Cross-chain bridge-oriented transaction tracking method and system
By integrating blockchain data crawling and machine learning models, smart contracts are generated and feature vectors are fused, solving the problems of tampering and privacy leakage in cross-chain transactions, realizing trusted tracking of cross-chain bridge transactions, and improving the accuracy and security of tracking.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INSTITUTE OF INFORMATION ENGINEERING CHINESE ACADEMY OF SCIENCES
- Filing Date
- 2025-12-05
- Publication Date
- 2026-05-01
AI Technical Summary
Cross-chain transactions face challenges such as untrusted cross-chain bridge contracts, diverse on-chain communication protocols, and weak inter-chain networks. These challenges include transaction tampering leading to unreliable tracking results, infection of risk control systems, and transaction hijacking resulting in user privacy leaks or malicious exploitation. Existing technologies cannot achieve high-security protection for the integrity and confidentiality of cross-chain transaction tracking.
By integrating blockchain data crawling, key transaction feature extraction, and machine learning model recognition modules, feature value smart contracts and recognition result smart contracts are generated. Combined with structural feature extractors and semantic feature extractors, feature vector fusion and integrity verification are performed. By utilizing a multi-chain data acquisition environment and edge smart contract calls, cross-chain bridge transaction tracking is achieved.
It achieves trusted verification of cross-chain bridge transaction features, integrity protection for deposit transaction identification, and provides anti-tampering function for transaction pair matching. It builds a secure and trusted cross-chain bridge transaction tracking environment, improving the accuracy and robustness of cross-chain transaction tracking.
Smart Images
Figure CN121961564A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of blockchain cross-chain transaction monitoring technology, and in particular to a transaction tracking method and system for cross-chain bridges. Background Technology
[0002] The booming development of the multi-chain blockchain ecosystem has given rise to cross-chain bridges, a crucial infrastructure that enables assets to flow freely between different public chains, making business scenarios more flexible, efficient, and closer to user needs. However, due to the heterogeneous data structures and significant differences in confirmation times among different chains, the "source chain deposit → target chain withdrawal" process in cross-chain transactions is inherently fragmented and anonymized. Therefore, cross-chain transaction tracking for regulatory and risk control purposes has attracted widespread attention from researchers.
[0003] Cloud-based collaboration for cross-chain regulation tightly integrates on-chain data crawling, off-chain feature engineering, and machine learning model inference. Through a model training model in the cloud and real-time tracking off-chain, cloud and blockchain complement each other, integrating resources to distribute high-precision cross-chain transaction tracking models to the risk control end. This achieves the decentralization of cloud computing power and the extension of tracking capabilities, such as abnormal fund monitoring in regulatory technology and anti-money laundering decisions in DeFi scenarios. Furthermore, containerized microservices, as a widely used lightweight technology with excellent characteristics such as portability and rapid startup, are used by cross-chain tracking systems as a common carrier for cloud-based model distribution and an important tool for deploying tracking services at the risk control end. Container technology provides the best solution to the complexities of cross-chain data management needs, accelerating the development of cross-chain regulatory collaboration.
[0004] In environments characterized by untrusted cross-chain bridge contracts, diverse on-chain communication protocols, and weak inter-chain networks, various security and compliance issues pose significant challenges to cross-chain tracing. For example, cross-chain transactions may be hijacked or tampered with due to insecure data transmission between chains, leading to malicious actors forging transaction proofs or using member reasoning attacks to deduce original user privacy data. To address these risks, current technical solutions mostly employ controllable regulatory systems based on third-party blockchain analytics companies, but these cannot guarantee the complete trustworthiness of third parties in production environments. With the rapid development of zero-knowledge proof and trusted execution environment technologies, some research has also attempted regulatory systems based on decentralized verification, but these still cannot achieve high-level security protection for the integrity and confidentiality of cross-chain transaction tracing, nor build a trustworthy cross-chain regulatory environment. Therefore, under the current technological background, cross-chain transaction tracing still faces security issues such as unreliable tracing results due to cross-chain transaction tampering or the poisoning of risk control systems, and user privacy leaks or malicious exploitation due to cross-chain transaction hijacking. Figure 1 The diagram illustrates a scenario where cross-chain bridge transactions are tampered with and hijacked.
[0005] 1. Tampering with cross-chain transactions can lead to unreliable tracking results or compromised risk control systems.
[0006] When a cloud-trained tracking model is transmitted off-chain to a risk control node, a man-in-the-middle attack could occur through the inter-chain HTTPS channel, potentially leading to malicious actions such as replacing model parameters, implanting backdoor weights, or modifying inference thresholds. If the risk control system lacks a secure and reliable integrity verification mechanism before loading the model, it could result in risks such as missed detection of abnormal accounts and wrongful blocking of legitimate addresses. Therefore, integrity verification of the tracking model by the risk control system is crucial under cross-chain supervision.
[0007] 2. Hijacking of cross-chain transactions can lead to the leakage of user privacy or malicious exploitation.
[0008] Insecure inter-chain communication protocols can lead to the interception of cross-chain transaction data packets. When tracking models are distributed or transaction features are transmitted back from the cloud, malicious third parties or nodes can intercept sensitive data and use member inference attacks to reconstruct user address association graphs, resulting in serious privacy breaches. Furthermore, the tracking model can be used to illegally bypass regulations. Therefore, under cross-chain regulation, the confidentiality protection of tracking models is equally crucial.
[0009] 3. Semantic features lack on-chain binding and encryption protection, making them easy to intercept or forge.
[0010] While there are methods that use pre-trained models such as BERT to extract semantic features of transaction functions, their embedding vectors are easily poisoned by attackers through replay, replacement, or fine-tuning of the vectors, thereby misleading subsequent identification and matching results.
[0011] In summary, how to simultaneously address the issues of malicious tampering and privacy leaks in cross-chain transaction tracking models, achieve model integrity verification and confidentiality protection under cross-chain supervision, and build a trustworthy cross-chain transaction tracking environment has become an urgent technical problem to be solved. Summary of the Invention
[0012] This invention proposes a transaction tracking method and system for cross-chain bridges. The system integrates modules such as blockchain data crawling, key transaction feature extraction, machine learning model identification, and automatic transaction pair matching. It can also be integrated as a core module for anomaly account detection and transaction tracking tools, thereby solving problems such as difficulty in transaction tracing, difficulty in identifying anomaly accounts, and low tracking accuracy in blockchain cross-chain bridge transaction tracking.
[0013] To achieve the above objectives, the technical solution of the present invention includes the following:
[0014] A transaction tracing method for cross-chain bridges, applied in a cloud environment, the method comprising: The system negotiates with the cross-chain bridge to generate a feature value smart contract and an identification result smart contract, and uploads them to the blockchain. The access control function of the identification result smart contract includes the relationship between the cross-chain bridge, the contract address, and the feature extraction component, which includes a structural feature extractor and a semantic feature extractor. For a cross-chain bridge transaction generated in this cross-chain bridge, collect the original transaction data of the cross-chain bridge transaction; The structural feature vector of the original transaction data is extracted based on the structural feature extractor and the semantic feature extractor. With semantic feature vector Next, feature vector fusion is performed; Deposit transactions are identified based on the fused feature vector, and the fused feature vector and the deposit transaction identification result are uploaded to the blockchain using feature value smart contracts and identification result smart contracts, respectively. This allows the edge device to obtain the transaction tracking results of the cross-chain bridge transaction based on the feature value smart contracts, identification result smart contracts, fused feature vectors, and deposit transaction identification results in the blockchain.
[0015] Furthermore, the generation process of the feature extraction component includes: The system receives the contract address information and related metadata of the cross-chain bridge, and assigns monitoring addresses to the cross-chain bridge based on the cloud's monitoring requirements for the cross-chain bridge, in order to build a cross-chain bridge address database; wherein, the related metadata includes: contract address and chain ID; Based on the cross-chain bridge address database, structural feature extractors and semantic feature extractors are generated using structural embedding algorithms and word embedding algorithms, respectively.
[0016] Furthermore, the raw transaction data of this cross-chain bridge transaction is collected, including: Cross-chain bridges utilize the open-source Scrapy framework to build a multi-chain data acquisition environment based on distributed crawler nodes; In this multi-chain data collection environment, crawler nodes are used to collect the raw transaction data of the cross-chain bridge transaction.
[0017] Furthermore, the edge device, based on the feature value smart contract, the identification result smart contract, the fused feature vector, and the deposit transaction identification result in the blockchain, obtains the transaction tracking results of the cross-chain bridge transaction, including: Download the feature value smart contract and the recognition result smart contract from the blockchain; Based on this feature value, the smart contract obtains the fused feature vector in the blockchain; Verify the integrity of the fused feature vector; If the integrity of the fused feature vector is verified, the smart contract uses the recognition results to obtain the deposit transaction recognition results; If the deposit transaction identification result shows that the cross-chain bridge transaction is a deposit transaction, obtain the transaction log of the cross-chain bridge; Based on the transaction log, the source chain deposits and target chain withdrawals are matched to obtain the tracking results of the cross-chain bridge transaction.
[0018] Furthermore, the integrity of the fused feature vector is verified, including: Collect the raw transaction data of the cross-chain bridge transaction and extract the feature extraction component from the smart contract of the identification result; Based on the original transaction data, the structural feature vector of cross-chain bridge transactions is extracted using a feature extraction component. With semantic feature vector ; Fusion structural feature vector With semantic feature vector The integrity determination result of the fused feature vector is obtained based on the fusion result.
[0019] Furthermore, based on this transaction log, source chain deposits and target chain withdrawals are matched to obtain the tracking results of the cross-chain bridge transaction, including: Extract the receiving address, transaction type, asset type, timestamp, and amount of all cross-chain bridge transactions from the transaction log; The transaction is matched based on the receiving address, transaction type, asset type, timestamp, and amount to obtain the tracking results of the cross-chain bridge transaction.
[0020] A transaction tracking system for cross-chain bridges, the system comprising: The cloud, used for: The system negotiates with the cross-chain bridge to generate a feature value smart contract and an identification result smart contract, and uploads them to the blockchain. The access control function of the identification result smart contract includes the relationship between the cross-chain bridge, the contract address, and the feature extraction component, which includes a structural feature extractor and a semantic feature extractor. For a cross-chain bridge transaction generated in this cross-chain bridge, collect the original transaction data of the cross-chain bridge transaction; The structural feature vector of the original transaction data is extracted based on the structural feature extractor and the semantic feature extractor. With semantic feature vector Next, feature vector fusion is performed; Deposit transactions are identified based on the fused feature vector, and the fused feature vector and deposit transaction identification results are uploaded to the blockchain using feature value smart contracts and identification result smart contracts, respectively. At the edge, it is used to obtain the transaction tracking results of the cross-chain bridge transaction based on the feature value smart contract, the identification result smart contract, the fused feature vector and the deposit transaction identification result in the blockchain; Blockchain is used to store feature value smart contracts, recognition result smart contracts, fused feature vectors, and deposit transaction recognition results.
[0021] An electronic device includes: a processor and a memory storing computer program instructions; the processor, when executing the computer program instructions, implements the transaction tracking method for cross-chain bridges as described above.
[0022] A computer-readable storage medium, characterized in that the computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the transaction tracking method for cross-chain bridges as described above.
[0023] A computer program product, characterized in that, when the computer program product is run on a computer device, it causes the computer device to execute the transaction tracking method for cross-chain bridges described above.
[0024] Compared with the prior art, the positive effects of the present invention are as follows: 1) A transaction feature extraction method based on structural embedding and word embedding is proposed to solve the problem of low tracking accuracy caused by inaccurate cross-chain transaction feature extraction, and to realize reliable verification of cross-chain bridge transaction feature extraction; 2) A deposit transaction identification method based on machine learning is proposed to solve the problem of incomplete tracking links caused by the difficulty in identifying deposit transactions in cross-chain bridges, and to achieve integrity protection of cross-chain bridge transaction tracking; 3) A transaction pairing algorithm based on multi-dimensional features is proposed, which provides anti-tampering and secure transmission functions for transaction pair pairing, solves the problem of inaccurate cross-chain transaction link restoration, realizes fine-grained management of transaction pair pairing, and further improves the accuracy protection capability of cross-chain bridge transaction tracking. 4) Design a cross-chain bridge transaction tracking system based on multi-chain data collection and intelligent identification model, realize the integrity verification and accuracy protection of cross-chain bridge transaction tracking, and build a safe and reliable cross-chain bridge transaction tracking environment.
[0025] 5) Add on-chain hash binding and simple encrypted transmission to BERT semantic embedding to reduce the risk of vector poisoning and man-in-the-middle tampering, and improve the robustness of cross-chain transaction tracking with low overhead. Attached Figure Description
[0026] Figure 1 This diagram illustrates a scenario where transaction tracing and abnormal account identification are difficult in cross-chain bridge transaction tracking according to an embodiment of the present invention.
[0027] Figure 2 This is a flowchart of a transaction tracking method for cross-chain bridges according to an embodiment of the present invention.
[0028] Figure 3 It is a feature vector block structure.
[0029] Figure 4 To identify the block structure of the results. Detailed Implementation
[0030] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.
[0031] The transaction tracking system for cross-chain bridges disclosed in this embodiment consists of a cloud, an edge, and a blockchain. The cloud includes a cross-chain bridge address management component, a feature extraction component, a raw data acquisition component, a structural feature extraction component, a semantic feature extraction component, a feature fusion component, a deposit transaction identification component, and an on-chain data management component. The edge includes a smart contract invocation component, a transaction pair matching component, and an abnormal account tracking component. The blockchain mainly includes a consensus mechanism and smart contracts.
[0032] The following example illustrates the invention using a multi-chain data collection environment built on a distributed crawler in the cloud with n cross-chain bridges. Specifically, the cross-chain bridges Celer and Multichain need to monitor transaction activities on the Ethereum chain, while the cross-chain bridge PolyNetwork needs to monitor transaction activities on the BSC chain. Figure 2 As shown, the transaction tracking for cross-chain bridges of the present invention mainly includes the following steps.
[0033] Step 1: Building a multi-chain data acquisition environment and managing cross-chain bridge addresses.
[0034] A distributed crawler system is built on the cloud and edge based on node authentication, authorization, and permission communication using the Scrapy framework. The edge sends the contract address information of the cross-chain bridge to the cross-chain bridge address management component in the cloud. The cross-chain bridge address management component assigns monitoring addresses to the cross-chain bridge based on this information, stores them in a structured database, and generates a cross-chain bridge address library.
[0035] In this embodiment, the specific processing flow for constructing a multi-chain data acquisition environment and managing cross-chain bridge addresses includes the following steps 1.1 to 1.2.
[0036] Step 1.1: Building a multi-chain data acquisition environment based on distributed crawlers in the cloud and cross-chain bridge.
[0037] The cloud-based and cross-chain bridges Celer, Multichain, and PolyNetwork utilize the open-source Scrapy framework to build a multi-chain data acquisition environment based on distributed crawlers. This system can achieve parallel data acquisition from multiple public chains such as Ethereum, BSC, and Polygon. Through mechanisms such as API key rotation, node load balancing, and request frequency control, it improves the stability and efficiency of data acquisition. Table 1 shows the configuration parameters of the multi-chain data acquisition environment, illustrating key parameters such as API configuration, request frequency limits, and data acquisition scope for different public chains.
[0038] Table 1. Configuration Parameters for Multi-Chain Data Acquisition Environment public chain name API Interface Request frequency limit Data collection range Number of API keys Node weight Ethereum Etherscan 5 times / second Blocks, transactions, events 3 1:1:1 BSC Bscscan 5 times / second Blocks, transactions, events 3 1:1:1 Polygon PolygonScan 5 times / second Blocks, transactions, events 1 1 Avalanche Snowtrace 5 times / second Blocks, transactions, events 2 1:1 Arbitrum Arbiscan 5 times / second Blocks, transactions, events 2 1:1 In this embodiment, the cloud and all cross-chain bridge nodes have passed API authentication, authorization management, and communication permissions, and unanimously selected Ethereum, BSC, and Polygon public chains as data collection targets. The cloud and all cross-chain bridges negotiate to write feature value smart contracts and recognition result smart contracts and upload them to the blockchain. All participants in the distributed crawler download the two smart contracts to their local machines.
[0039] Step 1.2: Cross-chain bridge address management.
[0040] After the cloud and cross-chain bridges form a multi-chain data collection environment, the cross-chain bridges send their respective contract address information and related metadata (including contract address, chain ID, etc.) to the cloud. Upon receiving this information, the cross-chain bridge address management component in the cloud, based on the cloud's monitoring requirements for the cross-chain bridges, assigns Ethereum chain monitoring addresses to the cross-chain bridges Celer and Multichain. Then, it stores the address information in a structured database and generates an address library. , Assign monitoring addresses for the BSC chain to the cross-chain bridge PolyNetwork, using the address database. As a carrier of monitoring addresses, it ultimately generates a cross-chain bridge address database. , , The cloud uploads the cross-chain bridge address database information to the access control function in the smart contract containing the identification results. In this way, the cloud can determine the correspondence between different cross-chain bridges and different contract addresses based on the cross-chain bridge address database, and synchronize the relevant information to the configuration file, achieving fine-grained management of cross-chain bridge monitoring from the cloud.
[0041] The system supports monitoring of various cross-chain bridges, including mainstream ones such as Celer, Multichain, and PolyNetwork. Table 2 shows the cross-chain bridge monitoring configuration, displaying the monitoring parameters and configuration information for different cross-chain bridges.
[0042] Table 2 Cross-Bridge Monitoring Configuration Table Cross-chain bridge name Supported chain number Contract Quantity Number of monitoring addresses Data acquisition frequency Tracking accuracy Celer 15+ 50+ 200+ real time 72.3% Multichain 20+ 80+ 300+ real time 70.8% PolyNetwork 10+ 30+ 150+ real time 71.5% Stargate 8+ 25+ 100+ real time 69.2% Step 2: Feature extraction group generation and raw data collection.
[0043] The cloud-based feature extraction component generates corresponding feature extraction groups based on the cross-chain bridge address database, and then sends them to the raw data collection component. This component uses feature extraction algorithms to collect multi-chain transaction data to obtain the raw transaction data.
[0044] Step 2.1: Feature extraction group generation.
[0045] The cloud-based cross-chain bridge address management component generates a cross-chain bridge address library. , , This information is sent to the feature extraction component, which then uses a structure embedding algorithm to generate a structure feature extractor based on the cross-chain bridge address database information. A semantic feature extractor is generated using a word embedding algorithm. and form a feature extraction group ( , ).
[0046] In one embodiment, Table 3 is a performance comparison table of feature extraction algorithms, showing the performance of different feature extraction algorithms in terms of accuracy, processing speed, and memory usage. Considering the complexity and diversity of cross-chain transaction features, this invention selects the Motif counting algorithm as the structural feature extraction method and the Word2Vec algorithm as the semantic feature extraction method.
[0047] Table 3 Performance Comparison of Feature Extraction Algorithms Algorithm Name Feature Dimension accuracy Processing speed (ms / tx) Memory usage (MB) Applicable Scenarios Motif count 16-dimensional 72.3% 15 128 Structural feature extraction Word2Vec 36-dimensional 68.5% 22 256 Semantic feature extraction BERT 768 dimensions 75.1% 85 512 High-level semantic features GraphSAGE 64-dimensional 73.8% 45 384 Graph structure features Node2Vec 128 dimensions 71.2% 38 320 Node embedding features It should be noted that the cloud-generated transaction feature extraction group is used to determine the relationship between the cross-chain bridge (i.e., the edge), the contract address, and the feature extraction group, and to synchronize this relationship information to the access control function in the smart contract of the recognition result. After obtaining the recognition result smart contract, edge users can only read and write the recognition result within the granted scope of "cross-chain bridge + contract address + feature extraction group". The contract updates the access control list accordingly to achieve fine-grained permission isolation. Specifically, the "cross-chain bridge, contract address, feature extraction group" triple is manifested as follows: the same cross-chain bridge can correspond to multiple contract addresses on multiple chains (one-to-many); each address is bound to only one structure extractor and one semantic extractor (address-method one-to-one); different bridges can reuse the same extraction method (many-to-one); edge users can only read and write the recognition result within the granted scope of "cross-chain bridge + contract address + feature extraction group", and the contract updates the access control list accordingly to achieve fine-grained permission isolation.
[0048] Step 2.2: Raw data collection.
[0049] After the feature extraction group is generated in the cloud ( , The feature extraction component will ( , The data is sent to the raw data acquisition component. The raw data acquisition component first uses distributed crawler nodes to pull blocks, transaction receipts, and event logs of relevant contracts from chains such as Ethereum, BSC, and Polygon in real time. After parsing, cleaning, and standardization, the raw transaction data is formed. .
[0050] Step 3: Generation of structural feature vectors and semantic feature vectors.
[0051] After generating the transaction feature extraction group in the cloud, the transaction feature extraction group ( , The original transaction data is used to extract features, generating structural feature vectors and semantic feature vectors. Since cross-chain transaction data consists of multiple transaction fields, each containing relevant sensitive information, the cloud utilizes the transaction feature extraction method based on structural embedding and word embedding proposed in this invention. Extract the structural features of the transaction data and utilize... The semantic features of transaction data are extracted to form a complete feature vector. To ensure privacy during feature extraction, this method can also extract features directly without parsing the transaction data in plaintext. Cloud-based feature extraction using structural embedding and word embedding-based methods aims to alleviate computational burden, address risks of data and privacy breaches, and provide security, efficiency, and privacy protection for edge-based feature extraction.
[0052] Step 3.1: Extraction of structural feature vectors.
[0053] The cloud sends the raw transaction data from the raw data acquisition component to the structural feature extraction component. This component uses the Motif counting algorithm to calculate the raw transaction data and obtain the structural feature vector.
[0054] In this embodiment, the raw data acquisition component in the cloud generates raw transaction data. Then, the raw transaction data is sent to the structural feature extraction component. The structural feature extraction component uses the Motif counting algorithm to calculate and generate a 32-dimensional structural feature vector for all raw transaction data, which serves as the structural feature representation of the transaction data. The raw data acquisition component sends all structural feature vectors to the on-chain data management component according to their corresponding relationships.
[0055] Step 3.2: Semantic feature vector extraction.
[0056] The cloud sends the raw transaction data from the raw data acquisition component to the semantic feature extraction component. This component uses the Word2Vec algorithm to calculate the semantic feature vector from the raw transaction data.
[0057] In this embodiment, the raw data acquisition component in the cloud generates raw transaction data. Then, the raw transaction data is sent to the semantic feature extraction component. The semantic feature extraction component uses the Word2Vec algorithm to calculate and generate a 32-dimensional semantic feature vector for all raw transaction data, which serves as the semantic feature representation of the transaction data. The raw data acquisition component sends all semantic feature vectors to the on-chain data management component according to their corresponding relationships.
[0058] Step 4: Feature vector fusion and deposit transaction recognition.
[0059] At the edge, a feature fusion component is used to fuse two types of feature vectors, generating a fused feature vector for subsequent deposit transaction identification and transaction pairing. Higher dimensionality of the fused feature vector leads to greater accuracy, but also longer training and inference times. This scheme considers three dimensions: feature dimension, training time, and recognition accuracy. Experiments were conducted to compare three common feature dimensions: 16-dimensional, 32-dimensional, and 64-dimensional. Table 4 shows the comparison results for the three feature dimensions. As shown in Table 4, considering all factors, 32-dimensional features perform optimally. Therefore, the transaction feature extraction method based on structural embedding and word embedding proposed in this embodiment uses a 32-dimensional feature vector to calculate the fused feature vector, generating a 32-dimensional feature representation.
[0060] Table 4 Comparison results of the three feature dimensions In this embodiment, a feature fusion component is used to fuse structural feature vectors and semantic feature vectors to generate a fused feature vector. Next, the deposit transaction recognition component is used to identify deposit transactions from the fused feature vector, and the recognition results are output. Finally, the feature value smart contract and the recognition result smart contract are used to upload the fused feature vector and recognition result to the blockchain.
[0061] Specifically, after receiving the data, the on-chain data management component calls the feature value smart contract interface, invokes the "write" function within the feature value smart contract, and then achieves consensus among all participants through the blockchain's consensus mechanism, and proceeds according to... Figure 3 The feature vector block structure is stored on the chain; After the feature vectors are stored on the blockchain in the cloud, the on-chain data management component calls the "write" function of the smart contract containing the recognition results through its interface, updates the access control information in the smart contract, and then reaches a consensus among all participants through the blockchain's consensus mechanism, proceeding as follows: Figure 4 The identification result block structure, and the on-chain storage block structure; Step 5: Transaction tracking.
[0062] Step 5.1: After receiving the feature vector, the edge device uses the smart contract to call the feature value smart contract interface in the component, and reads the feature vector from the chain. Then, the edge device uses the same feature extraction algorithm locally to calculate the original transaction data collected by the edge device, and compares whether the feature vectors of the two are consistent, thus completing the integrity verification of the feature vector.
[0063] In this embodiment, after receiving the feature vector, the edge device calls the feature value smart contract call interface in the component through the smart contract, and calls the "read" function of the feature value smart contract to read the fused feature vector on the chain. Then, the edge device uses the same feature extraction algorithm locally to calculate the fused feature vector on the original transaction data. If the calculation is consistent, the feature extraction has not been tampered with.
[0064] Step 5.2: After the edge device completes the integrity comparison of the feature vectors, it needs to obtain the corresponding recognition results from the blockchain before it can perform transaction pair matching. The edge device calls the smart contract for the recognition results. The smart contract first uses the access permission function to determine whether the edge user who wants to read the recognition results meets the access permission requirements based on the correspondence between "cross-chain bridge, address, and feature extraction method". If the user meets the access permission requirements, the user can read the recognition results; otherwise, access is denied.
[0065] Step 5.3: After obtaining the fused feature vector and recognition results at the edge, the recognition results are configured in the transaction pair pairing file. This eliminates the need for re-extraction of features and allows for direct transaction pair pairing. The transaction pair pairing algorithm employs a multi-rule matching strategy, including five core rules: receiving address matching, transaction type matching, asset type matching, timestamp matching, and amount matching.
[0066] In this embodiment, the edge computing platform utilizes a transaction pairing component to automatically pair source chain deposits with target chain withdrawals based on multi-dimensional features such as receiving address, amount, and timestamp. These features are all derived from standardized transaction logs parsed on-chain. The receiving address and amount are directly extracted from the transaction input and event logs, and the timestamp is taken from the block header, ensuring data integrity and immutability. For transactions identified as deposit transactions, the system automatically searches for corresponding withdrawal transactions, forming a complete cross-chain transaction chain. The pairing process employs a multi-rule matching strategy, cascading according to priority based on pairing rules: receiving address → transaction type → asset type → timestamp → to ensure the accuracy of the pairing results.
[0067] Finally, after all edge terminals have completed the pairing of trading pairs, the abnormal account tracking component is used to integrate the tracking results with the abnormal account discovery tool, enabling cross-chain fund flow tracking of suspicious accounts, which further ensures the integrity and accuracy of cross-chain bridge transaction tracking.
[0068] In summary, this invention proposes a cross-chain bridge transaction tracking method based on multi-chain data acquisition and intelligent recognition models, and designs a corresponding system. By constructing a multi-chain data acquisition environment and generating a cross-chain bridge address database, it provides transaction feature extraction methods based on structural embedding and word embedding, supports reliable verification of cross-chain bridge transaction feature extraction, provides a deposit transaction recognition method based on machine learning, achieves integrity protection for cross-chain bridge transaction tracking, and provides a transaction pairing algorithm based on multi-dimensional features, enabling fine-grained management of transaction pair pairing. This provides integrity verification and accuracy protection for cross-chain bridge transaction tracking, constructing a reliable and highly accurate cross-chain bridge transaction tracking environment.
[0069] The above embodiments are provided merely for the purpose of describing the present invention and are not intended to limit the scope of the invention. The scope of the invention is defined by the appended claims. Various equivalent substitutions and modifications made without departing from the spirit and principles of the invention should be covered within the scope of the invention.
Claims
1. A transaction tracking method for cross-chain bridges, characterized in that, Applied to a cloud environment, the method includes: The system negotiates with the cross-chain bridge to generate a feature value smart contract and an identification result smart contract, and uploads them to the blockchain. The access control function of the identification result smart contract includes the relationship between the cross-chain bridge, the contract address, and the feature extraction component, which includes a structural feature extractor and a semantic feature extractor. For a cross-chain bridge transaction generated in this cross-chain bridge, collect the original transaction data of the cross-chain bridge transaction; The structural feature vector of the original transaction data is extracted based on the structural feature extractor and the semantic feature extractor. With semantic feature vector Next, feature vector fusion is performed; Deposit transactions are identified based on the fused feature vector, and the fused feature vector and the deposit transaction identification result are uploaded to the blockchain using feature value smart contracts and identification result smart contracts, respectively. This allows the edge device to obtain the transaction tracking results of the cross-chain bridge transaction based on the feature value smart contracts, identification result smart contracts, fused feature vectors, and deposit transaction identification results in the blockchain.
2. The method according to claim 1, characterized in that, The generation process of the feature extraction component includes: The system receives the contract address information and related metadata of the cross-chain bridge, and assigns monitoring addresses to the cross-chain bridge based on the cloud's monitoring requirements for the cross-chain bridge, in order to build a cross-chain bridge address database; wherein, the related metadata includes: contract address and chain ID; Based on the cross-chain bridge address database, structural feature extractors and semantic feature extractors are generated using structural embedding algorithms and word embedding algorithms, respectively.
3. The method according to claim 1, characterized in that, Collect the raw transaction data for this cross-chain bridge transaction, including: Cross-chain bridges utilize the open-source Scrapy framework to build a multi-chain data acquisition environment based on distributed crawler nodes; In this multi-chain data collection environment, crawler nodes are used to collect the raw transaction data of the cross-chain bridge transaction.
4. The method according to claim 1, characterized in that, The edge terminal obtains the transaction tracking results of the cross-chain bridge transaction based on the feature value smart contract, the identification result smart contract, the fused feature vector, and the deposit transaction identification result in the blockchain, including: Download the feature value smart contract and the recognition result smart contract from the blockchain; Based on this feature value, the smart contract obtains the fused feature vector in the blockchain; Verify the integrity of the fused feature vector; If the integrity of the fused feature vector is verified, the smart contract uses the recognition results to obtain the deposit transaction recognition results; If the deposit transaction identification result shows that the cross-chain bridge transaction is a deposit transaction, obtain the transaction log of the cross-chain bridge; Based on the transaction log, the source chain deposits and target chain withdrawals are matched to obtain the tracking results of the cross-chain bridge transaction.
5. The method according to claim 4, characterized in that, Verify the integrity of the fused feature vector, including: Collect the raw transaction data of the cross-chain bridge transaction and extract the feature extraction component from the smart contract of the identification result; Based on the original transaction data, the structural feature vector of cross-chain bridge transactions is extracted using a feature extraction component. With semantic feature vector ; Fusion structural feature vector With semantic feature vector The integrity determination result of the fused feature vector is obtained based on the fusion result.
6. The method according to claim 4, characterized in that, Based on this transaction log, source chain deposits and target chain withdrawals are matched to obtain the tracking results of the cross-chain bridge transaction, including: Extract the receiving address, transaction type, asset type, timestamp, and amount of all cross-chain bridge transactions from the transaction log; The transaction is matched based on the receiving address, transaction type, asset type, timestamp, and amount to obtain the tracking results of the cross-chain bridge transaction.
7. A transaction tracking system for cross-chain bridges, characterized in that, The system includes: The cloud, used for: The system negotiates with the cross-chain bridge to generate a feature value smart contract and an identification result smart contract, and uploads them to the blockchain. The access control function of the identification result smart contract includes the relationship between the cross-chain bridge, the contract address, and the feature extraction component, which includes a structural feature extractor and a semantic feature extractor. For a cross-chain bridge transaction generated in this cross-chain bridge, collect the original transaction data of the cross-chain bridge transaction; The structural feature vector of the original transaction data is extracted based on the structural feature extractor and the semantic feature extractor. With semantic feature vector Next, feature vector fusion is performed; Deposit transactions are identified based on the fused feature vector, and the fused feature vector and deposit transaction identification results are uploaded to the blockchain using feature value smart contracts and identification result smart contracts, respectively. At the edge, it is used to obtain the transaction tracking results of the cross-chain bridge transaction based on the feature value smart contract, the identification result smart contract, the fused feature vector and the deposit transaction identification result in the blockchain; Blockchain is used to store feature value smart contracts, recognition result smart contracts, fused feature vectors, and deposit transaction recognition results.
8. An electronic device, characterized in that, The electronic device includes: a processor and a memory storing computer program instructions; the processor, when executing the computer program instructions, implements the transaction tracking method for cross-chain bridges as described in any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the transaction tracking method for cross-chain bridges as described in any one of claims 1-6.
10. A computer program product, characterized in that, When the computer program product is run on a computer device, the computer device performs the transaction tracking method for cross-chain bridges as described in any one of claims 1-6.