Cloud office data authority control method based on block chain

Abstract

The invention discloses a cloud office data authority control method based on a block chain, and the method comprises the following steps: S1, collecting user behavior data, carrying out the preprocessing, building a user behavior model, and analyzing the behavior mode of a user through an improved mixed attention network; s2, collecting identity verification data of the user, and analyzing the identity verification data; s3, dynamically adjusting the access authority of the user to the cloud office data, and intelligently updating the authority according to the work demand and the authority condition; s4, recording permission change information and data access logs of all users in the block chain; s5, abnormal behaviors of the user are monitored in real time, and potential security threats are identified; and S6, automatically executing the authority control rule through the block chain smart contract. According to the method, deep learning and block chain technologies are combined, intelligent dynamic authority management and abnormal behavior detection are realized, and the method has the advantages of high security, high flexibility and transparent auditing.

Description

Technical Field

[0001] The present invention relates to the field of cloud computing technology, and in particular to a cloud office data authority control method based on blockchain. Background Art

[0002] With the rapid development of information technology, cloud computing has become the primary platform for information management and data storage in modern enterprises. This is particularly true in the cloud office sector, where businesses and employees increasingly rely on cloud platforms for daily work and communication. While this cloud-based work environment offers significant convenience, it also poses new security risks. Data permission control, identity authentication, and data access auditing have become core issues that urgently need to be addressed in cloud office. Currently, many enterprises still rely on traditional data permission control methods based on static permission models. This approach is unable to adapt to the complex modern work environment and ever-changing business needs, resulting in lagging and inflexible permission management, significantly impacting data security and system efficiency.

[0003] Traditional permission management systems typically use a static permission allocation model, meaning that when a user obtains permission, the administrator configures fixed access rights for them. A significant drawback of this approach is its lack of flexibility. When a user's role or work requirements change, the static permission model cannot be dynamically adjusted in a timely manner, potentially leading to problems with too many or too few permissions. Furthermore, static permission allocation cannot intelligently respond to user behavior patterns or real-time needs, creating the risk of over-authorization or abuse of permissions. For businesses that need to respond quickly to market changes, the efficiency and security of the static permission model fall far short of meeting requirements.

[0004] In terms of identity authentication, traditional authentication methods, such as username and password verification, although widely used, have poor security and are vulnerable to attacks. Security issues such as password leakage and brute force cracking are emerging in an endless stream. Especially in remote office and distributed work environments, password systems appear to be more vulnerable. Although some companies have begun to adopt biometric technologies such as two-factor authentication, facial recognition, and fingerprint recognition to enhance security, these technologies still have problems such as complex operation, large computing resource usage, and long response time, and cannot fully and effectively prevent illegal access. In addition, traditional authentication methods usually rely on centralized identity authentication systems, which means that once the system is attacked, it may lead to the exposure of the data security of the entire enterprise.

[0005] In addition, with the increase in data access volume and frequent changes in user permissions, how to record and audit data access behavior has also become a major problem in cloud office systems. Currently, data access log records on many cloud platforms are still incomplete, opaque, and difficult to trace. As a result, when data leaks or security incidents occur, it is difficult for enterprises to trace the specific responsible persons and event details through logs. Traditional log recording systems mostly use centralized storage and management methods, which makes log information susceptible to tampering and forgery, reducing the credibility and effectiveness of data audits.

[0006] A common problem with existing technologies in the cloud office field is the lack of comprehensive and intelligent management of functions such as data permissions, identity authentication, and abnormal behavior monitoring. Existing permission control systems are often based on static permission models and traditional identity authentication mechanisms, and cannot be dynamically adjusted according to user behavior patterns, work needs, and system changes. This makes it difficult for enterprises to cope with the ever-changing permission requirements and security challenges in complex cloud office environments. Especially in the context of large-scale distributed offices and remote offices, traditional permission management and identity authentication systems cannot effectively ensure data security, reduce operational risks, and cannot achieve full monitoring and auditing of the entire data access process, posing huge potential risks to enterprises.

[0007] In order to make up for the shortcomings of existing technologies, blockchain technology has gradually been widely used in the fields of permission management and data auditing due to its characteristics of being tamper-proof, decentralized, and traceable. The application of blockchain can ensure the transparency and tamper-proofness of data access and permission change records, effectively improving data security. At the same time, the application of artificial intelligence technologies such as deep learning in abnormal behavior monitoring also provides new solutions for dynamic permission adjustment and user identity authentication. By combining deep learning and blockchain technology, intelligent, dynamic and automated permission management can be realized in cloud office environments, which can not only greatly improve data security, but also improve the system's response speed and management efficiency.

[0008] However, the existing cloud office data permission control solutions based on blockchain and artificial intelligence still have some technical bottlenecks. How to achieve real-time permission adjustment and intelligent abnormal behavior detection while ensuring system performance, and how to more accurately and efficiently integrate blockchain technology to achieve complete data auditing are still difficult problems that need to be solved urgently by current technology.

[0009] Therefore, how to provide a blockchain-based cloud office data permission control method is an urgent problem that technicians in this field need to solve. Summary of the Invention

[0010] One purpose of the present invention is to propose a cloud office data permission control method based on blockchain. The present invention combines deep learning, improved hybrid attention network and blockchain technology to innovatively realize intelligent dynamic permission management, identity authentication, abnormal behavior monitoring and data auditing functions. It adjusts data access rights in real time through intelligent analysis of user behavior patterns, and uses blockchain to ensure the non-tamperability of permission changes and log records. It has the advantages of high security, strong flexibility, rapid response and transparent auditing.

[0011] A blockchain-based cloud office data authority control method according to an embodiment of the present invention includes the following steps:

[0012] S1. Collect user behavior data, perform preprocessing, and establish a user behavior model. The user behavior model analyzes user behavior patterns through an improved hybrid attention network and extracts user behavior features at different levels.

[0013] S2. Collect the user's authentication data, analyze the user's authentication data, verify the user's identity, and grant or revoke permissions based on the verification results;

[0014] S3. Based on the extracted user behavior characteristics and the user behavior model, dynamically adjust the user's access rights to cloud office data, and intelligently update the rights according to the real-time changing work needs and permission conditions;

[0015] S4. Record all users’ permission change information and data access logs in the blockchain;

[0016] S5. Monitor users' abnormal behavior in real time, identify potential security threats, and automatically trigger alarms when abnormal behavior is detected, restricting the permissions of users with abnormal behavior;

[0017] S6. Automatically execute permission control rules through blockchain smart contracts, automatically assign, revoke or adjust user permissions under specific conditions, and reduce human intervention and operational errors.

[0018] Optionally, the identity verification data includes facial recognition, voice recognition and biometrics.

[0019] Optionally, the S1 specifically includes:

[0020] S11. Collect user behavior data, including user access history, access time, operation frequency, project progress, work content, device usage, geographic location, and access path;

[0021] S12. Preprocessing the user behavior data, wherein the preprocessing includes removing invalid data, filling missing data, and standardizing and normalizing the data;

[0022] S13. Establish a user behavior model, input the preprocessed user behavior data into the user behavior model, and extract user behavior features at different levels through an improved hybrid attention network. The improved hybrid attention network includes a self-attention mechanism, a graph convolutional network, and a multi-level attention mechanism.

[0023] S14. In the graph convolutional network, a graph structure is constructed and a graph convolution operation is performed. The graph structure represents the relationship between user behaviors and the dependency between nodes:

[0024]

[0025] Among them, H (l+1) represents the node features of the l+1 layer, H (l) represents the node features of the lth layer, W (l) represents the weight matrix of the lth layer, b (l) represents the bias term, σ represents the nonlinear activation function, Represents the normalized adjacency matrix, which represents the relationship between nodes in the graph structure;

[0026] S15. Use the self-attention mechanism to calculate the correlation between each time step of user behavior data:

[0027]

[0028] Among them, α ij represents the correlation between the i-th time step and the j-th time step, that is, the attention weight, exp represents the natural exponential function, Q i represents the query vector at the i-th time step, K j represents the key vector at the i-th time step, represents the transpose operation of the key vector at the i-th time step, represents the transpose operation of the key vector at the kth time step, T represents the transpose operation, n represents the total number of time steps, γ ij Indicates an additional offset based on user behavior habits, used to introduce additional contextual information;

[0029] S16. The correlation between each time step calculated by the self-attention mechanism is weighted and combined to generate a weighted representation of each time step:

[0030]

[0031] Among them, Z i represents the weighted representation of the i-th time step, α ij represents the attention weight, V j represents the value vector at the jth time step, δi Represents the bias term calculated by the multi-level attention mechanism, which is used to enhance the effect of information fusion;

[0032] S17. The node features obtained through the graph convolutional network are fused with the output of the self-attention mechanism to generate the final user behavior representation:

[0033]

[0034] Among them, H final represents the final user behavior, λ c Represents the weight of each channel, H c represents the output of the cth channel, Z i represents the weighted representation of the i-th time step, ζ c represents the learning bias term for additional optimization through the graph convolutional network, ⊕ represents the splicing operation, and C represents the number of channels;

[0035] S18. Use the fully connected layer to further process the final user behavior representation, extract high-order features, and generate a user behavior feature vector:

[0036]

[0037] Among them, V user represents the final user behavior feature vector, σ represents the nonlinear activation function, W represents the weight matrix of the fully connected layer, and H final represents the final user behavior representation, b represents the bias term, Represents an adjustment factor to further optimize the accuracy of permission control.

[0038] Optionally, the S3 specifically includes:

[0039] S31. Calculate the permission requirements of each user based on the user behavior feature vector generated by the user behavior model:

[0040]

[0041] Among them, P i represents the permission requirement of the i-th user, W f and W′ f represents the weight matrix, V user represents the user behavior feature vector, b f represents the bias term, α represents the adaptive adjustment factor, and exp represents the natural exponential function;

[0042] S32. Calculate the user's permission level based on the permission requirements and the system's permission control rules:

[0043]

[0044] Among them, L i Indicates the permission level of the i-th user, W p Represents the weight matrix of the permission control policy, b p Represents the bias term, using the Sigmoid function to ensure that the output of the permission level is between 0 and 1, reflecting the proportion of different permissions;

[0045] S33, the authority level L i Combined with the user's authentication result, the user's final access rights are calculated:

[0046] A i =min(L i ,I i )+(1-min(L i ,I i ))·φ i ;

[0047] Among them, A i Indicates the access rights of the i-th user, min(L i ,I i ) represents the minimum value between the permission level and the user's authentication result, I i Indicates the user's authentication result, φ i Represents an adjustment factor, which weights and adjusts permissions based on the user's historical behavior and behavior patterns:

[0048]

[0049] Among them, γ represents the hyperparameter, W phi Represents the historical behavior weight matrix;

[0050] S34. Dynamically adjust permission requirements based on real-time changing work requirements and permission conditions:

[0051]

[0052] Among them, P′ i Indicates the adjusted permission requirements, W g Represents the weight matrix of the mapping function, D i Represents real-time data related to user work, C i represents the complexity of the task, b g represents the bias term, Represents element-by-element multiplication, which is used to simultaneously consider the impact of work data and task complexity on permission requirements;

[0053] S35. Recalculate the user's permission level based on the adjusted permission requirements:

[0054]

[0055] Among them, L' i Indicates the adjusted permission level, P' i Indicates the adjusted permission requirements;

[0056] S36. Adjust the user's access rights based on the user's dynamic permission requirements and the final calculated permission level:

[0057]

[0058] Among them, A' i Represents the adjusted access rights, L' i Indicates the adjusted permission level. Represents an adjustment factor that weights permissions based on the user's latest behavior history.

[0059] Optionally, the S4 specifically includes:

[0060] S41. When a user performs a data access operation, the detailed information of each access operation is recorded, including the user ID, access time, type of accessed data, scope of accessed data, and access permission type:

[0061] L i =[ID i ,T i ,D i ,P i ];

[0062] Among them, L i Indicates the access log of the i-th user, ID i Represents the user's unique identifier, T i Indicates the timestamp of the access operation, D i Indicates the data type and data range to be accessed, P i Indicates the type of permission the user has in this operation;

[0063] S42. When recording a change in user permissions, record detailed information about the change in user access permissions, including user ID, original permission value, new permission value, permission change time, and reason for change:

[0064] L ′ i =[ID i ,T ′ i ,P old,i ,P new,i ,R i ];

[0065] Among them, L ′i Indicates user permission change log, ID i Represents the user's unique identifier, T ′ i Indicates the timestamp of permission change, P old,i Indicates the permission value before the user operates, P new,i Indicates the new permission value after the user operation, R i Indicates the reason for the change in authority;

[0066] S43. Encrypt the user access log and the user authority change log:

[0067]

[0068] in, and They represent the encrypted access log and permission change log respectively, Enc represents the encryption operation, and K represents the encryption key;

[0069] S44. Use blockchain technology to record encrypted access logs and permission change logs in the blockchain:

[0070]

[0071] Among them, B i and B′ i Respectively represent the B′ i The access log and permission change log of each user are recorded in the blockchain. Block represents the blockchain. i and T′ i Indicates the timestamp of the record, H i and H′ i Indicates the hash value of the corresponding block, and Represents the encrypted access log and permission change log respectively;

[0072] S45. Ensure that each access log and permission change log is tamper-proof and traceable in the blockchain:

[0073] Hash(B i )=H i andHash(B′ i )=H′ i ;

[0074] Among them, Hash represents hash operation, H i and H′ i Indicates the hash value of the block;

[0075] S46. Each time a log record is made in the blockchain, the state of the blockchain is dynamically updated and the new block is added to the chain:

[0076] B i+1 =AddBlock(B i ,H i ,T i );

[0077] Among them, B i+1 It represents the record of the access log of the i+1th user in the blockchain, AddBlock represents the block addition operation, and H i Indicates the hash value of the previous block, B i Represents the access log record of the i-th user in the blockchain.

[0078] The beneficial effects of the present invention are:

[0079] First, by combining deep learning and an improved hybrid attention network, the present invention can effectively identify user behavior patterns through intelligent user behavior modeling, and dynamically adjust permissions based on multi-dimensional data such as user access habits, work content, and project progress. This innovative function can not only address the problem that traditional static permission models cannot respond to changes in user needs in real time, but can also adjust permissions according to real-time changing working conditions and task requirements, thereby improving the flexibility and adaptability of the system.

[0080] Secondly, traditional identity authentication methods often have problems such as poor security and complex operation. The present invention combines multiple identity authentication technologies such as facial recognition, voice recognition and biometrics to construct a multi-factor identity authentication mechanism. This method enhances the security of user identity authentication, and through the intelligent application of deep learning technology in identity authentication, it improves the efficiency and accuracy of the authentication process, greatly reduces the loopholes and errors in traditional methods, and improves the overall security of the system.

[0081] In addition, in terms of permission management and data auditing, blockchain technology is used to record all user permission changes and data access logs to ensure the transparency and non-tamperability of the permission control process. This blockchain-based audit mechanism can not only trace back to specific operation records, but also provide enterprises with detailed audit evidence when security incidents occur, reducing the risk of forgery and tampering in traditional centralized log systems, thereby enhancing the security and credibility of data access. With the decentralized nature of blockchain, the security and data consistency of the entire system are effectively guaranteed.

[0082] Finally, this invention accurately identifies potential security threats by monitoring abnormal user behavior in real time. It then automatically executes permission control rules through smart contracts, enabling timely responses to abnormal behavior and adjustments to access rights. Compared to traditional methods, this anomaly detection system based on deep learning and blockchain technology offers greater accuracy and real-time performance. It can detect and respond to threats in real time during data access, reducing manual intervention and operational errors and improving the system's automation level.

[0083] In summary, the present invention realizes the intelligent, flexible and automated control of cloud office data permissions by combining deep learning, blockchain and smart contract technology. Compared with traditional technologies, the present invention not only improves the accuracy and security of permission management, but also improves the response speed and overall efficiency of the system. It has strong technical innovation and practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0085] Figure 1 This is a flowchart of a blockchain-based cloud office data authority control method proposed by the present invention;

[0086] Figure 2 This is a schematic diagram of the application of an improved hybrid attention network in the user behavior model of the blockchain-based cloud office data permission control method proposed in the present invention. DETAILED DESCRIPTION

[0087] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0088] refer to Figure 1 and Figure 2 , a cloud office data permission control method based on blockchain, comprising the following steps:

[0089] S1. Collect user behavior data, perform preprocessing, and establish a user behavior model. The user behavior model analyzes user behavior patterns through an improved hybrid attention network and extracts user behavior features at different levels.

[0090] S2. Collect the user's authentication data, analyze the user's authentication data, verify the user's identity, and grant or revoke permissions based on the verification results;

[0091] S3. Based on the extracted user behavior characteristics and the user behavior model, dynamically adjust the user's access rights to cloud office data, and intelligently update the rights according to the real-time changing work needs and permission conditions;

[0092] S4. Record all users’ permission change information and data access logs in the blockchain;

[0093] S5. Monitor users' abnormal behavior in real time, identify potential security threats, and automatically trigger alarms when abnormal behavior is detected, restricting the permissions of users with abnormal behavior;

[0094] S6. Automatically execute permission control rules through blockchain smart contracts, automatically assign, revoke or adjust user permissions under specific conditions, and reduce human intervention and operational errors.

[0095] In this embodiment, the identity verification data includes facial recognition, voice recognition and biometrics.

[0096] In this embodiment, S1 specifically includes:

[0097] S11. Collect user behavior data, including user access history, access time, operation frequency, project progress, work content, device usage, geographic location, and access path;

[0098] S12. Preprocessing the user behavior data, wherein the preprocessing includes removing invalid data, filling missing data, and standardizing and normalizing the data;

[0099] S13. Establish a user behavior model, input the preprocessed user behavior data into the user behavior model, and extract user behavior features at different levels through an improved hybrid attention network. The improved hybrid attention network includes a self-attention mechanism, a graph convolutional network, and a multi-level attention mechanism.

[0100] S14. In the graph convolutional network, a graph structure is constructed and a graph convolution operation is performed. The graph structure represents the relationship between user behaviors and the dependency between nodes:

[0101]

[0102] Among them, H (l+1) represents the node features of the l+1 layer, H (l) represents the node features of the lth layer, W (l) represents the weight matrix of the lth layer, b (l) represents the bias term, σ represents the nonlinear activation function, Represents the normalized adjacency matrix, which represents the relationship between nodes in the graph structure;

[0103] S15. Use the self-attention mechanism to calculate the correlation between each time step of user behavior data:

[0104]

[0105] Among them, α ij represents the correlation between the i-th time step and the j-th time step, that is, the attention weight, exp represents the natural exponential function, Q i represents the query vector at the i-th time step, K j represents the key vector at the i-th time step, represents the transpose operation of the key vector at the i-th time step, represents the transpose operation of the key vector at the kth time step, T represents the transpose operation, n represents the total number of time steps, γ ij Indicates an additional offset based on user behavior habits, used to introduce additional contextual information;

[0106] S16. The correlation between each time step calculated by the self-attention mechanism is weighted and combined to generate a weighted representation of each time step:

[0107]

[0108] Among them, Z i represents the weighted representation of the i-th time step, α ij represents the attention weight, V j represents the value vector at the jth time step, δ i Represents the bias term calculated by the multi-level attention mechanism, which is used to enhance the effect of information fusion;

[0109] S17. The node features obtained through the graph convolutional network are fused with the output of the self-attention mechanism to generate the final user behavior representation:

[0110]

[0111] Among them, H final represents the final user behavior, λ c Represents the weight of each channel, H c represents the output of the cth channel, Z i represents the weighted representation of the i-th time step, ζ c represents the learning bias term for additional optimization through the graph convolutional network, ⊕ represents the splicing operation, and C represents the number of channels;

[0112] S18. Use the fully connected layer to further process the final user behavior representation, extract high-order features, and generate a user behavior feature vector:

[0113]

[0114] Among them, V user represents the final user behavior feature vector, σ represents the nonlinear activation function, W represents the weight matrix of the fully connected layer, and H final represents the final user behavior representation, b represents the bias term, Represents an adjustment factor to further optimize the accuracy of permission control.

[0115] In this embodiment, S3 specifically includes:

[0116] S31. Calculate the permission requirements of each user based on the user behavior feature vector generated by the user behavior model:

[0117]

[0118] Among them, P i represents the permission requirement of the i-th user, W f and W′ f represents the weight matrix, V user represents the user behavior feature vector, b f represents the bias term, α represents the adaptive adjustment factor, and exp represents the natural exponential function;

[0119] S32. Calculate the user's permission level based on the permission requirements and the system's permission control rules:

[0120]

[0121] Among them, L i Indicates the permission level of the i-th user, W p Represents the weight matrix of the permission control policy, b p Represents the bias term, using the Sigmoid function to ensure that the output of the permission level is between 0 and 1, reflecting the proportion of different permissions;

[0122] S33, the authority level L i Combined with the user's authentication result, the user's final access rights are calculated:

[0123] A i =min(L i ,I i )+(1-min(L i ,I i ))·φ i ;

[0124] Among them, A i Indicates the access rights of the i-th user, min(L i ,I i) represents the minimum value between the permission level and the user's authentication result, I i Indicates the user's authentication result, φ i Represents an adjustment factor, which weights and adjusts permissions based on the user's historical behavior and behavior patterns:

[0125]

[0126] Among them, γ represents the hyperparameter, W phi Represents the historical behavior weight matrix;

[0127] S34. Dynamically adjust permission requirements based on real-time changing work requirements and permission conditions:

[0128]

[0129] Among them, P′ i Indicates the adjusted permission requirements, W g Represents the weight matrix of the mapping function, D i Represents real-time data related to user work, C i represents the complexity of the task, b g represents the bias term, Represents element-by-element multiplication, which is used to simultaneously consider the impact of work data and task complexity on permission requirements;

[0130] S35. Recalculate the user's permission level based on the adjusted permission requirements:

[0131]

[0132] Among them, L' i Indicates the adjusted permission level, P' i Indicates the adjusted permission requirements;

[0133] S36. Adjust the user's access rights based on the user's dynamic permission requirements and the final calculated permission level:

[0134]

[0135] Among them, A' i Represents the adjusted access rights, L' i Indicates the adjusted permission level. Represents an adjustment factor that weights permissions based on the user's latest behavior history.

[0136] In this embodiment, the S4 specifically includes:

[0137] S41. When a user performs a data access operation, the detailed information of each access operation is recorded, including the user ID, access time, type of accessed data, scope of accessed data, and access permission type:

[0138] L i =[ID i ,T i ,D i ,P i ];

[0139] Among them, L i Indicates the access log of the i-th user, ID i Represents the user's unique identifier, T i Indicates the timestamp of the access operation, D i Indicates the data type and data range to be accessed, P i Indicates the type of permission the user has in this operation;

[0140] S42. When recording a change in user permissions, record detailed information about the change in user access permissions, including user ID, original permission value, new permission value, permission change time, and reason for change:

[0141] L ′ i =[ID i ,T ′ i ,P old,i ,P new,i ,R i ];

[0142] Among them, L ′ i Indicates user permission change log, ID i Represents the user's unique identifier, T ′ i Indicates the timestamp of permission change, P old,i Indicates the permission value before the user operates, P new,i Indicates the new permission value after the user operation, R i Indicates the reason for the change in authority;

[0143] S43. Encrypt the user access log and the user authority change log:

[0144]

[0145] in, and They represent the encrypted access log and permission change log respectively, Enc represents the encryption operation, and K represents the encryption key;

[0146] S44. Use blockchain technology to record encrypted access logs and permission change logs in the blockchain:

[0147]

[0148] Among them, B i and B′ i Respectively represent the B′ i The access log and permission change log of each user are recorded in the blockchain. Block represents the blockchain. i and T′ i Indicates the timestamp of the record, H i and H′ i Indicates the hash value of the corresponding block, and Represents the encrypted access log and permission change log respectively;

[0149] S45. Ensure that each access log and permission change log is tamper-proof and traceable in the blockchain:

[0150] Hash(B i )=H i andHash(B′ i )=H′ i ;

[0151] Among them, Hash represents hash operation, H i and H′ i Indicates the hash value of the block;

[0152] S46. Each time a log record is made in the blockchain, the state of the blockchain is dynamically updated and the new block is added to the chain:

[0153] B i+1 =AddBlock(B i ,H i ,T i );

[0154] Among them, B i+1 It represents the record of the access log of the i+1th user in the blockchain, AddBlock represents the block addition operation, and H i Indicates the hash value of the previous block, B i Represents the access log record of the i-th user in the blockchain.

[0155] Example 1:

[0156] In order to verify the feasibility of the present invention in implementation, the present invention was applied to the cloud office platform of a large multinational enterprise. The daily operations of the enterprise involve a large amount of sensitive data access and frequent cross-departmental collaboration. As the scale of the enterprise continues to expand, the management's requirements for data permission control and access security are increasing. The traditional static permission management method can no longer meet the needs of real-time dynamic adjustment. The enterprise has encountered sensitive data leakage due to permission abuse or permission expiration, which seriously affected the company's reputation and operational security. To solve this problem, the enterprise decided to introduce an intelligent permission management system based on blockchain and deep learning.

[0157] In actual applications, corporate employees need to perform various data processing, file sharing, project collaboration and other operations through cloud office platforms every day. Each employee will have different data access rights based on their position, work content and project needs. In traditional management methods, once permissions are assigned, it is difficult to adjust them dynamically, which leads to permission management problems between employees in different departments. What is more serious is that some employees access data that is not within their job scope without timely approval, creating the risk of data leakage.

[0158] In order to improve this situation, the cloud office data permission control method based on blockchain technology of the present invention was introduced into the enterprise's cloud office platform. By introducing intelligent dynamic permission adjustment and abnormal behavior monitoring mechanism, the system can automatically adjust permissions in real time according to user behavior patterns and work needs and ensure that all access operations can be traced and audited.

[0159] First, the system collects and preprocesses employees' behavioral data to establish a behavioral model for each employee. These behavioral data include employee access history, access time, operation frequency, work content, equipment usage, etc. To ensure the accuracy of the data, the system cleans and standardizes these data, thereby providing high-quality input data for subsequent behavioral analysis.

[0160] After establishing a user behavior model, the system analyzes employee behavior using an improved hybrid attention network. By leveraging self-attention, graph convolutional networks, and multi-level attention mechanisms, the system extracts multi-layered features from employee behavior, accurately identifying employee behavior patterns in specific work environments. These patterns not only help the system identify employees' daily work habits but also automatically adjust permissions based on real-time work needs. For example, when an employee requires temporary access to more sensitive data for a specific project, the system dynamically increases their access permissions. Once the project concludes, these permissions are automatically revoked, avoiding the risks of over-authorization and abuse.

[0161] At the same time, the system adopts a multi-factor identity authentication method for employee identity verification. Through technologies such as facial recognition, voice recognition and biometrics, it ensures that the identity verification of each employee is more secure and reliable. After successful identity verification, the system automatically grants or revokes permissions based on behavioral models, greatly reducing the possibility of manual intervention and operational errors.

[0162] Furthermore, to ensure that every permission change and data access is transparently recorded and tamper-proof, all permission change information and data access logs are recorded on the blockchain. The introduction of blockchain technology ensures the immutability and traceability of log information. When the system detects abnormal behavior, it automatically triggers an alarm and restricts the employee's access rights. At the same time, all abnormal behavior and permission adjustments are automatically recorded and stored on the blockchain, ensuring data security and audit transparency.

[0163] Table 1 Experimental data comparison table

[0164]

[0165] In terms of the average response time for permission adjustments, before implementation, the system took approximately 48 hours to complete a single permission adjustment. After implementation, this response time was significantly reduced to just 5 minutes, a 90% reduction. This improvement significantly improves the efficiency of permission management, especially in work environments with rapidly changing business needs. It can adapt to dynamic adjustments to employee permissions in real time, avoiding potential work inefficiencies or data security risks caused by delayed permission adjustments.

[0166] The improvement in the accuracy of abnormal behavior identification also demonstrates the intelligence level of the system. Before implementation, the accuracy of abnormal behavior identification was only 75%, but after implementation, the recognition accuracy increased to 98%. This improvement is due to the deep learning model and multi-level behavior analysis mechanism adopted by this invention, which enables the system to more accurately identify employees' abnormal behavior and automatically adjust permissions through smart contracts, avoiding security risks caused by human misjudgment or omissions.

[0167] In terms of the number of data access security incidents, before the implementation, the company experienced approximately 12 security incidents per month. These incidents were usually related to employee abuse of authority or unauthorized access to data. However, after implementation, the system was able to automatically detect and block unauthorized access, successfully eliminating the occurrence of security incidents and achieving a 100% improvement. Through the introduction of blockchain technology, all permission changes and data access are transparently recorded and audited, greatly improving data security.

[0168] In addition, before the implementation of the system, there were about 6 cases of permission abuse per month, mainly due to the abuse of permissions that were not revoked in time by employees. After the implementation of the system, these problems were completely resolved and the number of permission abuses dropped to 0. This was due to the system's automated permission revocation mechanism. Especially after the project is completed, the system can automatically reclaim expired or no longer needed permissions, thereby avoiding unnecessary permission leakage and abuse.

[0169] The traceability of data auditing has been improved after implementation. Through blockchain technology, all access operations are clearly recorded, ensuring the transparency and non-tamperability of data auditing. This not only improves the traceability of data access, but also provides strong security protection for enterprises. Especially in the event of data leakage or security issues, the blockchain log can be used to quickly trace the source, find the responsible person and take corresponding measures.

[0170] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A cloud office data authority control method based on blockchain, characterized in that: The steps include: S1. Collect user behavior data, perform preprocessing, and establish a user behavior model. The user behavior model analyzes user behavior patterns through an improved hybrid attention network and extracts user behavior features at different levels. S2. Collect the user's authentication data, analyze the user's authentication data, verify the user's identity, and grant or revoke permissions based on the verification results; S3. Based on the extracted user behavior characteristics and the user behavior model, dynamically adjust the user's access rights to cloud office data, and intelligently update the rights according to the real-time changing work needs and permission conditions; S4. Record all users’ permission change information and data access logs in the blockchain; S5. Monitor abnormal user behavior in real time, identify potential security threats, and automatically trigger alarms when abnormal behavior is detected, restricting the permissions of users with abnormal behavior; S6. Automatically execute permission control rules through blockchain smart contracts, automatically assign, revoke or adjust user permissions under specific conditions, and reduce human intervention and operational errors.

2. A cloud office data authority control method based on blockchain according to claim 1, characterized in that: The authentication data includes facial recognition, voice recognition and biometrics.

3. A cloud office data authority control method based on blockchain according to claim 1, characterized in that: Said S1 specifically includes: S11. Collect user behavior data, including user access history, access time, operation frequency, project progress, work content, device usage, geographic location, and access path; S12. Preprocessing the user behavior data, wherein the preprocessing includes removing invalid data, filling missing data, and standardizing and normalizing the data; S13. Establish a user behavior model, input the preprocessed user behavior data into the user behavior model, and extract user behavior features at different levels through an improved hybrid attention network. The improved hybrid attention network includes a self-attention mechanism, a graph convolutional network, and a multi-level attention mechanism. S14. In the graph convolutional network, a graph structure is constructed and a graph convolution operation is performed. The graph structure represents the relationship between user behaviors and the dependency between nodes: Among them, H (l+1) represents the node features of the l+1 layer, H (l) represents the node features of the lth layer, W (l) represents the weight matrix of the lth layer, b (l) represents the bias term, σ represents the nonlinear activation function, Represents the normalized adjacency matrix, which represents the relationship between nodes in the graph structure; S15. Use the self-attention mechanism to calculate the correlation between each time step of user behavior data: Among them, α ij represents the correlation between the i-th time step and the j-th time step, that is, the attention weight, exp represents the natural exponential function, Q i represents the query vector at the i-th time step, K j represents the key vector at the i-th time step, represents the transpose operation of the key vector at the i-th time step, represents the transpose operation of the key vector at the kth time step, T represents the transpose operation, n represents the total number of time steps, γ ij Indicates an additional offset based on user behavior habits, used to introduce additional contextual information; S16. The correlation between each time step calculated by the self-attention mechanism is weighted and combined to generate a weighted representation of each time step: Among them, Z i represents the weighted representation of the i-th time step, α ij represents the attention weight, V j represents the value vector at the jth time step, δ i Represents the bias term calculated by the multi-level attention mechanism, which is used to enhance the effect of information fusion; S17. The node features obtained through the graph convolutional network are fused with the output of the self-attention mechanism to generate the final user behavior representation: Among them, H final represents the final user behavior representation, λ c Represents the weight of each channel, H c represents the output of the cth channel, Z i represents the weighted representation of the i-th time step, represents the learning bias term for additional optimization through the graph convolutional network, represents the splicing operation, and C represents the number of channels; S18. Use the fully connected layer to further process the final user behavior representation, extract high-order features, and generate a user behavior feature vector: Among them, V user represents the final user behavior feature vector, σ represents the nonlinear activation function, W represents the weight matrix of the fully connected layer, and H final represents the final user behavior representation, b represents the bias term, Represents an adjustment factor to further optimize the accuracy of permission control.

4. A cloud office data authority control method based on blockchain according to claim 1, characterized in that: The S3 specifically includes: S31. Calculate the permission requirements of each user based on the user behavior feature vector generated by the user behavior model: Among them, P i represents the permission requirement of the i-th user, W f and W′ f represents the weight matrix, V user represents the user behavior feature vector, b f represents the bias term, α represents the adaptive adjustment factor, and exp represents the natural exponential function; S32. Calculate the user's permission level based on the permission requirements and the system's permission control rules: Among them, L i Indicates the permission level of the i-th user, W p Represents the weight matrix of the permission control policy, b p Represents the bias term, using the Sigmoid function to ensure that the output of the permission level is between 0 and 1, reflecting the proportion of different permissions; S33, the authority level L i Combined with the user's authentication result, the user's final access rights are calculated: A i =min(L i ,TO i )+(1-min(L i ,TO i ))·φ i 4 Among them, A i Indicates the access rights of the i-th user, min(L i ,I i ) represents the minimum value between the permission level and the user's authentication result, I i Indicates the user's authentication result, φ i Represents an adjustment factor, which weights and adjusts permissions based on the user's historical behavior and behavior patterns: Among them, γ represents the hyperparameter, W phi Represents the historical behavior weight matrix; S34. Dynamically adjust permission requirements based on real-time changing work requirements and permission conditions: Among them, P′ i Indicates the adjusted permission requirements, W g Represents the weight matrix of the mapping function, D i Represents real-time data related to user work, C i represents the complexity of the task, b g represents the bias term, Represents element-by-element multiplication, which is used to simultaneously consider the impact of work data and task complexity on permission requirements; S35. Recalculate the user's permission level based on the adjusted permission requirements: Among them, L' i Indicates the adjusted permission level, P' i Indicates the adjusted permission requirements; S36. Adjust the user's access rights based on the user's dynamic permission requirements and the final calculated permission level: Among them, A' i Indicates the adjusted access rights, L' i Indicates the adjusted permission level. Represents an adjustment factor that weights permissions based on the user's latest behavior history.

5. A cloud office data authority control method based on blockchain according to claim 1, characterized in that: The S4 specifically includes: S41. When a user performs a data access operation, the detailed information of each access operation is recorded, including the user ID, access time, type of accessed data, scope of accessed data, and access permission type: L i =[ID i ,T i ,D i ,P i ]; Among them, L i Indicates the access log of the i-th user, ID i Represents the user's unique identifier, T i Indicates the timestamp of the access operation, D i Indicates the data type and data range to be accessed, P i Indicates the type of permission the user has in this operation; S42. When recording a change in user permissions, record detailed information about the change in user access permissions, including user ID, original permission value, new permission value, permission change time, and reason for change: L′ i =[ID i ,T′ i ,P old,i ,P new,i ,R i ]; Among them, L′ i Indicates user permission change log, ID i represents the user's unique identifier, T′ i Indicates the timestamp of permission change, P old,i Indicates the permission value before the user operates, P new,i Indicates the new permission value after the user operation, R i Indicates the reason for the change in authority; S43. Encrypt the user access log and the user authority change log: in, and They represent the encrypted access log and permission change log respectively, Enc represents the encryption operation, and K represents the encryption key; S44. Use blockchain technology to record encrypted access logs and permission change logs in the blockchain: Among them, B i and B′ i Respectively represent the B′ i The access log and permission change log of each user are recorded in the blockchain. Block represents the blockchain. i and T′ i Indicates the timestamp of the record, H i and H′ i Indicates the hash value of the corresponding block, and Respectively represent the encrypted access log and permission change log; S45. Ensure that each access log and permission change log is tamper-proof and traceable in the blockchain: Hash(B i )=H i andHash(B′ i )=H′ i ; Among them, Hash represents hash operation, H i and H′ i Indicates the hash value of the block; S46. Each time a log record is made in the blockchain, the state of the blockchain is dynamically updated and the new block is added to the chain: B i+1 =AddBlock(B i ,H i ,T i ); Among them, B i+1 It represents the record of the access log of the i+1th user in the blockchain, AddBlock represents the block addition operation, and H i Indicates the hash value of the previous block, B i Represents the access log record of the i-th user in the blockchain.

Images