A method and system for recording and tracing operational behavior across all dimensions.
By constructing an independent risk control audit program and an isolated service cluster, and adopting the RDA rights confirmation encryption algorithm and the blockchain dual-path rights confirmation certificate system, the problems of data isolation and security protection, insufficient credibility of evidence storage, weak adaptability to multiple scenarios, and insufficient granularity of traceability have been solved, thus achieving full-chain compliance auditing and efficient traceability.
Patent Information
- Application Number
- CN202611303674.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-26
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies in data auditing systems have shortcomings in data isolation and security protection, insufficient credibility of evidence storage, weak adaptability to multiple scenarios, insufficient granularity of traceability, and imperfect compliance control across the entire chain, making it difficult to meet the unified auditing needs of multi-track operations.
An independent risk control audit program and isolated service cluster are constructed. The RDA rights confirmation encryption algorithm and blockchain dual-path rights confirmation certificate system are adopted to generate irreversible electronic change ledgers, realize full-dimensional operation behavior recording and data change traceability, support one-click retrieval of single and batch data full life cycle modification records, and combine blockchain and NFG two-way binding mechanism to ensure data immutability and realize full-chain compliance audit.
It achieves secure isolation between sensitive data and audit data, enhances the credibility of evidence data, reduces expansion costs, improves traceability efficiency and compliance control capabilities, and meets the unified audit needs of multi-track operations.
Smart Images

Figure CN123116355A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and in particular to a method and system for recording all-dimensional operational behaviors and tracing data changes. Background Technology
[0002] In existing technologies, operation recording is typically achieved by embedding a log module within the business system. This module stores basic information such as the operation account, operation time, and operation type, supporting simple operation queries and post-event traceability. Some solutions enhance log tamper-proof capabilities by introducing hash verification or optimize log query efficiency through hierarchical storage, thereby improving the availability of the auditing system to some extent.
[0003] However, existing technologies still have many shortcomings: First, there are shortcomings in data isolation and security protection. Existing audit modules are mostly deployed on the same cluster as business systems, sharing computing and storage resources. The massive amount of raw business data can easily crowd out the operating resources of the audit system, and audit response delays are likely to occur in high-concurrency scenarios. At the same time, the audit process needs to read complete business data, making it difficult to achieve physical isolation between raw business data and audit data, which poses a risk of leakage of sensitive business data and fails to meet strict data security compliance requirements.
[0004] Second, the evidence is not credible enough and has limited anti-tampering capabilities. Existing operation logs are mostly stored in business-side databases, and relying solely on access control to restrict log modification does not provide irreversible evidence preservation capabilities. Log data is susceptible to tampering and deletion, making it difficult to serve as reliable audit evidence in the event of violations or data disputes, and also making it impossible to accurately identify the responsible party.
[0005] Third, the system has weak adaptability to multiple scenarios and high expansion costs. The business system architecture, data format, and interface specifications of different business tracks vary greatly. Existing audit systems are mostly customized for single business scenarios. When accessing new tracks, the system architecture needs to be significantly modified. It cannot achieve rapid plug-in access, and horizontal expansion is difficult and has a long implementation cycle, making it difficult to adapt to the unified audit requirements of multiple business tracks.
[0006] Fourth, the granularity of traceability is insufficient, resulting in low audit efficiency. Existing log systems mostly only support single-record queries and cannot automatically connect the entire lifecycle of a single data entry from creation to all changes. They also struggle to achieve hierarchical traceability based on the three-tiered business entity structure of manufacturers, wholesalers, and stores. Furthermore, abnormal operations lack unified quantitative judgment standards, requiring auditors to manually identify each case, leading to low efficiency in locating violations and an inability to quickly pinpoint the responsible party.
[0007] Fifth, the end-to-end compliance management system is incomplete. Existing solutions mainly focus on the retention of operation records, lacking standardized account operation and maintenance process verification and end-to-end compliance rule management capabilities. Compliance standards for different scenarios are difficult to implement uniformly, making it impossible to form a systematic compliance audit capability covering the entire operation process, and thus failing to support the enterprise's full-domain compliance management needs.
[0008] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention
[0009] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of the invention.
[0010] According to one aspect of this application, a method for recording and tracing all-dimensional operational behavior and data changes is provided, including: acquiring all operational behaviors related to account creation, permission allocation, account activation / deactivation, business data modification, POS cashier transactions, and asset distribution for multi-level business entities; simultaneously collecting operational account, login IP, terminal device identifier, and precise operation time information; deploying an independent risk control audit program, a dedicated database, and an isolated service cluster, physically isolated from the business time-series processing program, synchronizing only the unique identifier of business data, and not storing the complete original business data stream; adopting a single-base plug-in architecture to adapt to multi-track offline operational audit scenarios; using a complete RDA rights confirmation encryption algorithm to generate solidified evidence records, building a blockchain and NFG dual-path rights confirmation certificate system, constructing a two-way binding mechanism between rights confirmation certificates and audit ledgers, automatically binding all-dimensional operational entity information when data changes, and generating unbreakable records. The system features a reverse electronic change log; it provides one-click retrieval of single and batch data modification records throughout their entire lifecycle, fully restoring the entire data change process, retaining complete traceability fields, and accurately identifying the responsible parties for data tampering and abnormal changes; the traceability audit logs are logically partitioned and isolated according to the operating entities (A-end, B-end, and C-end); business terminals only have log appending and writing permissions, and cannot modify or delete generated historical logs; the system maintenance terminal can perform log archiving and migration, but cannot tamper with the original log content; log writing synchronously generates corresponding RDA hash digests and NFG trusted digital ownership certificates, relying on certificate constraints to prohibit tampering with generated historical logs; all backend maintenance operation logs are solidified, multi-level account maintenance processes are standardized, a complete online data compliance audit architecture is built, and the audit cluster is dedicated to carrying out evidence storage work, separating business data storage tasks.
[0011] Another aspect of this application is a full-dimensional operation behavior recording and data change tracing system, the system being configured to execute the above-described full-dimensional operation behavior recording and data change tracing method by executing the executable instructions.
[0012] According to another aspect of this application, an electronic device includes: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the above-described method for recording all-dimensional operational behavior and tracing data changes by executing the executable instructions.
[0013] According to another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a second processor, implements the above-described method for recording all-dimensional operational behaviors and tracing data changes.
[0014] The beneficial effects of this application are as follows: Addressing the audit and traceability needs of multi-track offline operations, it establishes an independent audit cluster completely physically isolated from the business time-series processing system. It only synchronizes unique identifiers for business data in one direction, achieving secure isolation between sensitive business data and audit data from the underlying architecture, while ensuring independent scheduling of audit resources unaffected by business traffic. It adopts a standardized, single-base, pluggable architecture, quickly adapting to the audit access requirements of different business tracks through unified interface specifications and compatibility verification mechanisms. Combining RDA rights confirmation encryption algorithms and a consortium blockchain two-way anchoring mechanism, it generates an irreversible electronic change ledger with dual guarantees of local encryption and on-chain immutability. It supports one-click retrieval of single and batch full lifecycle change links, with accompanying quantitative anomaly judgment rules to automatically identify risky operations and accurately pinpoint responsible parties. Simultaneously, it solidifies all operation and maintenance logs and standardizes multi-level account processes, constructing a comprehensive compliance audit system covering the entire operation chain.
[0015] This application aims to achieve a physically isolated architecture that cuts off the path of business data leakage from the bottom layer, independently allocates audit computing power and storage resources, and ensures stable audit response even in high-concurrency scenarios, thus balancing data security and operational efficiency.
[0016] Furthermore, the dual ownership confirmation mechanism of RDA encryption and blockchain significantly enhances the credibility and tamper-proof capabilities of the stored data. The stored records serve as reliable audit evidence, providing effective support for data disputes and liability determination. The plug-in adaptation model significantly reduces the cost of accessing and transforming new business models, allowing for rapid deployment without restructuring the system architecture and significantly improving the scalability of unified auditing across multiple business sectors. It achieves automatic connection and quantitative anomaly identification throughout the entire lifecycle, reducing manual screening workload and significantly improving the speed of violation location and source tracing audit efficiency. Standardized compliance management across the entire chain enables full-process verification of accounts, operations, and boundaries, comprehensively supporting the enterprise's full-domain compliance management needs.
[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description
[0018] Figure 1 This document illustrates a flowchart of a method for recording and tracing all aspects of operational behavior and data changes, provided in an embodiment of this application. Figure 2 The diagram shows a schematic representation of a comprehensive operational behavior recording and data change tracing system provided in an embodiment of this application. Detailed Implementation
[0019] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0020] In one implementation, Figure 1 The diagram illustrates a flowchart of a method for recording all-dimensional operational behavior and tracing data changes according to an embodiment of this application.
[0021] S101 acquires all operational behaviors related to account creation, permission allocation, account activation / deactivation, business data modification, POS cashier transactions, and multi-level business entity asset profit sharing, and simultaneously collects information such as the operation account, login IP, terminal device identifier, and precise operation time.
[0022] In one implementation, the hierarchical collection of full-scenario operational behaviors covers three major categories of operational events: online account maintenance, business data modification, and offline business transactions and profit sharing. Different capture channels are used for different types of operations, and event hierarchical marking is completed according to the level of the business entity.
[0023] The first category is online account maintenance operations, including three core sub-operations: account creation, permission allocation, and account startup / shutdown. These operations are captured by attaching lightweight data collection probes to all write operation interfaces in the backend management system. The probes only intercept metadata information from the operation requests and do not intrude on the business logic execution flow. When an interface receives an account-related write operation request, the probe automatically determines the operation type and reads the hierarchical attributes of the current account to classify and label the event at different levels, such as headquarters, region, and store. For example, when a backend administrator submits a request to create a new store operator account, the interface probe captures the write operation event in real time, determines it as an account creation type, identifies the store-level entity to which the account belongs, and includes the event in the corresponding level of data collection.
[0024] The second category is business data modification operations. These operations are captured by monitoring the change logs of the business database. Only the event type of the data change and the corresponding unique identifier of the business data are captured; the complete business data content is not read, strictly adhering to the principle of minimal data collection. When a new, modified, or deleted operation is detected in a core business data table, the monitoring component automatically extracts the unique primary key and operation type of the changed data, generating a corresponding operation event. For example, when store product pricing data is updated, the database change monitoring component captures the update event of the corresponding data table, extracts the unique identifier and modification operation type of the product data, completes event collection, and does not read the specific numerical value of the product price or other detailed business information throughout the process.
[0025] The third category is offline operational operations, including POS cashier transactions and multi-level operating entity asset profit sharing. This type of operation relies on a single-base plug-in data acquisition adapter to achieve multi-track access. Business systems from different offline operating tracks connect through standardized plug-in interfaces. All operational events are accessed by the data acquisition module in a unified format, while automatically associating the hierarchical relationships of multi-level operating entities. Multi-level operating entities are stored in a tree-like hierarchical structure. Each operating entity corresponds to a globally unique entity identifier, and the entity attributes include a field for the identifier of its superior entity. Lower-level operating entities are associated with their superior entities level by level, while the top-level entity has no superior identifier. The hierarchical association logic is as follows: when an operational event is collected, the system reads the lowest-level entity identifier synchronized from the business side, recursively queries all superior entity identifiers upwards along the tree structure, automatically completes the full hierarchical relationship of the entities, and attaches all entity identifiers at all levels to the corresponding operational event.
[0026] The hierarchical mapping verification rule is as follows: each subordinate entity must be associated with one and only one direct superior entity. Records whose hierarchical relationship does not meet the one-to-one constraint will automatically trigger verification anomalies and be pushed to the operations side for review and correction. Entity hierarchical data is fully synchronized once daily at midnight, and incremental synchronization is triggered in real time when there are daily entity changes to ensure that the entity hierarchical relationship on the audit side is consistent with that on the business side. For example, when a store in the retail sector completes a checkout transaction, the corresponding sector plugin of the POS terminal synchronizes the transaction operation event to the collection module according to a unified protocol, and automatically associates it with the superior entity identifier of the store, realizing the event collection of multi-level entity hierarchical association.
[0027] When each operation event triggers data collection, the system automatically and synchronously captures four core traceability elements. The operation event is bound to the traceability element one by one through the unique primary key of the event. All elements are calibrated and standardized using corresponding algorithm logic to ensure the accuracy and immutability of the traceability information.
[0028] The first element is the account information. The system extracts the unique identifier of the account from the identity authentication context of the current session, and automatically associates the account with the business entity level and permission scope to complete the mapping and binding of account and entity information. For example, when a store operator initiates a permission adjustment operation, the system parses the operator's unique account identifier from the identity credentials of the current session, and matches the store entity to which the account belongs with the corresponding superior wholesaler entity to complete the collection and archiving of account-level traceability information.
[0029] The second element is the login IP address. The system extracts the accessing public IP address from the source address field of the network request message, and simultaneously uses the IP address location resolution database to match the location information, generating a corresponding IP location tag, which is then stored in the traceability field set. For example, when an administrator logs into the backend through the office network to modify business data, the system extracts the corresponding public IP address from the request message, matches it to obtain the location information of the corresponding city, and simultaneously stores the IP address and location information in the traceability field of that operation record.
[0030] The third element is the terminal device identification element. The system employs a multi-feature weighted fusion device fingerprint generation algorithm. Based on more than ten dimensions of feature data, including browser features, operating system information, hardware parameters, and canvas features, hash encoding and weighted calculations are performed sequentially to generate a unique device fingerprint code of fixed length. The device fingerprint algorithm selects twelve core terminal features as the basis for calculation: browser user agent, operating system version, screen resolution, time zone information, language settings, number of CPU cores, graphics card information, canvas rendering features, audio fingerprint, local storage support status, plugin list, and font list. Feature weights are set according to discrimination: canvas rendering feature weight is 0.2, audio fingerprint weight is 0.15, graphics card information weight is 0.15, and the remaining nine features each have a weight of 0.05, with a total weight of 1.
[0031] The operation process is as follows: First, each feature value is standardized into a string, converted to lowercase, and redundant spaces are removed. Then, each standardized feature value is hashed using SHA256 to generate a corresponding single feature hash value. Next, all single hash values are weighted and concatenated according to the weight of each feature. Finally, the concatenated result is hashed again to output a unique device fingerprint code in 32-bit hexadecimal format.
[0032] When a terminal first connects to the system, an initial device fingerprint is generated. Subsequent operations recalculate the fingerprint and compare it with existing fingerprints. If the matching rate reaches 95% or higher, the device is considered the same, ensuring that device-level trajectory association remains possible even after the terminal clears its cache. This algorithm avoids the vulnerability of single device identifiers to tampering or deletion through multi-dimensional feature fusion. Even if the terminal clears its local cache, accurate association of operation trajectories at the device level can still be achieved. For example, when a terminal initiates an operation request for the first time, the system automatically collects more than ten terminal environmental features, generates a unique 32-bit device fingerprint code through weighted hash calculation, and all subsequent operations initiated by this terminal are bound to the same device identifier, achieving full device-level operation trajectory chaining.
[0033] The fourth element is precise operation time. The system employs a periodic clock skew correction algorithm, synchronizing with a standard time source every five minutes, calculating the deviation between the local cluster clock and the standard time, and performing real-time deviation correction on the collected operation timestamps, achieving millisecond-level time accuracy. This algorithm avoids the problems of client-side time tampering and inconsistencies between clocks on different nodes, ensuring the global consistency and authenticity of all operation times. For example, when an operation event arrives at the audit cluster, the local clock reads the initial time value, and the system corrects it using the clock skew value calculated in the current period, obtaining the calibrated precise operation time, which is then stored in the traceability field of that record.
[0034] All collected operation events and traceability elements are associated and aggregated using the event's unique primary key to generate a standardized raw dataset of operation events. The system synchronously performs data integrity verification, checking each record for completeness of the four traceability elements. Records lacking any element automatically trigger a completion mechanism, initiating an element completion request to the corresponding business node. Records failing to complete are marked as abnormal events and archived separately. The standardized dataset that passes verification is then transferred to the next stage's isolated storage module, providing complete and standardized foundational data for subsequent encrypted rights confirmation and ledger generation.
[0035] S102 deploys independent risk control audit programs, dedicated databases, and isolated service clusters, which are physically isolated from business time-series processing programs. It only synchronizes unique identifiers for business data and does not store complete original business data streams. It adopts a single-base plug-in architecture to adapt to offline operational audit scenarios across multiple tracks.
[0036] In one implementation, based on the core requirements of secure isolation of business data, independent audit computing power, compatibility with business audits across the entire industry, and stable operation of the system with low latency, an independent risk control audit program, a dedicated evidence storage database, and a distributed isolation service cluster are built. An underlying deployment architecture in which the business time-series processing program and the audit cluster are completely physically isolated is constructed. Lightweight data interaction constraints are formulated, which only synchronize unique identifiers of business data in one direction, do not store data on the ground, and do not store complete original business data streams. This avoids the dual risks of massive business data crowding out audit storage resources and leakage of original business data. The architecture is built around hardware-level physical separation as its core implementation path. The business time-series processing program and the risk control audit program run in separate physical server clusters. Each cluster is equipped with its own dedicated evidence storage database and independent computing resources. At the network link level, a single data path is constructed through a one-way data gateway. The one-way data gateway is implemented using one-way optical shutter hardware, which is physically separated into a business-side transmitter and an audit-side receiver. The two ends transmit optical signals only through optical fiber in one direction, with no reverse physical link or electrical connection. The business cluster and the audit cluster belong to two completely independent network domains with no other network connection paths. The optical shutter is the only data path between the two clusters. The gateway only allows the business side to output data to the audit side. There is no reverse data transmission channel, which blocks the possibility of audit-side data flowing back to the business side at the physical link level. Data synchronization employs an event-driven model. Synchronization tasks are triggered in real-time when data changes or operations occur on the business side. Synchronization messages are written to a local asynchronous message queue and then pushed to the auditing side via a one-way gateway, with a single message transmission latency not exceeding 50 milliseconds. Both clusters are uniformly connected to a standard NTP time service, performing clock synchronization every 5 minutes, with time deviation controlled within 10 milliseconds to ensure global timing consistency of operational events. A one-way transmission verification algorithm is used in the data interaction process to ensure data transmission compliance at the field level.
[0037] The one-way transmission verification algorithm comprises three core processing units: a whitelist configuration unit, a field-by-field verification unit, and an interception audit unit. These three units are connected in a serial, hierarchical manner. The whitelist configuration unit is the upstream foundational layer, outputting a set of rules for allowed field transmission. The field-by-field verification unit is the intermediate execution layer, reading rules from the top and outputting verification results. The interception audit unit is the downstream archiving layer, receiving interception information and generating permanent audit records. The algorithm execution steps are divided into four steps. First, the whitelist configuration unit predefines globally unique data identifiers as the only fields allowed for transmission, simultaneously configuring two types of verification rules: field length and encoding format, and setting a threshold of 5% for the percentage of abnormal fields in a single data packet. Second, the field-by-field verification unit receives the raw data packets transmitted across the cluster, breaks down each independent field by byte boundaries, matches it against the whitelist rules sequentially, and simultaneously verifies whether the field format and length conform to preset specifications. Third, fields that match the whitelist and have compliant formats are allowed to transmit, while business detail fields that do not match the whitelist are intercepted and discarded. Simultaneously, the percentage of abnormal fields within a single packet is calculated; if the percentage exceeds a preset threshold, the entire packet is intercepted and a system alarm is pushed. The fourth step involves the interception audit unit automatically recording the occurrence time, source business node, and interception field type of each interception event, and simultaneously generating an immutable transmission verification log for archiving and retention.
[0038] For example, when a store completes a POS transaction, the business system generates a complete business data packet containing a unique transaction number, product details, transaction amount, and payment method. The entire data processing and storage process is completed within the business cluster. During cross-cluster synchronization, after the data packet enters the one-way gateway, the one-way transmission verification algorithm breaks down all fields byte-by-byte, matching only the unique transaction number (a whitelist field). Business fields such as product details, amount, and payment method are all intercepted and discarded. If the percentage of abnormal fields in a single packet exceeds 5%, the algorithm directly intercepts the entire packet and triggers an alarm. No complete original business data stream enters the audit cluster throughout the process, thus mitigating the risk of business data leakage at the source and preventing massive amounts of transaction details from consuming audit storage resources, ensuring the audit system operates without storage bloat pressure over the long term.
[0039] Based on the design requirements of dedicated audit independent hosting, rapid reuse across multiple industry scenarios, and non-intrusive horizontal scaling, a layered and decoupled design is implemented. The overall task of irreversible evidence storage across all operations is functionally decomposed, dividing it into three core independent execution modules: underlying physically isolated cluster deployment, lightweight synchronization of unique business identifiers, and single-base plug-in scenario adaptation. The unidirectional data flow logic between modules, upstream and downstream call sequence, and permission isolation rules are clearly defined, forming a standardized architecture execution system where physically isolated evidence storage, simplified identifier synchronization, and multi-track plug-in adaptation operate in a closed loop. The architecture follows the design principles of dedicated audit independent hosting, rapid reuse across multiple industry scenarios, and non-intrusive horizontal scaling. The overall task of irreversible evidence storage across all operations is functionally decomposed into three core independent execution modules: the underlying physically isolated cluster deployment module, the lightweight synchronization module of unique business identifiers, and the single-base plug-in scenario adaptation module. These three modules adopt a unidirectional chain call relationship, with data strictly flowing from the business side to the synchronization module, then to the base adaptation module, and finally merging into the evidence storage unit of the isolated cluster. There are no reverse data calls throughout the entire process. The modules interact through standardized message interfaces. Each module retains only the minimum permissions required for its own function. The entire system is equipped with a module permission minimization verification algorithm to ensure clear boundaries of rights and responsibilities for each module at the calling level.
[0040] The module permission minimization verification algorithm comprises three hierarchical units: a permission matrix configuration unit, a call request parsing unit, and an unauthorized access interception and handling unit. The permission matrix configuration unit is the basic configuration layer, storing the functional permission boundary rules for all modules. The call request parsing unit is the intermediate logic layer, responsible for parsing the target resource and operation type of cross-module calls. The unauthorized access interception and handling unit is the execution output layer, responsible for intercepting unauthorized calls and generating audit logs. These three units are connected hierarchically: the configuration layer outputs permission rules downwards, the parsing layer reads the rules upwards and outputs the judgment results downwards, and the handling layer receives the results and executes the corresponding operations. The algorithm execution steps are divided into four steps. The first step pre-configures functional permission matrices for the three core modules, dividing permission ranges according to three types of resources: computing power scheduling, data transmission, and scenario access, and configuring a daily cross-domain call threshold of ten times for a single module. The second step, upon receiving a cross-module call request, extracts three core pieces of information: the caller module identifier, the target resource type, and the operation type. The third step performs a dimension-by-dimensional matching verification between the parsing results and the permission matrix to determine whether the target resource and operation type fall within the caller's functional boundaries. If the fourth step matches successfully, the call request is allowed; if the match fails, the request is blocked. At the same time, the number of unauthorized calls to this module is accumulated. When the accumulated number reaches the daily threshold, a module-level permission freeze alarm is triggered.
[0041] For example, when adding a new wholesale profit-sharing audit scenario, only the corresponding profit-sharing scenario adaptation plugin needs to be developed in the single-base pluggable scenario adaptation module. After the plugin is connected to the base through a standardized interface, it automatically reuses the existing business identifier synchronization channel and underlying isolated cluster resources. There is no need to adjust the cluster deployment architecture, nor to modify the transmission rules of the synchronization module. If the scenario adaptation module accidentally initiates a call request to adjust the cluster storage quota during development, the module's permission minimization verification algorithm will resolve that the target resource belongs to the computing power scheduling category and exceeds the scenario access functional boundary of the module. It will directly intercept the call request and accumulate the number of unauthorized accesses, preventing non-functional modules from touching the underlying resources, preventing the spread of single-point failures to affect the entire chain operation, and truly achieving non-intrusive horizontal scaling of the architecture.
[0042] Combining the requirements of unalterable audit data, low-latency system verification, rapid deployment across multiple offline operational scenarios, and conflict-free plugin compatibility, this system leverages hardware-level physical isolation mechanisms to define independent operating boundaries for the two sets of computing and storage resources for business and auditing. Based on the principle of minimum data transmission, the business side is limited to outputting only globally unique data identifiers. A unified single-base standardized plugin interface is used to set adaptation specifications for auditing scenarios across eight major tracks of offline operational assets, transactions, and profit sharing. The system comprehensively outputs four types of structured architectural information: detailed cluster hardware isolation deployment rules, business data synchronization granularity constraints, multi-track plugin compatibility judgment rules, and plugin dynamic loading and scheduling standards. The single-base standardized plugin interface uses the HTTP communication protocol, with data format uniformly encoded as JSON. All plugins must adhere to unified request and response field specifications. Required fields for the plugin request interface include the plugin track identifier, operation event type, unique business data identifier, and operation timestamp. Required fields for the plugin response interface include the processing status code, event reception result, and error message description. All field naming rules, data types, and length limits follow a unified standard.
[0043] Plugin invocation follows a request-response pattern. After the platform sends an event retrieval request to a plugin, the plugin must return a response within three seconds. If no response is returned within three seconds, a retry is triggered. Two timeouts result in the plugin being marked as abnormal and an alarm being triggered. The plugin lifecycle includes four states: loading, running, paused, and unloading. The platform manages plugin states according to unified scheduling rules. Newly added plugins undergo loading verification and compatibility testing before being officially integrated and running. Deactivating plugins involves pausing and unloading operations. The unloading process automatically cleans up the corresponding plugin's temporary data without affecting the normal operation of the platform and other plugins.
[0044] The core functions of the plugin are limited to three categories: first, track-specific data format conversion, mapping heterogeneous data from different business systems to the base's standard format; second, standardized mapping of operation events, aligning differentiated operation types from various tracks to the base's unified event classification; and third, compliance rule adaptation, loading the corresponding track's industry compliance verification rules. The plugin does not access core capabilities such as core encryption operations, the underlying evidence storage database, or blockchain nodes, interacting only with the base through standardized interfaces. Dynamic plugin loading employs a class loader isolation mechanism, with each plugin corresponding to an independent class loading instance. Class data between plugins is completely isolated and cannot access each other. Plugins run within a sandboxed permission environment, only granting access to standardized interface calls, prohibiting direct access to the local file system, underlying database, and system network resources. Plugin versions use semantic versioning rules, ensuring backward compatibility with plugins with minor version number differences, while rejecting loading plugins with inconsistent major version numbers to prevent system anomalies caused by version incompatibility.
[0045] The architecture implementation phase incorporates four constraints: immutability of audit data, low latency of system verification, rapid deployment across multiple offline operational scenarios, and conflict-free plugin compatibility. It outputs four types of structured architecture information based on three execution mechanisms. The first type uses hardware-level physical isolation to define the independent operational boundaries of the business and audit resources, corresponding to detailed cluster hardware isolation deployment rules. The second type limits the granularity of business-side output data based on the principle of minimum data transmission, corresponding to business data synchronization granularity constraints. The third type uses a unified single-base standardized plugin interface to set scenario adaptation specifications for each track, corresponding to multi-track plugin compatibility and adaptation judgment rules. The fourth type combines real-time cluster load status to formulate plugin loading and scheduling rules, corresponding to dynamic plugin loading and scheduling standards. This phase is accompanied by plugin compatibility matching algorithms and dynamic load scheduling algorithms, ensuring implementation effectiveness from two dimensions: access admission and operational optimization.
[0046] The plugin compatibility matching algorithm comprises four processing units: an interface field matching unit, a data format matching unit, a call sequence matching unit, and a comprehensive scoring unit. The first three units execute in parallel, calculating the matching scores for each of the three dimensions. The results are then aggregated into the comprehensive scoring unit for weighted fusion. The algorithm execution process consists of five steps. The first step pre-determines the weighting of the three dimensions: interface field dimension (0.5), data format dimension (0.3), and call sequence dimension (0.2), with a plugin access matching threshold set at 85%. The second step, the interface field matching unit, extracts all interface fields of the plugin to be integrated and compares them one by one with the standard interface fields of the platform. It calculates the percentage of overlapping fields, which is the raw score for that dimension. The third step, the data format matching unit, verifies the compatibility of the plugin's output event data structure, field types, and encoding formats with the platform standard, calculating the score for that dimension based on the compliance percentage of field types. The fourth step, the call sequence matching unit, verifies the compatibility of the plugin's request-response sequence, retry mechanism, and timeout settings with the platform's scheduling rules, converting the timeline deviation value to obtain the score for that dimension; a larger deviation results in a lower score. The fifth step, the comprehensive scoring unit, calculates the total matching degree according to preset weights. If the total score reaches the admission threshold, it is judged as compatible and allowed to be connected to the base; if it does not reach the threshold, it outputs a list of incompatible items to guide the optimization and adjustment of the plugin.
[0047] For example, when integrating with a retail audit scenario, an access check is first performed using a plugin compatibility matching algorithm. After the check, the interface field overlap rate is 92%, the data format compliance rate is 96%, and the call timing deviation meets the requirements, corresponding to a score of 80%. The total matching degree calculated by weighting is 90.8 points, which exceeds the access threshold of 85%. Therefore, the plugin is determined to be compatible and can be integrated into the platform, thus avoiding plugin conflicts that could affect system stability at the integration level.
[0048] The dynamic load scheduling algorithm comprises three serial processing units: a real-time load acquisition unit, a quota calculation unit, and a dynamic adjustment execution unit. The acquisition unit continuously collects operational data from each track, feeds it into the quota calculation unit to recalculate the computing power quota, and finally, the execution unit implements the adjustment. The algorithm execution process consists of five steps. The first step presets core operational parameters: operation concurrency weight is 0.6, cluster computing power utilization weight is 0.4, the single-round computing power adjustment step size is 5% of the total quota for a single track, and the evidence verification latency warning threshold is 20 milliseconds. The second step involves the acquisition unit collecting three types of operational data every ten seconds: operation concurrency, average evidence verification latency, and computing power utilization of the corresponding plugins for each track. The third step involves the quota calculation unit calculating the comprehensive load score for each track based on the preset weights; a higher score indicates greater load pressure, while simultaneously verifying whether the current average latency exceeds the warning threshold. The fourth step involves the execution unit sorting the tracks according to the comprehensive scores, increasing the computing power quota for the track with the highest load by the adjustment step size, and revoking the computing power quota for the track with the lowest load, ensuring that the total computing power resources do not exceed the cluster's total limit during the adjustment process. The fifth step is to trigger the priority scheduling mechanism when the verification delay of a certain track exceeds the warning threshold, and directly add a two-step computing power quota to the track to ensure the low latency operation requirements of the system.
[0049] For example, during the lunch rush, the concurrency of checkout operations in the catering sector increased significantly. The real-time load monitoring unit detected that the certificate verification latency in this sector rose to 25 milliseconds, exceeding the 20-millisecond warning threshold. The quota calculation unit combined concurrency and computing power utilization to calculate a comprehensive score, and the catering sector ranked first among all sectors. The dynamic adjustment execution unit triggered priority scheduling, directly adding 10% of the computing power quota to the catering sector, while simultaneously reclaiming the computing power resources corresponding to the building materials sector with the lowest current load, ensuring that the certificate verification latency in the catering sector quickly dropped back to within 20 milliseconds. During operation, business data synchronization in all scenarios strictly adhered to the granularity constraint of unique identifiers, ensuring that the overall architecture always met the design goals of secure isolation, stable operation, and rapid expansion.
[0050] S103 uses a complete RDA-based encryption algorithm to generate solidified evidence records, establishes a dual-path rights confirmation certificate system of blockchain and NFG, constructs a two-way binding mechanism between rights confirmation certificates and audit ledgers, automatically binds full-dimensional operation entity information when data changes, and generates an irreversible electronic change ledger.
[0051] In one implementation, a multi-layered, irreversible evidence-gathering technology foundation is built upon RDA-based encryption of target information, a blockchain-based distributed evidence storage unit, an NFG certificate generation and management unit, and a real-time collection component for all-dimensional operational elements. This foundation unifies the standardized data field encapsulation specifications for electronic audit ledgers and dynamically and adaptively optimizes core parameters such as element collection thresholds, encryption iteration counts, block on-chain intervals, and NFG certificate generation frequency for online account maintenance, POS transactions, multi-level business entity profit sharing, and multi-track offline operations with diverse and differentiated scenarios. The overall technology foundation consists of four core units: the RDA encryption unit, the blockchain distributed evidence storage unit, the NFG certificate management unit, and the all-dimensional operational element collection unit. These four units are connected hierarchically. The collection unit is the upstream data entry point, outputting standardized operational element data downwards. The RDA encryption unit is the intermediate processing layer, receiving the collected data and generating local encrypted digests. The NFG certificate management unit and the blockchain evidence storage unit are the downstream rights confirmation layers, generating ownership certificates and on-chain evidence storage based on the encrypted digests, respectively, forming a dual-path rights confirmation guarantee. The base system standardizes the data field encapsulation specifications for electronic audit ledgers, clarifies the fixed field composition and data format requirements for all evidence records, and provides a scenario parameter adaptive optimization algorithm to dynamically adjust core operating parameters for different business scenarios, balancing evidence security and system operating efficiency.
[0052] The RDA rights confirmation encryption algorithm comprises three serial processing modules: a plaintext preprocessing unit, a multi-round iterative encryption unit, and a digest generation and verification unit. The algorithm execution steps are divided into three steps. In the first step, the plaintext preprocessing unit sequentially concatenates and fills in the three types of raw data: operation subject information, unique identifier of business data, and precise timestamp, processing them into plaintext blocks of fixed length. The filling process uses a random padding mechanism to avoid the leakage of plaintext features.
[0053] The RDA (Real-Time Data Acquisition) encryption algorithm is an irreversible digest generation algorithm designed for audit and evidence preservation scenarios. The plaintext block length is fixed at 256 bits, and the final output is a 256-bit fixed-length encrypted verification digest. The plaintext preprocessing stage's padding rules are as follows: First, the unique identifier of the operation subject, the unique identifier of the business data, and the millisecond-level timestamp are concatenated in a fixed order to form the original plaintext string. If the length after concatenation is less than 256 bits, randomly generated meaningless characters are added to the end of the string. The padding length is equal to the difference between 256 bits and the original string length. The padding characters are randomly selected from a set of uppercase and lowercase letters and numbers. After padding, the entire plaintext block is input into the encryption unit.
[0054] The obfuscation operation in the multi-round iterative encryption stage employs byte-level permutation rules. Each round uses an independent round key to perform a byte-by-byte XOR operation on the plaintext blocks. The permutation table contains a full permutation mapping of 256 bytes. This table is generated based on the global root key using an improved Fisher-Yates shuffling algorithm. Using the SHA256 hash value of the root key as a random seed, bit-by-bit swaps are performed sequentially on the byte sequence from 0 to 255, ultimately generating a fixed 256-byte full permutation mapping table. Each root key corresponds to a unique permutation table, ensuring the reproducibility and consistency of the encryption operation. After each round, all bytes are rearranged according to the permutation table. The diffusion operation uses linear feedback shift logic to diffuse the results of the first half of the blocks to the second half. The diffusion operation uses an 8-stage linear feedback shift register, with the feedback polynomial being... During the operation, the 256-bit plaintext is divided into 128 bits each, and a byte-by-byte shift feedback operation is performed on the lower 128 bits. The result of the operation is XORed with the higher 128 bits byte by byte to complete one diffusion iteration. Each round of encryption performs two diffusion operations to ensure that small changes in the plaintext can be diffused to all block bytes, and to ensure that a single byte change in the plaintext will cause at least half of the digest bits to change.
[0055] The round key is generated by a derivation operation using the global root key, the current iteration number, and the operation timestamp. The derivation operation logic is as follows: the global root key string, the hexadecimal representation of the current iteration number, and the millisecond-level operation timestamp are concatenated in a fixed order; a SHA256 hash operation is performed on the concatenation result; and the first 32 bytes of the hash result are used as the round key for the current round. Different iteration numbers correspond to completely independent round keys, with no direct derivation relationship. The round keys for each iteration are independent of each other, with a default baseline of sixteen iterations, which can be dynamically adjusted according to the security level of the scenario. The verification logic in the verification stage is as follows: the original data to be verified undergoes identical preprocessing and multiple rounds of encryption operations; the newly generated digest is compared byte-by-byte with the existing digest; if they are completely identical, the data is determined to be unaltered; if any byte difference exists, the data is determined to have been tampered with. The complete verification process consists of three steps. The first step is to extract three types of original fields corresponding to the data to be verified: the unique business identifier, the operation subject information, and the timestamp. The second step is to perform plaintext preprocessing according to the same padding rules as during encryption, reuse the same root key and the permutation table to perform the same number of encryption iterations, and generate a verification digest. The third step is to perform a byte-by-byte full comparison between the verification digest and the stored RDA digest. If all bytes match completely, the verification passes. If any bit difference occurs, it is determined that the evidence data has been tampered with, and an anomaly alarm is triggered simultaneously.
[0056] The second step, a multi-round iterative encryption unit, performs obfuscation and diffusion operations according to the configured number of iterations. Each iteration automatically replaces the round key and operation permutation rules. Through multiple rounds of superimposed operations, the encryption result is ensured to be irreversible, making it impossible to recover the original plaintext information from the digest. The third step, a digest generation unit, outputs a fixed-length encrypted verification digest. The verification unit can re-encrypt the original data and compare the generated digest with the existing digest to verify whether the stored data has been tampered with.
[0057] The full lifecycle management of NFG ownership certificates follows this process: For scenarios such as asset profit sharing and core data changes, a second hash operation is performed based on the unique identifier of business data, the RDA encrypted verification digest, and the identity information of the initial operating entity to generate a globally unique NFG ownership certificate. The certificate contains the corresponding equity attributes, profit sharing ratio rules, and effective period of the operating asset, completing the one-to-one binding of the certificate with the initial ownership entity, and the binding record is synchronously written to the audit ledger.
[0058] An independent NFG credential hosting node is established, which is physically isolated from the audit cluster and business cluster, and synchronizes credential status identifiers only through a standardized interface. The hosting node centrally manages all NFG credentials, recording the entire status of credential activation, freezing, pledging, and cancellation in real time. All status changes must be executed after authorization verification, and the entire hosting operation is traced and a record is generated.
[0059] When ownership changes occur, such as asset transfer or permission assignment, the system first verifies the operating permissions of the initiating party, the validity of the target certificate, and its current status. Once the verification is successful, the ownership subject of the NFG certificate is updated. The entire ownership information before and after the change, the operating entity, the change time, and the reason for the change are fully retained. An ownership change record is generated and synchronously uploaded to the blockchain to ensure that the entire ownership transfer is traceable.
[0060] When a profit-sharing settlement event is triggered, the system automatically reads the current ownership and built-in profit-sharing rules of the corresponding NFG voucher, and automatically calculates the profit distribution details for each ownership entity based on the current period's revenue data. The profit-sharing result is bound to the unique identifier of the corresponding NFG voucher and the profit-sharing operation event ID for evidence storage, and is simultaneously written to the audit ledger and blockchain nodes to support full-chain traceability of profit-sharing and dispute verification.
[0061] The scenario parameter adaptive tuning algorithm comprises three serial processing units: a scenario load monitoring unit, a parameter adaptation calculation unit, and a parameter dynamic distribution unit. The algorithm execution steps are divided into four steps. The first step presets baseline operating parameters for the entire scenario. The baseline element collection threshold for the online account maintenance scenario is full data collection, the baseline value for the number of encryption iterations is sixteen, and the baseline value for the block upload interval is two seconds. Different tracks can set independent baseline values according to business attributes. The second step, the load monitoring unit, collects three types of operating data for each scenario at a fixed period: operation concurrency, single data volume, and average latency of evidence verification. The collection period is set to thirty seconds. The third step, the parameter adaptation calculation unit, calculates adjustment coefficients based on the current load data. When operation concurrency increases, the number of encryption iterations is proportionally reduced, and the block upload interval and NFG certificate batch generation interval are increased. When concurrency decreases, the parameters are adjusted in the opposite direction. Simultaneously, it checks whether the evidence verification latency exceeds the warning threshold. If the latency exceeds the threshold, the number of encryption iterations is preferentially reduced to ensure performance. The fourth step, the parameter dynamic distribution unit, synchronizes the adjusted parameters to the encryption and evidence storage module of the corresponding scenario. The parameter adjustment process does not interrupt the normal evidence storage business, achieving seamless dynamic optimization.
[0062] For example, during a retail promotion, the concurrency of POS transactions at stores tripled compared to normal days. The scenario load monitoring unit detected a significant increase in concurrency and a verification latency of 28 milliseconds, exceeding the 20-millisecond warning threshold. Based on this, the parameter adaptation calculation unit calculated adjustment coefficients, reducing the number of encryption iterations from 16 to 12 and the block upload interval from two seconds to three seconds. After the parameters were issued, the verification latency quickly returned to within the warning threshold. After the promotion ended and concurrency decreased, the algorithm automatically adjusted the parameters back to the baseline value, ensuring the encryption strength during off-peak periods.
[0063] This entire evidence preservation process connects multiple stages, including triggering business data change events, automatic capture of full-dimensional operational entity information, generation and solidification of RDA encrypted digests, generation and binding of NFG certificate hashes with ownership initialization, on-chaining of blockchain-based ownership certificates, bidirectional anchoring of on-chain hashes and NFG certificates, bidirectional binding of certificates with local audit ledgers, ownership custody registration, ownership change verification, revenue distribution accounting, persistent archiving of irreversible ledgers, and traceability indexing, constructing a complete closed-loop business chain. It establishes a multi-dimensional association mapping operation logic for multiple data types, including change event ID, operator identifier, access IP, terminal device code, precise timestamp, RDA encrypted verification digest, blockchain block hash, unique NFG certificate identifier, and business entity tag. This complete evidence preservation process connects multiple stages, including triggering business data change events, automatic capture of full-dimensional operational entity information, generation and solidification of RDA encrypted digests, generation and ownership management of NFG certificates, on-chaining of blockchain-based ownership certificates, bidirectional anchoring binding, persistent archiving of irreversible ledgers, and traceability indexing, forming a fully closed-loop business chain.
[0064] The blockchain module is deployed using a consortium blockchain architecture, with nodes distributed across three independent physical servers in the audit cluster. The consensus mechanism employs a practical Byzantine fault-tolerant algorithm, ensuring a single block consensus confirmation time of no more than two seconds. On-chain data packets contain only two core fields: a unique identifier for the change event and an RDA encrypted verification digest. They do not carry any original business data, guaranteeing that business data will not be leaked through on-chain nodes. Each on-chain transaction contains five fixed fields: a unique transaction sequence number, a change event ID, an RDA encrypted verification digest, a local ledger storage location number, and a timestamp. All fields are serialized and packaged into a transaction body. The consensus mechanism is configured as a three-node PBFT with one fault-tolerant node. The block generation interval is fixed at 2 seconds, and a single transaction must be confirmed by at least two nodes before being written into a block. Blocks adopt a chain structure, with each block header containing the hash value of the previous block, ensuring that on-chain data is immutable.
[0065] The specific logic of the two-way anchoring and binding between the blockchain and NFG credentials is as follows: When the local audit ledger is generated, the blockchain hash field and the NFG credential identifier field are automatically reserved; after the RDA summary is generated, the NFG credential generation and blockchain on-chain process are triggered simultaneously; after the NFG credential is generated and the initial ownership binding is completed, the unique identifier of the credential is written back to the local ledger; after the block is generated and consensus confirmation is completed, the system automatically extracts the block hash value corresponding to the transaction and writes it back to the blockchain hash field of the corresponding local ledger record, completing the positive binding from the local ledger to the on-chain credential and the NFG credential. At the same time, the transaction memo field of the on-chain block is written with the storage location number of the corresponding local ledger and the unique identifier of the NFG credential, and the corresponding block hash and local ledger index are recorded in the NFG credential metadata, realizing the two-way association between the three. After the two-way binding is completed, the full-chain evidence storage information can be retrieved through any unique identifier.
[0066] The specific logic of the two-way anchoring binding is as follows: When the local audit ledger is generated, a blockchain hash field is automatically reserved; after the RDA digest is generated, the system packages the event identifier and encrypted digest into an on-chain transaction and submits it to the blockchain node for consensus execution and on-chain processing; after the block is generated and consensus confirmation is completed, the system automatically extracts the block hash value corresponding to the transaction and writes it back to the blockchain hash field of the corresponding local ledger record, completing the forward binding from the local ledger to the on-chain certificate. At the same time, the transaction memo field of the on-chain block is written with the storage location number of the corresponding local ledger, realizing the reverse association from the on-chain certificate to the local ledger. After the two-way binding is completed, both the local ledger and the on-chain certificate can be retrieved from each other through a unique identifier.
[0067] The complete evidence verification process consists of three steps. The first step is to input the unique identifier of the event to be verified, retrieve the corresponding record in the local ledger, and extract the stored RDA digest and block hash. The second step is to query the corresponding block on the blockchain node based on the block hash and extract the RDA digest and event ID stored on the chain. The third step is to compare the local digest and the on-chain digest byte by byte, and at the same time verify the consistency of the event ID. If the two match completely, the evidence is deemed valid. If either one does not match, the evidence is deemed to have been tampered with, and an abnormal verification log is automatically generated.
[0068] In the event of network anomalies, consensus failures, or other abnormal situations, the system automatically triggers a retry mechanism with a 10-second retry interval. If three retries fail, the record is marked as pending on-chain status and added to the processing queue. Once the network recovers, the record is automatically replenished in batches. After replenishment, two-way binding is performed to ensure that all evidence records ultimately achieve dual confirmation of rights. Batch replenishment is executed according to the time order of the pending on-chain queue, with a single batch not exceeding 100 records. Replenishment has a lower priority than real-time on-chain tasks and does not consume normal business evidence storage computing power. If on-chain failures occur for 6 consecutive hours, the system automatically triggers a degradation mode, temporarily retaining only the local RDA encrypted ledger and suspending blockchain writing. Once the link recovers, records are replenished in batches according to time order. During the degradation period, all evidence records are automatically marked as pending verification status, and after replenishment, they are updated to valid evidence.
[0069] The link is equipped with an eight-dimensional data association mapping operation logic. Using the unique identifier of the change event as the primary key, it constructs an association mapping network of eight types of core data generated at each stage of the entire link, enabling the retrieval of the entire link's evidence storage information with a single key.
[0070] The eight-dimensional data association mapping operation logic comprises three functional modules: a primary key generation unit, a data mounting unit, and an association verification unit. These three modules are arranged serially along the link. The operation execution steps are divided into four steps. First, when a business data change event is triggered, the primary key generation unit generates a globally unique long integer change event identifier according to the distributed snowflake rule. This identifier serves as the unified primary key for the entire evidence storage link, ensuring that the identifier generated across nodes is globally unique. Second, as the link progresses through each stage, the data mounting unit automatically mounts the data generated in the current stage and its corresponding attribute tags to the event primary key, sequentially mounting seven types of data: operator identifier, access IP, terminal device code, precise timestamp, RDA encrypted verification digest, blockchain block hash, NFG certificate unique identifier, and third-level business entity tag. Third, the association verification unit performs integrity verification at each mounting node, checking whether the binding relationship between the current data and the event primary key is unique, avoiding data mismatch or duplicate mounting. Data that fails verification triggers a retransmission mechanism. Fourth, the traceability index construction stage builds a multi-dimensional joint index based on all mounted data, supporting subsequent rapid retrieval by account, time, entity, and other dimensions.
[0071] For example, when a manufacturer initiates an asset sharing operation with a corresponding wholesaler, the system detects the change in business data and first generates a globally unique change event identifier as the primary key. Then, it automatically captures the account information, login IP address, terminal device code, and precise operation time, attaching these four types of data to the event primary key. Next, the RDA encryption unit performs encryption operations on the operation elements and business data identifier, generating a fixed-length RDA encrypted verification digest, which is simultaneously attached to the same primary key. Then, the NFG certificate generation and blockchain on-chain process is triggered synchronously: a unique NFG certificate is generated based on the encrypted digest and entity information, and ownership binding is completed. The blockchain node packages the encrypted digest and event identifier onto the chain, generates the corresponding block hash value after distributed consensus confirmation, and uploads it to the event primary key. The system automatically matches the multi-level operating entity tags corresponding to this sharing, completing the final data attachment. Ultimately, all data forms a complete association mapping relationship through the same event identifier, and the entire evidence record can be retrieved through any dimension field.
[0072] Using unique operator identification, device network traceability constraints, RDA encryption verification rules, the bidirectional mapping relationship between blockchain and NFG, multi-track scenario adaptation parameters, and hierarchical traceability dimensions as multi-dimensional weighted fusion calculation factors, the algorithm normalizes and integrates the raw data from the entire process of collection, encryption, on-chain, certificate generation, ownership transfer, and revenue calculation. This generates standardized, irreversible electronic change ledger structured evidence information with dual tamper-proof guarantees of local encryption and solidification, on-chain evidence storage, and NFG ownership certificates. The algorithm uses six weighted fusion calculation factors—unique operator identification, device network traceability constraints, RDA encryption verification rules, the bidirectional mapping relationship between blockchain and NFG, multi-track scenario adaptation parameters, and hierarchical traceability dimensions—to output structured evidence information through weight allocation and standardized calculation.
[0073] The multi-dimensional weighted fusion algorithm comprises three sequential processing modules: a factor standardization unit, a weight calculation unit, and a normalization and integration unit. The algorithm execution steps are divided into four steps. The first step pre-determines the weight percentages of six factors: the unique identifier of the operation identity has a weight of 0.15, the device network traceability constraint has a weight of 0.15, the RDA encryption verification rule has a weight of 0.25, the blockchain and NFG bidirectional mapping relationship has a weight of 0.25, the multi-track scenario adaptation parameter has a weight of 0.1, and the hierarchical traceability dimension has a weight of 0.1. The second step, the factor standardization unit, performs numerical standardization on each factor, mapping factors of different dimensions to a score range of zero to one. For example, a successful RDA encryption verification corresponds to a score of one, while an abnormal verification corresponds to a score of zero; a match between the blockchain hash and the local ledger corresponds to a score of one, while a failed match corresponds to a score of zero; a valid NFG credential status and matching ownership correspond to a score of one, while an abnormal credential corresponds to a score of zero. The third step, the weight calculation unit, multiplies the standardized scores of each factor by their corresponding weights and sums them to obtain the comprehensive evidence strength score for the evidence record. The fourth step, the unification and integration unit, takes event identifiers as its core and integrates all the original data, verification results, and intensity scores corresponding to all factors into a standardized electronic change ledger structure, which is then synchronously written into the persistent storage unit and the traceability index library.
[0074] For example, a product pricing modification initiated by a store operator enters the fusion calculation stage after completing full-chain evidence storage. A complete and valid operator identity corresponds to a standardized score of 1; complete device network information corresponds to a score of 1; successful RDA encryption verification corresponds to a score of 1; bidirectional matching between the blockchain hash and NFG certificate corresponds to a score of 1; compliant scenario adaptation parameters correspond to a score of 1; and complete hierarchical traceability labels correspond to a score of 1. After weighted calculation according to preset weights, the comprehensive evidence storage strength score for this record is the maximum score of 1. The final generated electronic change ledger contains complete traceability data, RDA encrypted digests, on-chain hash certificates, and NFG ownership certificates, while also marking the comprehensive evidence storage strength. Once the ledger is generated, it cannot be modified. Local data relies on RDA encrypted digest verification for tamper-proofing, on-chain certificates rely on a distributed consensus mechanism for tamper-proofing, and NFG certificates rely on the uniqueness of the custodian node and hash for tamper-proofing. Multiple safeguards work together to achieve the irreversible nature of the evidence storage record.
[0075] S104 provides one-click access to the entire lifecycle modification records of single and batch data, fully restoring the entire data change process, retaining complete traceability fields, and accurately identifying the responsible parties for data tampering and abnormal changes.
[0076] In one implementation, a full-link data traceability foundation is built based on a full-domain operation evidence storage index, a hierarchical traceability field storage unit, and a batch retrieval computing component. This foundation standardizes and unifies the encapsulation format of traceability query messages, dynamically adapting to single fine-grained queries, batch multi-dimensional searches, multi-level hierarchical traceability of business entities, concurrent query thresholds for multi-track business scenarios, and adjustable index pagination parameters. The full-link data traceability foundation consists of three core components: the full-domain operation evidence storage index, the hierarchical traceability field storage unit, and the batch retrieval computing component, which are connected hierarchically. The full-domain operation evidence storage index serves as the top-level entry unit, storing a multi-dimensional joint index of all evidence records and responsible for quickly matching query targets. The hierarchical traceability field storage unit forms the intermediate data layer, storing complete traceability field data categorized by business entity level to support hierarchical traceability requirements. The batch retrieval computing component is the bottom-level computing unit, handling parallel computing tasks in batch query scenarios and ensuring the execution efficiency of large-scale searches. The base system standardizes the encapsulation format of traceability query messages, clarifies the unified specifications of query conditions, pagination parameters, and return fields, and provides two core algorithms to achieve query verification and dynamic parameter optimization.
[0077] The first type is a standardized query message verification algorithm, which includes three sequential processing modules: a message parsing unit, a field compliance verification unit, and a format normalization processing unit. The algorithm execution steps are divided into three steps: First, a whitelist of query rules is preset, supporting five query dimensions: unique data identifier, operation account, time range, business entity, and anomaly tag. A maximum length threshold for a single query condition field is set, and the maximum data volume for a single batch query is configured to be one thousand records. Second, after receiving the query request, the message parsing unit disassembles all query conditions and control parameters in the request, matches them against the whitelist rules one by one, and verifies the field length and data format. Third, compliant fields are normalized, unifying the encoding format and data type. Query conditions exceeding the rules return parameter anomaly prompts, blocking non-compliant requests and preventing illegal queries from affecting system stability.
[0078] The second category is a dynamic concurrency scheduling algorithm for query scenarios, comprising three serial processing modules: a real-time concurrency monitoring unit, a threshold adaptation calculation unit, and a resource dynamic allocation unit. The algorithm execution steps are divided into four steps. The first step presets baseline concurrency parameters: the baseline concurrency threshold for a single fine-grained query is 200 times per second, and the baseline concurrency threshold for batch multi-dimensional retrieval is 20 times per second. The three-tiered business entity tiered traceability and eight major track scenario queries share a common concurrency pool, and index pagination defaults to returning 100 records per page. The second step, the monitoring unit, collects real-time concurrency and average response latency for each query scenario at second-level intervals, simultaneously calculating the access popularity of the current index pagination. The third step, the threshold adaptation calculation unit, adjusts the concurrency threshold and pagination parameters based on the current load data. When response latency increases, it lowers the batch query concurrency threshold and increases the number of records returned per page; when response latency is sufficient, it adjusts the parameters in the opposite direction, while simultaneously enabling a caching mechanism for high-frequency pagination data. The fourth step, the resource allocation unit, dynamically allocates computing resources to the computing components based on the adjustment results, prioritizing the response efficiency of single, precise queries.
[0079] For example, when auditors initiate a change tracking query for a single product pricing data item, the query message standardization verification algorithm first verifies the product unique identifier field in the query conditions, confirming that the format is compliant and within the whitelist range. After normalization, the data is then passed to the index. The dynamic concurrency scheduling algorithm detects that the current single query concurrency is low, allocates sufficient computing power to support fast retrieval, and finally returns the corresponding query results in milliseconds. When the audit team initiates a batch query for the monthly profit sharing records of all stores in the entire region, the algorithm verifies that the amount of data queried in a single session does not exceed the threshold limit of one thousand records. At the same time, it detects that the current batch query concurrency is high, automatically lowers the batch query concurrency threshold and adjusts the pagination parameters, and executes the retrieval in batches through parallel computing components to ensure the overall system response stability.
[0080] The system establishes a complete query and handling chain, encompassing change record index matching, full lifecycle time-series link assembly, complete source tracing field retrieval, entity association information matching, responsible entity locking, and source tracing result archiving. It also establishes a multi-dimensional association mapping operation logic for change events, time-series operation nodes, account and device source tracing fields, operational hierarchical entities, and abnormal risk tags. The index uses a composite index structure with a unique business data identifier as the primary key. Index items include four dimensions: event ID, operation timestamp, account identifier, and entity hierarchical tag. All change records are aggregated and stored according to the business primary key. During link assembly, the business primary key is used as the search condition to retrieve all change events corresponding to that data, automatically forming a complete chain from creation to the latest modification, sorted in ascending order of timestamp, without manual concatenation. In batch query scenarios, multi-threaded parallel retrieval is used, with execution sharded according to the business primary key. The query response time for a single batch of 1000 records does not exceed 3 seconds.
[0081] The entire query and handling process connects six complete steps: change record index matching, full lifecycle time-series link splicing, complete traceability field retrieval, entity association information matching, responsible entity identification, and traceability result archiving, forming an end-to-end closed-loop query link. This link is complemented by multi-dimensional association mapping operation logic. Using the core identifier of the query target as an anchor point, it constructs an association mapping network for five types of data: change events, time-series operation nodes, account and device traceability fields, business layer entities, and abnormal risk tags, enabling a single query to output complete information across the entire chain.
[0082] The multidimensional association mapping operation logic comprises four sequential functional modules: a primary key index matching unit, a time-series link assembly unit, a multidimensional data association unit, and an anomaly risk label generation unit. The operation execution steps are divided into four steps. First, after receiving the query conditions, the primary key index matching unit retrieves all matching change events from the evidence storage index and extracts the unique identifier set for each event. Second, the time-series link assembly unit sorts all matching change events in ascending order according to their precise operation timestamps, sequentially connecting them to form a full lifecycle time-series link of the target data from creation to the latest modification. The assembly process automatically verifies the continuity of timestamps, automatically marking missing nodes and filling in placeholders.
[0083] The third step, the multi-dimensional data association unit, uses each event identifier as a key to retrieve the corresponding traceability fields such as the operation account, login IP, and terminal device code. Simultaneously, it matches the corresponding three-tiered business entity information, associating all fields with the corresponding time-series nodes. Layered traceability follows a downward compatibility principle: manufacturer-level accounts can view the full operation traceability data of themselves and all their subordinate wholesalers and stores; wholesaler-level accounts can only view the traceability data of themselves and their subordinate stores, and cannot view data across wholesalers; store-level accounts can only view the operation records within their own store. During queries, the system automatically verifies the current account's entity-level permissions and automatically filters and blocks data exceeding the permission scope, ensuring compliance with the data boundaries of layered traceability. The fourth step, the abnormal risk label generation unit, performs risk assessment on operational behaviors in the chain based on preset abnormal judgment rules. Operations that meet abnormal characteristics are automatically labeled with abnormal risk tags. The judgment rules include four core judgment dimensions: operations outside of working hours, frequent modifications within a short period, login operations from uncommon devices, and unauthorized operations.
[0084] The specific quantitative rules for the four categories of anomaly judgment dimensions are as follows: Category 1: Operations outside of working hours. The default standard working hours are 8:00 AM to 8:00 PM daily. Operations falling within the range of 8:00 PM on the current day to 8:00 AM the following day are judged as abnormal operations outside of working hours. Category 2: Frequent modifications within a short period. The time window is set to ten minutes. If the same account initiates three or more modification operations on the same type of business data, it is judged as frequent modification abnormal behavior. Category 3: Login from infrequently used devices. Using a continuous 30-day statistical period, if a device initiates operations on the corresponding account less than three times within the period, it is judged as login operation from an infrequently used device. Category 4: Unauthorized operations. Based on a comparison with the account's default permission range, operations exceeding the account's current permission boundaries are judged as unauthorized operations. Unauthorized operations are divided into three levels: exceeding the scope of the current level is Level 1 unauthorized operation, cross-level operations are Level 2 unauthorized operation, and touching core system configuration permissions is Level 3 unauthorized operation. The system classifies risk levels based on the number of anomaly dimensions hit and the level of unauthorized access. Hitting a single low-risk dimension is considered low risk, hitting two or more dimensions or having a first-level unauthorized access is considered medium risk, and having a second-level or higher unauthorized access is considered high risk. Correspondingly, different levels of anomaly risk labels are generated.
[0085] Differentiated handling procedures correspond to different risk levels. Low-risk events are automatically archived in the anomaly log and pushed to the corresponding administrator for viewing. Medium-risk events trigger real-time alerts and are pushed to the auditors' to-do list, requiring review within 72 hours. High-risk events immediately trigger temporary account freezing and are simultaneously pushed to the headquarters audit team to initiate a special investigation. Responsibility determination follows the principle of attribution to the operating entity. For events handled independently by a single account, the account holder bears direct responsibility. For events handled collaboratively by multiple individuals, primary and secondary responsibilities are determined based on operating permissions and contributions, with the account executing the core modification operation being the primary responsible party.
[0086] For example, when auditors investigate abnormal price changes at a store, they first initiate a query by entering the product's unique identifier. The primary key index matching unit retrieves all five change records corresponding to the product from the index and extracts all event identifiers. The time-series link assembly unit sorts the five records in ascending order by operation timestamp, assembling a complete lifecycle link from the product's initial creation, two price adjustments, one promotional modification, to the final restoration of the original price, verifying that the timestamps are continuous and without missing information. The multi-dimensional data association unit sequentially retrieves the account information, IP address, device fingerprint, and operation time corresponding to each operation, while matching the store's upstream wholesaler and manufacturer information, associating all traceability fields to the corresponding operation nodes. The abnormal risk label generation unit detects that one of the price adjustments occurred during non-working hours in the early morning, and that the corresponding device was initiating this type of operation for the first time, automatically labeling the record with an abnormal risk tag. Finally, the system automatically matches the operator account and identity information corresponding to the abnormal operation, accurately identifying the responsible party.
[0087] The algorithm uses unique data identifiers, time-series operation timestamps, account / device identity fields, multi-level operational hierarchies, anomaly risk judgment rules, and batch retrieval concurrency constraints as weighting factors to perform multi-dimensional fusion calculations. This process normalizes and integrates the original traceability data from single and batch retrievals across the entire process, generating standardized structured query information for the entire data change traceability process. The original traceability data obtained from the entire retrieval process is normalized and integrated using a multi-dimensional weighted fusion algorithm, ultimately generating standardized structured query information for the entire data change traceability process. The algorithm uses six categories of indicators—unique data identifiers, time-series operation timestamps, account / device identity fields, multi-level operational hierarchies, anomaly risk judgment rules, and batch retrieval concurrency constraints—as weighting fusion calculation factors, and outputs well-structured results through weight allocation and standardized calculations.
[0088] The multi-dimensional weighted fusion algorithm comprises four sequential processing modules: factor standardization unit, weight configuration unit, fusion calculation unit, and result normalization and integration unit. The algorithm execution steps are divided into four steps. The first step pre-determines the weight percentages of six factors: unique data identifier (0.1), time-series operation timestamp (0.15), account / device identity field (0.2), multi-level business stratification dimension (0.2), anomaly risk judgment rule (0.25), and batch retrieval concurrency constraint (0.1). The second step, the factor standardization unit, performs numerical standardization on each factor, mapping factors of different dimensions to a score range of zero to one. For example, a perfect match of the unique data identifier corresponds to a score of one, while a fuzzy match corresponds to a score calculated based on the degree of matching. Higher anomaly risk levels correspond to higher scores, while no anomalies correspond to a score of zero. The third step, the fusion calculation unit, multiplies the standardized scores of each factor by their corresponding weights and sums them to obtain the comprehensive matching score for each query result. A higher matching degree indicates a higher degree of fit between the result and the query requirements. The fourth step, result unification and integration unit, sorts the results from high to low matching degree, and integrates the time sequence link, traceability field, main information, and exception label of each record into a standardized structured format, which is then synchronously written to the query result cache and operation archive log.
[0089] For example, when auditors perform a batch query for abnormal change records of all stores under a wholesaler in the current month, the system retrieves more than two hundred matching records, which then enter the fusion calculation stage. Each record is calculated with standardized scores for six factors, with the abnormal risk assessment factor having the highest weight; records with higher abnormal levels correspond to higher scores. After weighted calculation according to preset weights, all records receive a comprehensive matching score. The algorithm sorts the records by score from highest to lowest, prioritizing the display of high-risk change records. The final output structured query information includes a complete lifecycle change chain for each record, a full set of account and equipment traceability fields, information on the third-level operating entity, and information on the abnormal risk level and corresponding responsible party. The query results are displayed in a standardized format, supporting one-click export and archiving. The system also automatically retains the operator, query time, and query conditions for this query, forming a complete audit query archive.
[0090] S105, the source audit logs are logically partitioned and isolated according to the operating entities of end A, end B, and end C; business terminals only have the permission to append and write logs, and business terminals have no right to modify or delete the generated historical logs; the system operation and maintenance terminal can perform log archiving and migration, but cannot tamper with the original log content; the log writing synchronously generates the corresponding RDA hash digest and NFG trusted digital rights confirmation certificate, and the credential constraint prohibits the tampering with the generated historical logs.
[0091] In one implementation, the system logically partitions and stores all traceability audit logs according to three levels of business entities: A, B, and C. The three partitions correspond to the permission boundaries of different levels of business entities. Independent logical storage spaces are built within the storage resources of the same audit cluster, achieving clear division of data ownership without the need for additional physical storage devices.
[0092] The partitioned storage is based on a hierarchical tree architecture of operating entities. Each operating entity corresponds to a globally unique entity identifier, and the entity attributes include a field for the identifier of the superior entity. Lower-level operating entities are linked upwards to their superior entities, while top-level entities have no superior identifier. During the log writing phase, each operation event is automatically attached with a complete three-level entity tag. The system assigns the log to the corresponding logical partition of the lower-level entity based on the tag, and simultaneously creates a retrieval index in the higher-level entity partition. The original log data is stored only within the lower-level partition; the higher-level partition only retains the index mapping relationship. Partition access follows a downward compatibility principle: higher-level entities can view their own log data and the log data of all lower-level entities. Entities at the same level cannot access each other's data across partitions, and lower-level entities have no right to view the log content of higher-level or other peer-level entities.
[0093] Taking an offline chain store scenario as an example, A-end corresponds to the brand manufacturer's headquarters, B-end corresponds to regional wholesalers, and C-end corresponds to offline physical stores. When a store generates a POS transaction log, the system automatically stores the original log in the corresponding C-end store's dedicated logical partition based on the subject tag carried by the event. At the same time, it creates retrieval indexes in the B-end partition of its affiliated wholesaler and the A-end partition of its affiliated manufacturer. When the store account logs into the audit system, it can only view the log data within its own partition; when the corresponding wholesaler account logs in, it can view the logs of itself and all its subordinate stores; the manufacturer's headquarters account can access the log data of all partitions. Strict logical access boundaries are set between different partitions. All cross-partition access requests are subject to subject permission verification, and requests exceeding the permission scope are directly blocked, ensuring that the data boundaries of each operating entity are clear and controllable.
[0094] All business terminals only have append-only write permissions for logs and are not authorized to modify or delete existing historical logs, thus building the first line of defense against tampering at the terminal access level. Access control is implemented through a two-layer mechanism of interface permission isolation and write mode constraints. At the interface level, the log interface open to business terminals only supports new write requests and does not provide any entry point for modifying or deleting historical logs. The collection probes deployed on business terminals can only report new operation events and cannot initiate commands to modify historical records. At the write mode level, log storage adopts an append-only write-once-read-many mode. All logs are appended to the end of the corresponding partition's log file in chronological order. The system's underlying layer sets a read-only protection attribute for already written log data blocks, and does not support overwriting existing data blocks. Each time a write request arrives, the system first verifies the requester's terminal type and permission identifier. Requests identified as belonging to business terminals are only allowed to append write operations; other operation types are directly intercepted and an interception audit record is generated.
[0095] Taking a store POS terminal as an example, the lightweight data acquisition probe deployed on the terminal is only responsible for capturing operation events and reporting log data. When a cashier completes a transaction settlement, the probe captures the operation event and generates a standardized log, which is then submitted to the log writing interface of the audit cluster. After verifying the terminal's identity and operation type, the interface appends the log to the end of the corresponding C-end partition's log file in chronological order. Throughout this process, the terminal program cannot read the written historical log content, nor can it initiate requests to modify or delete historical records. Even if the terminal program is maliciously tampered with, it can only attempt to report false new logs, and cannot modify the historical records already written to disk. Furthermore, false logs will be identified and removed by subsequent verification processes due to the lack of a matching unique identifier for business data.
[0096] The system operations and maintenance (O&M) side has the authority to archive and migrate logs, but not the authority to modify the original log content. This ensures the maintainability of storage resources while preventing log tampering risks at the O&M level. O&M permissions are finely segmented, granting only permissions related to log lifecycle management, covering four types of operations: log archiving, migration, backup, and deletion of expired archive files. Editing permissions for the original log content are not granted to any O&M role. Archive and migration operations perform batch processing on complete log files, without modifying individual log entries within the file. The entire operation is automatically logged, generating corresponding O&M operation records stored in a dedicated O&M audit log. To prevent log content tampering during the archive and migration process, the system calculates an RDA encrypted checksum for the target log file before and after the operation. The operation is confirmed only if the two checksum values match perfectly; if a discrepancy is found, the operation is automatically terminated and an alarm is triggered.
[0097] Taking quarterly log archiving as an example, when operations and maintenance personnel need to migrate historical logs stored for more than 6 months from the online storage cluster to the low-cost archive storage cluster, they can only initiate batch archiving commands through the operations and maintenance management terminal, selecting log files within the corresponding time range for migration. After receiving the command, the system first calculates and records the RDA checksums for the target log files in batches, and then performs the file migration operation. After the migration is completed, the system recalculates the RDA summaries of the corresponding files in the archive storage, comparing them one by one with the summaries before the migration. If all summaries match, the archiving is marked as successful; files with discrepancies are automatically rolled back and an alarm is pushed. Throughout the entire process, operations and maintenance personnel cannot open or edit the internal content of any log file. All operations are recorded in the dedicated operations and maintenance audit log, and the operations and maintenance audit log itself also follows the rule of immutability, ensuring that the entire operations and maintenance process is traceable.
[0098] Simultaneously with each log entry written, the system generates a corresponding RDA hash digest and an NFG trusted digital ownership certificate. This dual ownership mechanism ensures the immutability of historical logs, providing verifiable technical evidence for log authenticity. After log writing, the system automatically extracts three core fields: the unique identifier of the operation subject, the unique identifier of the business data, and the millisecond-level precise timestamp. These are then fed into the RDA ownership encryption algorithm module to generate a hash digest. The algorithm execution consists of three stages. The first stage is plaintext preprocessing, where the three fields are concatenated in a fixed order, with randomly generated meaningless characters appended to the end, resulting in a fixed-length 256-bit plaintext block to prevent plaintext feature leakage. The second stage is multi-round iterative encryption, with a default of 16 rounds of iteration. Each round uses an independently derived round key to perform byte-level XOR obfuscation, combined with a permutation table to rearrange bytes, and then performs diffusion operations through an 8-level linear feedback shift register to ensure that a single byte change in the plaintext will cause more than half of the digest bits to change. The third stage is digest output, ultimately generating a fixed-length 256-bit irreversible encrypted verification digest, which is bound to the unique identifier of the corresponding log entry for storage.
[0099] The synchronously generated NFG trusted digital ownership certificate is obtained by performing a double hash operation based on the unique identifier of the business data, the RDA encrypted verification digest, and the identity information of the operating entity, ensuring global uniqueness. The certificate contains the ownership attributes of the corresponding operating entity. After binding the ownership with the initial operating entity, it is stored in an independent NFG certificate hosting node. The hosting node maintains physical isolation from the audit cluster and business cluster, synchronizing certificate status identifiers only through a standardized interface. Changes to the certificate's validity, freezing, and circulation status all require authorization verification and cannot be arbitrarily modified locally. A two-way binding relationship is established between log records, RDA hash digests, and NFG ownership certificates; information from the other two can be retrieved through any one of them.
[0100] Taking a store product pricing modification log as an example, when this operation log is written to the C-end partition, the system automatically extracts the corresponding fields, initiates RDA encryption, generates a unique 256-bit hash digest, and stores it in the log's appendix field. Simultaneously, based on this digest, business data identifier, and the operating store information, the system generates a unique NFG ownership certificate. This certificate records that the ownership of the log belongs to the corresponding store. The certificate is stored in an independently hosted node, and the unique identifier of the certificate is written back to the local log record, completing a two-way binding. When auditors subsequently verify the authenticity of this log, they only need to re-extract the core fields of the log and perform the exact same RDA encryption operation. The newly generated digest is compared byte-by-byte with the original digest attached to the log, and the corresponding NFG certificate is retrieved to verify its validity. If both match, it proves the log has not been tampered with; if any one does not match, it indicates the log has been modified, automatically triggering an anomaly alarm. This entire dual ownership mechanism ensures the log's immutability from two dimensions: local encryption and independent storage. Even if the local storage is compromised, authenticity verification can still be completed through the hosted certificate.
[0101] The above four steps are progressive, forming a complete protection system from storage architecture, terminal permissions, operation and maintenance boundaries to technical rights confirmation, ensuring that the traceability audit logs are true, complete and tamper-proof throughout the process, providing reliable data support for subsequent data change traceability and identification of responsible parties.
[0102] S106 solidifies all backend operation and maintenance logs, standardizes multi-level account operation and maintenance processes, builds a complete online data compliance audit architecture, and dedicates the audit cluster to the evidence storage work, separating business data storage tasks.
[0103] In one implementation, a comprehensive compliance audit technology foundation is built upon a log persistence engine, a multi-level account process standardization verification component, a physically isolated audit computing power scheduling unit, and a multi-track compliance rule plugin adapter. This foundation unifies the standard encapsulation message format for audit logs, account processes, and cluster resource interactions across the entire domain. It dynamically and adaptively optimizes adjustable core parameters for online account maintenance, POS transactions, multi-level business entity asset profit sharing, and differentiated offline business scenarios across multiple tracks, including log storage timeliness, multi-level account hierarchical verification thresholds, audit cluster storage resource allocation ratios, and business data isolation boundary control. The comprehensive compliance audit technology foundation consists of four core components: a log persistence engine, a multi-level account process standardization verification component, a physically isolated audit computing power scheduling unit, and a multi-track compliance rule plugin adapter. The four components are interconnected in a hierarchical manner. The bottom layer is the physically isolated underlying scheduling unit for audit computing power, responsible for the independent scheduling and boundary control of the entire cluster's computing power and storage resources, providing dedicated operating resources for all upper-layer components. The middle layer sets up a log solidification and persistence engine and a multi-level account process standardization verification component in parallel, responsible for the permanent solidification and storage of log data and the standardized verification of account operation and maintenance processes, respectively. The top layer is a multi-track compliance rule plugin adapter, responsible for the access and adaptation of differentiated compliance rules for different offline business tracks. The base uniformly defines the standard message encapsulation format for the interaction of full-domain audit logs, account processes, and cluster resources, clarifies the field composition and data specifications of all cross-component interactions, and provides two types of core algorithms to ensure message compliance and dynamic adaptation of operating parameters.
[0104] The first type is the message standardization verification algorithm, which includes three serial processing modules: a message parsing unit, a field rule verification unit, and a format normalization processing unit. The algorithm execution steps are divided into three steps: First, it presets a list of required fields and encoding specifications for unified interaction messages, configures the maximum length threshold for a single field and verification rules for missing required fields, and clarifies the unified encoding format for all data. Second, after receiving cross-component interaction messages, the parsing unit breaks down the message content field by field, checking whether each required field is complete, whether the field length exceeds the threshold, and whether the encoding format conforms to the unified specification. Third, it performs format normalization processing on compliant fields, unifying data types and encoding standards; non-compliant messages are directly intercepted and a verification failure result is returned; simultaneously, a message verification log is automatically generated and permanently stored.
[0105] The second category is the compliant scenario parameter adaptive tuning algorithm, which includes three serial processing modules: a scenario load collection unit, a parameter adaptation calculation unit, and a parameter synchronization and distribution unit. The algorithm execution steps are divided into four steps. The first step is to preset the benchmark operating parameters for the entire scenario. The log storage time for online account operation and maintenance scenarios is set to permanent retention. The benchmark threshold for multi-level account hierarchical verification is a three-level control mode. The three-level account permission boundaries are clearly defined as follows: the headquarters-level account has the highest permissions of full system configuration, full subject data viewing, and all account permission allocation; the regional / wholesaler-level account has permissions for account management, data viewing, and report export within its jurisdiction, but cannot modify the core system configuration or cross-regional data; the store-level account only has permissions for business operations and data viewing within its own store, and cannot adjust account permissions or perform cross-store operations. The account operation and maintenance verification process is as follows: the creation of subordinate accounts and the adjustment of permissions must be initiated by the superior account. The entire operation is recorded. Applications that exceed the scope of the current level's permissions are automatically rejected and require approval from a higher-level account. The baseline allocation ratio of audit cluster storage resources is 40% for online scenarios and 60% for offline scenarios. The business data isolation boundary is subject to strict control mode by default. Different business tracks can set independent baseline values according to industry compliance requirements.
[0106] The second step, the data collection unit, collects three core operational data points for each scenario on a minute-by-minute basis: operation concurrency, log data growth, and cluster computing power utilization. The third step, the parameter adaptation calculation unit, calculates adjustment coefficients based on the current load data. When log growth is too rapid, the storage quota for the corresponding track is increased; when account operation concurrency increases, the computing power allocation for hierarchical verification is optimized; simultaneously, the execution status of isolation boundaries is continuously verified, and the isolation control level is automatically strengthened when data out-of-bounds risks are detected. The fourth step, the parameter distribution unit, synchronizes the adjusted parameters to the corresponding functional components. Parameter adjustments are made without interrupting normal evidence storage and auditing operations, achieving seamless dynamic optimization.
[0107] For example, during large-scale promotional events in the retail sector, the volume of POS transactions at stores surges, resulting in a corresponding increase in audit logs that doubles compared to normal days. The scenario load collection unit detects this rapid increase in storage utilization in this sector, and the parameter adaptation calculation unit calculates and adjusts the coefficient accordingly, temporarily increasing the storage resource allocation ratio for the retail sector by five percentage points. Simultaneously, it maintains strict control over the business data isolation boundary to ensure that raw business data never flows into the audit side. After the promotional event ends, the algorithm automatically detects a decrease in transaction volume and log growth, and adjusts the resource allocation ratio back to the baseline value, ensuring that the overall resource utilization of the cluster remains balanced.
[0108] The logs employ an append-only WORM write mode, meaning that once written, the log file cannot be modified or deleted; only appending new records is supported. A hash chain mechanism is also used to solidify the logs, with each log record containing the hash value of the previous one. These records are linked together to form a complete hash chain, meaning that tampering with any log record will break all subsequent hashes. Every day at midnight, an RDA digest verification is performed on all logs, generating a unified notarized digest of the day's logs and synchronizing it on the blockchain to further ensure the immutability of the operation and maintenance logs.
[0109] This system establishes a complete closed-loop compliance handling chain, connecting real-time capture of all types of back-end operation and maintenance activities, persistent storage of encrypted logs, standardized verification of multi-level account operation and maintenance processes, independent computing power scheduling and allocation for the audit cluster, load shedding of raw business data stream storage, unified collection of audit rules across the entire domain, and long-term archiving of compliance files. It also establishes a multi-dimensional correlation mapping operation logic for six core data categories: operation and maintenance events, account hierarchical permissions, audit cluster computing power resources, business data isolation boundaries, operational tiers, and compliance risk tags. The entire compliance handling process connects seven complete stages: real-time capture of all types of back-end operation and maintenance activities, persistent storage of encrypted logs, standardized verification of multi-level account operation and maintenance processes, independent computing power scheduling and allocation for the audit cluster, load shedding of raw business data stream storage, unified collection of audit rules across the entire domain, and long-term archiving of compliance files, forming an end-to-end closed-loop compliance handling chain. The link is equipped with a six-dimensional data association mapping operation logic. Using the unique identifier of the operation and maintenance event as the primary key, it constructs an association mapping network for six types of core data: operation and maintenance event, account level permissions, audit cluster computing resources, business data isolation boundaries, business layer entities, and compliance risk tags, so that a single operation can be associated with full-dimensional compliance information.
[0110] The six-dimensional data association mapping operation logic comprises three sequential functional modules: an event primary key generation unit, a hierarchical data mounting unit, and an association integrity verification unit. The operation execution steps are divided into four steps. First, when the system captures a background operation, the event primary key generation unit generates a globally unique operation event identifier according to distributed rules. This identifier serves as the unified primary key for the entire compliance handling chain, ensuring that identifiers generated across nodes are globally unique. Second, as the handling chain progresses through each stage, the hierarchical data mounting unit automatically mounts the corresponding dimension data generated in the current stage to the event primary key, sequentially mounting account-level permission data, audit cluster computing resource allocation data, business data isolation boundary execution data, operational hierarchical entity data, and compliance risk label data. Third, the association integrity verification unit performs integrity verification at each mounting node, verifying the unique binding relationship between the current data and the event primary key to avoid data mismatch or duplicate mounting. Data that fails verification automatically triggers a completion mechanism; multiple completion failures are marked as abnormal compliance events and archived separately. Fourth, the compliance archive archiving stage constructs a multi-dimensional joint compliance index based on all mounted data, supporting subsequent multi-dimensional compliance statistics and audits by entity, time, and risk level.
[0111] For example, a regional administrator initiates a permission adjustment operation for a store operator account. The system captures this operation in real time and generates a globally unique event identifier as the primary key. During the process, firstly, account-level permission data is mounted to identify that this operation is a permission adjustment from a regional account to a store-level account, conforming to the hierarchical relationship of three-tier control. Next, computing resource data is mounted to record the audit cluster computing power quota and scheduling priority used in this operation. Then, isolation boundary data is mounted to confirm that this operation only synchronizes the unique identifier of business data, with no original business data flowing into the audit cluster beyond its boundaries. Next, operational hierarchical entity data is mounted, associating the corresponding regional, wholesaler, and store-level operational entity information. Finally, the full-domain audit rule verification finds that this permission adjustment exceeds the standard operational permission range of the store account, automatically labeling the event as high compliance risk. Ultimately, all six types of core data are bound to the same event identifier, forming a complete single compliance event record.
[0112] Using unique operational identifiers, multi-level account hierarchical control constraints, audit cluster resource scheduling and allocation rules, physical isolation boundaries between business and audit storage, log solidification timeliness control parameters, and differentiated compliance standards across multiple tracks as multi-dimensional weighted fusion calculation factors, the algorithm normalizes and integrates the raw data across the entire chain of log retention, process verification, cluster scheduling, and storage isolation. This generates a standardized, full-domain online data compliance audit structured system information with triple guarantees: log retention, standardized account processes, and dedicated isolation for audit computing power. The raw compliance data generated from the entire chain of processing is normalized and integrated through a multi-dimensional weighted fusion algorithm, ultimately generating a standardized, full-domain online data compliance audit structured system information with triple guarantees: log retention, standardized account processes, and dedicated isolation for audit computing power. The algorithm uses six categories of indicators—unique operational identifiers, multi-level account hierarchical control constraints, audit cluster resource scheduling and allocation rules, physical isolation boundaries between business and audit storage, log solidification timeliness control parameters, and differentiated compliance standards across eight tracks—as weighted fusion calculation factors. Through weight allocation and standardized calculation, it outputs a systematic compliance result.
[0113] The multi-dimensional weighted fusion algorithm comprises four sequential processing modules: a factor standardization unit, a weight configuration unit, a comprehensive compliance calculation unit, and a systematic integration output unit. The algorithm execution involves four steps. The first step pre-determines the weight percentages of six fusion factors: the unique identity identifier for operations and maintenance is weighted at 0.1; the multi-level account hierarchical control constraint is weighted at 0.2; the audit cluster resource scheduling and allocation rules are weighted at 0.15; the physical isolation boundary between business and audit storage is weighted at 0.25; the log solidification timeliness control parameter is weighted at 0.15; and the differentiated compliance standards for the eight tracks are weighted at 0.15. The second step, the factor standardization unit, performs numerical standardization on each factor, mapping factors of different dimensions to a score range of zero to one. Strict enforcement of physical isolation boundaries with no data out-of-bounds records corresponds to a score of one; the occurrence of data out-of-bounds risk corresponds to a score of zero; account operations fully complying with hierarchical control rules correspond to a score of one; and unauthorized operations correspond to scores calculated progressively based on the severity of the violation. The third step, the comprehensive compliance calculation unit, multiplies the standardized scores of each factor by their corresponding weights and sums them to obtain the comprehensive compliance score for a single compliance event. The higher the score, the stronger the overall compliance of the event. The fourth step, the systematic integration output unit, uses the event identifier as the core to integrate all the original data, verification results, and compliance scores corresponding to all factors into a standardized compliance file. At the same time, it performs multi-dimensional aggregation according to the level of the operating entity, the type of scenario, and the time period, ultimately forming a complete full-domain compliance audit system.
[0114] For example, when auditors check the quarterly compliance of all stores under a wholesaler, the system retrieves all compliance records of operations and maintenance within that period and calculates the overall compliance score for each record. One store pricing modification operation receives one point for a complete and valid identity identifier, one point for account level conforming to store operation permissions, one point for compliant resource scheduling execution, one point for no abnormalities at physical isolation boundaries, one point for permanently fixed log execution, and one point for meeting all track compliance standards. After weighting, the overall compliance score is a full one point, and it is marked as a low-risk compliance event. Another operation, an unauthorized adjustment of profit-sharing ratio, violates account hierarchical control rules, resulting in a corresponding factor score of only 0.4 points, and an overall compliance score of 0.8 points, marking it as a medium-risk compliance event. The system automatically sorts all events from highest to lowest compliance score, and performs hierarchical aggregation based on the three-tier operating entity, generating a complete quarterly compliance audit system report. The report includes core information such as the retention status of all operation and maintenance logs, the standardization execution rate of account processes, the resource isolation status of the audit cluster, and the compliance compliance rate of each track, forming a standardized compliance audit structured system covering all scenarios.
[0115] This application also provides an implementation method in which the traceability record event set includes business data change events and permission configuration change events such as permission granting, permission revocation, and permission level adjustment. Permission configuration change events are collected in real-time when changes occur in the permission control system, and share the same RDA hash digest and NFG trusted digital rights confirmation certificate storage and verification chain as business data change events. Specifically, the traceability record event set covers two core dimensions: business operations and account management. It includes business data change events, as well as three types of permission configuration change events: permission granting, permission revocation, and permission level adjustment, achieving full-scenario audit coverage of business data operations and account permission operations. Permission configuration change events are triggered for real-time collection when changes occur in the permission control system, and share a completely unified RDA hash digest generation and NFG trusted digital rights confirmation certificate storage and verification chain with business data change events, ensuring consistent storage standards, equivalent anti-tampering capabilities, and unified audit verification rules for both types of events.
[0116] The traceability record event set consists of business data change events and permission configuration change events. These two types of events cover all core operational scenarios in offline operational audits, eliminating any audit blind spots. Business data change events correspond to data operations at the operational level, encompassing all additions, modifications, and deletions of core business data, such as product information modifications, transaction data adjustments, and profit-sharing rule changes. These are the core objects of operational data traceability. Permission configuration change events correspond to permission operations at the account management level, specifically including three sub-types: permission granting, permission revokement, and permission level adjustment. This covers the entire lifecycle of account permissions from activation and adjustment to cancellation, compensating for the shortcomings of traditional audits that only focus on business operations and ignore permission changes, thus mitigating operational risks caused by unauthorized permission adjustments at the source. Both types of events are uniformly incorporated into the full-domain traceability audit system, following the same traceability field standards, evidence verification rules, and query retrieval processes. They are logically distinguished only by event type tags, together forming a complete, multi-dimensional set of operational traceability events.
[0117] The permission configuration change event is collected in real-time. The collection action is triggered synchronously with the change operation of the permission control system, without interfering with the business execution process of permission change. This ensures that the permission operation is executed normally while capturing all elements of the event. The collection channel relies on lightweight collection probes mounted on all write operation interface layers of the permission control system. The probes only intercept the metadata information of the operation request and do not participate in the core logic operation of permission change. When the permission control system performs any of the following operations: granting permission, revoking permission, or adjusting permission level, the probe instantly identifies the operation type and starts collection the moment the operation request arrives at the corresponding write interface, automatically extracting full-dimensional traceability elements and permission change-specific fields. The core information collected includes two categories: one is general traceability elements, namely, operation account identifier, login IP address, terminal device fingerprint, and precise operation timestamp, which are completely consistent with the collection standards for business data change events; the other is permission change-specific fields, namely, permission change subtype, unique identifier of the target account, permission scope or level identifier before and after the change, and business entity level information to which the permission belongs.
[0118] The data collection process synchronously performs subject hierarchy mapping verification. Each subordinate subject must be associated with one and only one direct superior subject. Records whose hierarchical relationships do not meet the constraints automatically trigger verification anomalies and are pushed to the operations side for review and correction. Subject hierarchy data is fully synchronized once a day at midnight, and incremental synchronization is triggered in real time when subjects change daily, ensuring that the subject hierarchy relationship on the audit side is consistent with that on the business side.
[0119] Taking a scenario where a regional wholesaler administrator grants pricing adjustment permissions to subordinate store operators as an example, when the administrator submits a permission grant request in the permission management backend, and the request reaches the write operation interface of the permission control system, the interface layer's collection probe instantly captures the operation event and determines the event type as permission grant. The probe simultaneously extracts four common traceability elements: the wholesaler administrator's unique account identifier, the current public IP address, the terminal device fingerprint, and a millisecond-level operation timestamp. It also extracts the target store operator's account identifier, the basic cashier permissions before the change, the newly added pricing adjustment permissions after the change, and the corresponding store's hierarchical information with its superior wholesaler. After all fields are collected, a hierarchical verification is automatically performed to confirm the compliance of the store's affiliation with the wholesaler, ultimately generating a standardized permission configuration change event dataset. The entire collection process is completed within milliseconds of the request reaching the interface, without delaying the execution of the permission grant operation. Event collection ends synchronously upon the completion of the operation.
[0120] All completed permission configuration change events, along with business data change events, enter the same evidence storage processing queue. The entire chain of evidence storage logic, including RDA hash digest generation, NFG trusted digital rights confirmation certificate generation, two-way binding archiving, and integrity verification, is fully reused. The two types of events use completely consistent encryption parameters, certificate rules, and verification standards, and are only distinguished by the event type identifier. The core evidence storage capabilities are identical.
[0121] The algorithm uses a shared RDA-based encryption method to generate hash digests for both types of events. Irreversible hash digests are generated using the same RDA encryption algorithm, with completely unified algorithm execution flow, parameter configuration, and strength standards. Only the core input fields are replaced according to the event type. The algorithm execution consists of three stages. The first stage is plaintext preprocessing, extracting three core fields: the unique identifier of the operation subject corresponding to the event, the unique identifier of the business or permission event, and the millisecond-level precise timestamp. These are concatenated into a raw string in a fixed order. If the length after concatenation is less than 256 characters, randomly generated meaningless characters are added to the end of the string. The padding characters are randomly selected from a set of uppercase and lowercase letters and numbers. The final result is a 256-character fixed-length plaintext block, preventing the leakage of plaintext features. The second stage is multi-round iterative encryption, with a default of 16 rounds of iteration. Each round uses an independently derived round key, which is derived from the global root key combined with the current iteration number and operation timestamp through SHA256 hashing. Round keys from different rounds are completely independent and unrelated. Each round of computation first performs byte-by-byte XOR obfuscation on plaintext blocks using the round key, then rearranges the bytes based on a 256-byte permutation table generated from the root key. Following this, two diffusion operations are performed using an 8-stage linear feedback shift register to ensure that a single byte change in the plaintext will cause at least half of the digest bits to change. The third stage is digest output, ultimately generating a 256-bit fixed-length irreversible encrypted checksum, which is bound to a unique identifier for the corresponding event and stored.
[0122] Taking the aforementioned permission granting event as an example, after the event dataset enters the evidence storage chain, the system extracts three core fields: the unique identifier of the wholesaler administrator account, the unique identifier of the permission change event, and the millisecond-level operation timestamp. These fields are then concatenated according to rules and supplemented with random characters to form a 256-bit plaintext group. After 16 rounds of complete obfuscation and diffusion iterations, a unique 256-bit RDA hash digest is generated. The encryption strength, iteration count, and operation rules of this digest are completely consistent with those of business data change events such as product pricing modifications. There is no differentiated encryption downgrade processing, ensuring that permission change records have the same anti-tampering foundation.
[0123] Both types of events share the NFG Trusted Digital Ownership Certificate system, adhering to the same rules for generation, hosting, and full lifecycle management of NFG Trusted Digital Ownership Certificates, and sharing the same independent hosting nodes and status control mechanisms. During the certificate generation phase, a second hash operation is performed based on the event's unique identifier, the generated RDA encrypted verification digest, and the operator's identity information to generate a globally unique NFG ownership certificate. The certificate contains the corresponding event's ownership attributes, recording the event's operating entity level and initial operator information, completing a one-to-one binding between the certificate and the initial ownership entity. After generation, the certificates are uniformly stored in an independent NFG certificate hosting node. The hosting node maintains physical isolation from the audit cluster and business cluster, synchronizing certificate status identifiers only through standardized interfaces. The hosting node records the entire status of all certificates in real time, including activation, freezing, transfer, and cancellation. All status changes require authorization verification before execution, and the entire operation is traceable and generates a record for evidence storage.
[0124] Using the example of the permission granting event described above, after the RDA hash digest is generated, the system performs a secondary hash operation by combining the unique identifier of the permission change event, the RDA verification digest, and the wholesaler administrator's identity information to generate the corresponding NFG authorization certificate. The certificate contains the ownership attribute of the permission change operation, clearly indicating its ownership by the corresponding wholesaler business entity. After initial ownership binding is completed, it is synchronously stored in an independent hosting node. The hosting node marks the certificate as effective in real time. If the permission is subsequently revoked, the corresponding certificate will be synchronously updated to an invalid state. The entire status change process is traced and a corresponding evidence record is generated. The entire certificate generation and hosting process fully reuses the NFG certificate management system for business data change events, eliminating the need to deploy separate hosting resources for permission events.
[0125] Both types of events share the same two-way binding and verification rules, uniformly executing the two-way binding mechanism between the evidence storage voucher and the audit ledger. Subsequent authenticity verification follows the same verification logic and judgment standards. The two-way binding execution logic is as follows: When the local audit ledger is generated, an RDA summary field and an NFG voucher identifier field are uniformly reserved; after the RDA summary is generated, it is written to the corresponding field, simultaneously triggering the NFG voucher generation process; after the NFG voucher completes its initial ownership binding, the voucher's unique identifier is written back to the corresponding field in the local ledger, completing the forward binding from the local ledger to the evidence storage voucher. Simultaneously, the NFG voucher metadata records the corresponding index information of the local ledger, realizing the reverse association from the voucher to the ledger. After the two-way binding is completed, the entire chain of evidence storage information can be retrieved using any unique identifier.
[0126] During subsequent verification of the log authenticity, both types of events employ the same verification process. First, the core fields of the event to be verified are extracted. Plaintext preprocessing and multiple rounds of encryption are then performed according to the same rules as during encryption, generating an RDA digest for verification. This digest is then compared byte-by-byte with the stored original digest. Simultaneously, the corresponding NFG credential is retrieved to verify the credential's status validity and ownership matching. If both verifications pass, the record is determined to be unaltered; if any one of them fails, the record is considered to be at risk of tampering, and an anomaly alarm is automatically triggered.
[0127] For example, when auditors verify the authenticity of the aforementioned permission grant records, they only need to extract the operation subject identifier, event unique identifier, and operation timestamp for the event. They then re-execute the RDA encryption operation according to the standard procedure, comparing the newly generated digest byte-by-byte with the original digest stored in the ledger. Simultaneously, they retrieve the corresponding NFG voucher to confirm its effectiveness and matching permissions. If both are true, the record is confirmed to be authentic and valid. This entire verification process is completely unified with the verification process for business data change records. Auditors do not need to master two sets of verification standards, ensuring the consistency and impartiality of audit judgments.
[0128] By incorporating permission configuration change events into the traceability record set and sharing a unified evidence storage and verification link with business data change events, on the one hand, full-scenario coverage of operation auditing is achieved, bringing account permission changes into the traceability scope and avoiding operational data risks caused by unauthorized permission adjustments from the source, forming a two-way audit closed loop for business operations and permission control. On the other hand, by reusing the evidence storage link, the uniformity and credibility of the evidence storage standards for the two types of events are ensured, avoiding the problems of high maintenance costs and inconsistent verification standards caused by multiple evidence storage systems. At the same time, all event records have the same strength of anti-tampering capabilities and can all serve as reliable evidence for compliance auditing and liability determination.
[0129] This application also provides an implementation method that interfaces with a multi-terminal data isolation storage and hierarchical access control system. The system receives standardized operation events from this system, carrying identity tags and allow / intercept markers. Events carrying allow markers are sent to this system for RDA hash calculation and generate NFG trusted digital authorization certificates for storage. Events carrying intercept markers are only stored in audit logs and do not generate business authorization certificates. Specifically, this system interfaces with the multi-terminal data isolation storage and hierarchical access control system using a one-way standardized interface. The upstream system is responsible for completing the entire process of terminal identity verification, access permission determination, and data isolation boundary verification, and then outputs all operation events to this system in a unified format. The interface follows a unified message specification, and all output events have a standardized structure, carrying two types of identification information: identity tags and allow / intercept markers.
[0130] Identity tags are used to label the full-dimensional identity attributes of events, specifically including four types of information: terminal type identifier, unique identifier of the operating account, the three-level business entity to which the account belongs, and the current permission level of the account. This supports the partitioned storage and hierarchical traceability of subsequent audit logs. The allow / block flag is a binary judgment flag, generated by the upstream permission system after verification according to preset hierarchical access rules. A flag indicating allow means the operation complies with the permission rules and can access business data; a flag indicating block means the operation exceeds the permission boundary and has been blocked by the permission system.
[0131] The event receiving process is equipped with a standardized message verification algorithm, which comprises three serial processing modules: a message parsing unit, a field compliance verification unit, and a format normalization processing unit. The algorithm execution consists of three steps: First, it pre-defines the list of required fields and encoding specifications for event messages, configures the maximum length threshold for a single field and verification rules for missing required fields, and clarifies the unified encoding format for all data. Second, after receiving event messages pushed by the upstream system, the parsing unit breaks down the message content field by field, verifying the completeness of the identity tag field and the allow / intercept flags, whether the field length exceeds the threshold, and whether the encoding format conforms to the unified specification. Third, it performs format normalization processing on compliant fields, unifying data types and encoding standards; non-compliant messages are directly intercepted and a verification failure result is returned; simultaneously, a message verification log is automatically generated and permanently stored.
[0132] Taking three types of terminal operations in offline business scenarios as an example: a store POS terminal initiates a cashier transaction request, a wholesaler administrator initiates a profit-sharing data query request through the PC backend, and a store operator initiates a product pricing modification request through a mobile terminal. These three requests first enter the upstream multi-terminal access control system. The system verifies the identity legitimacy, account permission scope, and data access boundaries of each terminal. The POS terminal's cashier operation complies with the store terminal's permissions and is approved; the wholesaler administrator's profit-sharing query complies with the corresponding level of permissions and is approved; the store operator's pricing modification exceeds the store account's permission boundaries and is blocked. Subsequently, the upstream system attaches corresponding identity tags and approval or blocking flags to the three events and pushes them to this system in a unified message format. After receiving the message, this system first verifies the compliance of all fields using a standardized validation algorithm. After confirming that the message format is correct, it is sent to the subsequent classification and processing stage.
[0133] Operation events carrying a release flag indicate that they have passed the compliance verification of the authorization system and are considered valid business operations. They will be fully sent to the system's evidence storage processing link to perform RDA hash calculations, generate NFG trusted digital ownership certificates, and complete the entire irreversible evidence storage process. This process uses the same evidence storage verification standards as business data change events. After an event enters the evidence storage link, three core fields are extracted: the unique identifier of the operation subject, the unique identifier of the business data, and the millisecond-level precise timestamp. These are then sent to the RDA ownership encryption algorithm module to generate a hash digest. The algorithm execution consists of three stages. The first stage is plaintext preprocessing, where the three types of fields are concatenated into a raw string in a fixed order. If the length after concatenation is less than 256 characters, randomly generated meaningless characters are added to the end of the string. These padding characters are randomly selected from a set of uppercase and lowercase letters and numbers. The final processing results in a 256-character fixed-length plaintext block, preventing the leakage of plaintext features. The second stage involves multi-round iterative encryption, with a default of 16 rounds of iteration. Each round uses an independently derived round key to perform byte-level XOR obfuscation, followed by byte rearrangement using a 256-byte permutation table. Then, two diffusion operations are performed using an 8-level linear feedback shift register to ensure that a single byte change in the plaintext will cause at least half of the digest bits to change. The third stage outputs a 256-bit fixed-length irreversible encrypted checksum, which is bound to a unique identifier for the corresponding event and stored.
[0134] After the RDA digest is generated, the NFG trusted digital ownership certificate generation process is initiated simultaneously. Based on the event unique identifier, the generated RDA encrypted verification digest, and the operator's identity information, a secondary hash operation is performed to generate a globally unique NFG ownership certificate. The certificate contains the corresponding operating asset's equity attributes and ownership details. After binding the ownership to the initial operator, it is stored in an independent NFG certificate hosting node. The hosting node maintains physical isolation from the audit cluster and business cluster, synchronizing certificate status identifiers only through standardized interfaces. Finally, a two-way binding is performed between the ownership certificate and the audit ledger. The local ledger records the RDA digest and the NFG certificate's unique identifier, while the NFG certificate metadata records the local ledger index information, enabling bidirectional retrieval.
[0135] Using the approved POS transaction event from the example above, after the event, carrying a release marker, enters the evidence storage chain, the system extracts three core fields: the store operator account identifier, the unique number of the corresponding transaction, and the millisecond-level operation timestamp. After plaintext preprocessing and 16 rounds of iterative encryption, a 256-bit RDA hash digest is generated. Simultaneously, a secondary hash operation is performed based on the event identifier, RDA digest, and store entity information to generate a corresponding NFG ownership certificate. This certificate contains the ownership attributes of the transaction and, after being bound to the corresponding store operating entity, is stored in an independent hosting node. Finally, the system completes the two-way binding of the RDA digest, NFG certificate identifier, and local audit ledger, officially making the complete evidence storage for this transaction effective and serving as a reliable audit basis for subsequent business transactions.
[0136] Operation events marked with an interception flag indicate that they have been identified as unauthorized access and blocked by the upstream permission system. They are not considered valid business operations and therefore only basic audit logs are retained. RDA hash calculations are not performed, NFG trusted digital rights certificates are not generated, and they are not included in the business rights ledger system. The audit logs for intercepted events only retain basic traceability information, specifically including four categories: unique event identifier, full information of the identity tag, interception timestamp, and interception reason classification. They are not associated with any business data rights attributes and do not have business rights confirmation effect. Log storage follows a unified partitioning and isolation rule, categorized into corresponding A-end, B-end, and C-end logical partitions according to the business entity level in the identity tag. An append-only write mode is used, supporting only new records; existing historical interception logs cannot be modified or deleted. Logs are also included in the full-domain audit index system, supporting retrieval by account, time, and entity level, for audit scenarios such as unauthorized access statistics, attack tracing, and permission vulnerability investigation.
[0137] Intercepted event logs are not included in the weighted calculation of business evidence strength, nor are they included in the core evidence set of the irreversible electronic change ledger. They are only used for auxiliary audit data archiving and retention. The log retention period is consistent with the business evidence ledger, and is permanently retained by default. The archiving and migration rules are exactly the same as those for ordinary audit logs. Integrity verification must be performed before and after migration to ensure that the log records are complete and traceable.
[0138] Using the example of the intercepted store operator pricing modification event from the previous example, this event, carrying an interception marker, will not trigger the RDA encryption and NFG credential generation process after entering the system. The system only extracts basic information from the event, such as the operator account identifier, store entity level, mobile terminal type, operation timestamp, and reason for unauthorized access, to generate an interception audit log, which is then categorized into the logical partition of the corresponding C-end store according to the entity level. This log only records the trace of this unauthorized access behavior for subsequent auditing and investigation of permission violations; it does not have any effect on confirming the ownership of business data, nor is it associated with any change ledger for product pricing data. When auditors investigate abnormal permissions later, they can retrieve this interception record to confirm that the unauthorized operation has been promptly blocked, but they cannot generate any evidence of business ownership confirmation from this record.
[0139] By connecting to the upstream access control system and implementing differentiated evidence storage, a deep linkage between access control and audit evidence storage is achieved. All valid business operations have complete dual authorization and evidence storage, with unified standards and consistent credibility, serving as reliable evidence for compliance audits and dispute verification. Furthermore, it achieves full audit coverage of all access behaviors, leaving complete traces even for blocked violations, supporting access vulnerability investigation and security incident tracing. Simultaneously, the categorized handling model avoids the waste of evidence storage and authorization resources caused by ineffective blocking events, ensuring the operational efficiency of core business evidence storage and the rational allocation of storage resources.
[0140] In one implementation, such as Figure 2 As shown, this application also provides a comprehensive operational behavior recording and data change tracing system, including: The full-domain operation element collection module 201 is used to collect all types of operation events, including account maintenance, business modification, POS cashier, and third-level distributor profit sharing. It simultaneously captures and archives four types of full-dimensional traceability metadata, including operation account, login IP, terminal device, and operation time. The isolated audit base deployment module 202 is used to build an independent risk control program, a dedicated database and an isolated service cluster. It sets data interaction rules that only synchronize unique business identifiers and do not store raw data streams. It adapts to offline operational audit scenarios across multiple tracks based on a single base plug-in architecture. The dual-rights confirmation ledger generation module 203 is used to carry the complete set of RDA rights confirmation encryption algorithms, build a two-way binding mechanism between blockchain and NFG dual-path rights confirmation certificates and audit ledgers, automatically associate the operation subject information when data changes, and output an tamper-proof electronic change ledger. The full-link traceability retrieval module 204 is used to provide one-click query capability for single and batch change records, connect time-series link restoration, extract full traceability fields, and match the responsible persons of abnormal behavior to complete the handling link; The traceability audit log hierarchical storage and anti-tampering module 205 is used to logically partition and isolate the traceability audit log according to the operating entities of end A, end B, and end C; business terminals only have the permission to append to and write logs, and business terminals have no right to modify or delete the generated historical logs; the system operation and maintenance terminal can perform log archiving and migration, but cannot tamper with the original log content; the log writing synchronously generates the corresponding RDA hash digest and NFG trusted digital ownership certificate, and the certificate constraint prohibits the tampering of the generated historical logs; The 206 module for comprehensive compliance management is used to solidify all operation and maintenance logs, unify multi-level account operation and maintenance standards, rely on an independent audit cluster to carry out evidence storage business, strip the storage load of business systems, and build an integrated data compliance audit system.
[0141] The computer-readable storage medium provided in the above embodiments of this application and the full-dimensional operation behavior recording and data change tracing method provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.
[0142] The various embodiments in this application are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for evaluating the full-dimensional operation behavior recording and data change tracing method, electronic device, electronic device, and readable storage medium are basically similar to the embodiments of the full-dimensional operation behavior recording and data change tracing method described above, and are therefore described simply. Relevant parts can be referred to in the descriptions of the embodiments of the full-dimensional operation behavior recording and data change tracing method described above.
Claims
1. A method for recording all-dimensional operational behaviors and tracing data changes, characterized in that, include: Acquire all operational behaviors related to account creation, permission allocation, account activation and deactivation, business data modification, POS cashier transactions, and multi-level business entity asset profit sharing, and simultaneously collect information such as the operation account, login IP, terminal device identifier, and precise operation time. Deploy independent risk control audit programs, dedicated databases and isolated service clusters, physically isolated from business time-series processing programs, only synchronize unique identifiers of business data, do not store complete original business data streams, and adopt a single-base plug-in architecture to adapt to offline operation audit scenarios across multiple tracks; A complete RDA-based encryption algorithm is used to generate solidified evidence records. A dual-path blockchain and NFG-based rights confirmation certificate system is built, and a two-way binding mechanism between rights confirmation certificates and audit ledgers is constructed. When data changes, full-dimensional operation entity information is automatically bound, and an irreversible electronic change ledger is generated. It provides one-click access to the entire lifecycle modification records of single and batch data, fully restores the entire data change process, retains complete traceability fields, and accurately identifies the responsible parties for data tampering and abnormal changes. The source tracing audit logs are logically partitioned and stored separately according to the operating entities at the A, B, and C ends. Business terminals only have the permission to append logs and have no right to modify or delete historical logs that have already been generated; the system maintenance terminal can perform log archiving and migration, but cannot tamper with the original log content; log writing synchronously generates the corresponding RDA hash digest and NFG trusted digital ownership certificate, and relies on certificate constraints to prohibit tampering with historical logs that have already been generated; Solidify all backend operation and maintenance logs, standardize multi-level account operation and maintenance processes, build a complete online data compliance audit architecture, dedicate an audit cluster to handle evidence storage, and separate business data storage tasks.
2. The method as described in claim 1, characterized in that, The method further includes: the traceability record event set includes business data change events and permission configuration change events such as permission granting, permission revocation, and permission level adjustment; permission configuration change events are collected in real time when changes occur in the permission control system, and share the same set of RDA hash digests and NFG trusted digital rights confirmation certificate storage and verification links with business data change events.
3. The method as described in claim 1, characterized in that, The method further includes: interfacing with a multi-terminal data isolation storage and hierarchical access control system, receiving standardized operation events output by the system carrying identity tags and release / interception markers; sending events carrying release markers into this system to perform RDA hash calculations and generate NFG trusted digital rights confirmation certificates to complete the notarization; events carrying interception markers only retain audit logs and do not generate business rights confirmation certificates.
4. The method as described in claim 1, characterized in that, Deploy independent risk control and auditing programs, dedicated databases, and isolated service clusters, physically isolated from business time-series processing programs. Only unique identifiers for business data are synchronized; complete original business data streams are not stored. A single-base, pluggable architecture is adopted to adapt to multi-track offline operational auditing scenarios, including: Based on the core requirements of secure isolation of business data, independent audit computing power, compatibility with business audits across the industry, and stable operation of the system with low latency, we built an independent risk control audit program, a dedicated evidence storage database, and a distributed isolation service cluster. We constructed an underlying deployment architecture that completely physically isolates the business time-series processing program from the audit cluster. We formulated lightweight data interaction constraints that only synchronize unique identifiers of business data in one direction, do not store data on the ground, and do not store complete original business data streams. This avoids the dual risks of massive business data crowding out audit storage resources and leakage of original business data. Based on the layered and decoupled design requirements of dedicated audit independent bearing, rapid reuse in multiple industry scenarios, and non-intrusive horizontal expansion of the architecture, the overall task of irreversible evidence storage for the entire domain operation is functionally decomposed, and three core independent execution modules are divided: underlying physical isolation cluster deployment, lightweight synchronization of unique business identifiers, and single-base plug-in scenario adaptation. The unidirectional data flow logic between modules, the upstream and downstream call sequence and permission isolation rules are clarified, forming a standardized architecture execution system in which physical isolation evidence storage, simplified identifier synchronization, and multi-track plug-in adaptation operate in a closed loop. Combining the requirements of unalterable audit data, low-latency system verification, rapid deployment across multiple offline business scenarios, and conflict-free compatibility of plugins, this system leverages hardware-level physical isolation mechanisms to define independent operating boundaries for the two sets of computing and storage resources for business and auditing. Based on the principle of minimum data transmission, it limits the business side to output only globally unique data identifiers. It also establishes adaptation specifications for auditing scenarios of offline operating assets, transactions, and profit sharing across eight major tracks using a unified single-base standardized plugin interface. The system fully outputs four types of structured architecture information: cluster hardware isolation deployment details, business data synchronization granularity constraints, multi-track plugin compatibility and adaptation judgment rules, and plugin dynamic loading and scheduling standards.
5. The method as described in claim 1, characterized in that, A complete RDA-based encryption algorithm is used to generate solidified evidence records, establishing a dual-path rights confirmation certificate system based on blockchain and NFG. A two-way binding mechanism between rights confirmation certificates and audit ledgers is constructed, automatically binding full-dimensional operation entity information when data changes, generating an irreversible electronic change ledger, including: Based on RDA-based encryption of target information, blockchain-based distributed evidence storage, NFG-based certificate generation and management, and real-time collection of all-dimensional operational elements, a multi-layered irreversible evidence storage technology foundation is built. Standardized data field encapsulation specifications for unified electronic audit ledgers are established, and core parameters such as the element collection threshold, encryption iteration count, block on-chain interval, and NFG certificate generation frequency are dynamically and adaptively optimized for online account maintenance, POS cashier transactions, multi-level business entity profit sharing, and multi-track offline operations in various differentiated scenarios. The system connects business data change event triggering, automatic capture of full-dimensional operation entity information, generation and solidification of RDA encrypted digests, generation and binding of NFG certificate hashes with ownership initialization, on-chaining of blockchain ownership certificates, bidirectional anchoring of on-chain hashes and NFG certificates, bidirectional binding of certificates with local audit ledgers, ownership custody registration, ownership change verification, revenue distribution accounting, persistent archiving of irreversible ledgers, and traceability index to build a complete closed-loop business link. It establishes a multi-dimensional association mapping operation logic for multiple types of data, including change event ID, operator identifier, access IP, terminal device code, precise timestamp, RDA encrypted verification digest, blockchain block hash, unique NFG certificate identifier, and business entity label. Using unique operator identity identifiers, device network traceability constraints, RDA encryption verification rules, blockchain and NFG bidirectional mapping relationship, multi-track scenario adaptation parameters, and hierarchical traceability dimensions as multi-dimensional weighted fusion calculation factors, the original data of the entire process of collection, encryption, on-chain, certificate generation, ownership transfer, and revenue accounting are normalized and integrated to generate standardized irreversible electronic change ledger structured evidence information with local encryption and solidification, and dual tamper-proof guarantee of on-chain and NFG certificates.
6. The method as described in claim 5, characterized in that, The full lifecycle management process for NFG ownership certificates specifically includes: Based on the unique identifier of business data, the RDA encrypted verification digest, and the identity information of the operating entity, a hash operation is performed to generate a globally unique NFG ownership certificate, completing the ownership binding between the certificate and the initial operating entity. The certificate carries the equity attributes and profit-sharing rules of the corresponding operating assets. An independent hosting node is set up to uniformly host and manage all NFG credentials, recording the validity, freezing, and circulation status of credentials in real time. The hosting node is physically isolated from the audit cluster and only synchronizes credential status identifiers through a standardized interface. When a change of ownership of an operating entity occurs, the operation authority of the initiator of the change and the validity of the target voucher are first verified. After the verification is passed, the ownership of the voucher is updated synchronously. The ownership information before and after the change, the operating entity and the change time are completely retained, and an ownership change record is generated. When a profit-sharing operation is triggered, the system automatically calculates the profit distribution details for each ownership entity based on the current ownership of the NFG certificate and the built-in profit-sharing rules. The profit-sharing results are then bound to the unique identifier of the corresponding NFG certificate for record-keeping, supporting full-chain traceability of profit-sharing.
7. The method as described in claim 1, characterized in that, It provides one-click retrieval of single and batch data modification records throughout their entire lifecycle, fully reconstructing the entire data change process, retaining complete traceability fields, and accurately identifying the responsible parties for data tampering and abnormal changes, including: Based on the full-domain operation evidence storage index library, hierarchical traceability field storage unit, and batch retrieval operation component, a full-link data traceability base is built. The standardized and unified traceability query message encapsulation format is dynamically adapted to single fine-grained queries, batch multi-dimensional searches, multi-level business entity hierarchical traceability, multi-track business scenario query concurrency thresholds, and adjustable index pagination parameters. The system connects change record index matching, splices the entire lifecycle time-series link, pulls complete traceability fields, matches subject association information, locks the responsible subject, and archives traceability results to form a complete query and handling chain. It establishes a multi-dimensional association mapping operation logic for change events, time-series operation nodes, account and device traceability fields, business hierarchical subjects, and abnormal risk labels. Using unique data identifiers, time-series operation timestamps, account and device identity fields, multi-level business hierarchical dimensions, abnormal risk judgment rules, and batch retrieval concurrency constraints as weighting factors, multi-dimensional fusion calculations are performed to unify and integrate the original traceability data of the entire chain of single and batch retrieval, and generate standardized data change traceability structured query information.
8. The method as described in claim 7, characterized in that, Solidify all backend operation and maintenance logs, standardize multi-level account operation and maintenance processes, build a complete online data compliance audit architecture, dedicate an audit cluster to handle evidence storage, and separate business data storage tasks, including: Based on a log solidification and persistence engine, a multi-level account process standardization verification component, a physical isolation underlying scheduling unit for audit computing power, and a multi-track compliance rule plug-in adapter, a full-link integrated compliance audit technology foundation is built. It unifies the standard encapsulation message format for audit logs, account processes, and cluster resource interaction across the entire domain. It dynamically and adaptively optimizes the adjustable core parameters for online account operation and maintenance, POS cashier transactions, multi-level business entity asset profit sharing, and differentiated scenarios of offline operations across multiple tracks, such as log storage timeliness, multi-level account hierarchical verification thresholds, audit cluster storage resource allocation ratios, and business data isolation boundary control. It connects the real-time capture of all types of back-end operation and maintenance, the persistent storage of encrypted logs, the standardized verification of multi-level account operation and maintenance processes, the independent computing power scheduling and allocation of audit clusters, the load stripping of business raw data stream storage, the unified collection of audit rules across the entire domain, and the complete closed-loop processing link of long-term archiving of compliance files. It establishes a multi-dimensional correlation mapping operation logic for six core data categories: operation and maintenance operation events, account level permissions, audit cluster computing power resources, business data isolation boundaries, business layered entities, and compliance risk labels. Using unique identity identifiers for operation and maintenance, multi-level account hierarchical control constraints, audit cluster resource scheduling and allocation rules, physical isolation boundaries between business and audit storage, log solidification and timeliness control parameters, and multi-track differentiated compliance standards as multi-dimensional weighted fusion calculation factors, the original data of the entire chain of log retention, process verification, cluster scheduling, and storage isolation are normalized and integrated to generate a standardized full-domain online data compliance audit structured system information with triple protection of log retention, standardized account processes, and dedicated isolation of audit computing power.
9. A comprehensive operational behavior recording and data change traceability system, characterized in that, The system is configured to execute the full-dimensional operation behavior recording and data change tracing method according to any one of claims 1 to 8 by executing executable instructions.
10. An electronic device, characterized in that, include: First processor; and memory for storing executable instructions of the first processor; The first processor is configured to execute the full-dimensional operation behavior recording and data change tracing method according to any one of claims 1 to 8 by executing the executable instructions.
