Data verification method and device of cross-border document, electronic equipment and storage medium

Abstract

The invention relates to the technical field of cross-border document data verification, and discloses a cross-border document data verification method and device, electronic equipment and a storage medium, and the method comprises the steps: extracting a key paragraph of a specified cross-border document; mapping the key paragraph into a preset cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relation is commodity description-coordination system code-local extension code; and verifying the cross-border document based on the three-level mapping relationship. The cross-border document processing method has the advantages that the document processing accuracy is improved, compliance verification becomes more efficient, missing detection and misjudgment caused by manual auditing are reduced, intellectualization, automation and standardization of cross-border document processing are achieved, and higher efficiency and safety are given to global trade.

Description

Technical Field

[0001] The present invention relates to the technical field of data verification of cross-border documents, and in particular to a method, device, electronic device and storage medium for data verification of cross-border documents. Background Art

[0002] Currently, the processing of cross-border warehousing documents primarily relies on manual review. However, due to the complex and ever-changing tariffs, HS codes, and certification requirements across different regions, operators are prone to errors during manual matching and verification, leading to customs clearance delays and increased fines. Manual review also suffers from significant drawbacks such as low efficiency and high error rates. Traditional technologies lack the ability to dynamically integrate and intelligently verify tariff data across regions, and are unable to adapt to tariff changes or regulatory updates in real time, leading to verification errors and increased compliance risks. Summary of the Invention

[0003] Based on this, it is necessary to address the data verification problem of existing cross-border documents and propose a data verification method, device, electronic device and storage medium for cross-border documents.

[0004] A data verification method for a cross-border document, the method comprising: Extract key paragraphs of designated cross-border documents; The key paragraph mapping is preset in the cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-harmonized system code-local extension code; Verifying the cross-border document based on the three-level mapping relationship; Before the step of mapping the key paragraphs into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship, the method further includes: Get tariff data for multiple preset regions; Extracting the preset three-level mapping relationship in the tariff data; The preset cross-border knowledge graph is constructed based on the preset three-level mapping relationship.

[0005] Furthermore, after the step of constructing the preset cross-border knowledge graph based on the preset three-level mapping relationship, the method further includes: Real-time monitoring of the tariff update status of each preset area; If it is monitored that the tariff update status of the target preset area is updated, then the updated tariff of the target preset area is obtained; Based on the updated tariff, the target three-level mapping relationship is extracted and the preset three-level mapping relationship corresponding to the preset cross-border knowledge graph is updated.

[0006] Furthermore, after the step of constructing the preset cross-border knowledge graph based on the preset three-level mapping relationship, the method further includes: Obtaining a mandatory authentication request for each of the preset areas; Obtaining the three-level mapping relationship to be authenticated corresponding to the preset cross-border knowledge graph based on the mandatory authentication request; Determining whether the three-level mapping relationship to be authenticated meets the mandatory authentication request; If the three-level mapping relationship to be authenticated does not satisfy the mandatory authentication request, generating a change prompt based on the three-level mapping relationship to be authenticated and the mandatory authentication request; Obtaining the changed content based on the change prompt; Modify the three-level mapping relationship to be authenticated based on the modified content to form a compliant three-level mapping relationship; The compliant three-level mapping relationship is input into the preset cross-border knowledge graph to update the corresponding three-level mapping relationship to be authenticated.

[0007] Furthermore, after the step of changing the to-be-certified three-level mapping relationship based on the changed content to form a compliant three-level mapping relationship, the method further includes: Acquire the identity of the operator who changes the three-level mapping relationship to be authenticated based on the change content and the change time; Generate a change log based on the change time, operator identity, the to-be-certified three-level mapping relationship, the compliant three-level mapping relationship, and the change prompt; The change log is uploaded to a preset blockchain for storage.

[0008] Furthermore, before the step of extracting the key paragraphs of the designated cross-border document, the method further includes: Collect multiple abnormal scenarios in advance to build an abnormal pattern library; Building a structured rule template based on the abnormal pattern library; Detecting the designated cross-border document based on the structured rule template to obtain non-compliant data and corresponding abnormality levels; Based on the abnormality level and the preset level and data acquisition method correspondence table, obtaining the target data acquisition method; The compliant data corresponding to the non-compliant data is obtained based on the target data acquisition method, and the non-compliant data in the cross-border document is replaced by the compliant data.

[0009] Furthermore, the step of extracting key paragraphs of a designated cross-border document includes: Divide the designated cross-border document into multilingual regions using a preset YOLOv5 model; Obtaining a corresponding OCR processing model according to the language type of each language region; Processing the corresponding language regions through each of the OCR processing models to obtain subtext data corresponding to each language region; The sub-text data are spliced ​​together to obtain the cross-border document text data; The key paragraphs in the cross-border document text data are extracted using a preset large language model.

[0010] Furthermore, before the step of splicing the sub-text data to obtain the cross-border document text data, the process further includes: Each of the sub-text data is input into a preset context-aware model to modify each portion of the sub-text data to obtain modified sub-text data.

[0011] A data verification device for a cross-border document, comprising: The first extraction module is used to extract key paragraphs of a specified cross-border document; A mapping module, configured to map the key paragraphs into a preset cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-Harmonized System code-local extension code; A verification module, configured to verify the cross-border document based on the three-level mapping relationship; The acquisition module is used to obtain tariff data for multiple preset areas; A second extraction module is used to extract the preset three-level mapping relationship in the tariff data; A construction module is used to construct the preset cross-border knowledge graph based on the preset three-level mapping relationship.

[0012] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps: Extract key paragraphs of designated cross-border documents; The key paragraph mapping is preset in the cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-harmonized system code-local extension code; Verifying the cross-border document based on the three-level mapping relationship; Before the step of mapping the key paragraphs into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship, the method further includes: Get tariff data for multiple preset regions; Extracting the preset three-level mapping relationship in the tariff data; The preset cross-border knowledge graph is constructed based on the preset three-level mapping relationship.

[0013] A computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to perform the following steps: Extract key paragraphs of designated cross-border documents; The key paragraph mapping is preset in the cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-harmonized system code-local extension code; Verifying the cross-border document based on the three-level mapping relationship; Before the step of mapping the key paragraphs into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship, the method further includes: Get tariff data for multiple preset regions; Extracting the preset three-level mapping relationship in the tariff data; The preset cross-border knowledge graph is constructed based on the preset three-level mapping relationship.

[0014] The beneficial effects of the present invention are as follows: by extracting key paragraphs of designated cross-border documents and mapping them to a preset cross-border knowledge graph, it ensures that the verification process can quickly obtain the three-level mapping relationship between product descriptions, harmonized system codes and local extension codes, thereby improving the accuracy of document processing and making compliance verification more efficient, reducing missed inspections and misjudgments caused by manual review, and realizing the intelligent, automated and standardized processing of cross-border documents, thus providing higher efficiency and security for global trade. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] in: Figure 1 A diagram illustrating an application environment of a method for verifying data of a cross-border document in one embodiment; Figure 2 Flowchart of a data verification method for a cross-border document in one embodiment; Figure 3 This is a structural block diagram of a data verification device for a cross-border document in one embodiment; Figure 4 FIG. 4 is a structural block diagram of an electronic device in an embodiment. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0018] Figure 1 This is a diagram of the data verification application environment for cross-border documents in an embodiment. Figure 1 This cross-border document data verification method is applied to a cross-border document data verification system. This cross-border document data verification system includes a terminal 110 and a server 120. Terminal 110 and server 120 are connected via a network. Terminal 110 can be a desktop terminal or a mobile terminal. The mobile terminal can be at least one of a mobile phone, tablet computer, or laptop computer. Server 120 can be implemented as a standalone server or a server cluster consisting of multiple servers. Terminal 110 is used to obtain a specified cross-border document, and server 120 is used to verify the cross-border document.

[0019] like Figure 2 As shown, in one embodiment, a method for verifying data of a cross-border document is provided. This method can be applied to both a terminal and a server. This embodiment uses the application of the terminal as an example. The method for verifying data of a cross-border document specifically includes the following steps: S1: Extract key paragraphs of designated cross-border documents; S2: mapping the key paragraphs into a preset cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-Harmonized System code-local extension code; S3: Verify the cross-border document based on the three-level mapping relationship; Before the step S2 of mapping the key paragraphs into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship, the method further includes: S101: Obtain tariff data for multiple preset areas; S102: extracting the preset three-level mapping relationship in the tariff data; S103: Constructing the preset cross-border knowledge graph based on the preset three-level mapping relationship.

[0020] As described in step S1, key paragraphs of the designated cross-border document are extracted. The input cross-border document is parsed to identify important information. Key paragraphs typically include product description, quantity, invoice number, shipper and consignee information, etc. Specifically, optical character recognition (OCR) technology can be used to convert paper or scanned documents into editable text format. In addition, combined with natural language processing (NLP) technology, the system can use specific semantic models, such as named entity recognition (NER) models, to identify and extract key fields related to finance and trade. To ensure the accuracy of the extraction, the recognition results can be post-processed, such as data cleaning and format correction, to improve data consistency.

[0021] As described in step S2 above, the key paragraphs are mapped into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship. After extracting the key paragraphs, this information is mapped into the preset cross-border knowledge graph. The preset cross-border knowledge graph usually contains data structures for relevant data and trade compliance in each region, mainly including a three-level mapping relationship of product description, Harmonized System Code and local extension code. First, the corresponding entry is searched in the knowledge graph based on the extracted product description. This process involves query operations on the knowledge graph and fine-grained comparison to ensure that the HS code (Harmonized System Code) and its local extension code of the corresponding product can be accurately matched. In this mapping process, the matching algorithm can be optimized through machine learning technology to improve accuracy. In addition, in order to cope with the differences in language and regulations between different regions, the system needs to dynamically adapt to the compliance requirements of each region to ensure the effectiveness and timeliness of the mapping relationship.

[0022] As described in step S3 above, the cross-border document is verified based on the three-level mapping relationship. After completing the acquisition of the three-level mapping relationship, the system will verify the cross-border document based on these relationships. The pre-set compliance logic is compared with the extracted data to detect whether the key information in the document complies with the tariff requirements of different regions. If the product description and the corresponding HS code and local extension code in the document can correctly match the relevant laws and regulations, the system will determine that the document is valid and enter the next processing link. If any discrepancy is found, the system will automatically generate an early warning message to prompt relevant personnel to make supplements or corrections.

[0023] As described in step S101 above, tariff data for multiple preset areas is obtained. In actual applications, in order to ensure accurate verification of cross-border documents, the system first needs to obtain the latest tariff data for multiple preset areas. When the tariff data is updated, it triggers the automatic update of the knowledge graph, which involves collecting relevant information from official data agencies or trade databases in each region. This process can be specifically done through an API interface, or by regularly downloading relevant information from an officially released database. These tariff data should include the latest HS codes, commodity classifications, tariff rates, etc. In addition, the system can also format and normalize the acquired data so that the subsequent steps can effectively utilize this information. In this way, a comprehensive and timely updated tariff data foundation is established to ensure sufficient data support for the subsequent knowledge graph construction.

[0024] As described in step S102 above, the pre-set three-level mapping relationships in the tariff data are extracted. After obtaining the tariff data, the system then needs to extract the required pre-set three-level mapping relationships. Specifically, this involves parsing the tariff data obtained from various regions and identifying structured information such as product descriptions, corresponding Harmonized System codes, HS codes, and their local extensions. Text mining and data column extraction techniques, combined with regular expressions or pattern matching algorithms, can be used to rapidly extract and classify the required information.

[0025] As described in step S103 above, the preset cross-border knowledge graph is constructed based on the preset three-level mapping relationship. The preset cross-border knowledge graph will serve as the data basis of the entire verification system, and perform structured modeling on the relationship between various commodities and their applicable regional tariffs. During the construction process, graph database technology may be used to store data in the form of nodes and edges, where nodes represent commodity descriptions, HS codes, local extension codes, etc., and edges represent the relationship connections between elements. This process needs to take into account flexibility and scalability so that it can adapt to the addition of new regional tariffs or commodity classifications in the future. At the same time, in order to improve query efficiency, the design of the graph also needs to comply with the optimization strategy of data indexing and retrieval, and improve the response speed of subsequent queries by establishing an effective index structure. Ultimately, the formed cross-border knowledge graph will provide real-time and accurate data support for the intelligent verification of subsequent documents.

[0026] In one embodiment, after the step S103 of constructing the preset cross-border knowledge graph based on the preset three-level mapping relationship, the method further includes: S1041: Real-time monitoring of the tariff update status of each of the preset areas; S1042: If it is monitored that the tariff update status of the target preset area is updated, then obtain the updated tariff of the target preset area; S1043: Extracting a target three-level mapping relationship based on the updated tariff and updating the preset three-level mapping relationship corresponding to the preset cross-border knowledge graph.

[0027] As described in steps S1041-S1043 above, a real-time monitoring mechanism is established to continuously track and obtain tariff updates for multiple pre-defined regions. This process can be implemented in a variety of ways. For example, the system can regularly access the official customs or relevant data bureau websites of each region to review the latest announcements and dynamic information. To improve monitoring efficiency, an API interface can be used to automatically obtain tariff data for each region. In addition to directly retrieving web information, the system can also receive notifications of relevant updates through email subscriptions or RSS feeds. Furthermore, databases published by each region can also serve as data sources. The frequency of monitoring updates typically depends on the characteristics of tariff changes in each region. Therefore, the frequency of inspections can be customized for different regions to more efficiently capture change information. Through this real-time monitoring mechanism, the system can ensure that it has the latest tariff developments and be prepared for subsequent updates, thereby reducing compliance risks and the possibility of misjudgment caused by tariff changes.

[0028] Specific target regions associated with monitored tariff updates are identified. When the tariff status for a region is confirmed to be "updated," the system automatically initiates a process to retrieve the latest tariff data. This process typically involves invoking a pre-defined API or data scraping module to access the latest tariff information provided by that country. The retrieved data format must be compatible with the system's internal data structure for efficient parsing and storage. Furthermore, when retrieving updated tariff information, the system must ensure data integrity and accuracy. This includes comparing changes between old and new data and reviewing specific legal documents or announcements to ensure that the updated information is valid and meets compliance requirements. For some regions, updated tariff information may contain multiple sub-items or appendices, requiring the use of pre-defined data analysis models to extract these contents and integrate them into existing data tables. Using the newly acquired updated tariff information, new three-level mapping relationships are extracted and updated accordingly with the existing cross-border knowledge graph. First, the system parses the updated tariff data to identify and extract the required three-level mapping relationships. While extracting the new mapping relationships, the system also integrates them with the existing knowledge graph to determine which mappings need to be updated. For example, a product's HS code may be replaced by a new classification in a new tariff schedule, a change that directly impacts the graph structure. Once these changes are made, the graph is updated to reflect the new tariff information. By systematically monitoring the status of tariff updates in each region, acquiring new tariffs, and extracting the corresponding three-level mapping relationships, the entire verification process remains up-to-date and compliant, thereby improving the accuracy of cross-border document processing and mitigating compliance risks.

[0029] In one embodiment, after the step S103 of constructing the preset cross-border knowledge graph based on the preset three-level mapping relationship, the method further includes: S1141: Obtaining a mandatory authentication request for each of the preset areas; S1142: Obtaining the to-be-authenticated three-level mapping relationship corresponding to the preset cross-border knowledge graph based on the mandatory authentication request; S1143: Determine whether the three-level mapping relationship to be authenticated meets the mandatory authentication request; S1144: If the three-level mapping relationship to be authenticated does not satisfy the mandatory authentication request, generating a change prompt based on the three-level mapping relationship to be authenticated and the mandatory authentication request; S1145: Acquire the changed content based on the change prompt; S1146: Modify the to-be-certified three-level mapping relationship based on the modified content to form a compliant three-level mapping relationship; S1147: Input the compliance three-level mapping relationship into the preset cross-border knowledge graph to update the corresponding three-level mapping relationship to be authenticated.

[0030] As described in step S1141 above, it is necessary to identify and collect mandatory certification requests for specific commodities in each pre-defined region. This involves collecting relevant regulations and requirements from official channels of regulatory agencies or trade management agencies in each region. It should be noted that tariff data refers to relevant information stored in official data agencies or trade databases in each region, including the latest HS codes, commodity classifications, tariff rates, and so on. Compulsory certification requests, on the other hand, are mandatory rules set by each region and are not reflected in the tariff data. For example, region A may require that imported electronic products of a certain type must meet rule A, while region B may require that products meet rule B. Such certification requests typically include detailed regulations and required documentation types, such as product manuals and safety test reports. Obtaining information on these mandatory certification requests requires a dynamic update mechanism that can promptly reflect changes in regional policies and ensure that the certification requests used by the system are up to date. Since mandatory certification requests are generally published through regional websites, industry associations, or international trade organizations, the system can configure an API interface or regularly access relevant websites to automatically extract relevant data. Furthermore, the acquired data needs to be standardized to ensure that certification request information from different regions enters the subsequent processing steps in the same format. This process ensures that in subsequent steps, the products to be certified can be effectively matched with the corresponding certification requirements, reducing compliance risks caused by incomplete or inaccurate information.

[0031] As described in step S1142 above, based on the mandatory certification request, the three-level mapping relationship to be certified corresponding to the preset cross-border knowledge graph is obtained. The obtained mandatory certification request is matched with the three-level mapping relationship to be certified in the preset cross-border knowledge graph, which contains three-level mapping information such as product description, harmonized system, and local extension code, to extract information related to the mandatory certification request. For example, the system needs to identify which product categories meet the mandatory certification request in a specific region and then obtain the corresponding three-level mapping relationship. Specifically, the knowledge graph can be queried to extract the necessary information based on the description of the certification request. This process may also involve the use of natural language processing (NLP) technology to ensure that the system can understand the complexity of the mandatory certification request and accurately match the relationship information. For example, the system may refine the broad category of "electronic devices" to determine which of all devices need to meet specific certification conditions. In this way, the three-level mapping relationship to be certified can be clearly identified, providing a basis for subsequent compliance judgment and potential changes.

[0032] As described in step S1143 above, determine whether the three-level mapping relationship to be certified meets the mandatory certification request. Evaluate whether the three-level mapping relationship to be certified meets the corresponding mandatory certification request, and the judgment process includes comparing the description of the product to be certified, the HS code and its local extension code to see whether they are consistent with the requirements in the mandatory certification request. For example, if the HS code of a product requires a specific certification, but the mapping relationship to be certified for the product does not contain the certification information, the system will determine that the relationship to be certified does not meet the requirements. Generally speaking, the system will use conditional judgment logic to check each field in the relationship to be certified one by one to confirm its integrity and compliance. In order to improve the accuracy of the judgment, the system can also integrate a dynamic rule engine to analyze and apply the compliance logic and certification requirements of different regions to ensure a comprehensive evaluation of the relationship to be certified. If it is found that certain requirements are missing or mismatched during this process, the system will issue a corresponding alarm and start the subsequent change prompt generation step to ensure that compliance issues can be identified and handled in a timely manner.

[0033] As described in step S1144 above, if the three-level mapping relationship to be authenticated does not meet the mandatory authentication request, a change prompt is generated based on the three-level mapping relationship to be authenticated and the mandatory authentication request. When the system recognizes that certain conditions in the authentication request are not met, a change prompt is automatically generated. These prompts will indicate which specific requirements are not met, thereby reminding relevant personnel to make necessary adjustments and supplements. The process of generating change prompts is to compare the relationship to be authenticated with the mandatory authentication request and identify missing or incorrect fields. For example, if the authentication request requires the provision of individual compliance documents, and the document type is not listed in the relationship to be authenticated, the system will automatically generate a prompt, requiring the user to supplement the "missing M code". These change prompts will be presented in a user-friendly format, such as displaying each unmet compliance item in a list format, to ensure that relevant personnel can quickly understand and take appropriate measures. This process not only improves the intelligence and response speed of the system, but also reduces the burden of manual verification, ensuring that compliance issues can be handled quickly, thereby ensuring the smooth progress of the entire cross-border transaction.

[0034] As described in step S1145 above, the change content is obtained based on the change prompt. According to the generated change prompt, the specific change content is further obtained. These change contents will include all the information and data that need to be supplemented, and clearly show the specific items that need to be corrected in the three-level mapping relationship to be certified. For example, if the system finds that the relationship to be certified lacks a specific certification document, such as "missing C field", this step will require the user to provide the corresponding certification information. In order to obtain these change contents, the system may first analyze each prompt in detail to ensure that the system understands the type of missing information and its importance. This process may include calling the API of the relevant database or regulatory agency to review the required document list or requirements. In addition, the system may also prompt the user to obtain the necessary documents from formal channels, such as sending a prompt requiring the user to upload a new certification code, compliance document or other necessary supporting materials. The accuracy and comprehensiveness of this process are also necessary conditions to ensure the smooth completion of subsequent correction operations, avoiding correction failures and subsequent verification delays due to incomplete information.

[0035] As described in step S1146 above, the three-level mapping relationship to be certified is changed based on the change content to form a compliant three-level mapping relationship. Using the obtained change content, the three-level mapping relationship to be certified is modified as necessary to ensure that it complies with all corresponding mandatory certification requests. This process not only involves a simple replacement of data, but may also require the regeneration of the entire mapping relationship based on the new information. For example, if the relationship to be certified is missing the M code, it is added to the relevant field. After completing this change, the system must ensure that the format and structure of all updated data still meet the preset standards so that it can continue to be effectively used in subsequent steps. In addition, the system can also trigger an automatic review mechanism to ensure that the converted compliant three-level mapping relationship complies with the latest laws and regulations, further reducing compliance risks. Through this process, the final compliant three-level mapping relationship will be considered to be in a verified state to ensure that the same compliance issues will not be faced in future trade activities.

[0036] As described in step S1147 above, the compliance three-level mapping relationship is input into the preset cross-border knowledge graph to update the corresponding three-level mapping relationship to be certified. The modified compliance three-level mapping relationship will be re-entered into the preset cross-border knowledge graph to replace the original three-level mapping relationship to be certified. This update process is to maintain the corrected data status in the knowledge graph so that the entire system can operate effectively based on the latest version of information. The update process usually needs to ensure the compatibility and correctness of the data so that the resulting mapping relationship can adapt to future query and verification needs. The verification process may include a series of automatically triggered data integrity checks to ensure that the new data meets the format and logic requirements. In order to ensure continuous monitoring of compliance, the process can also record all change operations for future audits and compliance checks. The new compliance three-level mapping relationship can provide real-time compliance support for the system's subsequent document processing, ensuring that in cross-border transactions, the relevant goods always meet the legal and regulatory requirements of each region, thereby reducing customs clearance risks, improving the efficiency of overall transactions, and providing comprehensive information support for intelligent verification and compliance management of cross-border documents.

[0037] In one embodiment, after step S1146 of modifying the to-be-certified three-level mapping relationship based on the modified content to form a compliant three-level mapping relationship, the method further includes: S11471: Obtaining the identity of the operator who changes the three-level mapping relationship to be authenticated based on the change content and the change time; S11472: Generate a change log based on the change time, operator identity, the three-level mapping relationship to be authenticated, the compliant three-level mapping relationship, and the change prompt; S11473: Upload the change log to a preset blockchain for storage.

[0038] As described in steps S11471-S11473 above, the identity of the operator who performed the change and the time of the operation are recorded. This information is not only collected for audit and compliance purposes, but also helps track change records, ensuring that any data adjustments or modifications can be audited and investigated. The operator identity typically includes a unique identifier (such as a user ID) and associated role (such as auditor or administrator), which helps identify the responsible party for performing the operation during subsequent inquiries. The change time ensures that all modifications are clearly timestamped, creating a timeline of change records and facilitating analysis of operations and decision-making processes during subsequent compliance audits. This information may be automatically extracted through the system's user authentication module (such as SSO single sign-on) and stored in the data change record. A detailed change log is generated by integrating this collected information. This log should include: change time, operator identity, the original three-level mapping relationship to be authenticated, the updated three-level mapping relationship to be compliant, and the change notification content. The purpose of generating a change log is to create an audit-friendly, transparent record that details the process and context of data modifications, facilitating subsequent compliance verification and issue tracing. When generating a change log, this information needs to be formatted into standardized text or structured data for easy storage and retrieval. Furthermore, this log must adhere to certain security standards to prevent unauthorized users from altering or deleting it, thereby reducing the risk of data tampering. Effective change log management is not only crucial for ensuring data transparency but also crucial for meeting relevant compliance requirements. The generated change log is uploaded to a pre-defined blockchain for persistent storage. Leveraging the immutability of blockchain technology, all change records, once added to the blockchain, cannot be modified or deleted, ensuring data integrity and reliability. Uploading the change log to the blockchain typically involves encrypting the log data to ensure only authorized personnel can access the relevant information. Blockchain technology can also be used to manage access rights. The upload process can be implemented through smart contracts, ensuring that uploads are automatically executed when conditions are met. Each record in the blockchain is timestamped and contains the operator's identity, ensuring that each change can be traced back to the responsible individual and the specific time. This process not only enhances data security but also meets relevant requirements, significantly impacting the legitimacy of audit roles and the transparency of data processing. In short, uploading the change log to the blockchain is an important step in ensuring the compliance and security of cross-border document data processing, and constitutes a trusted audit chain for the entire system.

[0039] In one embodiment, before step S1 of extracting the key paragraphs of the designated cross-border document, the method further includes: S001: Collect multiple abnormal scenarios in advance to build an abnormal pattern library; S002: Constructing a structured rule template based on the abnormal pattern library; S003: Detecting the designated cross-border document based on the structured rule template to obtain non-compliant data and corresponding abnormality levels; S004: Obtaining a target data acquisition method based on the abnormality level and the preset level and data acquisition method correspondence table; S005: Acquire compliant data corresponding to the non-compliant data based on the target data acquisition method, and replace the non-compliant data in the cross-border document with the compliant data.

[0040] As described in steps S001-S003 above, extensive data analysis and research are conducted to collect common exception scenarios in cross-border documents, for example, over 300 common exception scenarios. This process can be achieved by analyzing historical data, audit records, and business feedback, with the goal of identifying common errors, compliance issues, and legal risks that arise during document processing. By analyzing the characteristics of non-compliant documents, the system can effectively identify key exception scenarios, such as "HS code error" or "inconsistency between invoice amount and customs declaration." A comprehensive exception pattern library is then established, storing various types of exception samples and providing various attributes for subsequent analysis and classification. This data can be compiled from historical transactions and information collection systems, or referenced by regional or industry standards to ensure the comprehensiveness and specificity of the exception pattern library. The purpose of establishing the exception pattern library is to enhance the system's intelligent recognition capabilities, enabling it to identify potential issues in data processing through well-defined exception patterns, thereby enabling more automated compliance verification and intelligent decision-making.

[0041] The collected exception scenarios are converted into structured rule templates. The purpose of this is to represent complex exception patterns in a standardized format to facilitate subsequent detection and verification by the system. During the design process of structured rule templates, a unified format and syntax can be adopted to make them highly readable and maintainable. At the same time, they must also be compatible with the mapping relationship of the knowledge graph to ensure that they can effectively support subsequent data detection and verification work in practical applications. Based on the previously established structured rule templates, automated compliance testing is performed on designated cross-border documents. First, the system needs to extract relevant field information (such as product description, quantity, HS code, etc.) from the document, and then use the rule engine to apply each preset rule template to determine whether the extracted data complies with relevant laws and regulations and internal company policies.

[0042] During the detection process, the system will record the data items corresponding to the defects and assign an anomaly level to each non-compliance issue. The anomaly level is usually divided based on the severity of the violation, such as "minor violation", "medium violation", "serious violation", etc. This evaluation process may take into account multiple factors, such as the impact of the violation on the compliance of the relevant data, the legal risks involved, and the frequency of occurrence of the error. The purpose of recording these non-compliant data and anomaly levels is to take appropriate corrective measures and monitor risks later. Automated detection not only improves audit efficiency, but also ensures the consistency of the screening process and reduces subjective bias in manual operations. By identifying and recording non-compliant data and its anomaly level, the system lays the foundation for subsequent response measures and data correction.

[0043] As described in steps S004-S005 above, the optimal method for obtaining compliant data is determined based on the anomaly level of the acquired non-compliant data, in conjunction with a pre-defined table of anomaly levels and data acquisition methods. This table represents a rule-based mapping relationship that defines the data correction strategy and data source required for correcting a given anomaly level. For example, if the anomaly level is determined to be simple (e.g., a formatting error), the system automatically corrects the issue; however, if a complex issue (e.g., a cross-border HS code conflict) is identified, manual review and recording is performed in the case database. This approach allows the system to flexibly select the appropriate data acquisition method (e.g., querying the business database, contacting external partners to obtain certificates or reports, etc.) based on the specific anomaly type and severity, facilitating efficient subsequent corrective measures. Compliant data corresponding to the non-compliant data is obtained based on the determined target data acquisition method. For example, if the HS code in a cross-border document is marked as "incorrect," but the correct HS code is obtained as a replacement through external query, the system will perform the replacement operation for the non-compliant item. Specifically, the system will interact with relevant data sources to obtain compliance information, such as contacting relevant certification bodies to obtain updated compliance documents, or retrieving verified data from internal record libraries. At the same time, the system also needs to perform data verification to ensure that the compliant data obtained can effectively replace the original non-compliant data. After the data replacement is completed, the system will update the document content and record all operations that replace non-compliant data with compliant data to facilitate subsequent audits and verifications. Through this step, the system can not only ensure the compliance of the documents, but also lay a solid foundation for the effectiveness of subsequent processing. This enables cross-border documents to basically achieve intelligent and automated management while complying with international trade regulations and requirements, enhancing the stability and transparency of the entire process.

[0044] In one embodiment, step S1 of extracting key paragraphs of a designated cross-border document includes: S111: Divide the designated cross-border document into multilingual regions using a preset YOLOv5 model; S112: Obtaining a corresponding OCR processing model according to the language type of each language area; S113: Processing the corresponding language region through each of the OCR processing models to obtain subtext data corresponding to each language region; S114: Joining the sub-text data to obtain cross-border document text data; S115: Extract key paragraphs from the cross-border document text data using a preset large language model.

[0045] As described in steps S111-S113 above, a preset YOLOv5 model is used to perform object detection on designated cross-border documents and identify multilingual regions. YOLOv5 is a highly efficient object detection algorithm. Its trained model can quickly locate different objects in an image. The model was trained using a dataset containing 100,000 annotated images of multilingual documents. In the cross-border document scenario, objects include text blocks in different languages ​​and various graphic elements (such as seals, barcodes, and icons). First, the system passes the input document image to the YOLOv5 model for processing. The model extracts features using its convolutional neural network (CNN) architecture and identifies text regions in the image based on existing training data. For each identified region, the model not only returns the location (identified box) but also provides a confidence score to determine whether it is the target region. After detection, the system labels the document based on the recognition results, dividing it into different language regions. The system analyzes the language type of each identified text region and selects an appropriate optical character recognition (OCR) processing model for each region. A multilingual OCR model library can be predefined, each optimized for processing text in a specific language. This automatically separates languages ​​and performs OCR by region for F-documents, integrating variant recognition models. Generally, languages ​​written left to right may utilize specialized OCR models, while mainstream languages ​​can use common OCR tools such as Tesseract. The system selects and loads the appropriate OCR processing model based on the characteristics of the recognized text region and language identifiers (e.g., Unicode range, characteristic character set). The primary task of OCR technology is to convert character information in an image into editable text. During processing, the system passes the image of each language region to the corresponding OCR model for parsing. This process generates text blocks, extracting key information such as invoice numbers, product names, and prices. During execution, the system also performs post-processing steps, such as text correction and formatting, to eliminate errors that may have occurred during the OCR process (e.g., character recognition confusion, spelling errors, etc.). The resulting subtext data is stored in a structured format for subsequent analysis and processing.

[0046] As described in S114-S115 above, the previously extracted subtext data from various language regions is integrated and spliced ​​into a complete cross-border document text data set. The system needs to arrange the splicing order of the subtexts based on the text differentiation between different languages ​​and the layout of the original document. During splicing, the text can be connected based on preset templates or rules, or directly according to the order of the identified regions, ensuring that the final text maintains a natural order. To consider the contextual structure of the text, the system should also check the grammatical and logical integrity of the splicing process during this process to ensure that the final result conforms to linguistic conventions. After splicing is completed, the synthesized cross-border document text data will serve as the basis for subsequent key paragraph extraction. The pre-set large language model (LLM) will be used to process the integrated cross-border document text data to extract key paragraphs and information. The key to this process is to effectively utilize the LLM's natural language understanding capabilities to enable it to identify key information from complex, structured text. Large language models, such as XLM-Roberta or mT5, have demonstrated excellent performance in processing multilingual data and can adapt to the contextual semantics of different languages. Given a given text input, LLM returns an in-depth analysis of the text and can extract specific key pieces of information (such as invoice information, product numbers, certification requirements, etc.) based on the specified task. This process can often be precisely controlled by defining specific extraction rules or task objectives. For example, the model can be instructed to identify and extract all paragraphs relevant to commodity customs clearance and return these paragraphs and their content. This ensures the accurate extraction of important information and provides a reliable basis for subsequent compliance checks, data verification, and processing, ensuring efficient and compliant cross-border document processing. Overall, these steps complement each other, from identifying different languages ​​to extracting key information, greatly improving cross-border document processing capabilities.

[0047] In one embodiment, before the step S114 of concatenating the sub-text data to obtain the cross-border document text data, the following steps are further included: S1131: Input each of the sub-text data into a preset context-aware model to correct each portion of the sub-text data to obtain corrected sub-text data.

[0048] As described in step S1131 above, each extracted subtext data item is input into a pre-set context-aware model to correct various portions of the subtext data. First, the text data entering the context-aware model is the original text extracted during the OCR process. During the conversion process, these texts may have spelling errors or incorrect formatting due to character recognition errors or fuzzy image quality. By analyzing these individual text segments and leveraging pre-trained language understanding capabilities, the text is corrected. For example, when processing a specific language region, the model can understand the grammatical rules of that language, thereby identifying incorrect vocabulary and correcting it. Similarly, for other languages ​​with unique grammatical logic, the model can also provide appropriate correction suggestions based on the grammar and context. This processing is not limited to word-level corrections but also includes optimizing the entire sentence structure. For example, if a subtext paragraph contains logical inconsistencies within the context, the model will provide reasonable correction suggestions based on its understanding of the surrounding context to ensure the semantic coherence of the entire paragraph. After the correction is completed, the corrected subtext data is stored to replace the original subtext, providing a more reliable text data foundation for the subsequent splicing process. The automation and intelligence of this process improves the efficiency of cross-border document processing and reduces data compliance risks caused by text errors.

[0049] Reference Figure 3 The present invention also provides a data verification device for cross-border documents, the device comprising: The first extraction module 902 is used to extract key paragraphs of a specified cross-border document; A mapping module 904 is configured to map the key paragraphs into a preset cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-Harmonized System code-local extension code; A verification module 906, configured to verify the cross-border document based on the three-level mapping relationship; An acquisition module 908 is used to acquire tariff data of multiple preset areas; The second extraction module 910 is used to extract the preset three-level mapping relationship in the tariff data; Construction module 912 is used to construct the preset cross-border knowledge graph based on the preset three-level mapping relationship.

[0050] In one embodiment, the cross-border document data verification device further includes: A tariff update status monitoring module, configured to monitor the tariff update status of each of the preset areas in real time; An updated tariff acquisition module is configured to acquire the updated tariff of a target preset area if it is detected that the tariff update status of the target preset area is updated; A preset three-level mapping relationship updating module is used to extract the target three-level mapping relationship based on the updated tariff and update the preset three-level mapping relationship corresponding to the preset cross-border knowledge graph.

[0051] In one embodiment, the cross-border document data verification device further includes: A compulsory authentication request acquisition module, configured to acquire a compulsory authentication request for each of the preset areas; A module for obtaining a three-level mapping relationship to be authenticated, configured to obtain a three-level mapping relationship to be authenticated corresponding to the preset cross-border knowledge graph based on the mandatory authentication request; A module for determining the three-level mapping relationship to be authenticated, configured to determine whether the three-level mapping relationship to be authenticated satisfies the mandatory authentication request; a change prompt generating module, configured to generate a change prompt based on the three-level mapping relationship to be authenticated and the compulsory authentication request if the three-level mapping relationship to be authenticated does not satisfy the compulsory authentication request; A change content acquisition module, configured to acquire the change content based on the change prompt; A module for changing the three-level mapping relationship to be authenticated, configured to modify the three-level mapping relationship to be authenticated based on the modified content to form a compliant three-level mapping relationship; The compliance three-level mapping relationship input module is used to input the compliance three-level mapping relationship into the preset cross-border knowledge graph to update the corresponding three-level mapping relationship to be authenticated.

[0052] In one embodiment, the cross-border document data verification device further includes: a change time acquisition module, configured to acquire the identity of an operator who changes the three-level mapping relationship to be authenticated based on the change content and the change time; A change log generation module, configured to generate a change log based on the change time, the operator identity, the three-level mapping relationship to be authenticated, the three-level mapping relationship to be compliant, and the change prompt; The change log saving module is used to upload the change log to a preset blockchain for storage.

[0053] In one embodiment, the cross-border document data verification device further includes: An abnormal pattern library construction module is used to collect multiple abnormal scenarios in advance to build an abnormal pattern library; A structured rule template building module, used to build a structured rule template based on the abnormal pattern library; A designated cross-border document detection module is used to detect the designated cross-border document based on the structured rule template to obtain non-compliant data and corresponding abnormality levels; a target data acquisition method acquisition module, configured to acquire a target data acquisition method based on the abnormality level and a table corresponding to a preset level and a data acquisition method; The non-compliant data replacement module is used to obtain compliant data corresponding to the non-compliant data based on the target data acquisition method, and replace the non-compliant data in the cross-border document with the compliant data.

[0054] In one embodiment, the first extraction module 902 includes: A multilingual region division submodule, configured to divide the designated cross-border document into multilingual regions using a preset YOLOv5 model; An OCR processing model acquisition submodule is used to acquire a corresponding OCR processing model according to the language type of each language region; A language region processing submodule, configured to process the corresponding language region through each of the OCR processing models to obtain subtext data corresponding to each language region; The sub-text data splicing sub-module is used to splice the sub-text data to obtain the cross-border document text data; The key paragraph extraction submodule is used to extract key paragraphs from the cross-border document text data through a preset large language model.

[0055] In one embodiment, the first extraction module 902 further includes: The subtext data correction submodule is used to input each subtext data into a preset context-aware model to correct each part of the content in the subtext data to obtain corrected subtext data.

[0056] Figure 4 FIG1 shows an internal structure diagram of an electronic device in an embodiment. The electronic device may be a terminal, a server, or a computer device. Figure 4 As shown, the electronic device includes a processor, a memory, and a network interface connected via a system bus. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the electronic device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement a data verification method for cross-border documents. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can implement a data verification method for cross-border documents. It will be understood by those skilled in the art that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the electronic device to which the solution of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0057] In one embodiment, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps: Extract key paragraphs of designated cross-border documents; The key paragraph mapping is preset in the cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-harmonized system code-local extension code; Verifying the cross-border document based on the three-level mapping relationship; Before the step of mapping the key paragraphs into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship, the method further includes: Get tariff data for multiple preset regions; Extracting the preset three-level mapping relationship in the tariff data; The preset cross-border knowledge graph is constructed based on the preset three-level mapping relationship.

[0058] By extracting key paragraphs of designated cross-border documents and mapping them to a preset cross-border knowledge graph, the verification process is ensured to quickly obtain the three-level mapping relationship between product descriptions, harmonized system codes, and local extension codes, thereby improving the accuracy of document processing and making compliance verification more efficient. This reduces missed inspections and misjudgments caused by manual review, and realizes the intelligent, automated, and standardized processing of cross-border documents, bringing higher efficiency and security to global trade.

[0059] In one embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the processor performs the following steps: Extract key paragraphs of designated cross-border documents; The key paragraph mapping is preset in the cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-harmonized system code-local extension code; Verifying the cross-border document based on the three-level mapping relationship; Before the step of mapping the key paragraphs into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship, the method further includes: Get tariff data for multiple preset regions; Extracting the preset three-level mapping relationship in the tariff data; The preset cross-border knowledge graph is constructed based on the preset three-level mapping relationship.

[0060] By extracting key paragraphs of designated cross-border documents and mapping them to a preset cross-border knowledge graph, the verification process is ensured to quickly obtain the three-level mapping relationship between product descriptions, harmonized system codes, and local extension codes, thereby improving the accuracy of document processing and making compliance verification more efficient. This reduces missed inspections and misjudgments caused by manual review, and realizes the intelligent, automated, and standardized processing of cross-border documents, bringing higher efficiency and security to global trade.

[0061] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0062] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0063] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A data verification method for cross-border documents, characterized in that: The method comprises: Extract key paragraphs of designated cross-border documents; The key paragraph mapping is preset in the cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-harmonized system code-local extension code; Verifying the cross-border document based on the three-level mapping relationship; Before the step of mapping the key paragraphs into the preset cross-border knowledge graph to obtain the corresponding three-level mapping relationship, the method further includes: Get tariff data for multiple preset regions; Extracting the preset three-level mapping relationship in the tariff data; The preset cross-border knowledge graph is constructed based on the preset three-level mapping relationship.

2. The cross-border document data verification method according to claim 1, characterized in that: After the step of constructing the preset cross-border knowledge graph based on the preset three-level mapping relationship, the method further includes: Real-time monitoring of the tariff update status of each preset area; If it is monitored that the tariff update status of the target preset area is updated, then the updated tariff of the target preset area is obtained; Based on the updated tariff, the target three-level mapping relationship is extracted and the preset three-level mapping relationship corresponding to the preset cross-border knowledge graph is updated.

3. The data verification method for cross-border documents according to claim 1 is characterized in that: After the step of constructing the preset cross-border knowledge graph based on the preset three-level mapping relationship, the method further includes: Obtaining a mandatory authentication request for each of the preset areas; Obtaining the three-level mapping relationship to be authenticated corresponding to the preset cross-border knowledge graph based on the mandatory authentication request; Determining whether the three-level mapping relationship to be authenticated meets the mandatory authentication request; If the three-level mapping relationship to be authenticated does not satisfy the mandatory authentication request, generating a change prompt based on the three-level mapping relationship to be authenticated and the mandatory authentication request; Obtaining the changed content based on the change prompt; Modify the three-level mapping relationship to be authenticated based on the modified content to form a compliant three-level mapping relationship; The compliant three-level mapping relationship is input into the preset cross-border knowledge graph to update the corresponding three-level mapping relationship to be authenticated.

4. The data verification method for cross-border documents according to claim 3 is characterized in that: After the step of modifying the to-be-certified three-level mapping relationship based on the modified content to form a compliant three-level mapping relationship, the method further includes: Acquire the identity of the operator who changes the three-level mapping relationship to be authenticated based on the change content and the change time; Generate a change log based on the change time, operator identity, the to-be-certified three-level mapping relationship, the compliant three-level mapping relationship, and the change prompt; The change log is uploaded to a preset blockchain for storage.

5. The data verification method for cross-border documents according to claim 1, characterized in that: Before the step of extracting the key paragraphs of the designated cross-border document, the method further includes: Collect multiple abnormal scenarios in advance to build an abnormal pattern library; Building a structured rule template based on the abnormal pattern library; Detecting the designated cross-border document based on the structured rule template to obtain non-compliant data and corresponding abnormality levels; Based on the abnormality level and the preset level and data acquisition method correspondence table, obtaining the target data acquisition method; The compliant data corresponding to the non-compliant data is obtained based on the target data acquisition method, and the non-compliant data in the cross-border document is replaced by the compliant data.

6. The data verification method for cross-border documents according to claim 1, characterized in that: The steps of extracting key paragraphs of a designated cross-border document include: Divide the designated cross-border document into multilingual regions using a preset YOLOv5 model; Obtaining a corresponding OCR processing model according to the language type of each language region; Processing the corresponding language regions through each of the OCR processing models to obtain subtext data corresponding to each language region; The sub-text data are spliced ​​together to obtain the cross-border document text data; The key paragraphs in the cross-border document text data are extracted using a preset large language model.

7. The cross-border document data verification method according to claim 6, characterized in that: Before the step of splicing the sub-text data to obtain the cross-border document text data, the method further includes: Each of the sub-text data is input into a preset context-aware model to modify each portion of the sub-text data to obtain modified sub-text data.

8. A data verification device for cross-border documents, characterized in that: The device comprises: The first extraction module is used to extract key paragraphs of a specified cross-border document; A mapping module, configured to map the key paragraphs into a preset cross-border knowledge graph to obtain a corresponding three-level mapping relationship; wherein the three-level mapping relationship is product description-Harmonized System code-local extension code; A verification module, configured to verify the cross-border document based on the three-level mapping relationship; The acquisition module is used to obtain tariff data for multiple preset areas; A second extraction module is used to extract the preset three-level mapping relationship in the tariff data; A construction module is used to construct the preset cross-border knowledge graph based on the preset three-level mapping relationship.

9. A computer-readable storage medium, characterized in that A computer program is stored, and when the computer program is executed by a processor, the processor executes the steps of the data verification method for cross-border documents as claimed in any one of claims 1 to 7.

10. An electronic device, characterized in that: The device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the data verification method for cross-border documents as claimed in any one of claims 1 to 7.

Images