dynamic verification information to the target client using a dynamic two-factor authentication mechanism. Then, obtain the third dynamic verification information and the fourth dynamic verification information input by the target user on the target client, and perform a tertiary verification on the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information. Then, when the tertiary verification is passed, use an environmental risk index model to evaluate the operation environment risk of the target user to obtain an environmental risk assessment value, and when the environmental risk assessment value is less than a preset threshold, pass the ultimate verification of the target user. Finally, use an ABAC-RAdAC hybrid model to perform dynamic authorization on the target user and generate a temporary token for the target user. In this way, by constructing an intelligent hierarchical-evaluation-control closed-loop system, while ensuring compliance, the data sharing efficiency is greatly improved, which is particularly suitable for scenarios that require complex data collaboration such as smart cities and fintech, and has significant market application prospects.

[0161] After introducing the data sharing network security privacy hierarchical control trust access method of the exemplary embodiments of the present disclosure, next, refer to Figure 9 to describe the data sharing network security privacy hierarchical control trust access device 900 of the exemplary embodiments of the present disclosure.

[0162] Refer to Figure 9, a network security privacy hierarchical control trusted access device 900 for data sharing, comprising: a primary verification module 910, configured to obtain a data identifier input by a target user on a target client when receiving a data access request from the target user, and perform primary verification on the data identifier using a data fingerprint; a secondary verification module 920, configured to obtain the fingerprint of the target client when the primary verification is passed, and perform secondary verification on the target client according to the fingerprint of the target client; a tertiary verification module 930, configured to obtain the login information input by the target user on the target client when the secondary verification is passed, and send a first dynamic verification information and a second dynamic verification information to the target client using a dynamic two-factor authentication mechanism; obtain a third dynamic verification information and a fourth dynamic verification information input by the target user on the target client, and perform tertiary verification on the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information; an ultimate verification module 940, configured to evaluate the operation environment risk of the target user using an environmental risk index model when the tertiary verification is passed, obtain an environmental risk assessment value, and pass the ultimate verification of the target user when the environmental risk assessment value is less than a preset threshold; a token generation module 950, configured to perform dynamic authorization on the target user using an ABAC-RAdAC hybrid model, and generate a temporary token for the target user.

[0163] In one implementation, the data fingerprint is generated through the following process: the target client obtains the original data; the original data includes structured data and unstructured data; the unstructured data includes text data, image data, and audio-video data; the original data is input into a TensorFlow Lite model and a semantic classification engine to process the original data and output a privacy level label of the original data; the privacy level label includes L1 level, L3 level, and L5 level; the L1-level original data is encrypted and stored using the national cryptography SM4 algorithm, the L3-level original data is encrypted and stored using a homomorphic encryption combined with an attribute-based encryption algorithm, the L5-level original data is encrypted using the post-quantum cryptography NTRU algorithm and stored in combination with blockchain sharding; the original data is processed using a hash algorithm to output a first hash value; the metadata and the privacy level label are processed using a hash algorithm to output a second hash value; the first hash value and the second hash value are concatenated to obtain a data fingerprint and stored.

[0164] In one embodiment, the original data is input into a TensorFlow Lite model and a semantic grading engine to process the original data and output a privacy level label for the original data, including: inputting the original data into the TensorFlow Lite model, using the TensorFlow Lite model to extract features from the original data, and outputting a feature vector and metadata; inputting the feature vector into the semantic grading engine, using the TF-IDF algorithm, a BiLSTM neural network, and a knowledge graph to perform semantic analysis on the feature vector, and grading the feature vector according to the semantic analysis result to output a privacy level label for the original data.

[0165] In one embodiment, the fingerprint of the target client is generated through the following process: collecting the hardware layer information, software layer information, and network layer information of the target client, and using the SimHash algorithm to generate a unique device identifier for the target client.

[0166] In one embodiment, the three-level verification module 930 is configured to: send a first dynamic verification message to the target client based on the national cryptography SM2 algorithm; the first dynamic verification message includes a dynamic verification code; use a Mel spectrogram combined with a ResNet-18 model to send a second dynamic verification message to the target client; the second dynamic verification message includes voiceprint verification information.

[0167] In one embodiment, the three-level verification module 930 is configured to: compare the first dynamic verification message with a third dynamic verification message, and pass the verification if the first dynamic verification message is consistent with the third dynamic verification message; use the second dynamic verification message to be compared with a fourth dynamic verification message, and if the second dynamic verification message is consistent with the fourth dynamic verification message, extract the MFCC feature of the fourth dynamic verification message, and compare the MFCC feature of the fourth dynamic verification message with the MFCC feature of the target user in the database. When the similarity between the MFCC feature of the fourth dynamic verification message and the MFCC feature of the target user in the database is greater than a preset threshold, the verification is passed.

[0168] In one embodiment, the three-level verification module 930 is configured to: obtain the mouse movement trajectory and keyboard keystroke interval sequence of the target user; preprocess the mouse movement trajectory to generate a speed sequence; perform a discrete Fourier transform on the speed sequence and extract the main frequency; discretize the keyboard keystroke interval sequence, and generate corresponding state encodings for different intervals of the discretization result; construct a state transition matrix based on the state encodings, and input the state transition matrix into a Markov model, and use the Markov model to process the state transition matrix to obtain a prediction probability; determine whether the behavior of the target user is abnormal according to the main frequency and the prediction probability; when the behavior of the target user is not abnormal, pass the three-level verification.

[0169] In one embodiment, the ultimate verification module 940 is configured to: determine a network risk value according to VPN usage, TOR exit nodes, and WiFi encryption level; determine a geographical risk value according to the deviation between base station triangulation and GPS coordinates; determine a time risk value according to the DTW distance between the access time period and historical behavior; and determine an environmental risk assessment value according to the network risk value, geographical risk value, and time risk value.

[0170] In one embodiment, the token generation module 950 is configured to: adopt an ABAC-RAdAC hybrid model, determine the initial target access permission level of the target user according to the attributes of the target user, and adjust the initial target access permission level according to the environmental risk assessment value to obtain the final target access permission level; determine the corresponding target access data according to the final target access permission level; adopt an encryption algorithm to generate a root key for accessing the target access data, derive a session key according to the root key, and generate a temporary token according to the session key.

[0171] The above device is used to execute the method provided in the foregoing embodiment, and its implementation principle and technical effects are similar, which will not be elaborated here.

[0172] The above modules may be one or more integrated circuits configured to implement the above method, for example: one or more application specific integrated circuits (ASICs), or, one or more microprocessors, or, one or more field programmable gate arrays (FPGAs), etc. Again, when a certain module above is implemented in the form of a processing element dispatching program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processors that can call program code. Again, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0173] Figure 10 It is a schematic diagram of a computer device provided in an embodiment of the present application. This device may be integrated into a terminal device or a chip of a terminal device, and the terminal may be a computing device with data processing capabilities.

[0174] The device includes: a processor 1001, a storage medium 1002, and a bus 1003.

[0175] The storage medium 1002 stores program instructions executable by the processor 1001. When the computer device 1000 runs, the processor 1001 communicates with the storage medium 1002 via the bus 1003, and the processor 1001 executes the program instructions to implement the above method embodiments. The specific implementation manners and technical effects are similar and will not be elaborated here.

[0176] Optionally, the present invention further provides a program product, such as a computer-readable storage medium, including a program that is used to execute the above method embodiments when executed by a processor.

[0177] In several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, 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 displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of the apparatus or unit can be in electrical, mechanical or other forms.

[0178] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0179] In addition, each functional unit in various embodiments of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware, or in the form of a hardware plus software functional unit.

[0180] The above integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above software functional unit is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (English: processor) to execute some steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (English: Read-Only Memory, abbreviated as: ROM), a random access memory (English: Random Access Memory, abbreviated as: RAM), a magnetic disk, or an optical disc that can store program codes.

[0181] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A data sharing network security privacy hierarchical control trust access method, applied to a server, characterized in that: include: When receiving a data access request from a target user, obtaining a data identifier input by the target user in a target client, and performing primary verification on the data identifier using a data fingerprint; If the primary verification is passed, obtaining the fingerprint of the target client, and performing secondary verification on the target client according to the fingerprint of the target client; If the secondary verification is passed, the target user's login information inputted by the target client is obtained, and the first dynamic verification information and the second dynamic verification information are sent to the target client by using a dynamic two-factor authentication mechanism; Acquire the third dynamic verification information and the fourth dynamic verification information input by the target user at the target client, and perform three-level verification on the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information; In the case of passing the three-level verification, the operating environment risk of the target user is evaluated using an environmental risk index model to obtain an environmental risk evaluation value, and in the case that the environmental risk evaluation value is less than a preset threshold, the ultimate verification of the target user is passed; An ABAC-RAdAC hybrid model is adopted to dynamically authorize the target user and generate a temporary token for the target user.

2. The method according to claim 1, characterized in that The data fingerprint is generated by the following process: The target client obtains original data; the original data includes structured data and unstructured data; the unstructured data includes text data, image data, audio and video data; Input the original data into the TensorFlow Lite model and the semantic grading engine, process the original data, and output the privacy level label of the original data; the privacy level label includes L1, L3, and L5; The L1-level original data is encrypted and stored using the national secret SM4 algorithm, the L3-level original data is encrypted and stored using homomorphic encryption combined with the attribute-based encryption algorithm, and the L5-level original data is encrypted using the post-quantum cryptography NTRU algorithm, combined with blockchain sharding storage; Processing the original data using a hash algorithm to output a first hash value; Processing the metadata and the privacy level label using a hash algorithm to output a second hash value; The first hash value and the second hash value are concatenated to obtain the data fingerprint and store it.

3. The method according to claim 2, characterized in that The step of inputting the raw data into a TensorFlowLite model and a semantic grading engine, processing the raw data, and outputting a privacy level label of the raw data includes: Inputting the original data into the TensorFlow Lite model, performing feature extraction on the original data using the TensorFlow Lite model, and outputting a feature vector and the metadata; The feature vector is input into the semantic grading engine, and the feature vector is semantically analyzed using the TF-IDF algorithm, BiLSTM neural network, and knowledge graph. The feature vector is graded according to the semantic analysis result, and the privacy level label of the original data is output.

4. The method according to claim 1, characterized in that: The target client's fingerprint is generated by the following process: The hardware layer information, software layer information, and network layer information of the target client are collected, and a unique device identifier of the target client is generated using a SimHash algorithm.

5. The method according to claim 1, characterized in that The adopting the dynamic two-factor authentication mechanism to send the first dynamic verification information and the second dynamic verification information to the target client includes: Based on the national secret SM2 algorithm, sending the first dynamic verification information to the target client; the first dynamic verification information includes a dynamic verification code; The Mel spectrum diagram is combined with the ResNet-18 model to send the second dynamic verification information to the target client; the second dynamic verification information includes voiceprint verification information.

6. The method according to claim 1, characterized in that The verifying the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information includes: Comparing the first dynamic verification information with the third dynamic verification information, and passing the verification if the first dynamic verification information is consistent with the third dynamic verification information; The second dynamic verification information is compared with the fourth dynamic verification information. When the second dynamic verification information is consistent with the fourth dynamic verification information, the MFCC features of the fourth dynamic verification information are extracted, and the MFCC features of the fourth dynamic verification information are compared with the MFCC features of the target user in the database. When the similarity between the MFCC features of the fourth dynamic verification information and the MFCC features of the target user in the database is greater than a preset threshold, the verification is passed.

7. The method according to claim 1, characterized in that Before passing the third-level verification, the method further includes: Obtaining the target user's mouse movement trajectory and keyboard keystroke interval sequence; Preprocessing the mouse movement trajectory to generate a speed sequence; Performing discrete Fourier transform on the velocity sequence and extracting the main frequencies; Discretizing the keyboard keystroke interval sequence, and generating corresponding state codes for different intervals of the discretization result; Constructing a state transfer matrix based on the state code, inputting the state transfer matrix into a Markov model, and processing the state transfer matrix using the Markov model to obtain a prediction probability; Determine whether the behavior of the target user is abnormal based on the main frequency and the predicted probability; and pass the three-level verification when the behavior of the target user is not abnormal.

8. The method according to claim 1, characterized in that The environmental risk index model is used to evaluate the operating environment risk of the target user to obtain an environmental risk evaluation value, including: Determine network risk based on VPN usage, TOR exit nodes, and WiFi encryption level; Determine the geographic risk value based on the base station triangulation positioning and GPS coordinate deviation; Determine the time risk value based on the DTW distance between the access period and the historical behavior; The environmental risk assessment value is determined according to the network risk value, the geographical risk value, and the time risk value.

9. The method according to claim 1, characterized in that: The ABAC-RAdAC hybrid model is used to dynamically authorize the target user and generate a temporary token for the target user, including: Adopting the ABAC-RAdAC hybrid model, determining the initial target access permission level of the target user according to the attributes of the target user, and adjusting the initial target access permission level according to the environmental risk assessment value to obtain a final target access permission level; Determining corresponding target access data according to the final target access permission level; An encryption algorithm is used to generate a root key for accessing the target access data, a session key is derived based on the root key, and the temporary token is generated based on the session key.

10. A data sharing network security privacy hierarchical control trust access device, applied to a server, characterized in that: include: The primary verification module is configured to, upon receiving a data access request from a target user, obtain a data identifier input by the target user in a target client, and perform primary verification on the data identifier using a data fingerprint; A secondary verification module is configured to obtain a fingerprint of the target client when the primary verification is passed, and perform secondary verification on the target client according to the fingerprint of the target client; The third-level verification module is configured to, if the second-level verification is passed, obtain the login information input by the target user at the target client, and send the first dynamic verification information and the second dynamic verification information to the target client by using a dynamic two-factor authentication mechanism; obtain the third dynamic verification information and the fourth dynamic verification information input by the target user at the target client, and perform a third-level verification on the third dynamic verification information and the fourth dynamic verification information according to the first dynamic verification information and the second dynamic verification information; The ultimate verification module is configured to, when the three-level verification is passed, use an environmental risk index model to evaluate the operating environment risk of the target user to obtain an environmental risk evaluation value, and when the environmental risk evaluation value is less than a preset threshold, pass the ultimate verification of the target user; The token generation module is configured to adopt an ABAC-RAdAC hybrid model to dynamically authorize the target user and generate a temporary token for the target user.

Images