A customs service data structured management method and device, computer equipment group and storage medium

Abstract

The application relates to the technical field of computer application, and discloses a customs service data structured management method and device, a computer equipment group and a storage medium. In the method, the edge computing idea is used to process customs service data for the first time, terminal data collection and preprocessing can realize the de-redundancy entry of batch customs data to be inspected, the risk control type can be automatically judged, the risk identification result can be returned, refined effective information can be uploaded to a fusion result database, and a hot deployment synchronous maintenance service is used, so that the network bandwidth uploaded by each terminal data source is reduced, the processing pressure of a customs data center server is reduced, the management efficiency and precision of a customs service data system are improved, high-risk goods and illegal goods formed by high-risk goods and illegal goods can be timely and accurately blocked, the human cost of inspection at a customs port site is reduced, and efficient fusion of customs service data is realized.

Description

Technical Field

[0001] This invention relates to the field of computer application technology, and in particular to a method, apparatus, computer equipment group, and storage medium for structured management of customs business data. Background Technology

[0002] Customs data refers to data collected, stored, used, processed, transmitted, provided, and disclosed by customs during the performance of its duties, encompassing its entire lifecycle. It is a product of the development of customs' technological productivity. With economic globalization and the surge in international trade, the traditional manual processing of customs data faces challenges due to the diverse types, large quantities, and high complexity of the business data that the customs risk assessment system needs to handle. These challenges are characterized by cumbersome procedures, slow response times, low data reuse rates, and inconvenient storage. Therefore, accelerating digital development is an inevitable choice for customs to improve its overall capabilities.

[0003] Currently, there are many methods for structured management of customs business data, but they all focus on the lifecycle management of individual data entries, and have the following shortcomings: First, the customs declaration form recognition system is difficult to meet the requirements of fast customs clearance and risk control in terms of efficiency, accuracy, and load capacity. Due to differences in trade standards among countries and improper storage of unstructured customs data, it is difficult to identify the data. The customs declaration form recognition system needs to formulate comprehensive data processing guidelines and the ability to identify and read original customs data. Second, the data storage redundancy between the old and new systems at the General Administration of Customs and the provincial customs offices is high, and the hierarchical structure between systems is complex, making it difficult to reuse data and resulting in low efficiency in information sharing and collaboration among all parties. Third, the risk knowledge and risk model methods used and processed in the entire process are limited, and risk identification is time-consuming. Faced with a wide variety of business data, the management system cannot accurately analyze the overall risk of cargo transactions in real time. Fourth, the cluster needs to be suspended during risk model system updates, resulting in long waiting times for users or the need for hot deployment using scripts, which increases time complexity, prolongs cargo holding time at ports, and increases port costs for enterprises. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention aims to provide a method, apparatus, computer equipment group, and storage medium for structured management of customs business data. It is the first to propose using edge computing to process customs business data. Through terminal data collection and preprocessing, it enables the batch, de-redundant input of customs data awaiting inspection, automatically determines risk control types, returns risk identification results, and uploads refined and effective information to the fusion result database. Simultaneously, it maintains service synchronously through hot deployment, thereby reducing network bandwidth for data sources from various terminals, alleviating the processing pressure on the customs data center server, improving the management efficiency and accuracy of the customs business data system, and promptly and accurately blocking the risks posed by high-risk goods and illegal or non-compliant goods. This, in turn, reduces the manpower costs of on-site inspections at customs ports and achieves efficient fusion of customs business data.

[0005] In a first aspect, the present invention provides a method for structured management of customs business data, including a business object reading stage, an accompanying document reading stage, a business object association stage, a scenario discrimination stage, a parameter condition execution stage, a risk knowledge query stage, a risk model invocation stage, a knowledge graph query stage, a fusion result generation stage, and a fusion result message push stage.

[0006] The business object reading phase is used to fill in basic information and create the business object structure when reading business objects. There are two types of structures: one is a normal business object structure, which requires no further configuration in business object associations or scenario selection; simply adding the included data object structure is sufficient. The other is an association relationship type, which requires further configuration and selection of a one-to-one scenario. When adding the data object structure, both require configuring the structure type and selecting whether it is a parent table. The read result association includes two functions: creating, modifying, deleting, and querying basic information of business objects, and creating, modifying, deleting, and querying the configuration of business object association relationship fields.

[0007] The accompanying document reading stage is used for extracting unstructured data information. Users can complete the accompanying document reading stage task by following steps such as importing / updating templates, setting accompanying document filtering conditions, and selecting template status. This is achieved through algorithms such as image preprocessing, text detection, detection correction, character recognition, and information correction of the unstructured documents. The reading result association includes two parts: adding, modifying, deleting, and querying the query field mapping table, and adding, modifying, deleting, and querying the information field mapping table.

[0008] The business object association phase, connected to the business object reading phase, is used to read business association messages from the customs clearance business system message queue service and assemble the associated customs clearance message information into business object information according to the business object configuration rules. Considering the inconsistent arrival times and uncertain time spans of data messages required by an associated business queue, the information in this associated business queue needs to be persisted in the database for a certain period. Simultaneously, to improve the processing efficiency of subsequent data messages, the IDs of other objects that have not yet arrived in this association message need to be added to a waiting list, allowing for quick identification of whether subsequent business data messages are needed by other associated business queues. By listening for messages, if an association message is received, the business queue is initialized; if a waiting message is received, it is added to the business queue, and after checking that the assembly is complete, it is submitted to the scenario discrimination service phase for processing. For other types of messages, it is only necessary to determine whether the message contains information needed by a certain association message. If so, it is assembled into the corresponding association message data for further processing (determining whether the assembly is complete); otherwise, it only needs to be temporarily stored for easy retrieval.

[0009] The scenario identification stage, connecting the business object association stage, is used to find the corresponding scenario ID. The corresponding scenario ID is searched in the cache based on the data object ID. If it exists, it is returned directly; otherwise, the relevant information table (using the data object ID as the primary key) is searched. If the database search is successful, the scenario ID is cached and returned. If there is no corresponding record in the database, the associated data for that business object cannot be merged.

[0010] The parameter-related condition execution phase is connected to the scenario judgment phase, which is used to verify the incoming document parameters. This includes checking whether there are any empty parameter conditions in the parameter-related condition list, whether the context data parameters required to execute the parameter-related conditions are empty, etc. If the verification is abnormal, exception handling is performed. If the parameter verification passes, the necessary context data is prepared for the execution of the MVEL expression, that is, the context data parameters are filled into the context variables of the MVEL execution.

[0011] The risk knowledge query stage connects to the scenario judgment stage and is used to query and associate the mapping relationship between risk knowledge-specific data objects and specified data. By inputting query conditions for a fuzzy search, a result list is obtained. Clicking to view the specific structure reveals two types: one is a standard risk knowledge structure, requiring no further configuration; the other is a risk knowledge-specific data object association type, which requires further establishing the association between the specific data objects and the table structure to be associated, configuring query fields for them, and then extracting and configuring specific information fields for the already associated objects. The query results include two functionalities: viewing the risk knowledge structure and adding, modifying, deleting, and querying risk knowledge-specific data objects.

[0012] The risk model invocation phase connects to the scenario discrimination phase, where the API address of the risk model is configured. Clicking on the model invocation filter settings allows you to select a scenario and edit and save the MVEL expression filter conditions. For cases requiring the setting of risk model association types, the method for setting risk model associations is exactly the same as the method for configuring risk knowledge dedicated data objects. Because risk models primarily achieve risk scoring through deep learning algorithms, which is typically time-consuming, setting model invocation filter conditions and a timeout discard mechanism prevents excessively frequent model invocations from causing system bottlenecks.

[0013] Optionally, the risk model can be updated in real time through hot deployment synchronization. Compared with other methods, this method does not require service interruption. The shortest time to use the new model is the shortest hot deployment time of the cluster server, and the full deployment time is the longest hot deployment time of the cluster server, which improves the efficiency of cluster hot deployment. Abnormal downtime of a single server does not seriously affect the overall service effect.

[0014] The knowledge graph query stage connects to the scenario judgment stage and is used to view the field structure returned by the knowledge graph. For cases where it is necessary to set the knowledge graph association type, the knowledge graph association setting method is exactly the same as the risk knowledge dedicated data object configuration method. The knowledge graph association can query the knowledge graph data object, and can also perform functions such as creating, querying, modifying and deleting associated fields.

[0015] The fusion result generation stage connects the accompanying document reading stage, scenario identification stage, parameter-related condition execution stage, risk knowledge query stage, risk model invocation stage, and knowledge graph query stage. It is responsible for querying, processing, and generating results for the risk knowledge, knowledge graph, risk model, accompanying document extraction results, and parameter-related condition execution results that the system relies on during data fusion processing. If the query fails to return results immediately after calling the query interface (e.g., the interface uses an asynchronous return method, or the query fails), depending on the implementation of the query interface or the program configuration, additional result query interfaces can be called, or a certain period of time can be waited before re-requesting. Only when the attempt fails after reaching certain conditions is the result considered abnormal. If results are returned, the result data returned by each interface is fused and delivered to the fusion result message push stage.

[0016] The fusion result message push stage connects to the fusion result generation stage and is used for storing the risk identification results of customs clearance documents. A fusion requirement ID is generated based on the Snowflake algorithm. Fields such as batch number, application ID, loading method, scenario ID, and delay time of the message are parsed. The loading method is categorized as incremental or full operation. The effective time is obtained based on the delay time, and all the obtained basic information is stored in the database.

[0017] Secondly, the present invention provides a customs business data management device, including a data object structure maintenance module, a business object fusion setting module, an accompanying document fusion setting module, a risk knowledge fusion setting module, a risk model fusion setting module, a knowledge graph fusion setting module, a parameter table management module, and a scenario data fusion requirement synchronization module.

[0018] The data object structure maintenance module is used for designing data object structures. It includes functions for creating, editing, querying, and deleting basic information about data objects, as well as their common and special fields. It provides necessary functional support for other modules such as business object structure maintenance and risk model fusion settings.

[0019] The business object fusion setting module is used for maintaining the business object structure and associating business objects. A business object is an important carrier of data objects. A business object structure can contain multiple data object structures, and it is the smallest unit for transmitting a section of data in the customs clearance operation system.

[0020] The accompanying document fusion setting module is used to provide functions such as fusion setting management, issuing setting updates, extracting setting updates, template matching, and OCR processing. It adopts interactive methods such as message queues to provide technical support for the fusion of accompanying document information.

[0021] The risk knowledge fusion module is used to generate relevant knowledge from the internet through intelligent algorithms, which is produced during the construction of the risk knowledge base. This knowledge is primarily in the form of a knowledge table. The risk knowledge table uses public code table information such as enterprise codes and product codes as the primary key, ensuring that each piece of risk knowledge can be accurately associated with the entire order or a specific product item through the public code, thus ensuring successful fusion.

[0022] The risk model fusion settings module is used for creating / modifying risk models, setting risk model filtering conditions, and setting risk model associations.

[0023] The knowledge graph fusion settings module is used to view the knowledge graphs available in the system and to set the return results of various graph calls and the association methods with ordinary data objects.

[0024] The parameter table management module provides services for the parameter fusion process. Parameter table management mainly includes two parts: user parameter table management and public parameter table management. User parameter table management mainly has three functions: adding, querying, and deleting. Public parameter table management includes two functions: querying public parameter tables and viewing the public parameter table structure.

[0025] The scenario data fusion requirement synchronization module is used to parse and store fusion requirement messages. It queries the database table to obtain fusion requirement information and stores all fusion requirement information. Fusion requirement parsing is mainly divided into two categories: if the fusion requirement message loading method is full, all rule information is parsed and converted into fusion requirement information for storage and loading. If the fusion requirement message loading method is full, the old version corresponding to the fusion requirement message needs to be added, deleted, or modified first, and then the modified fusion requirement message is parsed and stored incrementally.

[0026] Thirdly, the present invention provides a computer device group, including a data terminal node group, a risk identification node group, a fusion result node group, and a memory and processor communicatively connected thereto.

[0027] The data terminal node group is used for data acquisition, OCR recognition, data preprocessing, business fusion settings, and scenario discrimination. It includes the data object structure maintenance module, business object structure maintenance module, and accompanying document fusion setting module of the customs business data management device described in the second aspect. It executes the business object reading stage, accompanying document reading stage, business object association stage, and scenario discrimination stage of the customs business data structured management method described in the first aspect. All data information is stored at the current data terminal node. Data terminal nodes do not communicate with each other; each data terminal node only communicates with the risk identification node group. All structured and unstructured data...

[0028] The risk identification node group is used for risk control and prediction of customs clearance documents. It includes the risk knowledge fusion setting module, risk model fusion setting module, knowledge graph fusion setting module, and parameter table management module of the customs business data management device described in the second aspect. It executes the parameter condition execution stage, risk knowledge query stage, risk model invocation stage, knowledge graph query stage, and fusion result generation stage of the customs business data structured management method described in the first aspect. Based on the document risk identification requirements, the risk identification node notifies the data terminal node of the field parameters to be sent, receives data information from the data terminal node, and completes the risk identification requirements. All risk data information is stored in the current risk identification node. The risk identification nodes communicate with each other using a load balancing gateway to complete functions such as two-way verification of electronic documents and synchronization of risk model training sample sets. Furthermore, based on the different response speeds and result accuracy of each module in the group, the risk identification node group consists of a simple requirement risk identification group and a complex requirement risk identification group. The simple requirement risk identification group completes the setting methods for risk knowledge and knowledge graphs and the invocation function of simple risk models, performing preliminary risk identification. The document field parameters of the identification results that fall within the confidence interval are sent to the complex demand risk identification group to complete the training results with higher granularity, so as to ensure the accuracy of the model training results.

[0029] The aforementioned fusion result node group is used to receive the fusion request results from the risk identification nodes and respond to their access requests. This includes the scenario data fusion request synchronization module of the customs business data management device described in the second aspect, and the fusion result message push phase of the customs business data structured management method described in the first aspect. The fusion result node group achieves high-performance distributed index querying and storage through a master-slave strategy of database sharding, table partitioning, and read-write separation.

[0030] The memory is used to store computer programs, and the processor is used to read the computer programs and execute the customs business data structured management method described in the first aspect.

[0031] Based on the foregoing invention, a novel customs smart data management process is also provided, comprising a structured data processing flow for documents, distribution of parameter-related data to risk identification nodes, and clustering of nationwide business structured data. By batch processing unstructured customs document data at the terminal node and storing it in the relational database of the current node in structured data form, and by storing data based on extracted document data features, database storage redundancy is effectively reduced, while the parameter extraction time for various risk identification requests is accelerated, improving data transmission efficiency between the data terminal node and the risk identification node. Through multi-level risk identification groups, the accuracy of document data identification results is filtered level by level. The simple-demand risk identification group returns document data results with low accuracy requirements or significantly exceeding the normal result range using a low-complexity, high-timeliness risk identification scheme. Data that fails to filter is judged by the complex-demand risk identification group using a high-complexity, higher-accuracy risk identification scheme. Based on this, the implemented multi-level load-balanced risk identification system can address the balance between real-time performance and accuracy in risk control. In addition, the system adopts a multi-source heterogeneous data fusion solution based on edge computing, and designs distributed node functions with the principle of high cohesion and low coupling to enhance the atomicity of each node and the availability of the system.

[0032] Fourthly, the present invention provides a storage medium storing instructions that, when executed on a computer system, perform the customs business data structured management method described in the first aspect.

[0033] Fifthly, the present invention provides a computer system program product containing instructions, which, when executed on a computer system, cause the computer system to perform the customs business data structured management method described in the first aspect. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a system architecture diagram of the customs business data structured management method provided by the present invention.

[0036] Figure 2 This is a functional module interaction diagram of the customs business data structured management method provided by the present invention.

[0037] Figure 3 This is a flowchart illustrating the business object association stage processing provided by the present invention.

[0038] Figure 4 This is a flowchart illustrating the execution of the parameter table lookup function provided by the present invention.

[0039] Figure 5 This is a flowchart illustrating the parameter-dependent condition execution service provided by the present invention.

[0040] Figure 6 This is a flowchart illustrating the configuration of the risk knowledge-specific data object provided by the present invention.

[0041] Figure 7 This is a flowchart illustrating the process of parsing and managing fusion demand messages provided by the present invention.

[0042] Figure 8 This is a schematic diagram of the process for updating old fusion requirement messages provided by the present invention.

[0043] Figure 9 This is a schematic diagram of the data fusion requirement completion process provided by the present invention.

[0044] Figure 10 This is a schematic diagram of the fusion data query process provided by the present invention.

[0045] Figure 11 This is a schematic diagram of the application service hot deployment process provided by the present invention.

[0046] Figure 12 This is a schematic diagram of the process for updating the risk model of the load balancing gateway provided by the present invention.

[0047] Figure 13 This is a schematic diagram of the structure of the computer device group provided by the present invention. Detailed Implementation

[0048] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be noted that while the description of these embodiments is intended to aid in understanding the invention, it does not constitute a limitation thereof. The specific structural and functional details disclosed herein are merely for describing exemplary embodiments of the invention. However, the invention can be embodied in many alternative forms and should not be construed as being limited to the embodiments described herein.

[0049] It should be understood that although the terms first, second, etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit, without departing from the scope of the exemplary embodiments of the invention.

[0050] It should be understood that the term "and / or" that may appear in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist simultaneously. The term " / and" that may appear in this document describes another relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " that may appear in this document generally indicates that the related objects before and after it are in an "or" relationship.

[0051] It should be understood that the terminology used herein is for describing particular embodiments only and is not intended to limit the exemplary embodiments of the invention. Where used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that when the terms “comprising,” “including,” “containing,” and / or “including” are used herein, they specify the presence of the stated features, quantities, steps, operations, units, and / or components, and do not exclude the presence or addition of one or more other features, quantities, steps, operations, units, components, and / or combinations thereof.

[0052] It should be understood that, in some alternative designs, the functions / actions may appear in a different order than those shown in the figures. For example, depending on the functions / actions involved, they may actually be performed concurrently, or sometimes the two figures shown consecutively may be performed in reverse order.

[0053] It should be understood that specific details are provided in the following description to facilitate a complete understanding of the exemplary embodiments. However, those skilled in the art will understand that the exemplary embodiments can be implemented without these specific details. For example, the system may be shown in block diagrams to avoid obscuring the example with unnecessary details. In other instances, well-known processes, structures, and techniques may be omitted with non-essential details to avoid obscuring the exemplary embodiments.

[0054] like Figures 1-10 As shown, the customs business data structured management method provided in the first aspect of this embodiment can be, but is not limited to, executed on processing devices with certain computing resources. The customs business data structured management method includes a business object reading stage, an accompanying document reading stage, a business object association stage, a scenario discrimination stage, a parameter condition execution stage, a risk knowledge query stage, a risk model invocation stage, a knowledge graph query stage, a fusion result generation stage, and a fusion result message push stage.

[0055] The business object reading phase may include, but is not limited to, the following steps S101 to S104:

[0056] S101. In the data terminal node business object fusion setting module, collect basic information such as the business object's Chinese and English names and description, select "ordinary business object" or "related business object" as the type, and submit it to the corresponding database. If the business object already exists in the database, skip this step.

[0057] S102. In the business object selected in S101, collect information such as the English name and field description of the business object fields, select the corresponding database data type, and submit it to the corresponding database. If the required field information already exists in the database, skip this step.

[0058] S103. Collect basic information such as the English and Chinese names and descriptions of the data objects, select "Ordinary Data Object" or "Special Data Object" as the type, and submit it to the corresponding database.

[0059] S104. In the business object selected in step S103, collect the data object field information, perform key-value pair matching, and submit it to the corresponding database.

[0060] The accompanying document reading stage may include, but is not limited to, the following steps S201 to S204:

[0061] S201. In the document fusion setting module attached to the data terminal node, source files of unstructured data and semi-structured data, including but not limited to images and documents, are collected.

[0062] S202. Accordingly, through algorithms such as image preprocessing, text detection, detection correction, character recognition, and information correction of unstructured documents, the basic information and field information of the business object are extracted and submitted to the corresponding database. If the business object already exists in the database, this step is skipped.

[0063] S203. Accordingly, through algorithms such as image preprocessing, text detection, detection correction, character recognition, and information correction of unstructured documents, basic information and field information of data objects are extracted and submitted to the corresponding database.

[0064] The business object association stage may include, but is not limited to, the following steps S301 to S304:

[0065] S301. Read the business relationship messages from the message queue service of the customs clearance business system, and save the information in the related business queue to the database for persistent processing for a certain period of time.

[0066] S302. Add the IDs of other objects that have not yet arrived in the association message to the waiting list, so that when other business data messages arrive later, it can be quickly located whether they are the data needed by other related business queues.

[0067] S303. By listening to messages, if a relationship message is received, the business queue is initialized. If a waiting message is received, it is added to the business queue. After checking that the assembly is complete, it is submitted to the scenario discrimination service stage for processing.

[0068] S304. For other types of messages, it is only necessary to determine whether the message is the information required in a certain related message. If so, it is assembled into the corresponding related message data for further processing (determine whether the assembly is complete). Otherwise, it is only necessary to temporarily store it for easy querying.

[0069] The scene discrimination stage may include, but is not limited to, the following steps S401 to S402:

[0070] S401. Search for the corresponding scenario ID in the cache based on the data object ID. If it exists, return it directly. If it does not exist, search the corresponding information table for the dedicated business (using the data object ID as the primary key).

[0071] S402. If the database lookup is successful, cache the scene ID and return it. If there is no corresponding record in the database, the data associated with the business object cannot be merged.

[0072] The parameter-related condition execution phase may include, but is not limited to, the following steps S501 to S502:

[0073] S501. Validate the incoming document parameters, including checking whether there are any empty parameters in the parameter condition list, and whether the context data parameters required to execute the parameter conditions are empty, etc.

[0074] S502. If the validation fails, exception handling is performed; if the parameter validation passes, the necessary context data for the execution of the MVEL expression is prepared, that is, the context data parameters are filled into the context variables of the MVEL execution.

[0075] The risk knowledge inquiry stage may include, but is not limited to, the following steps S601 to S603:

[0076] S601. Perform a fuzzy search by entering the search criteria, and then click to view the specific structure after obtaining the result list.

[0077] S602. If it is a normal risk knowledge structure, no further configuration is required.

[0078] S603. If it is a risk knowledge-specific data object association type, it is necessary to further establish the mapping relationship between the required risk knowledge object and the data object to be associated, configure query fields for them, and then perform specific information field extraction configuration for the two that have already been associated.

[0079] The risk model invocation phase may include, but is not limited to, the following steps S701 to S702:

[0080] S701. Configure the API address of the risk model, click on the model call filter settings, select the scenario, edit and save the MVEL expression filter conditions. If this configuration information already exists, skip this step.

[0081] S702. Establish the mapping relationship between the required risk model objects and the data objects to be associated, configure query fields for them, and then extract and configure specific information fields for the two that have been associated.

[0082] The knowledge graph query stage may include, but is not limited to, the following steps S801 to S802:

[0083] S801. Perform a fuzzy search by entering the search criteria, and then click to view the specific structure after obtaining the result list.

[0084] S802. Establish the mapping relationship between the required knowledge graph objects and the data objects to be associated, configure query fields for them, and then configure the specific information field extraction for the two that have been associated.

[0085] The fusion result generation stage may include, but is not limited to, the following steps S901 to S902:

[0086] S901. After calling the query interface, if the query fails to return results immediately (e.g., the interface uses an asynchronous return method, or the query fails), depending on the implementation of the query interface or the program configuration, you can consider calling an additional result query interface or waiting for a certain period of time before making a new request. Only when the query fails after reaching a certain condition is the result considered abnormal.

[0087] S902. If a result can be returned, the result data returned by each interface is merged and delivered to the fusion result message push stage.

[0088] The fusion result message push stage may include, but is not limited to, the following steps S1001 to S1002:

[0089] S1001. Generate a fusion requirement ID based on the snowflake algorithm, and parse the batch number, application ID, loading method, scene ID, delay time and other fields of the message.

[0090] S1002. Based on the loading method, it is divided into incremental and full operations. The effective time is obtained based on the delay time, and all the obtained basic information is stored in the database.

[0091] After step S502 and before step S601, the method further includes: for the current risk knowledge base, if there is a required risk knowledge or risk knowledge field information to be updated, then the corresponding risk knowledge or risk knowledge field is updated first to obtain the latest risk knowledge base.

[0092] After step S602 and before step S701, the method may include, but is not limited to, the following steps S1101 to S1106:

[0093] S1101. For the current risk model library, if there is a need to update the risk model information, first update the corresponding risk model fields to obtain the latest risk model library, such as... Figures 11-12 As shown, the application service process of the corresponding risk model also needs to be redeployed.

[0094] S1102. The risk model server that needs to be updated is remotely called to upload the new risk model file. After the server successfully receives the file, it is immediately redeployed.

[0095] S1103. Once a server successfully updates its service, it will notify the load balancing gateway of the model version update. The load balancing gateway will then update its current model version and discard all packets returned by the old model.

[0096] S1104. After other servers complete hot deployment, update the model version in the update message.

[0097] S1105. To ensure availability, a message queue cluster is added before the load balancer to prevent message congestion.

[0098] S1106. During this period, if the first server crashes, the consumption of new services will be suspended, but as other servers are updated and added, the congestion will be alleviated.

[0099] After step S702 and before step S801, the method further includes: for the current knowledge graph library, if there is a required update of knowledge graph or knowledge graph field information, then first update the corresponding knowledge graph or knowledge graph field to obtain the latest knowledge graph library.

[0100] like Figure 2As shown, the second aspect of this embodiment provides a virtual device for implementing the customs business data structured management method described in the first aspect, including a data object structure maintenance module, a business object fusion setting module, an accompanying document fusion setting module, a risk knowledge fusion setting module, a risk model fusion setting module, a knowledge graph fusion setting module, a parameter table management module, and a scenario data fusion requirement synchronization module.

[0101] The data object structure maintenance module is used for designing data object structures. It includes functions for creating, editing, querying, and deleting basic information about data objects, as well as their common and special fields. It provides necessary functional support for other modules such as business object structure maintenance and risk model fusion settings.

[0102] The business object fusion setting module is used for maintaining the business object structure and associating business objects. A business object is an important carrier of data objects. A business object structure can contain multiple data object structures, and it is the smallest unit for transmitting a section of data in the customs clearance operation system.

[0103] The accompanying document fusion setting module is used to provide functions such as fusion setting management, issuing setting updates, extracting setting updates, template matching, and OCR processing. It adopts interactive methods such as message queues to provide technical support for the fusion of accompanying document information.

[0104] The risk knowledge fusion module is used to generate relevant knowledge from the internet through intelligent algorithms, which is produced during the construction of the risk knowledge base. This knowledge is primarily in the form of a knowledge table. The risk knowledge table uses public code table information such as enterprise codes and product codes as the primary key, ensuring that each piece of risk knowledge can be accurately associated with the entire order or a specific product item through the public code, thus ensuring successful fusion.

[0105] The risk model fusion settings module is used for creating / modifying risk models, setting risk model filtering conditions, and setting risk model associations.

[0106] The knowledge graph fusion settings module is used to view the knowledge graphs available in the system and to set the return results of various graph calls and the association methods with ordinary data objects.

[0107] The parameter table management module provides services for the parameter fusion process. Parameter table management mainly includes two parts: user parameter table management and public parameter table management. User parameter table management mainly has three functions: adding, querying, and deleting. Public parameter table management includes two functions: querying public parameter tables and viewing the public parameter table structure.

[0108] The scenario data fusion requirement synchronization module is used to parse and store fusion requirement messages. It queries the database table to obtain fusion requirement information and stores all fusion requirement information. Fusion requirement parsing is mainly divided into two categories: if the fusion requirement message loading method is full, all rule information is parsed and converted into fusion requirement information for storage and loading. If the fusion requirement message loading method is full, the old version corresponding to the fusion requirement message needs to be added, deleted, or modified first, and then the modified fusion requirement message is parsed and stored incrementally.

[0109] The working process, working details and technical effects of the aforementioned device provided in the second aspect of this embodiment can be found in the customs business data structured management method described in the first aspect, and will not be repeated here.

[0110] like Figure 13 As shown, the third aspect of this embodiment provides a computer equipment group for implementing the customs business data structured management method described in the first aspect, including a data terminal node group, a risk identification node group, a fusion result node group, and a memory and processor communicatively connected thereto.

[0111] The data terminal node group is used for data acquisition, OCR recognition, data preprocessing, business fusion settings, and scenario discrimination. It includes the data object structure maintenance module, business object structure maintenance module, and accompanying document fusion setting module of the customs business data management device described in the second aspect. It executes the business object reading stage, accompanying document reading stage, business object association stage, and scenario discrimination stage of the customs business data structured management method described in the first aspect. All data information is stored in the current data terminal node; the data terminal nodes do not communicate with each other, and each data terminal node only communicates with the risk identification node group.

[0112] The risk identification node group is used for risk control and prediction of customs clearance documents. It includes the risk knowledge fusion setting module, risk model fusion setting module, knowledge graph fusion setting module, and parameter table management module of the customs business data management device described in the second aspect. It executes the parameter condition execution stage, risk knowledge query stage, risk model invocation stage, knowledge graph query stage, and fusion result generation stage of the customs business data structured management method described in the first aspect. Based on the document risk identification requirements, it notifies the data terminal nodes of the field parameters to be sent, receives data information from the data terminal nodes, and completes the risk identification requirements. All risk data information is stored in the current risk identification node. The risk identification nodes communicate using content delivery network technology to complete functions such as two-way verification of electronic documents and synchronization of risk model training sample sets. Furthermore, based on the different response speeds and result accuracy of each module in the group, the risk identification node group consists of a simple requirement risk identification group and a complex requirement risk identification group. The simple requirement risk identification group completes the setting methods for risk knowledge and knowledge graphs and the invocation function of simple risk models, performing preliminary risk identification. The document field parameters of the identification results falling within the confidence interval are sent to the complex demand risk identification group to obtain more granular training results, thereby ensuring the accuracy of the model training results. Based on this, the implemented multi-level load balancing risk identification system can address the balance between real-time performance and accuracy in risk control.

[0113] The aforementioned fusion result node group is used to receive the fusion request results from the risk identification nodes and respond to their access requests. This includes the scenario data fusion request synchronization module of the customs business data management device described in the second aspect, and the fusion result message push phase of the customs business data structured management method described in the first aspect. The fusion result node group achieves high-performance distributed index querying and storage through a master-slave strategy of database sharding, table partitioning, and read-write separation.

[0114] The memory is used to store computer programs, and the processor is used to read the computer programs and execute the customs business data structured management method described in the first aspect.

[0115] The working process, working details and technical effects of the aforementioned computer equipment group provided in the third aspect of this embodiment can be found in the customs business data structured management method described in the first aspect, and will not be repeated here.

[0116] This fourth aspect of the embodiment provides a storage medium for storing instructions containing the customs business data structured management method described in the first aspect. Specifically, the storage medium stores the corresponding instructions, and when the instructions are executed on a computer, the customs business data structured management method described in the first aspect is performed. The storage medium refers to a data storage medium, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or Memory Sticks.

[0117] The working process, working details and technical effects of the aforementioned storage medium provided in the fourth aspect of this embodiment can be found in the customs business data structured management method described in the first aspect, and will not be repeated here.

[0118] The fifth aspect of this embodiment provides a computer program product containing instructions that, when executed on a computer system, cause the computer system to perform the customs business data structured management method described in the first aspect.

[0119] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0120] Finally, it should be noted that this invention is not limited to the optional embodiments described above, and anyone can derive other various forms of products under the guidance of this invention. The specific embodiments described above should not be construed as limiting the scope of protection of this invention, which should be determined by the claims, and the specification can be used to interpret the claims.

Claims

1. A method for structuring management of customs service data, characterized by, The business object reading stage, the attached document reading stage, the business object association stage, the scene discrimination stage, the involved condition execution stage, the risk knowledge query stage, the risk model calling stage, the knowledge graph query stage, the fusion result generation stage, and the fusion result message pushing stage are included. The business object reading stage includes steps S101-S104: S101. In the data terminal node business object fusion setting module, collect business object basic information, select the type as "ordinary business object" or "association relationship business object", and submit to the corresponding database. If the business object already exists in the database, skip this step. S102. In the selected business object in S101, collect business object field information, select the corresponding database data type, and submit to the corresponding database. If the required field information already exists in the database, skip this step. S103. In the data terminal node data object structure maintenance module, collect data object basic information, select the type as "ordinary data object" or "professional data object", and submit to the corresponding database. S104. In the data terminal node, collect data object field information, perform key-value pair matching, and submit to the corresponding database. The attached document reading stage includes steps S201-S204: S201. In the data terminal node attached document fusion setting module, collect source files including unstructured data and semi-structured data such as images and documents. S202. Through image preprocessing, text detection, detection correction, character recognition, and information correction algorithms of unstructured documents, extract business object basic information and business object field information, and submit to the corresponding database. If the business object already exists in the database, skip this step. S203. Through image preprocessing, text detection, detection correction, character recognition, and information correction algorithms of unstructured documents, extract data object basic information and data object field information, and submit to the corresponding database. The business object association stage includes steps S301-S304: S301. In the customs business system message queue service, read the business association relationship message, save the information in the association business queue to the database for a certain duration of persistent processing. S302. Put the other association relationship business object IDs that have not arrived in the waiting list to facilitate the quick positioning of whether the other business data message is needed for other association business queues when it arrives. S303. Through message listening, if the association relationship message is received, initialize the business queue, if the message in the waiting list is received, add it to the business queue, check the completion of the assembly, and submit to the scene discrimination service stage for processing. S304. For other types of messages, only need to judge whether the message is the information needed in a certain association relationship message, if so, assemble it into the corresponding association relationship message data for the next step processing, otherwise, only need to temporarily store it for easy query; The scene discrimination stage includes steps S401-S402: S401. Find the corresponding scene ID according to the data object ID from the cache, return directly if it exists, and find the special business corresponding information table if it does not exist; S402. If the database search is successful, cache the scene ID and return, if there is no corresponding record in the database, the business object associated data cannot be fused; The involved condition execution stage includes steps S501-S502: S501. Check the incoming document parameters, including judging whether the involved condition list is empty, and whether the context data parameter required for executing the involved condition is empty; S502. If the check is abnormal, perform exception handling; If the parameter check is passed, start preparing the context data required for MVEL expression execution, that is, fill the context data parameter into the context variable of MVEL execution; The risk knowledge query stage includes steps S601-S603: S601. Perform fuzzy query by inputting query conditions, get the result list, and click to view the specific structure; S602. If it is a normal risk knowledge structure, no further configuration is needed; S603. If it is a risk knowledge special data object association relationship type, further mapping relationship between the required risk knowledge object and the data object to be associated needs to be established, and query fields are configured for them, and specific information field extraction configuration is performed for the two that have established association; The risk model calling stage includes steps S701-S702: S701. Configure the API address of the risk model, click the model calling filter condition filter setting, select the scene, edit and save the MVEL expression filter condition, if the configuration information already exists, this step is skipped; S702. Establish the mapping relationship between the required risk model object and the data object to be associated, and configure the query fields for them, and perform specific information field extraction configuration for the two that have established association; The knowledge graph query stage includes steps S801-S802: S801. Perform fuzzy query by inputting query conditions, get the result list, and click to view the specific structure; S802. Establish the mapping relationship between the required knowledge graph object and the data object to be associated, and configure the query fields for them, and perform specific information field extraction configuration for the two that have established association; The fusion result generation stage includes steps S901-S902: S901. After calling the query interface, if the query cannot return the result immediately, according to the implementation or program configuration of the query interface, additionally call the additional result query interface or wait for a certain period of time and then request again, until it is considered as a result exception when it fails after trying to reach a certain condition; S902. If the result can be returned, fuse the result data returned by each interface to generate the fusion result message pushing stage; The fusion result message pushing stage includes steps S1001-S1002: S1001. Generate the fusion requirement ID according to the snowflake algorithm, parse the batch number, application ID, loading mode, scene ID, and delay time fields of the message; S1002. According to the loading mode, it is divided into incremental and full operation, and the effective time of the effect is obtained according to the delay time, and the obtained basic information is stored into the database.

2. The method for structuring management of customs business data according to claim 1, wherein, After the step S602 and before the step S701, the method further comprises steps S1101-S1106: S1101. For the current risk model library, if there is required risk model information update, first update the corresponding risk model field to obtain the latest risk model library, and the application service process of the corresponding risk model also needs to be re-deployed; S1102. The risk model server that needs to be updated is called remotely, and the new risk model file is uploaded, and the server successfully receives and immediately re-deploys the server; S1103. Once the server update service is successful, the server will notify all other servers of the message of successful update of itself, the load balancing gateway updates the current model version, and discards all old model returned messages; S1104. After the other servers complete the hot deployment, the model version in the update message is updated; S1105. To ensure availability, a message queue cluster is added in front of the load balancer to ensure that messages are not congested; S1106. During this period, if the first server crashes, message consumption will be suspended, but as other servers are updated and joined, the congestion situation will be better alleviated.

3. An apparatus for structurally managing customs service data, for implementing the method for structurally managing customs service data according to claim 1 or 2, characterized by, The data object structure maintenance module, the business object fusion setting module, the accompanying document fusion setting module, the risk knowledge fusion setting module, the risk model fusion setting module, the knowledge graph fusion setting module, the parameter table management module, and the scene data fusion demand synchronization module are included. The data object structure maintenance module is used for the design of data object structure, and includes the new creation, editing, query, and deletion functions of basic information of data objects and ordinary fields and special fields contained therein; and provides necessary function support for other modules. The business object fusion setting module is used for business object structure maintenance and business object association; the business object is an important carrier of data objects, and a business object structure contains multiple data object structures, which is the smallest unit of data transmission in the customs clearance system; The accompanying document fusion setting module is used for providing fusion setting management, issuing setting update, extracting setting update, template matching, and OCR processing function services, and adopts a message queue interaction mode to provide technical support for accompanying document information fusion; The risk knowledge fusion setting module is used for related knowledge generated from the risk knowledge base construction work through intelligent algorithms from the Internet, which exists in the form of a knowledge table; The risk knowledge table takes public code table information of enterprise codes and commodity codes as the primary key, ensures that various risk knowledge can be accurately associated with complete orders or specific commodity items through public coding, and ensures fusion success; The risk model fusion setting module is used for risk model new creation / modification, risk model filter condition setting, and risk model association setting; The knowledge graph fusion setting module is used for viewing available knowledge graphs of the system and setting return results of various graph calls and association modes with ordinary data objects. The parameter table management module is used to provide services for parameter fusion; the parameter table management includes user parameter table management and public parameter table management, wherein the user parameter table management has three functions of adding, querying and deleting, and the public parameter table management includes two functions of public parameter table querying and public parameter table structure viewing; The scene data fusion requirement synchronization module is used to analyze and store the fusion requirement message, query the analyzed information in the database table, obtain the fusion requirement information, and store all the fusion requirement information. The fusion requirement analysis is divided into two categories. If the fusion requirement message loading mode is full, all rule information is analyzed and converted into fusion requirement information for storage and loading. If the fusion requirement message loading mode is incremental, the old version corresponding to the fusion requirement message needs to be operated in addition, deletion and modification, and then the modified fusion requirement message is analyzed and stored in an incremental manner.

4. A computer device group for structured management of customs business data, characterized by The data terminal node group, the risk identification node group, the fusion result node group, and the memory and processor connected in communication, the memory is used to store the computer program, and the processor is used to read the computer program and execute the structured management method of the customs business data as claimed in any one of claims 1-2.

5. The computer equipment group for the structured management of customs operations data according to claim 4, characterized in that it comprises: The data terminal node group is used for data acquisition, OCR identification, data preprocessing, business fusion setting and scene judgment, and includes the data object structure maintenance module, the business object fusion setting module and the accompanying document fusion setting module of the customs business data structured management device as claimed in claim 3, and executes the business object reading stage, the accompanying document reading stage, the business object association stage and the scene judgment stage of the structured management method of the customs business data as claimed in any one of claims 1-2. All data information is stored in the current data terminal node, and each data terminal node does not communicate with each other, and each data terminal node only communicates with the risk identification node group.

6. The computer equipment group for the structured management of customs operations data according to claim 4, characterized in that it comprises: The risk identification node group is used for risk control and prediction of customs documents, and includes the risk knowledge fusion setting module, the risk model fusion setting module, the knowledge graph fusion setting module and the parameter table management module of the customs business data structured management device as claimed in claim 3, and executes the involved condition execution stage, the risk knowledge query stage, the risk model calling stage, the knowledge graph query stage and the fusion result generation stage of the structured management method of the customs business data as claimed in any one of claims 1-2; according to the document risk identification requirement, the data terminal node needs to send the field parameter, receives the data information of the data terminal node, and completes the risk identification requirement. All risk data information is stored in the current risk screening node, and the risk screening nodes communicate with each other using content distribution network technology to complete the functions of electronic document bidirectional verification and risk model training sample set synchronization. In addition, according to the different response speed and result accuracy of each module in the group, the risk screening node group is composed of a simple demand risk screening group and a complex demand risk screening group. The simple demand risk screening group completes the setting method of risk knowledge, knowledge graph and the calling function of simple risk model, and performs preliminary risk screening. The document field parameters of the screening results falling into the confidence interval are sent to the complex demand risk screening group to complete the training results with higher granularity, so as to ensure the result accuracy of model training.

7. The computer equipment group for the structured management of customs operations data according to claim 4, characterized in that it comprises: The fusion result node group is used for receiving the fusion demand results of the risk screening nodes and simultaneously responding to the request access, including the scene data fusion demand synchronization module of the customs business data structured management device as claimed in claim 3, and performing the fusion result message pushing stage of the customs business data structured management method as claimed in any one of claims 1-2. The fusion result node group realizes high-performance distributed index query and storage through the master-slave strategy of database and table separation and read-write separation.

8. A storage medium, characterized by The storage medium has instructions stored thereon, and when the instructions run on the computer, the customs business data structured management method as claimed in any one of claims 1-2 is executed.

Images