Data property right registration method and system for dynamic encryption evidence storage based on block chain

Abstract

The invention provides a block chain-based dynamic encryption evidence storage data property right registration method and system, and relates to the technical field of block chains, and the method comprises the steps: carrying out the trusted measurement of to-be-registered data through constructing a trust transfer network; establishing a dynamic verification mechanism and selecting a verification path; constructing a data block association graph based on verification node importance to realize distributed storage; and finally, submitting the storage index and the verification summary information to a block chain network to complete evidence storage and generate a registration certificate. According to the invention, the credibility and verification efficiency of data property right registration are improved, and the security and integrity of evidence storage data are enhanced.

Description

Technical Field

[0001] The present invention relates to blockchain technology, and in particular to a data property rights registration method and system based on blockchain-based dynamic encrypted evidence storage. Background Art

[0002] The current data property registration method mainly relies on centralized institutions for management, and confirms data ownership through simple digital signatures and timestamps. However, this approach faces challenges in data credibility measurement, dynamic verification, and storage security.

[0003] Current data property registration and evidence storage technologies have obvious deficiencies. First, existing technologies lack a multi-dimensional quantitative assessment mechanism for data credibility. Most of them adopt a single-dimensional trusted proof method, which cannot fully reflect the trusted transmission relationship of data in different trust networks, resulting in a weak trust foundation for data property rights recognition. Secondly, existing verification mechanisms are mostly static in design, which makes it difficult to cope with scenarios where data value and security threats change dynamically. There is a lack of effective coordination mechanisms between verification units, and they are unable to perceive anomalies and make corresponding adjustments, which reduces the reliability of verification results. Finally, existing distributed storage solutions fail to fully consider the importance of verification nodes and the integrity requirements of verification paths. Storage strategies lack elastic adaptability and it is difficult to strike a balance between storage efficiency and security. At the same time, the indexing mechanism and the verification process lack deep integration, which affects the efficiency of data retrieval and verification.

[0004] To solve the above problems, it is necessary to design a data property registration method based on blockchain-based dynamic encrypted evidence storage, realize multi-dimensional data credibility measurement by building a trust transfer network, establish a dynamic verification mechanism to improve verification reliability, and adopt distributed storage and adaptive indexing mechanisms to enhance the security and availability of data evidence storage, thereby providing more reliable and efficient technical support for data property registration. Summary of the Invention

[0005] The embodiments of the present invention provide a data property rights registration method and system based on blockchain-based dynamic encrypted evidence storage, which can solve the problems in the existing technology.

[0006] A first aspect of an embodiment of the present invention provides a data property rights registration method based on blockchain-based dynamic encrypted evidence storage, comprising:

[0007] receiving a data property registration request, wherein the data property registration request includes the data to be registered and the owner's identity information;

[0008] Constructing a trust transfer network to perform multi-dimensional trust measurement on the data to be registered, including constructing a trust transfer graph, recursively calculating a trust decay coefficient, and generating a trusted data packet containing a trust feature vector;

[0009] A dynamic verification mechanism is established based on the trusted data packet. Through the weight adjustment mechanism of multi-verification unit collaboration and anomaly perception, combined with deep reinforcement learning, the verification path is adaptively selected to generate a verification data packet containing the verification path and verification results.

[0010] Distributed storage is performed on the verification data packets, a data block association graph is constructed based on the importance of verification nodes and the integrity requirements of verification paths, and a storage index is generated using an adaptive multi-level storage strategy and a verification-aware indexing mechanism;

[0011] Submit the storage index and verification summary information to the blockchain network to complete the evidence storage and generate a registration certificate containing the owner's identity information.

[0012] In an optional embodiment,

[0013] The credibility measurement of the data to be registered includes:

[0014] Decomposing the generation process of the data to be registered into multiple operation points, and setting a trust probe at each operation point;

[0015] The trust probe collects multi-dimensional trust factors including operational behavior characteristics, data flow paths, and environmental context;

[0016] Constructing a trust transfer graph based on the multi-dimensional trust elements, wherein the nodes of the trust transfer graph represent operation points and the edges represent trust transfer relationships;

[0017] In the trust transfer graph, a trust decay coefficient from the data source to the current node is obtained by recursive calculation, and the trust decay coefficient is dynamically adjusted as the trust transfer chain is extended;

[0018] A trust threshold is set based on the trust decay coefficient. When the trust decay coefficient of a node is lower than the trust threshold, it is determined that a trust transmission interruption exists at the node. For the node where the trust transmission interruption occurs, an adaptive trust repair mechanism is triggered to restore the trust link by reconstructing the multi-path trust value.

[0019] The trust decay coefficient of the reconstructed trust link is recalculated to verify the reconstructed trust transfer state; based on the verified trust transfer graph and the updated trust decay coefficient, a trusted data packet containing a trust feature vector is generated.

[0020] In an optional embodiment,

[0021] For nodes where trust transmission is interrupted, an adaptive trust repair mechanism is triggered to restore the trust link through multi-path trust value reconstruction, including:

[0022] Obtain the historical trust sequence and associated node set of the interrupted node in the trust transfer graph;

[0023] Selecting a node with a trust value higher than a first preset threshold from the associated node set as a starting node; adding a trusted relay node to the starting node to generate an extended path; calculating the similarity between the extended paths, and retaining the extended path with the higher trust value when the similarity between any two extended paths is greater than a preset similarity threshold, to obtain a candidate trust transfer path;

[0024] Performing a trust evaluation on each path in the candidate trust transfer path, including: calculating the structural trust of the node based on the position information of the node in the trust transfer graph; calculating the temporal trust of the node based on the historical trust sequence; calculating the comprehensive trust of the node by combining the structural trust and the temporal trust; and combining the comprehensive trust of all nodes on the path with a path length penalty factor to obtain the path trust;

[0025] Calculating the path stability of each candidate trust transfer path based on the degree of fluctuation of the historical trust sequence; calculating the reconstruction cost based on the computational overhead and resource consumption;

[0026] A comprehensive path score is calculated based on path stability, reconstruction cost, and the trustworthiness of the path. A reconstruction deviation is calculated based on historical reconstruction results, and each score weight in the comprehensive path score is dynamically adjusted based on the reconstruction deviation. The path with the highest comprehensive path score is selected as the reconstructed path, and a new trust transfer link is established.

[0027] In an optional embodiment,

[0028] Establishing a dynamic authentication mechanism includes:

[0029] The trust feature vector is input into a verification network, which includes multiple verification units; each verification unit performs a verification task based on different dimensions of the trust feature vector to generate a verification score; the verification units collaborate using a dynamic weight mechanism based on anomaly perception, and establish a behavioral baseline value for the verification unit based on the verification accuracy change interval and verification time series characteristic parameters; the verification behavior of the verification unit is monitored in real time, and when it is detected that the verification behavior deviates from the baseline value, the credibility degradation factor of the verification unit is calculated based on the degree of abnormality of the verification behavior, and the verification weight is updated using a nonlinear mapping method based on the credibility degradation factor and the historical verification accuracy.

[0030] Adopting a deep reinforcement learning algorithm, taking the verification score and the trust decay coefficient as state inputs, and the verification accuracy and computational overhead as reward functions, the optimal verification path is adaptively selected;

[0031] When new verification scenarios emerge, the verification experience of multiple verification units is aggregated through federated learning to achieve dynamic evolution of verification strategies;

[0032] The verification task is executed based on the optimal verification path, and the execution result of the verification task and the optimal verification path are encapsulated into a verification data packet.

[0033] In an optional embodiment,

[0034] Using a deep reinforcement learning algorithm, taking the verification score and the trust decay coefficient as state inputs, and the verification accuracy and computational overhead as reward functions, the optimal verification path is adaptively selected, including:

[0035] Combining the verification scores and trust decay coefficients of multiple verification units to form a state vector;

[0036] Inputting the state vector into the policy network to obtain an initial verification path, executing the verification task according to the verification unit in the initial verification path, and calculating the verification reward value based on the execution result; calculating the verification reward value includes: obtaining the verification confidence of the verification unit; inputting the verification confidence into a sigmoid function to obtain a confidence adjustment coefficient; and multiplying the credibility degradation factor, the verification unit result accuracy, and the confidence adjustment coefficient to obtain the verification reward value;

[0037] Multiply the verification calculation time by the path length penalty value and combine them with the resource consumption vector to obtain the calculation cost penalty value;

[0038] Calculating a diversity bonus value based on the degree of difference between the initial verification path and the set of historical verification paths;

[0039] A comprehensive reward value is obtained by weighted combination of the verification reward value, the computational overhead penalty value, the diversity reward value, and the policy entropy;

[0040] Inputting the state vector and the comprehensive reward value into the policy network to perform verification path optimization, the verification path optimization comprising: calculating an advantage function value based on a temporal difference of an immediate reward and a state value function; performing a gradient update on policy network parameters based on the advantage function value; and adjusting the gradient update using an adaptive learning rate, the adaptive learning rate being calculated by first-order moment estimation and second-order moment estimation;

[0041] The optimized policy network output is used as the optimal verification path.

[0042] In an optional embodiment,

[0043] The verification path optimization specifically includes:

[0044] Extracting the current instantaneous reward from the comprehensive reward value sequence; inputting the state vector into the value function network to obtain the current state value and the next state value; using the difference between the instantaneous reward and the next state value adjusted by the discount factor as the time series difference error; combining the time series difference error with the exponentially weighted historical time series difference error to obtain the advantage function value;

[0045] Calculate the log-likelihood of the current verification path relative to the policy network parameters; multiply the log-likelihood by the advantage function value to obtain the policy gradient; perform an exponential sliding average on the policy gradient to obtain a first-order moment estimate; perform an exponential sliding average on the square of the policy gradient to obtain a second-order moment estimate;

[0046] performing deviation correction on the first-order moment estimate and the second-order moment estimate; obtaining an adaptive learning rate by dividing the first-order moment estimate by the sum of the square root of the second-order moment estimate and a second preset threshold;

[0047] The adaptive learning rate is multiplied by the policy gradient to update the policy network parameters.

[0048] In an optional embodiment,

[0049] Distributed storage of the verification data packet and generation of a storage index include:

[0050] Parsing the verification path and verification results in the verification data packet, extracting the verification dependency and the importance of the verification node;

[0051] Building a data block association graph based on the verification dependency relationship, wherein the edge weights in the data block association graph are determined by the importance of the verification node and the verification path integrity requirement;

[0052] Designing an adaptive storage strategy based on the topological structure of the data block association graph, including: using multiple copies of storage for target verification nodes, using erasure code storage for verification path branches, and using single copies of storage for independent verification nodes;

[0053] Build a dynamic distributed storage network and calculate the credibility of storage nodes based on their verification capabilities and historical performance.

[0054] Allocate storage tasks according to the adaptive storage strategy and node credibility;

[0055] A verification-aware index tree is constructed to record the storage location, verification path, and node credibility information of the data block. The verification-aware index tree is adaptively updated as the verification path changes dynamically.

[0056] A second aspect of an embodiment of the present invention provides a data property rights registration system with dynamic encrypted evidence storage based on blockchain, including:

[0057] The first unit is configured to receive a data property registration request, wherein the data property registration request includes the data to be registered and the owner's identity information;

[0058] The second unit is used to build a trust transfer network and perform multi-dimensional trust measurement on the data to be registered, including building a trust transfer graph, recursively calculating the trust decay coefficient, and generating a trusted data packet containing a trust feature vector;

[0059] The third unit is used to establish a dynamic verification mechanism based on the trusted data packet, adaptively select a verification path through the coordination of multiple verification units and anomaly perception weight adjustment mechanism, combined with deep reinforcement learning, and generate a verification data packet containing the verification path and verification results;

[0060] A fourth unit is configured to perform distributed storage on the verification data packets, construct a data block association graph based on the importance of verification nodes and the integrity requirements of verification paths, and generate a storage index using an adaptive multi-level storage strategy and a verification-aware indexing mechanism;

[0061] The fifth unit is used to submit the storage index and verification summary information to the blockchain network to complete the evidence storage and generate a registration certificate containing the owner's identity information.

[0062] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0063] processor;

[0064] a memory for storing processor-executable instructions;

[0065] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0066] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0067] The data property registration method based on blockchain dynamic encryption and evidence storage provided by the present invention realizes the reliable traceability of data sources and the quantitative expression of trust relationships by constructing a trust transmission network and multi-dimensional trust measurement, effectively improving the accuracy of data credibility assessment and solving the problem of single trust assessment in traditional data property registration.

[0068] The present invention adopts a weight adjustment mechanism of multi-verification unit collaboration and anomaly perception, combined with deep reinforcement learning to adaptively select verification paths, which not only improves the verification efficiency, but also can perceive and respond to abnormal situations in the verification process in real time, enhance the security and reliability of the data property registration process, and overcome the defects of the traditional method of fixed verification mechanism and lack of adaptability.

[0069] The present invention constructs a data block association graph based on the importance of verification nodes and the integrity requirements of verification paths, adopts an adaptive multi-level storage strategy and a verification-aware indexing mechanism, realizes efficient management and fast retrieval of distributed storage, and at the same time ensures the immutability of evidence and the clarity of ownership through blockchain technology, providing technical guarantees for data property rights protection and effectively solving the problems of low storage efficiency and insufficient proof of ownership in traditional data property rights registration. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 This is a flowchart of a data property rights registration method based on dynamic encryption and evidence storage in blockchain according to an embodiment of the present invention;

[0071] Figure 2 This is a comparison chart of the reconstruction success rate of different trust repair strategies under the comprehensive scores of each path;

[0072] Figure 3 The present invention is a flow chart of the adaptive distributed storage based on verification perception. DETAILED DESCRIPTION

[0073] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0074] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0075] Figure 1 This is a flow chart of a data property rights registration method based on dynamic encryption and evidence storage in blockchain according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0076] In an optional embodiment, a data property rights registration method based on dynamic encrypted evidence storage in blockchain is characterized by including:

[0077] receiving a data property registration request, wherein the data property registration request includes the data to be registered and the owner's identity information;

[0078] Constructing a trust transfer network to perform multi-dimensional trust measurement on the data to be registered, including constructing a trust transfer graph, recursively calculating a trust decay coefficient, and generating a trusted data packet containing a trust feature vector;

[0079] A dynamic verification mechanism is established based on the trusted data packet. Through the weight adjustment mechanism of multi-verification unit collaboration and anomaly perception, combined with deep reinforcement learning, the verification path is adaptively selected to generate a verification data packet containing the verification path and verification results.

[0080] Distributed storage is performed on the verification data packets, a data block association graph is constructed based on the importance of verification nodes and the integrity requirements of verification paths, and a storage index is generated using an adaptive multi-level storage strategy and a verification-aware indexing mechanism;

[0081] Submit the storage index and verification summary information to the blockchain network to complete the evidence storage and generate a registration certificate containing the owner's identity information.

[0082] In an optional implementation, performing credibility measurement on the data to be registered includes:

[0083] Decomposing the generation process of the data to be registered into multiple operation points, and setting a trust probe at each operation point;

[0084] The trust probe collects multi-dimensional trust factors including operational behavior characteristics, data flow paths, and environmental context;

[0085] Constructing a trust transfer graph based on the multi-dimensional trust elements, wherein the nodes of the trust transfer graph represent operation points and the edges represent trust transfer relationships;

[0086] In the trust transfer graph, a trust decay coefficient from the data source to the current node is obtained by recursive calculation, and the trust decay coefficient is dynamically adjusted as the trust transfer chain is extended;

[0087] A trust threshold is set based on the trust decay coefficient. When the trust decay coefficient of a node is lower than the trust threshold, it is determined that a trust transmission interruption exists at the node. For the node where the trust transmission interruption occurs, an adaptive trust repair mechanism is triggered to restore the trust link by reconstructing the multi-path trust value.

[0088] The trust decay coefficient of the reconstructed trust link is recalculated to verify the reconstructed trust transfer state; based on the verified trust transfer graph and the updated trust decay coefficient, a trusted data packet containing a trust feature vector is generated.

[0089] For example, the entire process of data property rights formation, transfer, and confirmation is analyzed and divided into multiple key operational points. Taking the registration of digital work property rights as an example, these operational points include work creation, version iteration, copyright declaration, copyright transfer, use authorization, and blockchain evidence storage. When identifying operational points, the system makes judgments based on changes in property rights status, transfers of ownership, and differences in registration behavior, ensuring coverage of key nodes in property rights formation.

[0090] At each operation point, the system deploys trust probes to collect trust factors. Trust probes are data collection components embedded in the property registration platform and deployed in a distributed manner across the blockchain network. Trust probes collect three types of trust factors: operational behavior characteristics, data flow paths, and environmental context. Operational behavior characteristics include information such as the property registration operation type, timestamp, and the identity of the operating entity; data flow paths record the chain of transmission of property data between different rights holders; and environmental context includes characteristics of the registration terminal, blockchain node information, and network environment characteristics.

[0091] Taking digital artwork property registration as an example, the trust probe deployed by the system during the artwork creation process collects the artist's identity authentication information, the fingerprint characteristics of the creation tools, the work's metadata, and timestamp proof. During the property transfer process, the trust probe collects information such as the digital signature of the transfer agreement, the identity verification results of both parties to the transaction, and the trust status of the transaction platform, providing a trusted foundation for subsequent blockchain evidence storage.

[0092] The trust transfer graph is a model of trust relationships in the data property registration process, where nodes represent property operation points and edges represent trust transfer relationships. When constructing the trust transfer graph, the system first identifies the dependencies between each operation point in the property transfer process and establishes initial connections. It then determines the direction of edges based on the time sequence of property changes, data dependencies, and the association of rights holders. Finally, it calculates edge weights based on the blockchain consensus mechanism, digital signature verification results, and timestamp proof.

[0093] Taking the blockchain digital copyright registration platform as an example, a trust chain is formed from the creator's submission of original content to platform verification and then to blockchain storage. The strength of the trust between the creation node and the verification node depends on the integrity of the creation proof, the originality verification results, and the validity of the digital signature. When the creation proof is authenticated by the blockchain timestamp service and the submission process uses multi-signature verification, the system assigns a higher trust strength to this connection, such as 0.95.

[0094] The trust decay coefficient represents the degree of trust loss during the property rights registration process. The calculation process begins at the data source node, with the initial trust decay coefficient set to 1, indicating complete trust. For each subsequent node, its trust decay coefficient is determined by the trust decay coefficient of the predecessor node and the trust transfer strength of the connecting edge. The trust decay coefficient decreases dynamically as the property rights transfer chain lengthens, and the extent of the decrease is affected by the strength of the blockchain consensus.

[0095] In an intellectual property transaction scenario, during the trust transfer process from the original creator to the copyright trading platform and then to the final rights transferee, if the trust transfer strength between the blockchain verification node and the creator is 0.98 (using an authoritative timestamp service and a multi-party witness mechanism), and the trust transfer strength between the rights transferee and the verification node is 0.96 (based on the results of automatic execution of smart contracts), the rights transferee's trust in the property data will decay by approximately 0.941. This means that trust is lost by 5.9% during the property transfer process.

[0096] The system sets a trust threshold based on the trust decay coefficient to identify any trust disruptions during the property registration process. The trust threshold is dynamically set based on the property type, value level, and regulatory requirements. For high-value digital asset properties, the system sets a higher trust threshold, such as 0.90; for general digital content properties, a lower trust threshold, such as 0.75, is used. When a node's trust decay coefficient falls below the set trust threshold, the system determines that the node has experienced a trust disruption.

[0097] A trust transfer interruption means that the credibility of the property registration at that node cannot be adequately guaranteed. Causes of this interruption include insufficient proof of ownership, blockchain node verification failure, abnormal smart contract execution, or an incomplete chain of preceding property rights. For example, during the NFT artwork property registration process, if the blockchain transaction record of a transfer node indicates on-chain verification failure or a disputed smart contract execution result, the system will determine that the trust decay coefficient of that node is below the threshold, indicating a trust transfer interruption.

[0098] For nodes where trust transfer is interrupted, the system triggers an adaptive trust repair mechanism. Through a multi-path trust value reconstruction approach, the system generates multiple candidate trust transfer paths and selects the optimal path to repair the trust link. In the data property rights transaction system, if trust is interrupted during the registration process, the system reconstructs the trust link through multiple paths, including main chain evidence records, side chain detailed information, and cross-chain indexes, to ensure the credibility of the property rights registration process.

[0099] For the reconstructed trust link, the system recalculates the trust decay coefficient and verifies the reconstructed trust transfer status. The verification process includes checking the integrity of the blockchain's evidence records, the consistency of the property rights declaration, and the reliability of the signature proof. If the reconstructed trust decay coefficient exceeds the trust threshold and the blockchain verification status meets the property rights registration requirements, the trust restoration is considered successful.

[0100] Based on the verified trust transfer graph and the updated trust decay coefficient, the system generates a trusted data package containing a trust feature vector. This trust feature vector represents a multi-dimensional representation of the trustworthiness of property rights data, including scores for the credibility of the property source, the integrity of the ownership transfer, and the validity of blockchain evidence. These scores comprehensively reflect the trustworthiness of property rights data from its generation to blockchain evidence storage. In the digital content property rights registration scenario, the system-generated trust feature vector includes scores of 0.97 for creator identity credibility, 0.98 for property rights claim integrity, and 0.99 for blockchain evidence validity, providing a foundation for subsequent property rights verification and blockchain evidence storage.

[0101] This invention innovatively introduces a blockchain-driven trust transfer graph model and dynamic decay mechanism, enabling precise quantification of the trust status of property rights at different stages. Through an adaptive trust repair mechanism, the system effectively addresses the challenge of trust transfer interruptions during property registration, improving the reliability of blockchain-based evidence storage. This enhances the authenticity and authority of digital property registration, provides a solid foundation for blockchain-based dynamic encrypted evidence storage, and effectively addresses the technical issues of incomplete trust assessment and limited evidence storage methods in traditional property registration.

[0102] In an optional embodiment, for a node where trust transfer is interrupted, triggering an adaptive trust repair mechanism to restore the trust link by reconstructing multi-path trust values ​​includes:

[0103] Obtain the historical trust sequence and associated node set of the interrupted node in the trust transfer graph;

[0104] Selecting a node with a trust value higher than a first preset threshold from the associated node set as a starting node; adding a trusted relay node to the starting node to generate an extended path; calculating the similarity between the extended paths, and retaining the extended path with the higher trust value when the similarity between any two extended paths is greater than a preset similarity threshold, to obtain a candidate trust transfer path;

[0105] Performing a trust evaluation on each path in the candidate trust transfer path, including: calculating the structural trust of the node based on the position information of the node in the trust transfer graph; calculating the temporal trust of the node based on the historical trust sequence; calculating the comprehensive trust of the node by combining the structural trust and the temporal trust; and combining the comprehensive trust of all nodes on the path with a path length penalty factor to obtain the path trust;

[0106] Calculating the path stability of each candidate trust transfer path based on the degree of fluctuation of the historical trust sequence; calculating the reconstruction cost based on the computational overhead and resource consumption;

[0107] A comprehensive path score is calculated based on path stability, reconstruction cost, and the trustworthiness of the path. A reconstruction deviation is calculated based on historical reconstruction results, and each score weight in the comprehensive path score is dynamically adjusted based on the reconstruction deviation. The path with the highest comprehensive path score is selected as the reconstructed path, and a new trust transfer link is established.

[0108] For example, a historical trust sequence is a record of changes in the trust value of a node over a period of time, typically stored as a timestamp and a corresponding trust value tuple. For digital artwork property registration, the system records trust value changes over the past 30 days, such as (2023-09-01T10:00:00, 0.92), (2023-09-05T14:30:00, 0.88), etc. The associated node set includes upstream and downstream nodes that have a direct trust transfer relationship with the interrupted node. The system determines these associated nodes by traversing the edges connected to the interrupted node in the trust transfer graph.

[0109] From the set of associated nodes, the system selects nodes with a trust value higher than a first preset threshold as the starting node. In practical applications, the first preset threshold is usually set to a higher value, such as 0.85, to ensure that the selected starting node has a sufficiently high credibility. For example, in a blockchain copyright registration platform, if the set of associated nodes includes an original submission node (trust value 0.95), a copyright review node (trust value 0.88), a blockchain evidence node (trust value 0.92), and a rights transfer node (trust value 0.78), the system will select three nodes with a trust value higher than 0.85 as the starting nodes.

[0110] After selecting the starting nodes, the system adds trusted relay nodes to these nodes to generate extended paths. Trusted relay nodes are nodes with high trust values ​​in the trust transfer graph and can provide alternative verification paths. The system selects appropriate relay nodes based on their connectivity, verification capabilities, and trust history. In the digital content property rights registration system, trusted relay nodes include content review servers, timestamp authentication services, and multi-signature verification nodes. Starting from each starting node, the system can construct up to five different extended paths, each containing two to three relay nodes.

[0111] The system calculates the similarity between the generated extended paths to prevent redundant paths from being selected. Path similarity is calculated based on the proportion of common nodes in the paths. When the similarity between any two extended paths exceeds a preset similarity threshold (usually set to 0.7), the system retains the extended path with the higher trustworthiness and eliminates the other. For example, if path A contains nodes {N1, N3, N5, N7} and path B contains nodes {N1, N2, N5, N8}, the two paths share two nodes and have a similarity of 0.5, which is below the threshold of 0.7, so both paths are retained. After similarity screening, the system obtains a set of candidate trust transfer paths.

[0112] For each candidate trust transfer path, the system performs a multi-dimensional trust evaluation. First, the structural trust of each node in the path is calculated, which is the importance of the node's position in the trust transfer graph. Structural trust considers factors such as node connectivity, centrality, and critical path proportion. For example, in a blockchain property registration system, consensus nodes connecting multiple verification links typically have a high structural trust, such as 0.90; while edge verification nodes have a lower structural trust, such as 0.70.

[0113] The system calculates a node's temporal trustworthiness based on its historical trust sequence, reflecting the historical stability of the node's trust value. Temporal trustworthiness considers the average level, fluctuations, and trends of trustworthiness. Nodes with stable and high average trustworthiness receive higher temporal trustworthiness, while nodes with significant fluctuations or a downward trend receive lower temporal trustworthiness. In the digital content rights management platform, blockchain nodes that have consistently provided verification services receive a temporal trustworthiness of 0.95, while nodes that have recently experienced frequent verification fluctuations receive only a temporal trustworthiness of 0.75.

[0114] The overall trustworthiness is derived through a weighted combination, with weights adjusted based on the application scenario. In the property registration system, structural trustworthiness is assigned a weight of 0.4, and temporal trustworthiness is assigned a weight of 0.6. For each node on a path, the overall trustworthiness is calculated. The overall trustworthiness of all nodes is then combined with a path length penalty factor to determine the path trustworthiness. The path length penalty factor decreases with increasing path length to prevent excessively long paths from receiving excessively high scores. For example, a path of length 3 has a penalty factor of 0.95, while a path of length 6 has a penalty factor of 0.85.

[0115] Path stability reflects the robustness of a path in the face of changes in the trust environment. Paths with low volatility have higher stability scores, while paths containing multiple nodes with high historical volatility have lower stability. For example, in a blockchain evidence storage system, a path consisting of multiple backbone nodes receives a stability score of 0.92, while a path containing multiple edge nodes receives a stability score of only 0.78.

[0116] The reconstruction cost takes into account factors such as the number of verification operations required, the number of blockchain interactions, and the need for cross-chain verification. In the property registration platform, paths that require multiple cross-chain verifications have a higher reconstruction cost, such as 0.85; while paths that only require verification on the main chain have a lower reconstruction cost, such as 0.30.

[0117] The system calculates a comprehensive path score based on path stability, reconstruction cost, and path trust. The comprehensive score is derived through a weighted combination, with initial weights of 0.5 for path trust, 0.3 for path stability, and 0.2 for reconstruction cost (negative contribution). For example, if a path has a trust of 0.88, a stability of 0.92, and a reconstruction cost of 0.40, its initial comprehensive score is 0.88 × 0.5 + 0.92 × 0.3 - 0.40 × 0.2 = 0.78.

[0118] The system calculates reconstruction bias based on historical reconstruction results and uses this bias to dynamically adjust scoring weights. Reconstruction bias reflects the success rate and accuracy of trust reconstruction using similar paths in the past. If a particular path has historically performed well in reconstruction, the system increases the weight of the associated feature; otherwise, it decreases it. For example, if historical data shows that the reconstruction success rate for highly stable paths is significantly higher than average, the system will adjust the stability weight from 0.3 to 0.4 while reducing the weights of other features.

[0119] The path with the highest overall path score is selected as the reconstruction path to establish a new trust link. In the digital asset property registration system, if the highest-scoring path includes "origin node → timestamp service → multi-signature verification → blockchain evidence storage," the system will use this path to replace the original interrupted path and re-establish the trust connection. After the reconstruction is complete, the system will record the characteristics and effects of the reconstructed path for future optimization of the reconstruction strategy.

[0120] Figure 2 This is a comparison chart of the reconstruction success rates of different trust repair strategies under the comprehensive scores of each path. It can be seen from the figure that the adaptive trust repair mechanism adopted by the present invention is significantly better than the existing technology in each score range, especially in the medium and high score range (0.4-0.8), the reconstruction success rate is significantly improved. When the comprehensive score of the path reaches 0.6, the reconstruction success rate of the present invention has reached 93%, while the static weight method is only 73%, and the single-path repair method is even lower at 47%. When the score reaches 1.0, the present invention can achieve 100% reconstruction success rate, which is much higher than other methods. This shows that the dynamic weight adjustment mechanism and multi-path trust reconstruction strategy of the present invention can effectively improve the accuracy and reliability of trust repair, especially when dealing with complex trust network environments.

[0121] The present invention's adaptive trust repair mechanism, based on a blockchain-based evidence storage system, effectively resolves the problem of trust transmission interruption through multi-path trust value reconstruction. Taking into account node structural characteristics, historical performance, and resource constraints, it introduces path similarity filtering and a comprehensive scoring mechanism to ensure the selection of the optimal reconstruction path. Through historical reconstruction effect analysis and dynamic weight adjustment, the system continuously optimizes reconstruction strategies, improving the accuracy and efficiency of trust repair. This adaptive trust repair mechanism significantly enhances the reliability and continuity of property rights data during blockchain evidence storage, providing strong technical support for high-trust registration of data property rights and effectively addressing the challenge of trust interruption in complex network environments.

[0122] In an optional implementation, establishing a dynamic verification mechanism includes:

[0123] The trust feature vector is input into a verification network, which includes multiple verification units; each verification unit performs a verification task based on different dimensions of the trust feature vector to generate a verification score; the verification units collaborate using a dynamic weight mechanism based on anomaly perception, and establish a behavioral baseline value for the verification unit based on the verification accuracy change interval and verification time series characteristic parameters; the verification behavior of the verification unit is monitored in real time, and when it is detected that the verification behavior deviates from the baseline value, the credibility degradation factor of the verification unit is calculated based on the degree of abnormality of the verification behavior, and the verification weight is updated using a nonlinear mapping method based on the credibility degradation factor and the historical verification accuracy.

[0124] Adopting a deep reinforcement learning algorithm, taking the verification score and the trust decay coefficient as state inputs, and the verification accuracy and computational overhead as reward functions, the optimal verification path is adaptively selected;

[0125] When new verification scenarios emerge, the verification experience of multiple verification units is aggregated through federated learning to achieve dynamic evolution of verification strategies;

[0126] The verification task is executed based on the optimal verification path, and the execution result of the verification task and the optimal verification path are encapsulated into a verification data packet.

[0127] The present invention provides a specific implementation method for establishing a dynamic verification mechanism, which realizes a verification system with high accuracy and low computational overhead through the collaborative work of dynamic weight adjustment, reinforcement learning and federated learning.

[0128] Exemplarily, the verification network includes multiple verification units. For example, five verification units can be set up, each responsible for processing different dimensions of the trust feature vector. Specifically, the first verification unit processes behavioral consistency features, the second verification unit processes temporal continuity features, the third verification unit processes spatial correlation features, the fourth verification unit processes interaction frequency features, and the fifth verification unit processes resource consumption features. Each verification unit analyzes the feature dimension it is responsible for through a deep neural network structure to generate a verification score between 0 and 1. For example, when user behavior conforms to historical patterns, the first verification unit outputs a high credibility score of 0.95; and when an abnormal jump in the user's location is detected, the third verification unit outputs a low credibility score of 0.35.

[0129] Verification units collaborate using a dynamic weighting mechanism based on anomaly detection. The system establishes a behavioral baseline for each verification unit, including the accuracy variation range and verification timing characteristic parameters. For example, the normal accuracy variation range for the second verification unit is 0.82 to 0.91, and the timing characteristic parameters include an average response time of 50 milliseconds and a standard deviation of 8 milliseconds. During actual operation, the system monitors the verification unit's verification behavior in real time. When a verification unit's behavior deviates from the baseline, the system deems it to be under attack or malfunctioning. In specific implementations, the system calculates the degree of abnormality in the verification behavior and, in turn, the unit's credibility degradation factor. For example, if the accuracy of the fourth verification unit suddenly drops from 0.88 to 0.65, and its response time increases to 200 milliseconds, the system calculates a credibility degradation factor of 0.6. The system then combines this credibility degradation factor with historical verification accuracy to update the verification weight using a nonlinear mapping function. In practical applications, a sigmoid function can be used as the nonlinear mapping function to reduce the original weight of 0.2 to 0.08.

[0130] The system uses deep reinforcement learning algorithms to adaptively select the optimal verification path.

[0131] When the system encounters a new verification scenario, it aggregates the verification experience of multiple verification units through federated learning. Each verification unit independently learns a local verification model and then uploads the model parameters (not the original data) to the central server. The central server aggregates these model parameters, updates the global verification model, and distributes the updated model to each verification unit.

[0132] After executing the verification task based on the selected optimal verification path, the system encapsulates the verification task execution results and the optimal verification path into a verification data packet. This data packet contains multiple fields: a verification result field (containing the comprehensive verification score and verification decision), a verification path field (recording the unit IDs and their weights participating in the verification), a verification timestamp field (recording the start and end time of the verification), a resource consumption field (recording CPU and memory usage), and a verification metadata field (recording information such as the verification scenario type and trust feature vector). For example, a typical verification data packet contains: verification score 0.87, "pass" verification decision, verification path "U1(0.3)-U3(0.4)-U5(0.3)" (indicating that units 1, 3, and 5 participate in the verification, with weights of 0.3, 0.4, and 0.3 respectively), start time "2023-06-01 14:30:22.156", end time "2023-06-01 14:30:22.289", CPU usage "12%", memory usage "45MB", and verification scenario "remote login behavior verification".

[0133] The present invention constructs a dynamic verification mechanism based on the collaboration of multiple verification units, realizes intelligent collaboration among verification units through dynamic weight adjustment of anomaly perception, combines deep reinforcement learning to adaptively select the optimal verification path, and introduces federated learning to realize the dynamic evolution of verification strategies; it significantly improves the accuracy and robustness of the verification process, can effectively cope with diverse data property rights verification scenarios, and provide highly reliable verification support for blockchain evidence storage.

[0134] In an optional embodiment, a deep reinforcement learning algorithm is used, the verification score and the trust decay coefficient are used as state inputs, and the verification accuracy and computational overhead are used as reward functions. Adaptive selection of the optimal verification path includes:

[0135] Combining the verification scores and trust decay coefficients of multiple verification units to form a state vector;

[0136] Inputting the state vector into the policy network to obtain an initial verification path, executing the verification task according to the verification unit in the initial verification path, and calculating the verification reward value based on the execution result; calculating the verification reward value includes: obtaining the verification confidence of the verification unit; inputting the verification confidence into a sigmoid function to obtain a confidence adjustment coefficient; and multiplying the credibility degradation factor, the verification unit result accuracy, and the confidence adjustment coefficient to obtain the verification reward value;

[0137] Multiply the verification calculation time by the path length penalty value and combine them with the resource consumption vector to obtain the calculation cost penalty value;

[0138] Calculating a diversity bonus value based on the degree of difference between the initial verification path and the set of historical verification paths;

[0139] A comprehensive reward value is obtained by weighted combination of the verification reward value, the computational overhead penalty value, the diversity reward value, and the policy entropy;

[0140] Inputting the state vector and the comprehensive reward value into the policy network to perform verification path optimization, the verification path optimization comprising: calculating an advantage function value based on a temporal difference of an immediate reward and a state value function; performing a gradient update on policy network parameters based on the advantage function value; and adjusting the gradient update using an adaptive learning rate, the adaptive learning rate being calculated by first-order moment estimation and second-order moment estimation;

[0141] The optimized policy network output is used as the optimal verification path.

[0142] Exemplarily, in a verification system, multiple verification units each have different verification scores and trust decay coefficients. In order to form a complete state representation, the system combines these parameters to form a state vector. Specifically, for n verification units, each verification unit i has a verification score score_i and a trust decay coefficient decay_i, and the state vector can be represented as [score_1, decay_1, score_2, decay_2, ..., score_n, decay_n]. For example, in a system containing 5 verification units, the state vector is [0.85, 0.92, 0.76, 0.88, 0.91, 0.95, 0.72, 0.83, 0.88, 0.90], where odd-numbered positions represent verification scores and even-numbered positions represent trust decay coefficients.

[0143] After the system inputs the state vector into the policy network, the network outputs an initial verification path. This path is a variable-length sequence representing the set of verification units that need to be executed in sequence. For example, for a system with five verification units, the initial verification path is [2,4,1], indicating that the second, fourth, and first verification units are executed in sequence. The system executes the verification task along this path and calculates the verification reward value based on the execution results.

[0144] The verification reward is calculated as follows: The system first obtains the verification confidence of the verification unit, which reflects the reliability of the verification result. For example, a verification unit returns a confidence value of 0.87. The system inputs this confidence into the sigmoid function for normalization, obtaining a confidence adjustment coefficient. If the confidence is 0.87, the adjustment coefficient calculated by the sigmoid function is 0.79. The system then multiplies the trust decay coefficient (e.g., 0.92), the verification accuracy (e.g., 0.85), and the confidence adjustment coefficient (0.79) to obtain a verification reward of 0.92 × 0.85 × 0.79 = 0.618.

[0145] The computational overhead penalty is calculated by taking into account both the time and resource consumption of the verification process. The system measures the verification computation time (e.g., 250 milliseconds) and multiplies it by the path length penalty (e.g., a path length of 3 corresponds to a penalty of 0.3), resulting in a time penalty of 75. Simultaneously, the system tracks various resource consumptions, such as CPU usage (0.4), memory usage (0.3), and network bandwidth (0.2), forming a resource consumption vector [0.4, 0.3, 0.2]. By setting the weight vector [0.5, 0.3, 0.2], the weighted sum is calculated as 0.4 × 0.5 + 0.3 × 0.3 + 0.2 × 0.2 = 0.3, resulting in a computational overhead penalty of 75 + 0.3 = 75.3.

[0146] To encourage the system to explore diverse verification paths, the system calculates a diversity bonus. This bonus is based on the difference between the current verification path and the set of historical verification paths. For example, if the set of historical paths includes the paths [2,4,1], [3,1,5], and [4,2,5], and the current path is [1,3,4], the system calculates the average edit distance between the paths to be 2.33, resulting in a diversity bonus of 0.233.

[0147] The system performs a weighted combination of the verification reward value (0.618), the computational overhead penalty value (normalized value of 75.3 is 0.753), the diversity reward value (0.233) and the policy entropy (such as 0.45, indicating the randomness of the policy), with weights of 0.6, -0.3, 0.2 and 0.1 respectively, to obtain a comprehensive reward value of 0.618×0.6-0.753×0.3+0.233×0.2+0.45×0.1=0.1709.

[0148] The present invention realizes the intelligent optimization of the verification path through a deep reinforcement learning algorithm, takes the verification score and trust decay coefficient as state input, comprehensively considers the verification reward, computational overhead and diversity, and constructs a complete reward mechanism; this technology can adaptively select the optimal verification path in a dynamically changing verification environment, balance verification accuracy and resource consumption, and effectively improve the verification efficiency and reliability of data property registration.

[0149] In an optional implementation, the verification path optimization specifically includes:

[0150] Extracting the current instantaneous reward from the comprehensive reward value sequence; inputting the state vector into the value function network to obtain the current state value and the next state value; using the difference between the instantaneous reward and the next state value adjusted by the discount factor as the time series difference error; combining the time series difference error with the exponentially weighted historical time series difference error to obtain the advantage function value;

[0151] Calculate the log-likelihood of the current verification path relative to the policy network parameters; multiply the log-likelihood by the advantage function value to obtain the policy gradient; perform an exponential sliding average on the policy gradient to obtain a first-order moment estimate; perform an exponential sliding average on the square of the policy gradient to obtain a second-order moment estimate;

[0152] performing deviation correction on the first-order moment estimate and the second-order moment estimate; obtaining an adaptive learning rate by dividing the first-order moment estimate by the sum of the square root of the second-order moment estimate and a second preset threshold;

[0153] The adaptive learning rate is multiplied by the policy gradient to update the policy network parameters.

[0154] Exemplarily, this embodiment describes in detail a specific implementation method for verification path optimization, including core technologies such as advantage function calculation, policy gradient estimation, and adaptive learning rate adjustment.

[0155] The verification path optimization process first extracts the current instantaneous reward from the comprehensive reward value sequence. The comprehensive reward value sequence is the reward data stream recorded by the system during the verification task execution, with each time point corresponding to a comprehensive reward value. The instantaneous reward reflects the real-time performance of the current verification path, including multiple factors such as verification accuracy, time cost, and resource consumption. In the digital artwork property rights verification scenario, if the current verification path successfully verifies the originality of the work with low time cost, the system records a higher instantaneous reward value, such as 0.85; if the verification result is uncertain or resource consumption is excessive, the instantaneous reward is lower, such as 0.45.

[0156] The system inputs the current state vector into the value function network to obtain the current state value and the next state value. The state vector is a combination of the verification scores and trust decay coefficients of multiple verification units, reflecting the comprehensive state of the current verification environment. The value function network is used to evaluate the value of different states. For example, in the property registration and verification system, the value function network usually adopts a multi-layer perceptron structure, containing 3-4 hidden layers, each layer containing 64-128 neurons, and using the ReLU activation function. The network input is the state vector, and the output is the state value estimate. For example, when the input state vector represents a combination of multiple high-trust verification units, the network outputs a higher state value, such as 0.82; when the input state vector contains multiple low-trust verification units, the output state value is lower, such as 0.38.

[0157] The discount factor is typically set between 0.9 and 0.99 to balance the importance of current rewards with future rewards. In data property rights verification scenarios, a discount factor of 0.95 means that the importance of future rewards decreases over time. Temporal difference error (TDE) reflects the gap between the estimated current state value and the actual reward and is a key signal for optimizing strategies. For example, if the immediate reward is 0.75, the current state value is 0.60, the next state value is 0.70, and the discount factor is 0.95, then the TDE is 0.75 + 0.95 × 0.70 - 0.60 = 0.82, indicating that the current state value is underestimated.

[0158] The system combines the temporal difference error with exponentially weighted historical temporal difference errors to derive the advantage function value. Exponential weighting uses a decay coefficient (typically 0.95) to smooth the historical error and reduce fluctuations during the optimization process. The advantage function value is a key concept in policy gradient methods, representing the degree of advantage of the current action relative to the average performance. In blockchain evidence verification systems, a positive advantage function value indicates that the current verification path is better than the average and its selection probability should be increased; a negative advantage function value indicates that the current path performs poorly and its selection probability should be decreased.

[0159] The policy network is a parameterized neural network used to generate verification path selection probabilities. It can adopt a multi-layer structure, consisting of 2-3 hidden layers, each containing 32-64 neurons. The output layer uses a softmax activation function to generate the selection probabilities for each verification unit combination. The log-likelihood reflects the logarithmic probability density of the current verification path in the policy distribution and is the basis for calculating policy gradients.

[0160] Policy gradients represent how to adjust the policy network parameters to increase the probability of high-reward actions. In a property rights verification system, if the advantage function value of a verification path combination is positive and large, the system will increase the probability of selecting that path; if the advantage function value is negative, the probability of selecting that path will be reduced.

[0161] To stabilize the training process, the system performs an exponential sliding average on the policy gradient to obtain a first-order moment estimate. This exponential sliding average uses a decay rate (typically 0.9) to weight the historical gradients, reducing the variance of the gradient estimate. Simultaneously, the system performs an exponential sliding average on the square of the policy gradient to obtain a second-order moment estimate, with a decay rate typically set to 0.999. The first-order moment estimate represents the momentum of the gradient, while the second-order moment estimate reflects the variance of the gradient.

[0162] The system performs bias correction on the first-order moment estimate and the second-order moment estimate to eliminate the problem of small estimates in the early stages of training. The bias correction is achieved by dividing by a coefficient close to 1, which approaches 1 as the number of training steps increases. After correction, the system divides the first-order moment estimate by the sum of the square root of the second-order moment estimate and a second preset threshold to obtain an adaptive learning rate. The second preset threshold is usually set to a small positive number, such as 0.00001, to prevent the denominator from approaching zero, which would cause the learning rate to be too large. This adaptive learning rate calculation method can automatically adjust the update step size for different parameters, using a smaller learning rate for frequently occurring features and a larger learning rate for rare features.

[0163] In the property rights verification system, if a parameter's historical gradient is large and stable, the system assigns a smaller learning rate, such as 0.001. For parameters with smaller or unstable historical gradients, the system assigns a larger learning rate, such as 0.01, to accelerate convergence. The system multiplies the adaptive learning rate by the policy gradient to update the policy network parameters. After the parameter update is complete, the policy network generates a new probability distribution for verification path selection, guiding the system to select a more optimal verification path in the next round of verification.

[0164] Policy network training typically uses batch updates, with each batch containing 16-32 verification samples. The system sets a verification cycle, such as updating the policy network after processing 100 property registration requests. As training progresses, the verification path selection gradually converges to the optimal one, improving verification accuracy while reducing resource consumption. For example, in a digital content property rights verification platform, an optimized verification path selection strategy was able to reduce average verification time by 30% and verification resource consumption by 25% while maintaining high verification accuracy.

[0165] This invention implements an intelligent verification path optimization mechanism based on deep reinforcement learning. Through temporal difference learning and adaptive learning rate adjustment, it effectively solves the efficiency and accuracy balance problem in traditional verification path selection. It comprehensively considers immediate rewards and long-term value, introduces an advantage function to evaluate action quality, and uses an adaptive learning rate mechanism to stabilize the training process, achieving efficient convergence and precise adjustment of the verification strategy. This intelligent verification path optimization technology significantly improves the verification efficiency and reliability of the data property registration system, reduces computing resource consumption, enhances the system's adaptability to complex verification scenarios, provides efficient and reliable verification support for blockchain evidence storage, and effectively solves the technical problems of inefficient path selection and resource waste in traditional property rights verification.

[0166] In an optional implementation, performing distributed storage on the verification data packet and generating a storage index includes:

[0167] Parsing the verification path and verification results in the verification data packet, extracting the verification dependency and the importance of the verification node;

[0168] Building a data block association graph based on the verification dependency relationship, wherein the edge weights in the data block association graph are determined by the importance of the verification node and the verification path integrity requirement;

[0169] Designing an adaptive storage strategy based on the topological structure of the data block association graph, including: using multiple copies of storage for target verification nodes, using erasure code storage for verification path branches, and using single copies of storage for independent verification nodes;

[0170] Build a dynamic distributed storage network and calculate the credibility of storage nodes based on their verification capabilities and historical performance.

[0171] Allocate storage tasks according to the adaptive storage strategy and node credibility;

[0172] A verification-aware index tree is constructed to record the storage location, verification path, and node credibility information of the data block. The verification-aware index tree is adaptively updated as the verification path changes dynamically.

[0173] For example, combined Figure 3 The present invention illustrates the verification-aware adaptive distributed storage flow chart. In a blockchain-based data property registration system, efficient distributed storage of verification data packets and the establishment of an indexing mechanism tailored to verification requirements are key to ensuring the reliability of evidence. This embodiment details the distributed storage and index generation method based on verification characteristics.

[0174] Parse the verification data packet and extract the verification path and verification result information. The verification data packet is a complete record of the property rights verification process, containing information such as the verification unit sequence, verification parameters, intermediate results, and final verification conclusions. The system parses the structured content of the verification data packet, identifies the dependencies between verification units, and constructs a verification dependency graph. At the same time, the system calculates the importance of the verification node based on the contribution, irreplaceability, and impact of the verification unit in the overall verification process. In the digital copyright registration scenario, the originality verification node usually has a higher importance, such as 0.9; while the importance of basic verification nodes such as format checking is relatively low, such as 0.3.

[0175] In the data block association graph, nodes represent verification data blocks, and edges represent dependencies between data blocks. The weight of an edge is determined by both the importance of the verification node and the integrity requirement of the verification path. The integrity requirement of the verification path reflects the integrity requirements for the data needed to recover a specific verification path, and is typically determined based on the criticality of the verification path, the complexity of the verification process, and the importance of the verification results. For example, in an NFT art property registration system, the edge weight connecting the artwork fingerprint extraction node and the blockchain evidence storage node is typically high, such as 0.85, indicating that the data association between these two nodes is crucial to the verification process; whereas the edge weight connecting the basic metadata verification node and the format standardization node is low, such as 0.40.

[0176] The system uses differentiated storage methods for different types of verification nodes: multiple replicas are used for target verification nodes, erasure coding is used for verification path branches, and single replicas are used for independent verification nodes. Target verification nodes are core nodes in the verification process and have a decisive impact on the verification results, such as identity verification nodes and ownership verification nodes. The system creates multiple replicas (typically 3-5) for these nodes, distributed across different storage nodes to ensure high availability. Verification path branches are the path segments connecting multiple verification nodes, providing a certain degree of redundancy and reconfigurability. The system uses erasure coding technology for this type of data, splitting the data into m data blocks and n check blocks. Retrieving any m blocks is sufficient to restore the original data. Common parameter combinations in property registration systems include (10,4) and (6,3), adjusting the redundancy based on the importance of the path. Independent verification nodes are independent verification nodes that can be executed independently and have minimal dependence on other nodes, such as format checking and integrity verification. The system uses single replicas for these nodes and can regenerate data from the verification source when necessary.

[0177] A dynamic distributed storage network consists of various node types, including dedicated storage nodes, blockchain nodes, and edge computing nodes. The system calculates node trustworthiness based on the node's verification capabilities and historical performance. Verification capabilities consider factors such as a node's computing resources, bandwidth, and supported verification algorithms; historical performance includes factors such as the node's online rate, response time, and data integrity maintenance record. Node trustworthiness is calculated by comprehensively evaluating these factors. For example, in a distributed storage network, a node that demonstrates long-term stability, rapid response, and support for multiple verification algorithms would receive a trustworthiness score of 0.95; whereas a newly joined node with limited resources would initially receive a trustworthiness score of only 0.60.

[0178] The allocation of storage tasks follows the rules of allocating important data to high-trust nodes, distributing related data blocks, and regional proximity. During the allocation process, the system considers factors such as the matching degree between the importance of the data block and the trustworthiness of the node, the correlation between data blocks, network topology, and geographical distribution. For example, in the digital work property registration system, the verification data of the original file is allocated to multiple high-trust nodes (such as 0.90 and above) for multi-copy storage; the intermediate results of the verification process are stored in multiple medium-trust nodes (such as 0.75-0.85) using erasure codes; and the basic metadata verification results are stored on a single node.

[0179] After completing the storage task allocation, the system builds a verification-aware index tree to record the storage location, verification path, and node credibility information of the data block. The verification-aware index tree is a multi-level index structure that is optimized for verification needs. The root node of the index tree contains verification summary information and path indexes; the middle-layer nodes are organized according to verification paths and dependencies; and the leaf nodes contain the storage location and access methods of specific data blocks. During the index construction process, the system considers the access frequency, criticality, and integrity requirements of the verification path, and places the indexes related to the frequently accessed verification paths close to the root node to improve access efficiency. For example, in the index tree of the music copyright registration platform, the index of the originality verification path is located in the first layer, which contains indexes of key nodes such as audio fingerprint verification and copyright database comparison; while the index of the format verification path is located at a deeper level.

[0180] The verification-aware index tree has adaptive update capabilities and can adjust as the verification path changes dynamically. When new verification units are added to the verification network, the performance of existing verification units changes, or the verification path is adjusted, the system will update the corresponding index structure. The update process is based on the principle of incremental adjustment, only modifying the affected index branches and maintaining the stability of other branches. For example, when the system introduces a new blockchain cross-chain verification unit, the index tree will add a corresponding index node and establish an association with the existing verification path; when the importance of a verification unit changes, the system will adjust the position and association weight of the relevant index node. The frequency of index updates is dynamically adjusted according to the rate of change of the verification network, and is usually updated after each major verification path adjustment or at a fixed period (such as every 1,000 verifications).

[0181] Once the index is generated, the system submits the stored index information to the blockchain network for evidence storage, ensuring the immutability and traceability of the index itself. The blockchain record of the stored index contains key information such as the index's hash value, timestamp, and index tree structure characteristics, providing the basis for subsequent verification data packet retrieval and integrity verification.

[0182] The present invention realizes intelligent hierarchical storage of verification data by constructing a data block association graph and verification path analysis. This solution innovatively introduces an adaptive storage strategy based on verification characteristics and a verification-aware index tree, effectively solving the problem of insufficient perception of verification requirements in traditional distributed storage systems. Through differentiated storage strategies and dynamically updated index structures, the system can significantly improve the access efficiency and recovery capabilities of verification data while ensuring data reliability. This verification-optimized storage index mechanism provides a solid data foundation for blockchain evidence storage, enhances the resilience of the data property registration system in the face of challenges such as node failures and network fluctuations, and provides efficient and reliable data support for property rights verification and evidence storage.

[0183] A second aspect of an embodiment of the present invention provides a data property rights registration system with dynamic encrypted evidence storage based on blockchain, including:

[0184] The first unit is configured to receive a data property registration request, wherein the data property registration request includes the data to be registered and the owner's identity information;

[0185] The second unit is used to build a trust transfer network and perform multi-dimensional trust measurement on the data to be registered, including building a trust transfer graph, recursively calculating the trust decay coefficient, and generating a trusted data packet containing a trust feature vector;

[0186] The third unit is used to establish a dynamic verification mechanism based on the trusted data packet, adaptively select a verification path through the coordination of multiple verification units and anomaly perception weight adjustment mechanism, combined with deep reinforcement learning, and generate a verification data packet containing the verification path and verification results;

[0187] A fourth unit is configured to perform distributed storage on the verification data packets, construct a data block association graph based on the importance of verification nodes and the integrity requirements of verification paths, and generate a storage index using an adaptive multi-level storage strategy and a verification-aware indexing mechanism;

[0188] The fifth unit is used to submit the storage index and verification summary information to the blockchain network to complete the evidence storage and generate a registration certificate containing the owner's identity information.

[0189] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0190] processor;

[0191] a memory for storing processor-executable instructions;

[0192] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0193] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0194] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0195] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data property rights registration method based on dynamic encryption and evidence storage in blockchain, characterized by: include: receiving a data property registration request, wherein the data property registration request includes the data to be registered and the owner's identity information; Constructing a trust transfer network to perform multi-dimensional trust measurement on the data to be registered, including constructing a trust transfer graph, recursively calculating a trust decay coefficient, and generating a trusted data packet containing a trust feature vector; A dynamic verification mechanism is established based on the trusted data packet. Through the weight adjustment mechanism of multi-verification unit collaboration and anomaly perception, combined with deep reinforcement learning, the verification path is adaptively selected to generate a verification data packet containing the verification path and verification results. Distributed storage is performed on the verification data packets, a data block association graph is constructed based on the importance of verification nodes and the integrity requirements of verification paths, and a storage index is generated using an adaptive multi-level storage strategy and a verification-aware indexing mechanism; Submit the storage index and verification summary information to the blockchain network to complete the evidence storage and generate a registration certificate containing the owner's identity information.

2. The method according to claim 1, characterized in that The credibility measurement of the data to be registered includes: Decomposing the generation process of the data to be registered into multiple operation points, and setting a trust probe at each operation point; The trust probe collects multi-dimensional trust factors including operational behavior characteristics, data flow paths, and environmental context; Constructing a trust transfer graph based on the multi-dimensional trust elements, wherein the nodes of the trust transfer graph represent operation points and the edges represent trust transfer relationships; In the trust transfer graph, a trust decay coefficient from the data source to the current node is obtained by recursive calculation, and the trust decay coefficient is dynamically adjusted as the trust transfer chain is extended; A trust threshold is set based on the trust decay coefficient. When the trust decay coefficient of a node is lower than the trust threshold, it is determined that a trust transmission interruption exists at the node. For the node where the trust transmission interruption occurs, an adaptive trust repair mechanism is triggered to restore the trust link by reconstructing the multi-path trust value. The trust decay coefficient of the reconstructed trust link is recalculated to verify the reconstructed trust transfer state; based on the verified trust transfer graph and the updated trust decay coefficient, a trusted data packet containing a trust feature vector is generated.

3. The method according to claim 2, characterized in that For nodes where trust transmission is interrupted, an adaptive trust repair mechanism is triggered to restore the trust link through multi-path trust value reconstruction, including: Obtain the historical trust sequence and associated node set of the interrupted node in the trust transfer graph; Selecting a node with a trust value higher than a first preset threshold from the associated node set as a starting node; adding a trusted relay node to the starting node to generate an extended path; calculating the similarity between the extended paths, and retaining the extended path with the higher trust value when the similarity between any two extended paths is greater than a preset similarity threshold, to obtain a candidate trust transfer path; Performing a trust evaluation on each path in the candidate trust transfer path, including: calculating the structural trust of the node based on the position information of the node in the trust transfer graph; calculating the temporal trust of the node based on the historical trust sequence; calculating the comprehensive trust of the node by combining the structural trust and the temporal trust; and combining the comprehensive trust of all nodes on the path with a path length penalty factor to obtain the path trust; Calculating the path stability of each candidate trust transfer path based on the degree of fluctuation of the historical trust sequence; calculating the reconstruction cost based on the computational overhead and resource consumption; A comprehensive path score is calculated based on path stability, reconstruction cost, and the trustworthiness of the path. A reconstruction deviation is calculated based on historical reconstruction results, and each score weight in the comprehensive path score is dynamically adjusted based on the reconstruction deviation. The path with the highest comprehensive path score is selected as the reconstructed path, and a new trust transfer link is established.

4. The method according to claim 1, wherein Establishing a dynamic authentication mechanism includes: The trust feature vector is input into a verification network, which includes multiple verification units; each verification unit performs a verification task based on different dimensions of the trust feature vector to generate a verification score; the verification units collaborate using a dynamic weight mechanism based on anomaly perception, and establish a behavioral baseline value for the verification unit based on the verification accuracy change interval and verification time series characteristic parameters; the verification behavior of the verification unit is monitored in real time, and when it is detected that the verification behavior deviates from the baseline value, the credibility degradation factor of the verification unit is calculated based on the degree of abnormality of the verification behavior, and the verification weight is updated using a nonlinear mapping method based on the credibility degradation factor and the historical verification accuracy. Adopting a deep reinforcement learning algorithm, taking the verification score and the trust decay coefficient as state inputs, and the verification accuracy and computational overhead as reward functions, the optimal verification path is adaptively selected; When new verification scenarios emerge, the verification experience of multiple verification units is aggregated through federated learning to achieve dynamic evolution of verification strategies; The verification task is executed based on the optimal verification path, and the execution result of the verification task and the optimal verification path are encapsulated into a verification data packet.

5. The method according to claim 4, characterized in that Using a deep reinforcement learning algorithm, taking the verification score and the trust decay coefficient as state inputs, and the verification accuracy and computational overhead as reward functions, the optimal verification path is adaptively selected, including: Combining the verification scores and trust decay coefficients of multiple verification units to form a state vector; Inputting the state vector into the policy network to obtain an initial verification path, executing the verification task according to the verification unit in the initial verification path, and calculating the verification reward value based on the execution result; calculating the verification reward value includes: obtaining the verification confidence of the verification unit; inputting the verification confidence into a sigmoid function to obtain a confidence adjustment coefficient; and multiplying the credibility degradation factor, the verification unit result accuracy, and the confidence adjustment coefficient to obtain the verification reward value; Multiply the verification calculation time by the path length penalty value and combine them with the resource consumption vector to obtain the calculation cost penalty value; Calculating a diversity bonus value based on the degree of difference between the initial verification path and the set of historical verification paths; A comprehensive reward value is obtained by weighted combination of the verification reward value, the computational overhead penalty value, the diversity reward value, and the policy entropy; Inputting the state vector and the comprehensive reward value into the policy network to perform verification path optimization, the verification path optimization comprising: calculating an advantage function value based on a temporal difference of an immediate reward and a state value function; performing a gradient update on policy network parameters based on the advantage function value; and adjusting the gradient update using an adaptive learning rate, the adaptive learning rate being calculated by first-order moment estimation and second-order moment estimation; The optimized policy network output is used as the optimal verification path.

6. The method according to claim 5, characterized in that The verification path optimization specifically includes: Extracting the current instantaneous reward from the comprehensive reward value sequence; inputting the state vector into the value function network to obtain the current state value and the next state value; using the difference between the instantaneous reward and the next state value adjusted by the discount factor as the time series difference error; combining the time series difference error with the exponentially weighted historical time series difference error to obtain the advantage function value; Calculate the log-likelihood of the current verification path relative to the policy network parameters; multiply the log-likelihood by the advantage function value to obtain the policy gradient; perform an exponential sliding average on the policy gradient to obtain a first-order moment estimate; perform an exponential sliding average on the square of the policy gradient to obtain a second-order moment estimate; performing deviation correction on the first-order moment estimate and the second-order moment estimate; obtaining an adaptive learning rate by dividing the first-order moment estimate by the sum of the square root of the second-order moment estimate and a second preset threshold; The adaptive learning rate is multiplied by the policy gradient to update the policy network parameters.

7. The method according to claim 1, characterized in that Distributed storage of the verification data packet and generation of a storage index include: Parsing the verification path and verification results in the verification data packet, extracting the verification dependency and the importance of the verification node; Building a data block association graph based on the verification dependency relationship, wherein the edge weights in the data block association graph are determined by the importance of the verification node and the verification path integrity requirement; Designing an adaptive storage strategy based on the topological structure of the data block association graph, including: using multiple copies of storage for target verification nodes, using erasure code storage for verification path branches, and using single copies of storage for independent verification nodes; Build a dynamic distributed storage network and calculate the credibility of storage nodes based on their verification capabilities and historical performance. Allocate storage tasks according to the adaptive storage strategy and node credibility; A verification-aware index tree is constructed to record the storage location, verification path, and node credibility information of the data block. The verification-aware index tree is adaptively updated as the verification path changes dynamically.

8. A data property rights registration system based on dynamic encryption and evidence storage based on blockchain, used to implement the method described in any one of claims 1 to 7, characterized in that: include: The first unit is configured to receive a data property registration request, wherein the data property registration request includes the data to be registered and the owner's identity information; The second unit is used to build a trust transfer network and perform multi-dimensional trust measurement on the data to be registered, including building a trust transfer graph, recursively calculating the trust decay coefficient, and generating a trusted data packet containing a trust feature vector; The third unit is used to establish a dynamic verification mechanism based on the trusted data packet, adaptively select a verification path through the coordination of multiple verification units and anomaly perception weight adjustment mechanism, combined with deep reinforcement learning, and generate a verification data packet containing the verification path and verification results; A fourth unit is configured to perform distributed storage on the verification data packets, construct a data block association graph based on the importance of verification nodes and the integrity requirements of verification paths, and generate a storage index using an adaptive multi-level storage strategy and a verification-aware indexing mechanism; The fifth unit is used to submit the storage index and verification summary information to the blockchain network to complete the evidence storage and generate a registration certificate containing the owner's identity information.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images