Delaration data processing method and device, equipment, storage medium and program product

Abstract

The invention relates to a declaration data processing method and device, equipment, a storage medium and a program product, and relates to the technical field of data processing.The method comprises the steps that in response to a target declaration service initiated by a declarator, a declaration data dimension architecture corresponding to a service type is inquired according to the service type of the target declaration service, determining a declaration data dimension required by the target declaration service; the declaration data dimension comprises a basic item dimension and a pre-filling item dimension; in response to the typing operation, obtaining basic item data corresponding to the basic item dimension; calling a pre-constructed decision model, and performing hierarchical matching processing on the pre-filled item dimension according to at least one of the basic item data and historical declaration data of the declarator to obtain pre-filled item data matched with the pre-filled item dimension; and according to the basic item data and the pre-filled item data, obtaining current declaration data of the target declaration service, and verifying the current declaration data to obtain verified declaration data. By adopting the method, the data processing efficiency can be improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a data processing method, apparatus, device, storage medium, and program product. Background Technology

[0002] In the processing of declarations across various industries, attention must be paid to both timeliness and accuracy. Applicants must accurately fill in all declaration elements according to regulations, ensuring no omissions or errors. The processing agency, as the reviewing body, must rigorously verify the declaration information to guarantee complete consistency between the data and the actual situation. However, current data processing methods require applicants to manually fill in each element, and processing agencies typically rely on manual review, resulting in low data processing efficiency. Summary of the Invention

[0003] Therefore, it is necessary to provide a data processing method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve data processing efficiency in response to the above-mentioned technical problems.

[0004] Firstly, this application provides a method for processing declaration data, including:

[0005] In response to a target declaration business initiated by the declarant, the declaration data dimension architecture corresponding to the business type of the target declaration business is queried to determine the multiple declaration data dimensions required for the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions.

[0006] In response to the input operation, the basic item data corresponding to the basic item dimension is obtained;

[0007] A pre-built decision model is invoked to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data that matches the pre-filled item dimensions; the historical application data is obtained based on the applicant's account information and the data source in the application data dimension architecture;

[0008] Based on the basic data and the pre-filled data, the current application data for the target application business is obtained, and the current application data is verified to obtain the verified application data.

[0009] In one embodiment, the pre-built decision model includes a rule engine model, a historical matching model, and a feature-weighted decision model. The step of calling the pre-built decision model to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data matching the pre-filled item dimensions, includes:

[0010] The rule engine model in the pre-built decision model is invoked, and the pre-filled item dimensions are matched according to the basic item data to obtain the rule matching result; the rule engine model includes a pre-built rule tree and a pre-built hash table;

[0011] If the rule matching result is the pre-built rule tree or the pre-built hash table, and there is target data that matches the pre-filled item dimension, then the target data is determined as the pre-filled item data.

[0012] In one embodiment, the declaration data processing method further includes:

[0013] If the rule matching result indicates that there is no target data that matches the pre-filled dimension, a tuple is determined based on the pre-filled dimension.

[0014] The historical application data of the applicant is filtered according to the tuple to obtain candidate historical data for at least one candidate application business.

[0015] If the candidate application business meets the preset association conditions, the candidate historical data is determined as the pre-filled item data; the preset association conditions are that the number of candidate application businesses is greater than a preset quantity threshold, and the confidence level between the candidate application business and the target application business is greater than a preset threshold.

[0016] In one embodiment, the declaration data processing method further includes:

[0017] If the candidate application does not meet the preset association conditions, the feature weighted decision model is invoked to extract features based on the applicant's historical application data, basic item data, and application data dimension architecture, thereby obtaining features in multiple preset dimensions.

[0018] Based on the features of each preset dimension and the preset weight coefficients of each preset dimension, the target fusion features are obtained, and the target fusion features are determined as the pre-filled data.

[0019] In one embodiment, the declaration data processing method further includes:

[0020] Obtain the declaration data dimensions for multiple declaration businesses of the same business type, and determine the sub-dimensions and data sources for each declaration data dimension;

[0021] Based on the sub-dimensions and data sources of each declared data dimension, a declared data dimension architecture is constructed.

[0022] In one embodiment, the declaration data processing method further includes:

[0023] Based on the aforementioned data dimension architecture, a rule engine model is constructed;

[0024] Based on the aforementioned data dimension architecture and the historical application data of each application business, a historical matching model is constructed;

[0025] Based on the declared data dimension architecture, historical declared data, and preset weight coefficients, a feature-weighted decision model is constructed.

[0026] Secondly, this application also provides a data processing apparatus, comprising:

[0027] The declaration dimension determination module is used to respond to a target declaration business initiated by the declarant, query the declaration data dimension architecture corresponding to the business type of the target declaration business, and determine the multiple declaration data dimensions required for the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions;

[0028] The basic data acquisition module is used to obtain the basic item data corresponding to the basic item dimension in response to the input operation;

[0029] The pre-filled data acquisition module is used to call a pre-built decision model and perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical declaration data to obtain pre-filled item data that matches the pre-filled item dimensions; the historical declaration data is obtained based on the applicant's account information and the data source in the declaration data dimension architecture;

[0030] The declaration data verification module is used to obtain the current declaration data of the target declaration business based on the basic item data and the pre-filled item data, and to verify the current declaration data to obtain the verified declaration data.

[0031] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0032] In response to a target declaration business initiated by the declarant, the declaration data dimension architecture corresponding to the business type of the target declaration business is queried to determine the multiple declaration data dimensions required for the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions.

[0033] In response to the input operation, the basic item data corresponding to the basic item dimension is obtained;

[0034] A pre-built decision model is invoked to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data that matches the pre-filled item dimensions; the historical application data is obtained based on the applicant's account information and the data source in the application data dimension architecture;

[0035] Based on the basic data and the pre-filled data, the current application data for the target application business is obtained, and the current application data is verified to obtain the verified application data.

[0036] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0037] In response to a target declaration business initiated by the declarant, the declaration data dimension architecture corresponding to the business type of the target declaration business is queried to determine the multiple declaration data dimensions required for the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions.

[0038] In response to the input operation, the basic item data corresponding to the basic item dimension is obtained;

[0039] A pre-built decision model is invoked to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data that matches the pre-filled item dimensions; the historical application data is obtained based on the applicant's account information and the data source in the application data dimension architecture;

[0040] Based on the basic data and the pre-filled data, the current application data for the target application business is obtained, and the current application data is verified to obtain the verified application data.

[0041] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0042] In response to a target declaration business initiated by the declarant, the declaration data dimension architecture corresponding to the business type of the target declaration business is queried to determine the multiple declaration data dimensions required for the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions.

[0043] In response to the input operation, the basic item data corresponding to the basic item dimension is obtained;

[0044] A pre-built decision model is invoked to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data that matches the pre-filled item dimensions; the historical application data is obtained based on the applicant's account information and the data source in the application data dimension architecture;

[0045] Based on the basic data and the pre-filled data, the current application data for the target application business is obtained, and the current application data is verified to obtain the verified application data.

[0046] The aforementioned data processing method, apparatus, computer equipment, computer-readable storage medium, and computer program product, in response to a target declaration business initiated by the declarant, query the declaration data dimension architecture corresponding to the business type of the target declaration business to determine multiple declaration data dimensions required for the target declaration business, wherein the declaration data dimensions include basic item dimensions and pre-filled item dimensions; in response to an input operation, obtain the basic item data corresponding to the basic item dimensions; invoke a pre-built decision model to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the declarant's historical declaration data to obtain pre-filled item data that matches the pre-filled item dimensions, wherein the historical declaration data is obtained based on the declarant's account information and the data source in the declaration data dimension architecture; based on the basic item data and the pre-filled item data, obtain the current declaration data of the target declaration business, and verify the current declaration data to obtain the verified declaration data. By constructing a standardized data dimension architecture for different business types and combining basic data collection with a hierarchical intelligent pre-filling mechanism based on decision models, the system achieves automated and accurate data collection for applications. This effectively reduces the burden on applicants in complex application scenarios and significantly improves application efficiency and user experience. Pre-filling application forms through multi-level matching logic fully utilizes applicants' historical application data and current basic data, greatly improving the accuracy and adaptability of the pre-filled results. Furthermore, a verification process is introduced after generating complete application data to further ensure data compliance and consistency. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating the payment declaration process for foreign exchange declaration business related to this technology;

[0049] Figure 2 A flowchart illustrating the foreign exchange reporting process for related technologies.

[0050] Figure 3 This is a flowchart illustrating a data processing method for one embodiment;

[0051] Figure 4 This is a structural block diagram of the data processing device in one embodiment;

[0052] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] As described in the background section, the data processing methods for related technologies suffer from low data processing efficiency. The inventors discovered that this problem arises because, according to the State Administration of Foreign Exchange regulations, banks must strictly adhere to two core requirements when handling related business: first, the timeliness requirement—banks must complete the submission of basic income and expenditure information before 12:00 PM on the first working day after the remittance or settlement is completed, and complete the submission of the application information within the first working day after the applicant provides the application information; failure to do so within the time limit will trigger compliance risks; second, the accuracy requirement—a dual responsibility system of "applicant submission + bank review" is implemented. The applicant must accurately fill in all application elements according to the specifications, ensuring no omissions or errors. The bank, as the reviewing entity, must strictly verify the application information to ensure that the application data is completely consistent with the actual transaction background, amount, nature, and other core elements. Figure 1The diagram illustrates the payment declaration process for foreign exchange declarations using relevant technologies. The declaring entity (customer) manually determines the declaration type (including declaration form type, such as "Overseas Remittance Application"; transaction code, such as "121010") through front-end channels such as online banking, direct bank-enterprise connections, or e-commerce platforms. This requires comprehensive judgment based on their business background and counterparty information, which can be confusing for non-professional customers. Customers manually fill in over 30 core elements, including: basic remittance information (amount, counterparty name / account number), declaration information (transaction code, counterparty country code, transaction remarks), and management information (customs declaration number, shipping document number). After completion, the information is submitted to the bank's payment system. The bank counter... After receiving the data, tellers manually verify the consistency of the declaration type (e.g., whether "transaction remarks 'imported equipment' matches code '121010'") and check the compliance of each element field by field. If the review fails, the customer's modification is rejected; if approved, the payment operation is executed. Every day at midnight, the payment system summarizes the previous day's declaration data and pushes it to the reporting system. Tellers need to perform a second verification across systems: errors in basic information (such as counterparty country code) need to be corrected in the payment system, and errors in declaration information (such as transaction code) need to be adjusted in the reporting system. After verification, the reporting system submits the data to the State Administration of Foreign Exchange in batches (daily verification time is relatively high). Figure 2As shown, the foreign exchange declaration processing flow for related technologies is provided. After the bank payment system receives an overseas remittance instruction (such as a SWIFT message), the teller manually determines the declaration type (such as "Foreign Income Declaration Form") and pushes the basic information of the receipt (amount, remitter's name) to the front-end channel. The customer manually fills in the declaration / management information (more than 30 elements similar to the payment process) and submits it to the payment system. The teller manually reviews the declaration information. If it passes, the payment is credited to the account; if it fails, it is rejected and modified. The subsequent data aggregation, secondary verification, and reporting process is consistent with the payment process. Therefore, the main problems of the existing foreign exchange declaration processing methods are mainly reflected in the following three aspects. First, the reliance on manual identification is high, with both error rate and timeliness risks being prominent: the existing solution mainly relies on tellers or customers to manually determine the declaration type. However, the international balance of payments declaration type needs to be comprehensively determined by combining multiple dimensions of data such as the declarant, trade background, and counterparty. The error rate of manual identification is high, requiring revisions and easily leading to timeouts and compliance risks. Second, the burden on customers is heavy, and the quality of source data is poor: Customers need to manually fill in more than 30 core elements of the international balance of payments declaration form (including transaction entity code, counterparty country code, customs declaration number, etc.), of which more than 10 are professionally coded elements (such as "settlement method" and "transaction code"). Non-professional customers have a high error rate when filling in the form, and tellers have a heavy workload for verification. Third, the logic of multiple systems is fragmented and the standards are not uniform: front-end channel systems such as online banking, direct bank-enterprise connection, and e-commerce platforms need to independently develop declaration type judgment modules. Due to the different understandings of international balance of payments declaration rules by developers of various systems, similar transactions are judged differently on different channels, which increases both system development and maintenance costs and compliance risks.

[0055] For the reasons mentioned above, this application provides a method for processing application data, which aims to improve the efficiency of application data processing.

[0056] In one embodiment, such as Figure 3 As shown, a method for processing declaration data is provided. This embodiment illustrates the application of this method to a server. It is understood that this method can also be applied to a terminal, or to a system including both a terminal and a server, and is implemented through the interaction between the terminal and the server. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. In this embodiment, the method includes the following steps:

[0057] Step S302: In response to the target declaration business initiated by the declarant, query the declaration data dimension architecture corresponding to the business type based on the business type of the target declaration business, and determine the multiple declaration data dimensions required for the target declaration business.

[0058] The reporting party can refer to the entity required to fulfill the reporting obligation, typically a client, including businesses or individuals. The target reporting business can be a specific reporting business currently initiated and being processed by the reporting party.

[0059] Among them, the business type can be a classification identifier used to distinguish different declaration businesses. Different business types correspond to different declaration forms, data item sets and regulatory logic.

[0060] The data dimension architecture for the application can be a pre-built data structure model for a specific business type, describing all the data dimensions and their sub-dimensions, sources, hierarchical relationships and mapping rules required for that type of application.

[0061] The data dimension of the declaration can be the smallest granular data unit category required to constitute a complete declaration; the data dimension of the declaration includes basic item dimension and pre-filled item dimension; the basic item dimension can be the dimension corresponding to the core declaration information manually entered or confirmed by the declarant. This type of dimension is difficult to predict directly through the model and requires customer participation to ensure compliance responsibility is fulfilled; the pre-filled item dimension can be the declaration information dimension that can be automatically recommended or filled by the algorithm model based on existing information (such as basic items, historical behavior).

[0062] Optionally, in response to a target declaration business initiated by the declarant, such as in response to a trigger operation performed by the declarant on a target declaration business in the business system, the server obtains the declarant's account information and the business type of the target declaration business, and then queries the declaration data dimension architecture corresponding to the business type based on the business type of the target declaration business to determine the multiple declaration data dimensions required for the target declaration business.

[0063] Step S304: In response to the input operation, obtain the basic item data corresponding to the basic item dimension.

[0064] Among them, the basic data can be the specific values ​​corresponding to the basic data dimensions that the applicant actually fills in or selects in the application form for the current target application business.

[0065] Optionally, the server responds to the applicant's input operation on the form for the target application business on the business page, and obtains the basic item data corresponding to each basic item dimension of the input or selection corresponding to the input operation.

[0066] Step S306: Invoke the pre-built decision model and perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical declaration data to obtain pre-filled item data that matches the pre-filled item dimensions.

[0067] Historical declaration data can be a complete collection of declaration records generated by the applicant when handling similar or related business in the past; historical declaration data is obtained based on the applicant's account information and the data source in the declaration data dimension architecture.

[0068] The pre-built decision model can be a composite algorithm model system for intelligent matching and recommendation of pre-filled item dimensions, which has been trained and deployed during the system initialization phase.

[0069] Among them, the pre-filled data can be candidate values ​​that are automatically generated by calling the decision model and combining the basic data and / or historical declaration data, and are intended to be used to fill the pre-filled data dimensions.

[0070] Optionally, after the server obtains the basic data filled in by the applicant, it initiates an intelligent filling process for each pre-filled item dimension, calls a pre-built decision model, and performs hierarchical matching processing on the pre-filled item dimension based on at least one of the basic data and the applicant's historical application data. This process adopts a hierarchical and progressive matching strategy to obtain pre-filled item data that matches the pre-filled item dimension.

[0071] Step S308: Based on the basic data and pre-filled data, obtain the current declaration data of the target declaration business, and verify the current declaration data to obtain the verified declaration data.

[0072] Optionally, the server integrates the application forms for the target application business based on the basic data and pre-filled data to obtain the current application data for the target application business. The server then verifies the current application data to obtain verified application data. Specifically, the server performs format verification, logical verification, and regulatory rule verification on the current application data. Format verification verifies the element coding rules, such as the range of transaction code enumeration values. Logical verification verifies data consistency, such as whether there are conflicts in the transaction type, transaction code, and transaction remarks. Regulatory rule verification verifies compliance with the latest regulatory requirements.

[0073] In the above-mentioned data processing method, the method responds to the target application business initiated by the applicant, queries the application data dimension architecture corresponding to the business type of the target application business, and determines multiple application data dimensions required for the target application business. The application data dimensions include basic item dimensions and pre-filled item dimensions. In response to the input operation, the basic item data corresponding to the basic item dimensions is obtained. A pre-built decision model is invoked to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data that matches the pre-filled item dimensions. The historical application data is obtained based on the applicant's account information and the data source in the application data dimension architecture. Based on the basic item data and the pre-filled item data, the current application data of the target application business is obtained, and the current application data is verified to obtain the verified application data. By constructing a standardized data dimension architecture for different business types and combining basic data collection with a hierarchical intelligent pre-filling mechanism based on decision models, the system achieves automated and accurate data collection for applications. This effectively reduces the burden on applicants in complex application scenarios and significantly improves application efficiency and user experience. Pre-filling application forms through multi-level matching logic fully utilizes applicants' historical application data and current basic data, greatly improving the accuracy and adaptability of the pre-filled results. Furthermore, a verification process is introduced after generating complete application data to further ensure data compliance and consistency.

[0074] In an exemplary embodiment, the pre-built decision model includes a rule engine model, a historical matching model, and a feature-weighted decision model. The pre-built decision model is invoked to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical declaration data, obtaining pre-filled item data that matches the pre-filled item dimensions, including:

[0075] The rule engine model in the pre-built decision model is invoked to perform matching processing on the pre-filled item dimension based on the basic item data to obtain the rule matching result. If the rule matching result is a pre-built rule tree or a pre-built hash table, and there is target data that matches the pre-filled item dimension, the target data is determined as the pre-filled item data.

[0076] The rule engine model can be a deterministic reasoning model that performs conditional judgment and data matching based on preset business rules. The rule engine model includes a pre-built rule tree and a pre-built hash table. The pre-built rule tree can be a nested judgment logic organized in a tree structure. The pre-built hash table can be a key-value pair cache table that maps common input combinations to standard output.

[0077] The rule matching result can be an intermediate result returned by the rule engine model after execution, indicating whether there is target data that meets the current input conditions.

[0078] Optionally, the server invokes the rule engine model in the pre-built decision model to perform matching processing on the pre-filled item dimensions based on the basic item data, thereby obtaining rule matching results. Specifically, the server performs data preprocessing on the basic item data, such as using the Jieba word segmentation algorithm for semantic parsing to obtain keywords, and converting the basic item data into structured features. For the converted structured features, the server performs queries according to the cross rules through the pre-built rule tree, and queries the pre-built hash table for the keywords to obtain rule matching results. If the rule matching results are in the pre-built rule tree or the pre-built hash table, and there is target data that matches the pre-filled item dimension, the target data is determined as the pre-filled item data.

[0079] In this embodiment, the server first performs semantic parsing and structured transformation on the basic data to achieve semantic recognition of unstructured information. Simultaneously, fields such as amount, currency, and counterparty information are transformed into standardized structured features. On one hand, the structured features are input into a pre-built rule tree, and judgments are made layer by layer according to the cross-logic between dimensions to ensure the accurate execution of complex business rules. On the other hand, for high-frequency fixed patterns, a pre-built hash table is quickly queried based on keywords obtained from word segmentation. When any path matches the target data corresponding to the current pre-filled item dimension, intelligent filling is immediately completed. This significantly reduces manual intervention while ensuring compliance and accuracy, improving the processing efficiency of declaration data and making it suitable for declaration scenarios with high repetition and clear rules.

[0080] In one exemplary embodiment, the declaration data processing method described above further includes:

[0081] If the rule matching result shows that there is no target data that matches the pre-filled item dimension, a tuple is determined based on the pre-filled item dimension; the historical declaration data of the applicant is filtered according to the tuple to obtain candidate historical data for at least one candidate declaration business; if the candidate declaration business meets the preset association conditions, the candidate historical data is determined as pre-filled item data.

[0082] The preset association conditions are that the number of candidate application business is greater than a preset number threshold, and the confidence level between the candidate application business and the target application business is greater than a preset threshold.

[0083] Among them, a tuple can be a set of key feature combinations used to filter historical declaration data, which is usually composed of the current pre-filled item dimension and related contextual information.

[0084] Among them, the candidate application business can be a past application record that is similar to the current target application business and is selected from the applicant's historical application data; the candidate historical data can be the historical application field value corresponding to the current pre-filled item dimension in the candidate application business.

[0085] Among them, confidence level can be a quantitative indicator that measures the similarity between candidate declaration business and target declaration business, and the value range is generally [0, 1].

[0086] Optionally, if the rule matching result shows no target data matching the pre-filled dimension, the server determines a tuple based on the pre-filled dimension. For example, each pre-filled dimension is treated as an element. In the case of foreign exchange-related business, the tuple can be a quintuple, such as "transaction entity account - counterparty account - payment / receipt type - currency - core keyword". Further, the server filters the applicant's historical declaration data according to the tuple, that is, it calculates the confidence level between each historical declaration record and the quintuple, and filters according to each confidence level and a pre-set confidence threshold to obtain candidate historical data for at least one candidate declaration business. If the candidate declaration business meets the preset association conditions, the candidate historical data is determined as pre-filled data.

[0087] In this embodiment, the server dynamically generates tuples based on current business characteristics and uses them as high-dimensional similarity retrieval conditions to accurately filter the applicant's historical application data. By calculating the confidence level between each historical application record and the tuple, and combining this with a preset threshold for filtering, highly relevant candidate application businesses are identified, thereby extracting the corresponding candidate historical data as the basis for pre-filling. When the number of candidate application businesses exceeds a preset threshold and the confidence level reaches a set standard, it is determined that the preset association conditions are met, and the corresponding historical data is determined as the current pre-filled data. This mechanism effectively overcomes the limitations of single-field matching, improves the continuity and accuracy of data pre-filling in complex or new business scenarios, and is especially suitable for situations where rules are not yet covered but there are comparable historical behaviors. It enhances the system's adaptability and intelligence level, and significantly improves application efficiency and user experience while ensuring compliance.

[0088] In one exemplary embodiment, the declaration data processing method described above further includes:

[0089] If the candidate application does not meet the preset association conditions, the feature weighted decision model is invoked to extract features based on the applicant's historical application data, basic item data, and application data dimension architecture, resulting in features of multiple preset dimensions. Based on the features of each preset dimension and the preset weight coefficients of each preset dimension, the target fusion feature is obtained and the target fusion feature is determined as the pre-filled item data.

[0090] Among them, the features of multiple preset dimensions can be structured feature variables that can be extracted from basic item data, historical declaration data and declaration data dimension architecture and can be used for modeling and analysis, including fused structured features, unstructured features and implicit features.

[0091] Optionally, if a candidate application does not meet the preset association conditions, the server extracts features based on the applicant's historical application data, basic data, and application data dimensional architecture. This yields features across multiple preset dimensions, including integrated structured features (such as enterprise entity, account type, etc.), unstructured features (such as high-frequency keywords), and implicit features (such as enterprise industry, concentration of historical application types, etc.). Based on the features of each preset dimension and their preset weight coefficients (e.g., 45% for integrated structured features, 35% for unstructured features, and 20% for implicit features), a weighted sum is calculated. The total score for each pre-filled dimension is then calculated according to the weighting rules. The integrated feature whose total score meets the preset value is selected as the target integrated feature, and this target integrated feature is designated as the pre-filled data. Understandably, if no matching data is found after the feature-weighted decision model matching process, the server issues a notification, prompting business personnel to make a judgment.

[0092] In this embodiment, the server comprehensively analyzes the historical application data, current basic data, and application data dimensional architecture of the applicant to perform deep feature extraction, constructing a multi-dimensional feature system covering structured features, unstructured features, and implicit behavioral features. Differential weight coefficients are set based on the actual contribution of each feature in different application scenarios, and the comprehensive score of each possible pre-filled item dimension is calculated through a weighted summation method. Finally, the optimal result whose total score meets a preset threshold is used as the target fusion feature and determined as the current pre-filled item data. This mechanism breaks through the dependence on explicit rules and highly similar historical records, maintaining a high accuracy rate for intelligent recommendations even in data-sparse or new business scenarios. It significantly improves the robustness and generalization ability of the system in complex and marginal situations, achieving a deep progression from experience-driven to feature-driven approaches, further improving the accuracy and efficiency of automatic pre-filling of application data.

[0093] In one exemplary embodiment, the declaration data processing method described above further includes:

[0094] Obtain the declaration data dimensions for multiple declaration businesses of the same business type, and determine the sub-dimensions and data sources of each declaration data dimension; construct the declaration data dimension architecture based on the sub-dimensions and data sources of each declaration data dimension.

[0095] Optionally, the server obtains the declaration data dimensions for multiple declaration businesses of the same type according to the management specifications of the declaration business, determines the sub-dimensions and data sources of each declaration data dimension, and constructs the declaration data dimension architecture based on the mapping relationship between the sub-dimensions and data sources of each declaration data dimension. For example, Table 1 provides a declaration data dimension architecture table for foreign exchange declaration business.

[0096] Table 1

[0097]

[0098] In this embodiment, based on the management specifications of the declaration business, multiple declaration instances of the same business type are systematically collected and analyzed. Common declaration data dimensions are automatically summarized, and the sub-dimensions of each dimension and their corresponding data sources are further refined, constructing a clear and logically complete declaration data dimension architecture. This architecture not only unifies the data standards for similar businesses and realizes modular and standardized management of declaration elements, but also clarifies the generation path of each data item, providing a unified metadata foundation for subsequent intelligent pre-filling, data verification, and cross-system collaboration, laying the groundwork for fully automated declaration processes.

[0099] In one exemplary embodiment, the declaration data processing method described above further includes:

[0100] Based on the declaration data dimension architecture, a rule engine model is constructed; based on the declaration data dimension architecture and historical declaration data of each declaration business, a historical matching model is constructed; based on the declaration data dimension architecture, historical declaration data and preset weight coefficients, a feature-weighted decision model is constructed.

[0101] Optionally, the server constructs a rule engine model based on the dimensional architecture of the declared data. Specifically, it constructs a dual matching mechanism of "structured feature combination logic + unstructured keyword index" for the dimensional architecture of the declared data. In the foreign exchange declaration business scenario, the structured side forms a rule tree through the cross rules of core fields such as transaction entity type, payment and receipt type, currency, counterparty country, and counterparty account type (e.g., "overseas institution free trade account ∩ payment ∩ foreign currency ∩ China ∩ overseas institution domestic foreign exchange account" for targeted domestic payment non-resident remittance scenarios). The unstructured side relies on the keyword hash index after Jieba word segmentation (e.g., "sea freight", "freight", and "import" match the "222012" international balance of payments declaration transaction code) to achieve fast query with the help of hash tables, directly covering high-frequency scenarios. The server constructs a historical matching model based on the dimensional architecture of the declaration data and historical declaration data for each declaration business. For example, it constructs a historical feature space based on the five-tuple of "transaction entity account - counterparty account - payment / receipt type - currency - core keywords," filters timeliness data through a 180-day time window, and reuses historical declaration types using a frequency confidence rule of "≥3 records and type consistency ≥85%" to build the historical matching model. In addition, based on the dimensional architecture of the declaration data and historical declaration data, it determines the preset weight coefficients and feature dimensions to construct a feature-weighted decision model. For example, it integrates structured features, unstructured features, and implicit features, extracts transaction features, and calculates the total score for each declaration type according to the weight rules. If the set score is met, the result is output; otherwise, it is handed over to manual judgment. Furthermore, the feature weights can be dynamically optimized based on the judgment accuracy (historical declaration data modification records).

[0102] In this embodiment, the server first constructs a rule engine model based on a dimensional architecture, innovatively employing a dual-path matching mechanism: on the structured side, a rule tree is built through the cross-rules of multi-dimensional fields to accurately identify complex business scenarios; on the unstructured side, keywords in transaction remarks are extracted using Jieba word segmentation, and a hash index is established to quickly match high-frequency declaration codes, effectively covering common high-incidence scenarios. Secondly, combining historical declaration data and the five-tuple feature space, a time window and confidence rules are set to construct a historical matching model, enabling accurate reuse of user habitual behaviors and enhancing personalized service capabilities. Finally, based on the same architecture and historical data, a feature-weighted decision model is further constructed, integrating structured, unstructured, and implicit features, calculating a comprehensive score according to preset weights, and outputting recommendation results when thresholds are met; otherwise, it proceeds to manual review. Simultaneously, it supports dynamic adjustment of weight coefficients based on historical modification records, continuously improving model accuracy.

[0103] In one exemplary embodiment, another method for processing declaration data is provided. Wherein:

[0104] Step 1: Construct a multi-dimensional data system for international balance of payments reporting. Based on the management standards of the State Administration of Foreign Exchange, a structured data system with 6 major categories and 28 sub-dimensions is compiled (as shown in Table 1). This system covers the full decision-making basis for identifying reporting types and the full scope of data collection, as well as the data sources and dimensional divisions, resulting in the multi-dimensional data system for international balance of payments reporting (reporting data dimension architecture).

[0105] Step 2: Construct an intelligent decision-making model.

[0106] Specifically, data preprocessing involves: semantic parsing of unstructured data (such as keywords related to the purpose of funds) using the "Jieba word segmentation algorithm" to transform it into structured features; using "historical mean filling + rule completion" for missing data (e.g., when "counterpartner country" is missing, it is automatically filled based on historical data of the same counterparty, or if no historical data exists, the counterparty country can be automatically identified through the counterparty bank); and standardizing all data (e.g., processing the "country" field into ISO3166 standard code values).

[0107] Three-level progressive decision-making model architecture: This model constructs a three-level progressive feature filtering architecture, achieving a balance between efficiency and accuracy by focusing on different scenarios in layers: 1) First level: Rule engine, which constructs a dual matching mechanism of "structured feature combination logic + unstructured keyword index" for standardized transactions. The structured side forms a rule tree through the cross rules of core fields such as transaction entity type, payment type, currency, counterparty country, and counterparty account type (e.g., "overseas institution free trade account ∩ payment ∩ foreign currency ∩ China ∩ overseas institution domestic foreign exchange account" for targeted domestic payment non-resident remittance scenarios). The unstructured side relies on the keyword hash index after Jieba word segmentation (e.g., "sea freight", "freight", "import" match the "222012" international balance of payments declaration transaction code), and uses a hash table to achieve fast query, directly covering high-frequency scenarios. 2) Second Level: Historical Relevance Matching. A historical feature space is constructed based on the five-tuple of "transaction entity account - counterparty account - payment / receipt type - currency - core keywords". Timeliness data is filtered through a 180-day time window, and historical declaration types are reused in combination with the frequency confidence rule of "≥3 records and type consistency ≥85%". 3) Third Level: Multi-dimensional Feature Dynamic Weighted Decision. Structured features (45%, such as enterprise entity, account nature, etc.), unstructured features (35%, high-frequency keywords, etc.), and implicit features (20%, such as enterprise industry, concentration of historical declaration types of the entity, etc.) are integrated. After extracting transaction features, the total score of each declaration type is calculated according to the weight rules. The result is output after the set score is met; otherwise, it is handed over to manual judgment. In addition, the feature weights can be dynamically optimized based on the judgment accuracy (modification records of historical declaration data).

[0108] Step 3: Unify the declaration decision engine. This encapsulates the aforementioned intelligent decision-making model as an API service, achieving unified logic across multiple channels. It supports high-concurrency calls, with millisecond-level interface response times, and returns parameters including "declaration type, transaction code, and pre-filled element list." Front-end channels such as online banking, direct bank-enterprise connections, and e-commerce platforms do not need to develop independent recognition logic; they only need to obtain results through the API call engine, ensuring consistent recognition across multiple channels for similar transactions.

[0109] Step 4: Automated data collection and real-time verification;

[0110] Data Collection: Customers only need to fill in basic information (remittance / payment amount, counterparty information, receiving bank, purpose of funds, etc.). After the front-end channel calls the decision engine, more than 20 professional elements are automatically pre-filled, including declaration type, transaction code, counterparty country code, settlement method, and regulatory compatibility identifier. Customers only need to verify and confirm and supplement less than 10 non-standardized elements. Real-time Verification: Format verification: Verifies element coding rules (such as the range of transaction code enumeration values); Logical verification: Verifies data consistency (such as whether there are conflicts in transaction type, transaction code, and transaction remarks); Regulatory rule verification: Verifies compliance with the latest regulatory requirements.

[0111] This embodiment utilizes technological innovation to construct a systematic solution that addresses three core issues: First, it simplifies the customer declaration process, reduces the complexity of filling out forms, and optimizes user experience while minimizing data errors at the source. Second, it introduces intelligent technology to automatically judge, verify, and collect declaration data, reducing operational biases from manual review and ensuring the accuracy and timeliness of declarations from a technical perspective. Third, it unifies the declaration logic standards across multiple access scenarios, avoiding declaration errors caused by differences in the technical capabilities of various entities in the development of multiple systems, misunderstandings of regulatory rules, or inconsistent implementation, thus forming a standardized compliance technology paradigm. Through these technical solutions, the State Administration of Foreign Exchange's regulatory requirements for "timely and accurate" declarations can be rigidly met, while also improving the compliance efficiency of banks' cross-border business and the quality of customer service.

[0112] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0113] Based on the same inventive concept, this application also provides a data processing apparatus for implementing the above-described data processing method for filing applications. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the data processing apparatus provided below can be found in the limitations of the data processing method described above, and will not be repeated here.

[0114] In one exemplary embodiment, such as Figure 4 As shown, a declaration data processing device 400 is provided, including: a declaration dimension determination module 401, a basic data acquisition module 402, a pre-filled data acquisition module 403, and a declaration data verification module 404, wherein:

[0115] The declaration dimension determination module 401 is used to respond to the target declaration business initiated by the declarant, query the declaration data dimension architecture corresponding to the business type of the target declaration business, and determine the multiple declaration data dimensions required by the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions;

[0116] The basic data acquisition module 402 is used to obtain the basic item data corresponding to the basic item dimension in response to the input operation;

[0117] The pre-filled data acquisition module 403 is used to call a pre-built decision model to perform hierarchical matching processing on the pre-filled item dimension based on at least one of the basic item data and the applicant's historical declaration data to obtain pre-filled item data that matches the pre-filled item dimension; the historical declaration data is obtained based on the applicant's account information and the data source in the declaration data dimension architecture.

[0118] The declaration data verification module 404 is used to obtain the current declaration data of the target declaration business based on the basic item data and the pre-filled item data, and to verify the current declaration data to obtain the verified declaration data.

[0119] Furthermore, in one embodiment, the pre-filled data acquisition module 403 is also used to call the rule engine model in the pre-built decision model, and perform matching processing on the pre-filled item dimension according to the basic item data to obtain a rule matching result; the rule engine model includes a pre-built rule tree and a pre-built hash table; if the rule matching result is that there is target data matching the pre-filled item dimension in the pre-built rule tree or the pre-built hash table, the target data is determined as the pre-filled item data.

[0120] Furthermore, in one embodiment, the pre-filled data acquisition module 403 is further configured to: determine a tuple based on the pre-filled item dimension when the rule matching result indicates that there is no target data matching the pre-filled item dimension; filter the historical declaration data of the applicant according to the tuple to obtain candidate historical data for at least one candidate declaration business; and determine the candidate historical data as the pre-filled item data when the candidate declaration business meets a preset association condition; wherein the preset association condition is that the number of candidate declaration businesses is greater than a preset quantity threshold, and the confidence level between the candidate declaration business and the target declaration business is greater than a preset threshold.

[0121] Furthermore, in one embodiment, the pre-filled data acquisition module 403 is also used to extract features based on the applicant's historical application data, basic item data, and application data dimension architecture when the candidate application does not meet the preset association conditions, to obtain features of multiple preset dimensions; to obtain target fusion features based on the features of each preset dimension and the preset weight coefficients of each preset dimension, and to determine the target fusion features as the pre-filled item data.

[0122] Furthermore, in one embodiment, the declaration dimension determination module 401 is also used to obtain declaration data dimensions of multiple declaration businesses of the same business type, and determine the sub-dimensions and data sources of each declaration data dimension; and construct a declaration data dimension architecture based on the sub-dimensions and data sources of each declaration data dimension.

[0123] Furthermore, in one embodiment, the declaration data processing device 400 further includes a model building module, used to build a rule engine model based on the declaration data dimension architecture; build a historical matching model based on the declaration data dimension architecture and historical declaration data of each of the declaration businesses; and build a feature-weighted decision model based on the declaration data dimension architecture, historical declaration data, and preset weight coefficients.

[0124] Each module in the aforementioned data processing device 400 can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0125] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores data such as the declaration data dimension structure, account information, and historical declaration data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a declaration data processing method.

[0126] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0127] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0128] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0129] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0130] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0131] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0132] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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 application.

[0133] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for processing declaration data, characterized in that, The method includes: In response to a target declaration business initiated by the declarant, the declaration data dimension architecture corresponding to the business type of the target declaration business is queried to determine the multiple declaration data dimensions required for the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions. In response to the input operation, the basic item data corresponding to the basic item dimension is obtained; A pre-built decision model is invoked to perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data that matches the pre-filled item dimensions; the historical application data is obtained based on the applicant's account information and the data source in the application data dimension architecture; Based on the basic data and the pre-filled data, the current application data for the target application business is obtained, and the current application data is verified to obtain the verified application data.

2. The method according to claim 1, characterized in that, The pre-built decision model includes a rule engine model, a historical matching model, and a feature-weighted decision model. The step of calling the pre-built decision model involves performing hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical application data, to obtain pre-filled item data that matches the pre-filled item dimensions, including: The rule engine model in the pre-built decision model is invoked, and the pre-filled item dimensions are matched according to the basic item data to obtain the rule matching result; the rule engine model includes a pre-built rule tree and a pre-built hash table; If the rule matching result is the pre-built rule tree or the pre-built hash table, and there is target data that matches the pre-filled item dimension, then the target data is determined as the pre-filled item data.

3. The method according to claim 2, characterized in that, The method further includes: If the rule matching result indicates that there is no target data that matches the pre-filled dimension, a tuple is determined based on the pre-filled dimension. The historical application data of the applicant is filtered according to the tuple to obtain candidate historical data for at least one candidate application business. If the candidate application business meets the preset association conditions, the candidate historical data is determined as the pre-filled item data; the preset association conditions are that the number of candidate application businesses is greater than a preset quantity threshold, and the confidence level between the candidate application business and the target application business is greater than a preset threshold.

4. The method according to claim 3, characterized in that, The method further includes: If the candidate application does not meet the preset association conditions, the feature weighted decision model is invoked to extract features based on the applicant's historical application data, basic item data, and application data dimension architecture, thereby obtaining features in multiple preset dimensions. Based on the features of each preset dimension and the preset weight coefficients of each preset dimension, the target fusion features are obtained, and the target fusion features are determined as the pre-filled data.

5. The method according to claim 1, characterized in that, The method further includes: Obtain the declaration data dimensions for multiple declaration businesses of the same business type, and determine the sub-dimensions and data sources for each declaration data dimension; Based on the sub-dimensions and data sources of each declared data dimension, a declared data dimension architecture is constructed.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: Based on the aforementioned data dimension architecture, a rule engine model is constructed; Based on the aforementioned data dimension architecture and the historical application data of each application business, a historical matching model is constructed; Based on the declared data dimension architecture, historical declared data, and preset weight coefficients, a feature-weighted decision model is constructed.

7. A data processing device for application submissions, characterized in that, The device includes: The declaration dimension determination module is used to respond to a target declaration business initiated by the declarant, query the declaration data dimension architecture corresponding to the business type of the target declaration business, and determine the multiple declaration data dimensions required for the target declaration business; the declaration data dimensions include basic item dimensions and pre-filled item dimensions; The basic data acquisition module is used to obtain the basic item data corresponding to the basic item dimension in response to the input operation; The pre-filled data acquisition module is used to call a pre-built decision model and perform hierarchical matching processing on the pre-filled item dimensions based on at least one of the basic item data and the applicant's historical declaration data to obtain pre-filled item data that matches the pre-filled item dimensions; the historical declaration data is obtained based on the applicant's account information and the data source in the declaration data dimension architecture; The declaration data verification module is used to obtain the current declaration data of the target declaration business based on the basic item data and the pre-filled item data, and to verify the current declaration data to obtain the verified declaration data.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images