A game account transaction method and system based on data consistency verification

Abstract

The application provides a game account transaction method and system based on data consistency verification, and relates to the technical field of data processing. The method comprises the following steps: collecting multi-modal original data to be used to generate a security target evidence document; pre-processing the multi-modal original data to obtain standard vector data; constructing a CC standard knowledge graph; associating the standard vector data and the CC standard knowledge graph to determine governance data; performing semantic analysis on the governance data by using a lightweight industry model fine-tuned in the CC security target field to obtain a key element set; performing semantic fusion on the key element set and a standard sentence based on a CC standard vector sentence library to generate standard content conforming to CC specifications; and outputting a security target evidence Word document conforming to CC standards according to the standard content conforming to CC specifications.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for game account transactions based on data consistency verification. Background Technology

[0002] With the rapid development of the online gaming industry, game accounts have gradually become an important carrier of character levels, equipment, skins, items, virtual currency, and other virtual assets. The account assets accumulated by users over a long period of gaming have a certain trading value, thus giving rise to various trading scenarios such as buying and selling, consignment, recycling, and escrow transactions of game accounts.

[0003] In the process of game account trading, the authenticity, completeness, and consistency of account asset data before and after the transaction are crucial factors affecting the credibility of the transaction. By collecting, anonymizing, comparing, and verifying account asset data, more reliable data can be provided to both buyers and sellers, enabling trading platforms to accurately determine the status of account assets and providing technical support for subsequent ownership transfer, fund flow, and order processing.

[0004] However, existing technologies still have significant shortcomings. Direct account link changes and private transactions lack a trusted third-party verification mechanism, making them prone to issues such as inaccurate account information, non-delivery after payment, or seller appeals for account recovery. Furthermore, manual verification relies on screenshot uploads and customer service visual judgment, which is inefficient, prone to errors, and makes it difficult to guarantee the consistency of account asset data. In addition, sensitive data such as real-name information, mobile phone numbers, and email addresses are easily exposed during transactions, making it difficult for existing solutions to simultaneously address compatibility, verification accuracy, privacy compliance, and the security of ownership transfer. Summary of the Invention

[0005] To address the lack of a trusted third-party verification mechanism in existing technologies for direct account link changes and private transactions, which can easily lead to issues such as inaccurate account information, non-delivery after payment, or seller appeals for account recovery, and to resolve the problems inherent in current technologies, manual verification relies on screenshot uploads and customer service visual judgment, which is inefficient, prone to errors, and fails to guarantee the consistency of account asset data. Furthermore, sensitive data such as real-name information, mobile phone numbers, and email addresses are easily exposed during transactions, and existing solutions struggle to simultaneously address the technical challenges of adaptability, verification accuracy, privacy compliance, and the security of ownership transfer.

[0006] The technical solution provided by this invention is as follows: S1: Collect multimodal raw data for the security target evidence document to be generated; S2: Preprocess the original multimodal data to obtain standard vector data; S3: Constructing a CC standard knowledge graph; S4: Link standard vector data and CC standard knowledge graph to determine governance data; S5: Through a lightweight industry-wide model finely tuned by CC security target domain, semantic analysis is performed on governance data to obtain a set of key elements; S6: Based on the CC standard vector sentence structure library, semantic fusion is performed on the key element set and standard sentence structure to generate standard content that conforms to CC specifications; S7: Based on the standard content conforming to the CC specification, output a Word document containing security target evidence that conforms to the CC standard.

[0007] The beneficial effects of the technical solution provided by this invention include: This invention utilizes a lightweight cross-game intermediate container and a non-intrusive data sniffing and collection method to collect account asset data without relying on official interfaces or actively logging into game servers. Combined with dynamic privacy desensitization and multimodal hash consistency verification, it automatically compares the initial and subsequent data collection to determine the authenticity, integrity, and consistency of account assets. Simultaneously, based on the verification results, it automatically executes ownership transfer, order rollback, or fund return, forming a secure, reliable, and automated transaction loop, thus improving the adaptability, security, and reliability of game account transactions. Attached Figure Description

[0008] Figure 1 A flowchart illustrating a game account transaction method based on data consistency verification, provided as an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a game account trading system based on data consistency verification, provided as an embodiment of the present invention. Detailed Implementation

[0009] S1: Collect multimodal raw data for the security target evidence document to be generated.

[0010] The "security objective evidence document to be generated" refers to the evidentiary document that needs to be generated subsequently according to the CC standard or the preset security objective specification. "Multimodal raw data" refers to the different types of basic data used to generate this document.

[0011] It should be noted that by uniformly collecting multimodal raw data for generating security target evidence documents, a complete data foundation can be provided for subsequent data preprocessing, knowledge graph association, semantic parsing, and document generation. This allows different types of data, such as text, tables, images, and rich text, to enter the same processing flow, avoiding the loss of subsequent generated content due to scattered data sources or inconsistent formats. This improves the completeness, accuracy, and automation of security target evidence document generation.

[0012] Specifically, In one possible implementation, the multimodal raw data includes: text, tables, images, and rich text.

[0013] S2: Preprocess the original multimodal data to obtain standard vector data.

[0014] Preprocessing refers to operations such as normalization, feature alignment, and anomaly removal on the original data. Standard vector data refers to the unified conversion of preprocessed data from different modalities into a computable high-dimensional vector representation, giving it a unified semantic space representation capability.

[0015] It should be noted that by normalizing, aligning features, and removing outliers from the original multimodal data, and mapping different types of data to standard vector data, data with inconsistent structures can be computed and analyzed in the same semantic space. This effectively improves the accuracy and stability of subsequent knowledge graph association, semantic parsing, and sentence matching, while reducing noise interference and improving overall processing efficiency and system robustness.

[0016] In one possible implementation, the preprocessing includes: normalization, feature alignment, and anomaly removal.

[0017] Specifically, data is collected through a non-intrusive data sniffing and acquisition module. This module consists of two main components: a local sniffing unit and a data acquisition and control unit. The structure, connection relationships, and functional rules of each component are as follows: The local sniffing unit is the module's data entry unit, responsible for the non-intrusive capture of local client account data. Its core attributes include the sniffing scope, acquisition method, and anti-circumvention rules. The sniffing scope only reads static asset data such as levels, equipment, and skins cached on the local client, and does not collect sensitive data such as account passwords or bound mobile phones, nor does it access sensitive areas of game memory. The acquisition method employs local process feature capture + lightweight UDP / TCP packet parsing technology, without sending any operation commands to the game server throughout the process. The sampling frequency is fixed at 1 time / 2 seconds, and the sampling duration is ≤10 seconds / time.

[0018] S3: Construct a CC standard knowledge graph.

[0019] The CC standard knowledge graph refers to a structured semantic knowledge system built around the CC (Common Criteria) security assessment standard. It is usually composed of seven ST (security objectives) modules, SFR / SAR security requirement rules, standard sentence structure specifications, and hierarchical and constraint relationships between modules. It is used to perform semantic modeling and rule expression of data elements in security objective documents.

[0020] It should be noted that by constructing a CC standard knowledge graph, the originally scattered and unstructured CC standard specifications are transformed into a structured knowledge network with hierarchical relationships and semantic connections. This enables the relationships between different security target modules to be uniformly expressed and calculated, thereby providing a unified knowledge foundation for subsequent data module attribution, rule binding, semantic parsing, and standard sentence matching. This improves the automation level of data governance, the consistency of rule execution, and the standardization of security target document generation.

[0021] In one possible implementation, the CC standard knowledge graph includes: the seven ST modules, SFR / SAR rules, CC standard syntax specifications, and the hierarchical relationships between the modules.

[0022] S4: Link the standard vector data and the CC standard knowledge graph to determine the governance data.

[0023] Among them, governance data refers to the data set that has been cleaned, classified and standardized after the above-mentioned association and rule constraints have been completed.

[0024] In one possible implementation, S4 specifically includes: S401: Perform similarity matching between the standard vector data and the entity nodes in the CC standard knowledge graph to obtain the data module affiliation and hierarchical position corresponding to the standard vector data.

[0025] S402: Combining the data module affiliation and hierarchical position, call the built-in reasoning rules in the CC standard knowledge graph to process and bind the standard vector data according to the rules.

[0026] S403: Based on the binding results of processing rules, clean up data noise, data with conflicting formats, and data with logical contradictions in standard vector data.

[0027] S404: In accordance with the CC standard specifications in the CC standard knowledge graph, the cleaned standard vector data is standardized by converting it into date, encoding and sentence structure to obtain governance data.

[0028] It should be noted that by semantically associating standard vector data with the CC standard knowledge graph, data from different sources and with different structures can be automatically mapped to the corresponding standard modules and hierarchical structures. Under the constraints of knowledge graph rules, data classification and normalization are completed, thereby achieving data noise filtering, logical conflict resolution, and structural consistency optimization. This improves the automation and standardization of data governance, and provides a high-quality, strongly structured data foundation for subsequent semantic parsing and standard content generation.

[0029] S5: Through a lightweight industry-wide model fine-tuned by CC security target domain, semantic analysis is performed on governance data to obtain a set of key elements.

[0030] Among them, the lightweight industry-wide model fine-tuned in the CC (Common Criteria) security objective domain refers to a model whose parameters are fine-tuned based on pre-trained lightweight large language models using CC (Common Criteria) security assessment data, enabling it to possess a specialized understanding of CC standard semantics, structure, and security objective content. The key element set refers to the set of core semantic information extracted from governance data, including key fields such as business entities, operational behaviors, protected objects, and scope of permissions.

[0031] It should be noted that by utilizing a lightweight industry-wide model fine-tuned by CC security target domain to perform deep semantic understanding and structured analysis of governance data, key business elements can be automatically extracted from complex standardized data. This achieves efficient conversion from rule-based data to semantic elements, thereby reducing the manual feature extraction process, improving the accuracy and consistency of element identification, and providing a high-quality semantic input foundation for subsequent standard sentence matching and security target content generation.

[0032] In one possible implementation, the set of key elements includes: business entity, operational behavior, protected object, and scope of permissions.

[0033] Specifically, a lightweight industry-wide model, finely tuned by CC security target domain instructions, is deployed to perform precise semantic deconstruction of the governed data; it automatically distinguishes between fixed redundant sentences in CC standards and user-defined key elements, and intelligently extracts core parameters such as business entities, operational behaviors, protected objects, and permission scopes, achieving accurate extraction of key elements without configuring regular expression rules.

[0034] S6: Based on the CC standard vector sentence structure library, semantic fusion is performed on key element sets and standard sentence structures to generate standard content that conforms to CC specifications.

[0035] The CC standard vector sentence structure library refers to a semantic retrieval library formed by representing the standard sentences in the Common Criteria (CC) security standard using vectorization methods. It is used to support sentence structure matching and invocation based on semantic similarity. Standard content refers to structured, semantically complete text content that conforms to standard expression forms and is generated in accordance with CC specification requirements.

[0036] It should be noted that by semantically matching and fusing the set of key elements with the standard sentence patterns in the CC standard vector sentence pattern library, the extracted core semantic information can be automatically embedded into the standard expression structure that conforms to the CC specification. This enables the automatic generation of standard text content from semantic elements, reduces the manual template writing process, improves the accuracy of sentence pattern matching and the consistency of generated content, and enhances the standardization and automation capabilities of CC specification document generation.

[0037] In one possible implementation, S6 specifically includes: S601: Convert the full CC standard syntax library into a vector index library to obtain the CC standard vector syntax library.

[0038] The CC standard full sentence library refers to the collection containing all the regulated expression sentences in the CC (Common Criteria) security standard. The vector index library refers to a database format that uses vectorized representations of standard sentences to construct an index structure, supporting efficient semantic retrieval.

[0039] S602: Based on the key element set, in the CC standard vector sentence pattern library, the semantic vector and sentence pattern vector corresponding to the key element set are matched with each other through the approximate nearest neighbor vector retrieval algorithm to obtain candidate standard sentence patterns.

[0040] Among them, the approximate nearest neighbor vector retrieval algorithm refers to an algorithm that quickly finds the closest target vector in a high-dimensional space by calculating the similarity between vectors. Candidate standard sentence patterns refer to a set of multiple standard sentence patterns that may conform to semantics, selected through preliminary semantic matching.

[0041] S603: Determine the optimal standard sentence structure from the candidate standard sentence structures based on sentence structure similarity, module affiliation, and hierarchical position.

[0042] Among them, the optimal standard sentence structure refers to the most matching sentence structure obtained by combining similarity, module affiliation and hierarchical constraints in the candidate set.

[0043] S604: The key element set and the optimal standard sentence structure are merged to obtain the initial standard content.

[0044] S605: Perform syntax verification and logical calibration on the initial standard content to obtain standard content that conforms to the CC specification.

[0045] It should be noted that by vectorizing the CC standard sentence structure and introducing an approximate nearest neighbor semantic retrieval mechanism, the key elements and standard sentence structure are efficiently matched. The optimal expression form is selected by combining multi-dimensional constraints, and the final standard content is generated through semantic fusion and grammatical calibration. This significantly improves the accuracy and efficiency of sentence structure matching, reduces reliance on manual template splicing and rule configuration, and makes the CC standard document generation process more automated, standardized and consistent.

[0046] S7: Based on the standard content conforming to the CC specification, output a Word document containing security target evidence that conforms to the CC standard.

[0047] Among them, the security target evidence Word document refers to a Word format document used to prove that the security target meets the requirements of the CC standard, and is used to carry the structured security target content and related evidence information.

[0048] In one possible implementation, S7 specifically includes: S701: Construct a tree-structured security target data model based on the standard content conforming to CC specifications and the hierarchical relationship of the CC standard knowledge graph.

[0049] Among them, the tree-structured security target data model refers to a hierarchical data organization structure built based on the hierarchical relationship of a knowledge graph, which is used to represent the hierarchical relationship of security target content.

[0050] S702: Bind the tree-structured security target data model with document sorting rules, style rules, and template placeholder mapping relationships to obtain the document data to be rendered.

[0051] The document data to be rendered refers to the intermediate document data after binding structured data and document rules.

[0052] S703: Stores the document data to be rendered in both vector and structured formats.

[0053] S704: Based on the hierarchical relationship of the CC standard knowledge graph, it dynamically loads the document data to be rendered through a streaming rich document rendering engine.

[0054] Among them, the streaming rich document rendering engine refers to a dynamic document generation engine that supports rendering while loading.

[0055] S705: Performs real-time rendering of multimodal raw data in dynamically loaded document data to be rendered.

[0056] Real-time rendering refers to the synchronous generation of document content during the data loading process.

[0057] S706: Automatically generates document table of contents, page numbers, headers, and footers based on real-time rendering results.

[0058] S707: Combining document table of contents, page numbers, headers, and footers, it outputs a Word document of security target evidence that conforms to the CC standard by loading, rendering, and outputting simultaneously.

[0059] It should be noted that by constructing a tree-structured data model based on the CC knowledge graph and combining it with document rules to bind structured data with templates, and by adopting a dual-format storage of vector and structured data as well as a streaming rich document rendering mechanism, dynamic loading and real-time generation of document content can be achieved. This avoids the delay problem of traditional static batch generation methods, improves document generation efficiency and system response speed, and ensures that the output results have higher stability and automation in terms of structural consistency, format standardization, and CC standard compliance.

[0060] In this embodiment of the invention, data acquisition adopts a Trusted Execution Environment (TEE) local feature extraction + homomorphic encryption feature upload to the cloud scheme. Feature extraction is completed within the hardware security zone of the data source, the original data does not leave the domain, and the encrypted data is uploaded to the cloud for processing, thereby achieving secure, private and efficient multi-source data access.

[0061] Data governance adopts a dynamic alignment scheme of neural symbolic reasoning and CC standard ontology, which integrates neural network perception and symbolic logic reasoning to automatically complete the dynamic adaptation of data with CC standards and can adapt to standard iteration without the need for pre-configured rules.

[0062] The feature extraction adopts a small-sample guided feature extraction scheme, which builds a CC domain-specific extraction mode based on a small number of labeled samples, and can achieve accurate extraction of key features in multiple scenarios without model fine-tuning.

[0063] The sentence generation adopts a constraint decoding compliant sentence generation scheme, which directly generates complete sentences that conform to CC specifications through dual constraints of syntax and logic. It does not rely on template libraries and can dynamically adapt to standard updates and expansions.

[0064] The data modeling adopts a hierarchical semantic forest dynamic modeling scheme, which constructs a flexible data model with a multi-branch tree structure, supports custom combination of document modules and layout expansion, and adapts to evaluation needs in multiple scenarios.

[0065] The document generation solution uses OOXML incremental streaming compilation to directly compile and output Word's underlying format without relying on a rendering engine, achieving lightweight, high-speed, and compliant document generation.

[0066] The overall solution adopts a distributed secure evidence generation architecture powered by zero-knowledge proof (ZKP), enabling end-to-end encrypted computation, compliance and verifiability, and cross-institutional joint evaluation, thus achieving privacy protection throughout the entire data lifecycle.

[0067] This invention constructs a dual-collection, dual-verification, one-way closed-loop flow mechanism: "Initial Hosting Collection → Desensitization → Initial Verification → Secondary Collection → Verification → Delivery / Revert." The entire process involves encrypted data transmission, no sensitive data exposure, no game intrusion, and no manual intervention, forming a traceable, verifiable, and closed-loop innovative transaction process. The main flow links are: Seller completes real-name authorization → Access layer cross-game lightweight intermediate container module (adaptive parsing of all game data + sandbox isolation hosting) → Access layer non-intrusive data sniffing and collection module (initial non-intrusive static asset collection, no login, no commands, no game server interaction) → Core processing layer dynamic privacy desensitization engine module (real-time sensitive information identification + irreversible desensitization processing) → Core processing layer multimodal hash consistency verification module (initial hash feature generation and benchmark value locking) → Core Processing layer full-scenario transaction adaptation module (transaction pattern recognition + order status scheduling) → Buyer completes payment and triggers secondary verification → Access layer non-intrusive data sniffing and collection module (secondary non-intrusive re-collection of data from the same source, collection dimensions / rules / frequency are completely consistent with the first time) → Core processing layer dynamic privacy desensitization engine module (secondary desensitization review to ensure no privacy residue) → Core processing layer multimodal hash consistency verification module (dual-collection data hash fusion comparison + consistency result judgment) → Core processing layer full-scenario transaction adaptation module (distributes execution instructions according to verification results) → Execution output layer interface-free ownership snapshot transfer module (verification passed: asset fingerprint binding to buyer + irreversible destruction of original account index; verification failed: order automatically rolled back + funds refunded via the original route) → Transaction complete loop (successful delivery / order termination).

[0068] The core modules and their operating mechanisms are as follows: The cross-game lightweight middleware container module's core function is to adapt account asset data across PC games, mobile games, web games, channel servers, and niche games, providing a secure and isolated lightweight hosting carrier for both parties in the transaction. It serves as the core data transfer node throughout the entire account transaction process, comprising two main components: a container isolation unit and a data adaptation unit. The structure, connections, and functional rules of each component are as follows: The container isolation unit is the core execution unit of the module, responsible for the sandbox isolation and hosted operation of account data. Its core attributes include container specifications, isolation rules, and operational control: Container specifications: Single container memory usage ≤ 50MB, container startup time ≤ 80ms, supports 1000+ concurrent instances, and the container process is independent of the game client and server, without mounting any game processes; Container resource allocation uses dynamic thresholds, ensuring that the peak resource usage within a single account transaction cycle does not exceed a preset threshold, triggering expansion protection if it does. Isolation rules employ kernel-level sandbox isolation technology to achieve physical isolation of data between buyers and sellers, prohibiting data exchange and cross-container access between containers; only anonymized account asset data is stored within containers, without retaining original account passwords, binding information, or other sensitive data; the container lifecycle is bound to the transaction process, and it is automatically destroyed immediately after the transaction is completed, leaving no residual data. Operational management categorizes container status into five types: initialization, hosting, verification, transfer, and destruction, with status changes synchronized to the transaction management center in real time; real-time alarms are triggered for abnormal container operation, automatically switching to a backup container; container operation logs are fully retained for a retention period of ≥5 years. The data adaptation unit is a module data compatibility processing unit responsible for automatic parsing and adaptation of multi-game data structures. Core attributes include adaptation scope, parsing rules, and data mapping: adaptation scope fully covers PC games, mobile games, web games, channel servers, and niche online games, with a data structure adaptive parsing coverage rate of 99%; it supports data adaptation for all categories of game accounts. The parsing rules utilize a built-in game asset feature parsing engine to automatically identify core asset fields such as level, equipment, and skins. Data without a fixed format is parsed using feature matching, achieving an accuracy rate of ≥99.5%. Data mapping uniformly maps non-standard asset fields to platform standard fields; the mapping relationship can be dynamically updated, and adding new games does not require restarting the module.

[0069] The non-intrusive data sniffing and collection module's core function is to collect static asset characteristic data of game accounts across the entire domain without logging into the game, triggering anti-cheat measures, or relying on official interfaces. This provides raw data input for subsequent verification and transactions. It comprises two main components: a local sniffing unit and a data collection and control unit. The structure, connection relationships, and functional rules of each component are as follows: The local sniffing unit is the module's data entry unit, responsible for the non-intrusive capture of local client account data. Its core attributes include sniffing scope, collection method, and anti-circumvention rules: Sniffing scope: It only reads static asset data such as levels, equipment, and skins cached on the local client, and does not collect sensitive data such as account passwords or bound mobile phones, nor does it access sensitive areas of game memory. Collection method: It uses local process feature capture + lightweight UDP / TCP packet parsing technology, without sending any operation commands to the game server throughout the process; the sampling frequency is fixed at 1 time / 2 seconds, and the sampling duration is ≤10 seconds / time. Anti-circumvention rules: It has a built-in game anti-cheat feature circumvention library. The collection behavior simulates the normal user client running state, with an anti-cheat trigger rate of 0; collection is immediately paused upon detecting a security check. The data acquisition and control unit is responsible for rule control and anomaly handling in the acquisition process. Its core attributes include acquisition switch, field filtering, frequency limits, and anomaly alarms: Acquisition switch: Acquisition function is enabled / disabled based on asset type. Rule changes require compliance approval from the risk control team, and operation records are encrypted and retained for ≥5 years. Field filtering: Required fields include game ID, account level, and core asset quantity; optional fields include fragments and limited-time items. Missing fields do not affect the process. Frequency limits: The first acquisition is executed immediately after seller authorization, and the second acquisition is triggered after buyer payment, with an interval of ≥30 seconds between acquisitions; the number of acquisitions per account is ≤5 within 24 hours. Anomaly alarms: Acquisition timeout ≥5 seconds, data missing rate ≥5%, or acquisition failure triggers both email and SMS alarms, automatically retrying twice; failed retry marks the account as abnormal.

[0070] The core function of the dynamic privacy masking engine module is to perform real-time sensitive information identification and masking processing on the collected raw account data, achieving data usability without visibility, ensuring privacy compliance throughout the transaction process, and providing downstream modules with masked and compliant data. It comprises two main components: a sensitive information identification unit and a masking execution unit. The structure, connection relationships, and functional rules of each component are as follows: The sensitive information identification unit, the module's privacy detection unit, is responsible for automatically scanning and identifying sensitive fields in the data. Its core attributes include the identification scope, identification algorithm, and identification threshold. The identification scope fully covers personal sensitive information such as real names, mobile phone numbers, email addresses, device IDs, and login IPs; it also identifies commercial sensitive data such as third-party bound accounts and payment records. The identification algorithm uses a sensitive field feature library + real-time vector matching algorithm, with millions of built-in sensitive feature samples, supporting fuzzy matching; the identification speed is ≤5ms / scan, and the time for a full scan of a single data point is ≤10ms. The identification threshold achieves a sensitive information identification accuracy of ≥99.8% and a missed identification rate of ≤0.01%; sensitive data is immediately marked upon identification. The data masking execution unit is responsible for the targeted masking and irreversible processing of sensitive data. Its core attributes include masking method, masking rules, and data control: Masking method: Mandatory sensitive fields are encrypted and masked, replaced with random masking identifiers; optional sensitive fields are filled with "Unauthorized display"; data cannot be reversed after masking. Masking rules: Masking is performed immediately after data collection, first identifying, then masking, and then transmitting, without storing the original sensitive data throughout the process; the integrity of asset data is not compromised. Data control: Masked data only circulates within the transaction module and is not displayed, exported, or shared externally; masking operation logs are fully recorded and retained for at least 5 years.

[0071] The core function of the multimodal hash consistency verification module is to perform multi-dimensional and accurate comparison of anonymized account data collected twice, determining the authenticity and consistency of account assets, and providing a core basis for transaction execution. It comprises two main components: a hash conversion unit and a consistency verification unit. The structure, connection relationships, and functional rules of each component are as follows: The hash conversion unit is responsible for converting multiple types of data into unique hash feature values. Its core attributes include conversion algorithms, feature encoding, and data fusion. For conversion algorithms, text asset data uses the SHA-256 algorithm to generate a 64-bit unique hash value; image screenshot data uses pHash-aware hashing to generate a 64-bit feature code; and behavioral feature data uses Hamming distance to calculate feature vectors. For feature encoding, the hash value is uniquely bound to the original asset data; even minor changes in the data will completely change the hash value, ensuring no collisions or duplications; the encoding process does not lose asset feature information. For data fusion, the three types of hash features—text, image, and behavior—are weighted and fused, with a weight ratio of 60% for text, 30% for image, and 10% for behavior. The consistency verification unit is responsible for hash feature comparison and result determination. Its core attributes include verification rules, threshold settings, and anomaly handling: Verification rules compare the comprehensive hash feature values ​​collected in the first and second rounds, performing precise matching field by field and feature by feature, fully automated without human intervention. Threshold settings: A comprehensive hash similarity ≥ 99.9% is considered consistent; below 99.9% is considered abnormal. A single verification takes ≤ 1.5 seconds, with an error rate ≤ 0.01%. Anomaly handling: Failed verification immediately terminates the transaction, automatically cancels the order, refunds funds, and adds the abnormal account to the risk control blacklist. Data that fails verification is stored in an anomaly pool, retained for 24 hours for manual review; failure to process within this period triggers a level three alarm.

[0072] The core function of the all-scenario transaction adaptation module is to adaptively match all types of transaction modes, including C2C, B2C, buyback, consignment, and escrow, automatically switching transaction processes and control rules. This allows a single system to adapt to all business scenarios. It comprises two main components: a pattern recognition unit and a process adaptation unit. The structure, connections, and functional rules of each component are as follows: The pattern recognition unit, the module's scenario determination unit, is responsible for automatically identifying transaction types and business rules. Its core attributes include recognition scope, determination rules, and parameter loading. Recognition scope fully covers five core transaction modes: C2C, B2C, account buyback, consignment, and escrow; it also supports custom transaction mode extensions. Determination rules automatically identify transaction modes based on seller identity, transaction target, fund custody method, and settlement rules, with 100% accuracy. Parameter loading automatically loads corresponding transaction parameters after pattern recognition, with a loading time of ≤100ms and real-time effectiveness. The process adaptation unit is responsible for the automated switching and execution of transaction processes. Its core attributes include process configuration, node control, and status synchronization. Process configuration allows for independent process nodes for different transaction modes, which can be flexibly configured, achieving 100% process adaptation. Node management: Real-time monitoring of process node execution status; automatic triggering of the next node upon completion of the previous node; automatic rollback upon execution failure. Status synchronization: Real-time synchronization of process status to the platform front-end, fund custody, and ownership transfer modules; full retention of transaction status logs for at least 5 years.

[0073] The core function of the interface-free ownership snapshot transfer module is to securely transfer account asset ownership and destroy original account permissions without relying on the game's official open interface, completely blocking the seller's recovery path and ensuring irreversible transaction settlement. It comprises three main components: an ownership snapshot unit, a permission transfer unit, and an index destruction unit. The structure, connection relationships, and functional rules of each component are as follows: The ownership snapshot unit is responsible for generating a unique ownership certificate for account assets. Its core attributes include snapshot generation, fingerprint binding, and certificate storage. Snapshot generation immediately generates a full-data feature snapshot of the account assets upon successful verification, with 100% data integrity and a generation time of ≤3 seconds. Fingerprint binding generates a unique asset feature fingerprint for the snapshot, which is bidirectionally bound to the buyer's real-name information and platform order number, ensuring 100% fingerprint uniqueness. Certificate storage uses encrypted hash storage and is stored in the platform's compliant evidence repository, with access only available for judicial audits and dispute resolution, and a storage period of ≥10 years. The Permission Transfer Unit is responsible for transferring asset permissions from the intermediate container to the buyer's account. Its core attributes include transfer rules, permission control, and execution timeliness: Transfer Rules: It does not interface with the official game interface; the transfer is achieved through permission redistribution within the intermediate container, completely transferring access rights, control rights, and disposal rights to the buyer's new account. Permission Control: After the transfer, the buyer has unique operating permissions; the seller and platform have no modification or access rights; permission change records are recorded on the blockchain in real time. Execution Timeliness: The entire ownership transfer process is automated, taking ≤3 seconds; automatic rollback occurs if the transfer fails. The Index Destruction Unit is responsible for the irreversible destruction of the original account's asset characteristic index. Its core attributes include destruction scope, execution rules, and security verification: Destruction Scope: It destroys the original seller account's asset characteristics, login permissions, and data association indexes, fully covering all recoverable paths. Execution Rules: The destruction operation is executed once, irreversibly and unrecoverable. Security Verification: After destruction, an automatic verification scan is performed to confirm no index residue; if verification passes, the transaction is marked as complete, and escrow funds are released.

[0074] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for game account trading based on data consistency verification, characterized in that, include: S1: Collect multimodal raw data for the security target evidence document to be generated; S2: Preprocess the original multimodal data to obtain standard vector data; S3: Constructing a CC standard knowledge graph; S4: Associate the standard vector data and the CC standard knowledge graph to determine the governance data; S5: Using a lightweight industry-wide model fine-tuned by the CC security target domain, semantic parsing is performed on the governance data to obtain a set of key elements; S6: Based on the CC standard vector sentence library, semantic fusion is performed on the set of key elements and standard sentences to generate standard content that conforms to CC specifications; S7: Based on the standard content conforming to the CC specification, output a Word document containing security target evidence that conforms to the CC standard.

2. The game account transaction method based on data consistency verification according to claim 1, characterized in that, The multimodal raw data includes: text, tables, images, and rich text.

3. The game account transaction method based on data consistency verification according to claim 1, characterized in that, The preprocessing includes: normalization, feature alignment, and anomaly removal.

4. The game account transaction method based on data consistency verification according to claim 1, characterized in that, The CC standard knowledge graph includes: seven ST modules, SFR / SAR rules, CC standard sentence structure specifications, and hierarchical relationships between the modules.

5. A method for game account trading based on data consistency verification according to claim 1, characterized in that, S4 specifically includes: S401: Perform similarity matching between the standard vector data and the entity nodes in the CC standard knowledge graph to obtain the data module affiliation and hierarchical position corresponding to the standard vector data; S402: Combining the data module affiliation and the hierarchical position, call the built-in reasoning rules in the CC standard knowledge graph to process and bind the standard vector data according to the rules; S403: Based on the binding results of the processing rules, clean the data noise, format conflict data and logical contradiction data in the standard vector data; S404: According to the CC standard specifications in the CC standard knowledge graph, the cleaned standard vector data is standardized by performing a transformation including date, encoding, and sentence structure to obtain the governance data.

6. A method for game account trading based on data consistency verification according to claim 1, characterized in that, The set of key elements includes: business entity, operational behavior, protected object, and scope of authority.

7. A method for game account trading based on data consistency verification according to claim 1, characterized in that, S6 specifically includes: S601: Convert the CC standard full-text sentence library into a vector index library to obtain the CC standard vector sentence library; S602: Based on the set of key elements, in the CC standard vector sentence pattern library, the semantic vector and sentence pattern vector corresponding to the set of key elements are matched with each other using an approximate nearest neighbor vector retrieval algorithm to obtain candidate standard sentence patterns; S603: Determine the optimal standard sentence structure from the candidate standard sentence structures based on sentence structure similarity, module affiliation, and hierarchical position; S604: Merge the set of key elements and the optimal standard sentence structure to obtain the initial standard content; S605: Perform syntax verification and logical calibration on the initial standard content to obtain the standard content that conforms to the CC specification.

8. A method for game account trading based on data consistency verification according to claim 1, characterized in that, Specifically, S7 includes: S701: Construct a tree-structured security target data model based on the standard content conforming to the CC specification and the hierarchical relationship of the CC standard knowledge graph; S702: Bind the tree-structured security target data model with document sorting rules, style rules, and template placeholder mapping relationships to obtain the document data to be rendered; S703: Store the document data to be rendered in both vector and structured formats; S704: Based on the hierarchical relationship of the CC standard knowledge graph, the document data to be rendered is dynamically loaded through a streaming rich document rendering engine; S705: Real-time rendering of multimodal raw data in dynamically loaded document data to be rendered; S706: Automatically generates document table of contents, page numbers, headers, and footers based on real-time rendering results; S707: Combining the document directory, page numbers, header, and footer, output a Word document of security target evidence that conforms to the CC standard by loading, rendering, and outputting simultaneously.

9. A game account trading system based on data consistency verification, characterized in that, include: Processor and memory; The memory stores programs or instructions that can run on the processor, and when the program or instructions are executed by the processor, they implement the steps of the game account transaction method based on data consistency verification as described in any one of claims 1 to 8.

10. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the game account transaction method based on data consistency verification as described in any one of claims 1 to 8.

Images