A method for comparing and analyzing customs manifest and declaration form based on neural network

Abstract

The present invention relates to the field of real-time comparison and analysis of customs manifest and declaration form, and particularly to a method for comparing and analyzing customs manifest and declaration form based on neural network, including: respectively obtaining the text data features of customs manifest and declaration form; establishing a customs manifest data association model based on neural network by using the text data features of the customs manifest, and establishing a declaration form data comparison model based on neural network by using the text data features of the declaration form; obtaining the comparison and analysis result of customs manifest and declaration form by using the customs manifest data association model and the declaration form data comparison model, optimizing the data matching logic through deep learning, reducing the misjudgment rate, improving the reliability of the comparison result, being able to process structured and unstructured data, adapting to different formats of manifest and declaration form, improving the versatility of the system, adopting a feedback verification mechanism, continuously optimizing the model, and improving the stability and adaptability for long-term use.

Description

Technical Field

[0001] The present invention relates to the field of real-time comparison and analysis of customs manifests and declaration forms, and particularly to a method for comparing and analyzing customs manifests and declaration forms based on neural networks. Background Art

[0002] With the rapid development of international trade, the complexity and data volume of customs supervision have increased sharply. The comparison and analysis of customs manifests and declaration forms have become an important link to ensure trade security and improve customs clearance efficiency. Traditional manual comparison methods are inefficient, error-prone, and difficult to meet the real-time processing requirements of large-scale data. Existing automated comparison systems are mostly based on rule matching or simple data verification algorithms, lacking the ability to deeply mine text data features and perform intelligent analysis, resulting in insufficient accuracy and adaptability of comparison results. In addition, the existing technology for comparing customs manifests and declaration forms is usually limited to single-dimensional data verification, lacking the ability to comprehensively model data positions, types, contents, and historical correlations, resulting in insufficient integrity and reliability of comparison results. Therefore, how to use neural network technology to construct an association model and a comparison model for customs manifests and declaration forms to achieve efficient and accurate automated analysis has become an urgent technical problem in the field of intelligent processing of customs data. Summary of the Invention

[0003] Aiming at the deficiencies of the existing technology, the present invention provides a method for comparing and analyzing customs manifests and declaration forms based on neural networks. By extracting the data feature contents of different customs documents and establishing a multi-level model based on neural networks, the analysis and comparison results can be obtained efficiently and accurately.

[0004] To achieve the above object, the present invention provides a method for comparing and analyzing customs manifests and declaration forms based on neural networks, including:

[0005] S1. Respectively obtain the text data features of the customs manifest and the declaration form;

[0006] S2. Establish a customs manifest data association model based on the text data features of the customs manifest using a neural network;

[0007] S3. Establish a declaration form data comparison model based on the text data features of the declaration form using a neural network;

[0008] S4. Obtain the comparison and analysis results of the customs manifest and the declaration form using the customs manifest data association model and the declaration form data comparison model.

[0009] Preferably, the step of respectively obtaining the text data features of the customs manifest and the declaration form includes:

[0010] Obtain the text data volume, text data type, and text data content of the real-time customs manifest;

[0011] Obtain the text data volume, text data type, and text data content of the real-time customs declaration form;

[0012] Obtain the customs manifest text table data according to the text data content of the real-time customs manifest;

[0013] Obtain the customs declaration form text table data according to the text data content of the real-time customs declaration form;

[0014] Use the text data volume, text data type, text data content, customs manifest text table data of the customs manifest and the text data volume, text data type, text data content, customs declaration form text table data of the real-time customs declaration form as the text data features of the customs manifest and the customs declaration form;

[0015] Among them, the customs manifest text table data includes the text table type, text table content, and text table arrangement order, and the customs declaration form text table data includes the text table type, text table content, and text table arrangement order.

[0016] Further, establishing a customs manifest data association model based on the neural network using the text data features of the customs manifest includes:

[0017] S2-1. Establish a customs manifest data analysis model based on the neural network using the text data features of the customs manifest corresponding to the text data volume, text data type, and text data content of the customs manifest;

[0018] S2-2. Establish a customs manifest data verification model based on the neural network using the customs manifest text table data;

[0019] S2-3. Use the customs manifest data analysis model and the customs manifest data verification model as the customs manifest data association model.

[0020] Further, establishing a customs manifest data analysis model based on the neural network using the text data features of the customs manifest corresponding to the text data volume, text data type, and text data content of the customs manifest includes:

[0021] Obtain the historical text data volume and historical text data content of the corresponding customs manifest according to the text data features of the customs manifest corresponding to the text data volume and text data content of the customs manifest respectively;

[0022] Use the historical text data content as the input, the historical text data volume as the output, and obtain the hidden layer structure according to the text data type of the customs manifest corresponding to the text data features of the customs manifest, and establish a basic corresponding model of the customs manifest data through training based on the neural network.

[0023] Further, establishing a customs manifest data verification model based on the neural network using the customs manifest text table data includes:

[0024] S2-2-1. Using the customs manifest text table data to obtain the data position markers of the customs manifest according to the text data characteristics of the customs manifest corresponding to the text data content;

[0025] S2-2-2. Using the customs manifest text table data to obtain the data corresponding markers of the customs manifest according to the text data characteristics of the customs manifest corresponding to the text data type;

[0026] S2-2-3. Using the customs manifest text table data as the input, the data position markers and data corresponding markers of the customs manifest as the output, and obtaining the hidden layer structure according to the text data characteristics of the customs manifest corresponding to the text data type, and training based on the neural network to establish an initial customs manifest data verification model;

[0027] S2-2-4. Obtaining the corresponding historical customs manifest text table data, historical data position markers and historical data corresponding markers of the customs manifest according to the customs manifest text table data;

[0028] S2-2-5. Inputting the historical customs manifest text table data into the initial customs manifest data verification model to obtain the initial customs manifest data verification result;

[0029] S2-2-6. Judging whether the initial customs manifest data verification result is corresponding to both the historical data position markers and historical data corresponding markers of the customs manifest. If so, using the initial customs manifest data verification model as the customs manifest data verification model. Otherwise, updating the customs manifest text table data with the non-corresponding initial customs manifest data verification result corresponding to the historical customs manifest text table data, and returning to S2-2-3.

[0030] Further, establishing a bill of entry data comparison model based on the neural network using the text data characteristics of the bill of entry includes:

[0031] S3-1. Establishing a bill of entry data self-verification model based on the neural network using the text data characteristics of the bill of entry;

[0032] S3-2. Performing feedback verification processing using the bill of entry data self-verification model to obtain the bill of entry data comparison model.

[0033] Further, establishing a bill of entry data self-verification model based on the neural network using the text data characteristics of the bill of entry includes:

[0034] S3-1-1. Obtain the historical text data volume, historical text data type, historical text data content, and historical customs declaration form text table data of the corresponding real-time customs declaration form according to the text data characteristics of the customs declaration form;

[0035] S3-1-2. Use the historical text data type and the text table type of the historical customs declaration form text table data as inputs, and the association result of the historical text data type and the text table type of the historical customs declaration form text table data as the output, and train a first-level proofreading model based on a neural network;

[0036] S3-1-3. Use the historical text data content and the text table content of the historical customs declaration form text table data as inputs, and the association result of the historical text data content and the text table content of the historical customs declaration form text table data as the output, and train a second-level proofreading model based on a neural network;

[0037] S3-1-4. Use the first-level proofreading model and the second-level proofreading model as the customs declaration form data self-proofreading model.

[0038] Further, using the customs declaration form data self-proofreading model to perform feedback verification processing to obtain a customs declaration form data comparison model includes:

[0039] S3-2-1. Use the text table arrangement order of the customs declaration form text table data as the model output verification label;

[0040] S3-2-2. Input the text data type of the real-time customs declaration form and the text table type of the customs declaration form text table data into the first-level proofreading model of the customs declaration form data self-proofreading model to obtain a first-level proofreading result;

[0041] S3-2-3. Input the text data type of the real-time customs declaration form and the text table type of the customs declaration form text table data into the second-level proofreading model of the customs declaration form data self-proofreading model to obtain a second-level proofreading result;

[0042] S3-2-4. Determine whether the first-level proofreading result corresponds to the text table arrangement order of the historical customs declaration form text table data. If so, execute S3-2-5; otherwise, obtain the text data type corresponding to the non-corresponding first-level proofreading result and the text table type of the customs declaration form text table data, update the historical text data type and the text table type of the historical customs declaration form text table data, and return to S3-1-2;

[0043] S3-2-5. Determine whether the text table arrangement order of the secondary proofreading result corresponds to that of the historical customs declaration form text table data. If so, use the customs declaration form data self-proofreading model as the customs declaration form data comparison model. Otherwise, obtain the text data content corresponding to the non-corresponding secondary proofreading result and the text table content of the customs declaration form text table data to update the text data content of the historical text data and the text table content of the historical customs declaration form text table data, and return to S3-1-3.

[0044] Further, the customs manifest and customs declaration form comparison and analysis results obtained by using the customs manifest data association model and the customs declaration form data comparison model include:

[0045] S4-1. Respectively obtain the real-time text data features of the customs manifest and the customs declaration form;

[0046] S4-2. Input the real-time text data features of the customs manifest into the customs manifest data association model to sequentially obtain the real-time customs manifest data basic correspondence result and the real-time customs manifest data verification result;

[0047] S4-3. Input the real-time text data features of the customs declaration form into the customs declaration form data comparison model to obtain the real-time customs declaration form data comparison result;

[0048] S4-4. Determine whether the real-time customs manifest data basic correspondence result and the real-time customs declaration form data comparison result are completely corresponding. If so, execute S4-5. Otherwise, output the non-corresponding content of the real-time customs manifest data basic correspondence result and the real-time customs declaration form data comparison result as the customs manifest and customs declaration form comparison and analysis result;

[0049] S4-5. Determine whether the real-time customs manifest data verification result and the real-time customs declaration form data comparison result are completely corresponding. If so, output the real-time customs manifest data basic correspondence result, the real-time customs manifest data verification result, and the real-time customs declaration form data comparison result as the customs manifest and customs declaration form comparison and analysis result. Otherwise, output the non-corresponding content of the real-time customs manifest data verification result and the real-time customs declaration form data comparison result as the customs manifest and customs declaration form comparison and analysis result.

[0050] Compared with the closest prior art, the beneficial effects of the present invention are:

[0051] Improve the comparison efficiency: Use the neural network to automatically extract data features and establish an association model, reduce manual intervention, and significantly improve the data processing speed;

[0052] Enhance the accuracy: Optimize the data matching logic through deep learning, reduce the misjudgment rate, and improve the reliability of the comparison result;

[0053] Adapt to complex data: It can process structured and unstructured data, adapt to manifests and customs declarations in different formats, and improve the generality of the system;

[0054] Dynamic optimization ability: Adopt a feedback verification mechanism to continuously optimize the model and improve the stability and adaptability for long-term use. Brief Description of the Drawings

[0055] Figure 1 It is a flowchart of a method for comparing and analyzing customs manifests and customs declarations based on a neural network provided by the present invention. Detailed Embodiments

[0056] The following further elaborates on the detailed embodiments of the present invention with reference to the accompanying drawings.

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0058] Embodiment 1:

[0059] The present invention provides a method for comparing and analyzing customs manifests and customs declarations based on a neural network, as Figure 1 shown, including:

[0060] S1. Respectively obtain the text data features of the customs manifest and the customs declaration;

[0061] S2. Based on the text data features of the customs manifest, establish a customs manifest data association model based on a neural network;

[0062] S3. Based on the text data features of the customs declaration, establish a customs declaration data comparison model based on a neural network;

[0063] S4. Use the customs manifest data association model and the customs declaration data comparison model to obtain the comparison and analysis result of the customs manifest and the customs declaration.

[0064] S1 specifically includes:

[0065] S1-1. Obtain the text data volume, text data type, and text data content of the real-time customs manifest;

[0066] S1-2. Obtain the text data volume, text data type, and text data content of the real-time customs declaration;

[0067] S1-3. Obtain the customs manifest text table data based on the text data content of the real-time customs manifest;

[0068] S1-4. Obtain the declaration form text table data based on the text data content of the real-time declaration form;

[0069] S1-5. Use the text data volume, text data type, text data content of the customs manifest, the customs manifest text table data, and the text data volume, text data type, text data content, and declaration form text table data of the real-time declaration form as the text data features of the customs manifest and the declaration form;

[0070] Among them, the customs manifest text table data includes the text table type, text table content, and text table arrangement order, and the declaration form text table data includes the text table type, text table content, and text table arrangement order.

[0071] S2 specifically includes:

[0072] S2-1. Establish a customs manifest data analysis model based on the neural network using the text data features of the customs manifest corresponding to the text data volume, text data type, and text data content of the customs manifest;

[0073] S2-2. Establish a customs manifest data verification model based on the neural network using the customs manifest text table data;

[0074] S2-3. Use the customs manifest data analysis model and the customs manifest data verification model as the customs manifest data association model.

[0075] In this embodiment, a method for comparing and analyzing a customs manifest and a declaration form based on a neural network, the specific definitions of the customs manifest data analysis model and the customs manifest data verification model include:

[0076] Data analysis model:

[0077] Input: Historical manifest data content;

[0078] Output: Historical manifest data volume;

[0079] Training method: Train based on a neural network (such as LSTM, Transformer) to establish a mapping relationship between the data volume and the content;

[0080] Data verification model:

[0081] Input: Manifest table data;

[0082] Output: Data position mark, data corresponding mark;

[0083] Training method: Use CNN or graph neural network (GNN) to process tabular data to ensure the matching of data positions and types.

[0084] S2-1 specifically includes:

[0085] S2-1-1. Obtain the historical text data volume and historical text data content of the corresponding customs manifest according to the text data volume and text data content of the customs manifest corresponding to the text data characteristics of the customs manifest.

[0086] S2-1-2. Use the historical text data content as the input, the historical text data volume as the output, and obtain the hidden layer structure according to the text data type corresponding to the text data characteristics of the customs manifest. Based on the neural network, train and establish a basic corresponding model for customs manifest data.

[0087] S2-2 specifically includes:

[0088] S2-2-1. Use the customs manifest text table data to obtain the data position mark of the customs manifest according to the text data content corresponding to the text data characteristics of the customs manifest.

[0089] S2-2-2. Use the customs manifest text table data to obtain the data corresponding mark of the customs manifest according to the text data type corresponding to the text data characteristics of the customs manifest.

[0090] S2-2-3. Use the customs manifest text table data as the input, the data position mark and data corresponding mark of the customs manifest as the output, and obtain the hidden layer structure according to the text data type corresponding to the text data characteristics. Based on the neural network, train and establish an initial customs manifest data verification model.

[0091] S2-2-4. Obtain the corresponding historical customs manifest text table data, historical data position mark and historical data corresponding mark of the customs manifest according to the customs manifest text table data.

[0092] S2-2-5. Input the historical customs manifest text table data into the initial customs manifest data verification model to obtain the initial customs manifest data verification result.

[0093] S2-2-6. Judge whether the initial customs manifest data verification result, historical data position mark and historical data corresponding mark of the customs manifest are all corresponding. If so, use the initial customs manifest data verification model as the customs manifest data verification model. Otherwise, use the non-corresponding initial customs manifest data verification result to update the customs manifest text table data corresponding to the historical customs manifest text table data, and return to S2-2-3.

[0094] S3 specifically includes:

[0095] S3-1. Establish a self-verification model for customs declaration form data based on the text data characteristics of the customs declaration form using a neural network;

[0096] S3-2. Use the self-verification model for customs declaration form data to perform feedback verification processing to obtain a comparison model for customs declaration form data.

[0097] In this embodiment, a method for comparing and analyzing customs manifest and customs declaration form based on a neural network. The specific definition of the comparison model for customs declaration form data includes:

[0098] Self-verification model:

[0099] First-level verification: Compare whether the data type matches the form type;

[0100] Second-level verification: Compare whether the data content matches the form content;

[0101] Feedback verification:

[0102] If the comparison result is inconsistent with the historical data, adjust the model parameters and retrain to improve accuracy.

[0103] S3-1 specifically includes:

[0104] S3-1-1. Obtain the historical text data volume, historical text data type, historical text data content, and historical customs declaration form text table data of the corresponding real-time customs declaration form according to the text data characteristics of the customs declaration form;

[0105] S3-1-2. Use the text table type of the historical text data type and the historical customs declaration form text table data as the input, and the association result of the text table type of the historical text data type and the historical customs declaration form text table data as the output, and train based on a neural network to obtain a first-level verification model;

[0106] S3-1-3. Use the text table content of the historical text data content and the historical customs declaration form text table data as the input, and the association result of the text table content of the historical text data content and the historical customs declaration form text table data as the output, and train based on a neural network to obtain a second-level verification model;

[0107] S3-1-4. Use the first-level verification model and the second-level verification model as the self-verification model for customs declaration form data.

[0108] S3-2 specifically includes:

[0109] S3-2-1. Use the text table arrangement order of the customs declaration form text table data as the model output verification label;

[0110] S3-2-2. Input the text data type of the real-time customs declaration form and the text table type of the customs declaration form text table data into the first-level verification model of the customs declaration data self-verification model to obtain the first-level verification result;

[0111] S3-2-3. Input the text data type of the real-time customs declaration form and the text table type of the customs declaration form text table data into the second-level verification model of the customs declaration data self-verification model to obtain the second-level verification result;

[0112] S3-2-4. Determine whether the arrangement order of the text tables of the first-level verification result corresponds to that of the historical customs declaration form text table data. If so, execute S3-2-5; otherwise, obtain the text data type corresponding to the non-corresponding first-level verification result and the text table type of the customs declaration form text table data, update the historical text data type and the text table type of the historical customs declaration form text table data, and return to S3-1-2;

[0113] S3-2-5. Determine whether the arrangement order of the text tables of the second-level verification result corresponds to that of the historical customs declaration form text table data. If so, use the customs declaration data self-verification model as the customs declaration data comparison model; otherwise, obtain the text data content corresponding to the non-corresponding second-level verification result and the text table content of the customs declaration form text table data, update the historical text data content and the text table content of the historical customs declaration form text table data, and return to S3-1-3.

[0114] S4 specifically includes:

[0115] S4-1. Obtain the real-time text data features of the customs manifest and the customs declaration form respectively;

[0116] S4-2. Input the real-time text data features of the customs manifest into the customs manifest data association model to obtain the real-time customs manifest data basic correspondence result and the real-time customs manifest data verification result in sequence;

[0117] S4-3. Input the real-time text data features of the customs declaration form into the customs declaration data comparison model to obtain the real-time customs declaration data comparison result;

[0118] S4-4. Determine whether the real-time customs manifest data basic correspondence result and the real-time customs declaration data comparison result are completely corresponding. If so, execute S4-5; otherwise, output the non-corresponding content of the real-time customs manifest data basic correspondence result and the real-time customs declaration data comparison result as the customs manifest and customs declaration form comparison analysis result;

[0119] S4-5. Determine whether the real-time customs manifest data verification result and the real-time customs declaration form data comparison result are exactly corresponding. If so, output the real-time customs manifest data basic corresponding result, the real-time customs manifest data verification result, and the real-time customs declaration form data comparison result as the customs manifest and customs declaration form comparison and analysis result. Otherwise, output the content where the real-time customs manifest data verification result and the real-time customs declaration form data comparison result do not correspond as the customs manifest and customs declaration form comparison and analysis result.

[0120] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0121] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0122] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device realizes the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0123] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0124] Finally, it should be noted that 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 above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent substitutions can still be made to the specific implementation manners of the present invention, and any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for comparing and analyzing customs manifest and customs declaration form based on neural network, characterized in that, Including: S1. Respectively obtain the text data features of the customs manifest and the customs declaration form; S1-1. Obtain the text data volume, text data type, and text data content of the real-time customs manifest; S1-2. Obtain the text data volume, text data type, and text data content of the real-time customs declaration form; S1-3. Obtain the customs manifest text table data according to the text data content of the real-time customs manifest; S1-4. Obtain the customs declaration form text table data according to the text data content of the real-time customs declaration form; S1-5. Use the text data volume, text data type, text data content, customs manifest text table data of the customs manifest and the text data volume, text data type, text data content, customs declaration form text table data of the real-time customs declaration form as the text data features of the customs manifest and the customs declaration form; Among them, the customs manifest text table data includes the text table type, text table content, and text table arrangement order, and the customs declaration form text table data includes the text table type, text table content, and text table arrangement order; S2. Use the text data features of the customs manifest to establish a customs manifest data association model based on a neural network; S2-1. Use the text data features of the customs manifest corresponding to the text data volume, text data type, and text data content of the customs manifest to establish a customs manifest data analysis model based on a neural network; S2-2. Use the customs manifest text table data to establish a customs manifest data verification model based on a neural network; S2-3. Use the customs manifest data analysis model and the customs manifest data verification model as the customs manifest data association model; S3. Use the text data features of the customs declaration form to establish a customs declaration form data comparison model based on a neural network; S3-1. Use the text data features of the customs declaration form to establish a customs declaration form data self-verification model based on a neural network; S3-2. Use the customs declaration form data self-verification model to perform feedback verification processing to obtain the customs declaration form data comparison model; S4. Use the customs manifest data association model and the customs declaration form data comparison model to obtain the comparison and analysis results of the customs manifest and the customs declaration form.

2. The method for comparing and analyzing the customs manifest and the declaration form based on a neural network according to claim 1, wherein Using the text data features of the customs manifest corresponding to the text data volume, text data type, and text data content of the customs manifest to establish a customs manifest data analysis model based on a neural network includes: Respectively obtain the historical text data volume and historical text data content of the corresponding customs manifest according to the text data features of the customs manifest corresponding to the text data volume and text data content of the customs manifest; Use the historical text data content as the input, the historical text data volume as the output, and obtain the hidden layer structure according to the text data type of the customs manifest corresponding to the text data features of the customs manifest, and perform training based on a neural network to establish a basic corresponding model of customs manifest data.

3. The method for comparing and analyzing the customs manifest and the declaration form based on a neural network according to claim 2, wherein Using the customs manifest text table data to establish a customs manifest data verification model based on a neural network includes: S2-2-1. Use the customs manifest text table data to obtain the data position mark of the customs manifest according to the text data content corresponding to the text data features of the customs manifest; S2-2-2. Obtain the data corresponding mark of the customs manifest according to the text data type corresponding to the text data characteristics of the customs manifest by using the customs manifest text table data; S2-2-3. Use the customs manifest text table data as the input, the data position mark and data corresponding mark of the customs manifest as the output, and obtain the hidden layer structure according to the text data type corresponding to the text data characteristics of the customs manifest, and establish an initial customs manifest data verification model based on neural network training; S2-2-4. Obtain the corresponding historical customs manifest text table data, historical data position mark and historical data corresponding mark of the customs manifest according to the customs manifest text table data; S2-2-5. Input the historical customs manifest text table data into the initial customs manifest data verification model to obtain the initial customs manifest data verification result; S2-2-6. Judge whether the initial customs manifest data verification result, the historical data position mark and historical data corresponding mark of the customs manifest are all corresponding. If so, use the initial customs manifest data verification model as the customs manifest data verification model. Otherwise, update the customs manifest text table data with the non-corresponding initial customs manifest data verification result corresponding to the historical customs manifest text table data, and return to S2-2-3.

4. The method for comparing and analyzing the customs manifest and the declaration form based on a neural network according to claim 1, characterized in that Establishing a self-verification model for declaration form data based on neural network by using the text data characteristics of the declaration form includes: S3-1-1. Obtain the historical text data volume, historical text data type, historical text data content and historical declaration form text table data corresponding to the real-time declaration form according to the text data characteristics of the declaration form; S3-1-2. Use the historical text data type and the text table type of the historical declaration form text table data as the input, and the correlation result of the historical text data type and the text table type of the historical declaration form text table data as the output, and perform training based on neural network to obtain a primary verification model; S3-1-3. Use the historical text data content and the text table content of the historical declaration form text table data as the input, and the correlation result of the historical text data content and the text table content of the historical declaration form text table data as the output, and perform training based on neural network to obtain a secondary verification model; S3-1-4. Use the primary verification model and the secondary verification model as the self-verification model for declaration form data.

5. The method for comparing and analyzing the customs manifest and the customs declaration form based on a neural network according to claim 4, characterized in that, Performing feedback verification processing by using the self-verification model for declaration form data to obtain a comparison model for declaration form data includes: S3-2-1. Use the text table arrangement order of the declaration form text table data as the model output verification label; S3-2-2. Input the text data type of the real-time declaration form and the text table type of the declaration form text table data into the primary verification model of the self-verification model for declaration form data to obtain a primary verification result; S3-2-3. Input the text data type of the real-time declaration form and the text table type of the declaration form text table data into the secondary verification model of the self-verification model for declaration form data to obtain a secondary verification result; S3-2-4. Determine whether the text table arrangement order of the first-level proofreading result corresponds to that of the historical customs declaration form text table data. If so, execute S3-2-5. Otherwise, obtain the text data types corresponding to the non-corresponding first-level proofreading results and the text table types of the customs declaration form text table data, update the text table types of the historical text data and the historical customs declaration form text table data, and return to S3-1-2; S3-2-5. Determine whether the text table arrangement order of the second-level proofreading result corresponds to that of the historical customs declaration form text table data. If so, use the customs declaration form data self-proofreading model as the customs declaration form data comparison model. Otherwise, obtain the text data content corresponding to the non-corresponding second-level proofreading results and the text table content of the customs declaration form text table data, update the text table content of the historical text data and the historical customs declaration form text table data, and return to S3-1-3.

6. The method for comparing and analyzing the customs manifest and the declaration form based on a neural network according to claim 2, wherein Obtaining the customs manifest and customs declaration comparison analysis result by using the customs manifest data association model and the customs declaration form data comparison model includes: S4-1. Obtain the real-time text data features of the customs manifest and the customs declaration form respectively; S4-2. Input the real-time text data features of the customs manifest into the customs manifest data association model to obtain the real-time customs manifest data basic correspondence result and the real-time customs manifest data verification result in sequence; S4-3. Input the real-time text data features of the customs declaration form into the customs declaration form data comparison model to obtain the real-time customs declaration form data comparison result; S4-4. Determine whether the real-time customs manifest data basic correspondence result and the real-time customs declaration form data comparison result are completely corresponding. If so, execute S4-5. Otherwise, output the non-corresponding content between the real-time customs manifest data basic correspondence result and the real-time customs declaration form data comparison result as the customs manifest and customs declaration comparison analysis result; S4-5. Determine whether the real-time customs manifest data verification result and the real-time customs declaration form data comparison result are completely corresponding. If so, output the real-time customs manifest data basic correspondence result, the real-time customs manifest data verification result, and the real-time customs declaration form data comparison result as the customs manifest and customs declaration comparison analysis result. Otherwise, output the non-corresponding content between the real-time customs manifest data verification result and the real-time customs declaration form data comparison result as the customs manifest and customs declaration comparison analysis result.

Images