Customs declaration code determination method

Abstract

The invention discloses a customs declaration code determination method. The method comprises the following specific steps: collecting information of commodities to be declarated; the collected commodity information is processed; establishing a rule engine based on customs coding rules; performing similarity comparison on the to-be-declarated commodity and the similar commodity with the determined code to achieve the purpose of verifying the rationality of the commodity code; and outputting the adjusted and verified customs declaration code as a final result. And generating an electronic document according to the customs declaration code, the related commodity information and the determined code basis, and storing the electronic document. According to the customs declaration code determination method, challenges caused by commodity complexity and diversity in international trade can be effectively handled, accuracy and efficiency of customs declaration code determination are remarkably improved, trade risks caused by coding errors are reduced, and smooth proceeding of the international trade is guaranteed. According to the method, the generated codes can be stored, and subsequent query and auditing are facilitated.

Description

Technical Field

[0001] The present invention relates to the technical field of customs declaration, in particular to a method for determining a customs declaration code. Background Art

[0002] Customs declaration is a crucial step in international trade. The HS Code is a standard code used to classify and count goods in international trade. It plays a key role in customs supervision, taxation, statistics, and the implementation of trade policies. With the continuous expansion of global trade and the increasing variety of traded goods, accurately determining customs declaration codes has become increasingly difficult. Currently, the determination of customs declaration codes relies primarily on the experience of customs declaration personnel and consulting the HS Code Manual. However, this traditional approach presents numerous challenges. Firstly, commodity information is complex and diverse, making it difficult for customs declaration personnel to accurately classify emerging and multifunctional commodities solely based on experience. For example, wearable devices with both smart communication and health monitoring capabilities have functional characteristics across multiple commodity categories, which can easily lead to coding errors. Secondly, the HS Code Manual is vast and frequently updated, making manual searches inefficient and prone to omissions and misunderstandings. Furthermore, subtle differences in the classification of certain commodities may exist between different countries and regions, further complicating the task of accurately determining customs declaration codes. Consequently, new methods for determining customs declaration codes have emerged on the market.

[0003] For example, Chinese patent publication number CN116776831B discloses a method and medium for determining customs declaration codes and constructing a decision tree, which can improve the classification coding of items while ensuring the accuracy of the classification coding. The method for determining customs codes includes: obtaining customs declaration information of the items to be processed; obtaining a description text of the items to be processed based on the customs declaration information; identifying content in the description text that belongs to a set dimension to obtain an identification result; matching the identification result with a decision tree, and determining at least one matching node in the decision tree corresponding to the identification result; generating a decision tree based on the correspondence between the description information of multiple items and the classification codes; wherein the nodes of the decision tree correspond to a classification code and carry the description information of the item corresponding to the classification code; the nodes of the decision tree include at least one matching node; and determining the customs code of the items to be processed based on the coding depth and the at least one matching node.

[0004] The customs code determination method provided by the aforementioned patent can set the code depth (i.e., the number of digits in the code) based on the detailed requirements of the officially published customs code, facilitating efficient positioning of the correct decision tree matching node and obtaining the required code. However, existing customs code determination methods fail to verify the code, making it prone to errors. This can lead to a series of problems, such as customs clearance delays and tax calculation errors, hindering the smooth progress of international trade and increasing trade costs and the burden on businesses. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for determining a customs declaration code, aiming to improve the problem that existing customs code determination methods cannot verify the code, resulting in coding errors, which in turn causes a series of problems such as customs clearance delays and tax calculation errors, affecting the smooth progress of international trade and increasing trade costs and corporate burdens.

[0006] The present invention is achieved in that: A method for determining a customs declaration code, the specific steps of the method are as follows: S100, collecting information on goods to be declared; S200: Processing the collected product information; S300. Establish a rule engine based on customs coding rules; S400: Compare the commodity to be declared with similar commodities with determined codes to verify the rationality of the commodity codes; S500: The adjusted and verified customs declaration code is output as the final result; S600: Generate an electronic document with the customs declaration code, related commodity information, and the basis for determining the code for storage.

[0007] Preferably, in step S100, the specific steps of collecting information are as follows: S110, configure the data collection interface to connect the trade order system and database; S120. Obtain basic information of the product from the trade order system, including product name, model, specifications, quantity, and price; S130. Extract more in-depth technical parameters and functional descriptions from product specifications or technical documents; S140. Obtain the trade mode, import and export countries or regions, and mode of transportation from the trade contract; S150. Clean the collected information, remove duplicate and useless information, and standardize the information.

[0008] Preferably, in step S200, the specific steps of processing the product information are as follows: S210, extracting keywords from the product information using a method based on a term frequency-inverse document frequency (TF-IDF) algorithm combined with part-of-speech tagging; S220: Construct product features based on keywords, and each keyword corresponds to a dimension in the feature vector, whose value can be binary (0 or 1, indicating whether the keyword exists) or assigned a certain weight value based on the importance of the keyword; S230, using a support vector machine learning algorithm, and using a large amount of accurately classified commodity data to train a classification model.

[0009] Preferably, in step S300, after the rule engine is established, when the intelligent classification module outputs the preliminary customs declaration code, the rule engine checks and adjusts the code according to the commodity information and customs coding rules.

[0010] Preferably, in step S400, the similarity comparison needs to be calculated. When calculating the similarity of goods, the feature vectors of the goods to be declared and the known goods are substituted into the formula: in, and are the characteristic vectors of the two commodities, and It is their model.

[0011] Preferably, in step S400, when encountering commodities that are difficult to accurately classify, fuzzy matching technology is used to fuzzy match commodity features with commodity features in the historical customs declaration case library. At the same time, the case learning mechanism will continuously update the case library and model based on new customs declaration cases.

[0012] Preferably, in step S500, the output coding format is strictly in accordance with the format prescribed by the customs, including coding at various levels such as chapter, item, and sub-item, to ensure the accuracy and standardization of the coding.

[0013] Preferably, in step S600, the stored electronic document is subsequently queried and audited in English, and the stored electronic document is also used as a new training sample to be fed back to the data analysis module to further optimize the classification model.

[0014] Preferably, it also includes S700, and an expert system is connected when the commodity code is determined. When a complex and difficult-to-determine commodity code appears during the output of the product code, the expert system will intervene to assist in determining the commodity code.

[0015] Preferably, in step S700, the expert system is composed of expert knowledge in the field of customs coding, and reviews the customs declaration codes of difficult commodities through rule reasoning and case reasoning.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The method for determining customs declaration codes of the present invention can effectively address the challenges brought about by the complexity and diversity of commodities in international trade, significantly improve the accuracy and efficiency of customs declaration code determination, reduce trade risks caused by coding errors, and ensure the smooth progress of international trade.

[0017] 2. The present invention can store the generated codes to facilitate subsequent query and audit. At the same time, the stored electronic documents can also be used as new training samples to feed back to the data analysis module to further optimize the classification model.

[0018] 3. The present invention is connected to an expert system, so when some complex products appear and the codes are difficult to determine, the codes of the complex products can be determined efficiently and quickly with the assistance of the expert system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flowchart of the method for determining the customs declaration code of the present invention; DETAILED DESCRIPTION

[0020] In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connect," "fixed," etc. should be understood broadly. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0021] The following is a further description with reference to the accompanying drawings and specific embodiments: Example 1 like Figure 1 As shown, a method for determining a customs declaration code, the specific steps of the method are as follows: S100. Collect information on the goods to be declared. The specific steps for collecting information are as follows: S110, configure the data collection interface to connect the trade order system and database; S120. Obtain basic information about the product from the trade order system, including product name, model, specifications, quantity, and price. For example, for a batch of electronic product orders, obtain detailed information such as the product model "XX brand smartphone, 128GB memory, 6.7-inch screen" S130. Extract more in-depth technical parameters and functional descriptions from product manuals or technical documentation. For products with special features, such as environmentally friendly and energy-saving equipment, collect information about their energy-saving principles and new technologies. Additionally, obtain appearance characteristics from product images and video footage, such as shape, color, and packaging. This information helps further determine the product's category.

[0022] S140. Obtain the trade method, import and export countries or regions, and mode of transportation from the trade contract; this information is also important for determining the customs declaration code. For example, commodities under certain trade methods may have specific coding rules, and bilateral or multilateral trade agreements between different countries may affect the classification of commodities.

[0023] S150. Clean the collected information to remove duplicate and useless information. For example, multiple data sources may contain product names, but the expressions may be slightly different. Data cleaning can unify them. At the same time, standardize the information, converting data in different formats to a unified format. For example, standardize the date format to "YYYY-MM-DD" and standardize quantity units, such as converting different weight units to kilograms.

[0024] S200: Process the collected product information. The specific steps for processing the product information are as follows: S210. Keywords are extracted from product information using a Term Frequency-Inverse Document Frequency (TF-IDF) algorithm combined with part-of-speech tagging. First, the product information text is segmented, and the TF-IDF value for each word is calculated. A word with a higher TF-IDF value is more representative of the product's characteristics. Part-of-speech tagging also focuses on nouns and adjectives. For example, for "high-tech, intelligent, and environmentally friendly lighting fixtures," keywords such as "high-tech," "intelligent," "environmentally friendly," and "lighting fixtures" are extracted. These keywords serve as an important basis for subsequent classification.

[0025] S220: Construct product features based on keywords. Each keyword corresponds to a dimension in the feature vector. The value can be binary (0 or 1, indicating whether the keyword exists) or weighted based on the importance of the keyword. For example, for lighting products, the keyword "lighting function" has a higher weight, while some auxiliary descriptive keywords have lower weights. In this way, product information is converted into a form that can be processed by computers.

[0026] S230: Using a support vector machine learning algorithm, a classification model is trained using a large amount of accurately classified commodity data. Labeled commodity feature vectors are used as training samples, and the optimal classification hyperplane is found by optimizing the objective function. In the application phase, the feature vectors of the commodity to be declared are input into the trained classification model, which then outputs a preliminary customs declaration code category.

[0027] S300. Establish a rules engine based on customs coding rules. Once the rules engine is established and the intelligent classification module outputs a preliminary customs declaration code, the rules engine checks and adjusts the code based on the commodity information and customs coding rules. The customs coding manual contains many provisions for special cases and commodity combinations. For example, for commodities containing multiple materials, there are different rules for classification by primary material or by value ratio. By developing a rule matching algorithm, the commodity information is matched with these rules and the preliminary code is revised as necessary.

[0028] S400: Compare the commodity to be declared with similar commodities with determined codes to verify the rationality of the commodity codes. The similarity comparison requires calculation. When calculating the commodity similarity, the feature vectors of the commodity to be declared and the known commodity are substituted into the formula: in, and are the characteristic vectors of the two commodities, and is their modulus. If the similarity exceeds a certain threshold (e.g., 0.8) and the customs declaration code of the known product is accurate, the rationality of the code of the product to be declared is further verified. If the similarity is low, the product information and code determination process are re-examined and necessary adjustments are made. When encountering products that are difficult to accurately classify, fuzzy matching technology is used to fuzzily match the product characteristics with those in the historical customs declaration case library. At the same time, the case learning mechanism continuously updates the case library and model based on new customs declaration cases.

[0029] S500, the adjusted and verified customs declaration code is output as the final result; the output code format is strictly in accordance with the format specified by the customs, including chapter, item, sub-item and other levels of coding to ensure the accuracy and standardization of the coding.

[0030] S600: The customs declaration code, related commodity information, and the basis for determining the code are generated into an electronic document for storage. The stored electronic document is then used for subsequent query and audit. The stored electronic document is also used as a new training sample to feed back to the data analysis module to further optimize the classification model.

[0031] Example 2 like Figure 1As shown, a method for determining a customs declaration code, the specific steps of the method are as follows: S100. Collect information on the goods to be declared. The specific steps for collecting information are as follows: S110, configure the data collection interface to connect the trade order system and database; S120. Obtain basic information about the product from the trade order system, including product name, model, specifications, quantity, and price. For example, for a batch of electronic product orders, obtain detailed information such as the product model "XX brand smartphone, 128GB memory, 6.7-inch screen" S130. Extract more in-depth technical parameters and functional descriptions from product manuals or technical documentation. For products with special features, such as environmentally friendly and energy-saving equipment, collect information about their energy-saving principles and new technologies. Additionally, obtain appearance characteristics from product images and video footage, such as shape, color, and packaging. This information helps further determine the product's category.

[0032] S140. Obtain the trade method, import and export countries or regions, and mode of transportation from the trade contract; this information is also important for determining the customs declaration code. For example, commodities under certain trade methods may have specific coding rules, and bilateral or multilateral trade agreements between different countries may affect the classification of commodities.

[0033] S150. Clean the collected information to remove duplicate and useless information. For example, multiple data sources may contain product names, but the expressions may be slightly different. Data cleaning can unify them. At the same time, standardize the information, converting data in different formats to a unified format. For example, standardize the date format to "YYYY-MM-DD" and standardize quantity units, such as converting different weight units to kilograms.

[0034] S200: Process the collected product information. The specific steps for processing the product information are as follows: S210. Keywords are extracted from product information using a Term Frequency-Inverse Document Frequency (TF-IDF) algorithm combined with part-of-speech tagging. First, the product information text is segmented, and the TF-IDF value for each word is calculated. A word with a higher TF-IDF value is more representative of the product's characteristics. Part-of-speech tagging also focuses on nouns and adjectives. For example, for "high-tech, intelligent, and environmentally friendly lighting fixtures," keywords such as "high-tech," "intelligent," "environmentally friendly," and "lighting fixtures" are extracted. These keywords serve as an important basis for subsequent classification.

[0035] S220: Construct product features based on keywords. Each keyword corresponds to a dimension in the feature vector. The value can be binary (0 or 1, indicating whether the keyword exists) or weighted based on the importance of the keyword. For example, for lighting products, the keyword "lighting function" has a higher weight, while some auxiliary descriptive keywords have lower weights. In this way, product information is converted into a form that can be processed by computers.

[0036] S230: Using a support vector machine learning algorithm, a classification model is trained using a large amount of accurately classified commodity data. Labeled commodity feature vectors are used as training samples, and the optimal classification hyperplane is found by optimizing the objective function. In the application phase, the feature vectors of the commodity to be declared are input into the trained classification model, which then outputs a preliminary customs declaration code category.

[0037] S300. Establish a rules engine based on customs coding rules. Once the rules engine is established and the intelligent classification module outputs a preliminary customs declaration code, the rules engine checks and adjusts the code based on the commodity information and customs coding rules. The customs coding manual contains many provisions for special cases and commodity combinations. For example, for commodities containing multiple materials, there are different rules for classification by primary material or by value ratio. By developing a rule matching algorithm, the commodity information is matched with these rules and the preliminary code is revised as necessary.

[0038] S400: Compare the commodity to be declared with similar commodities with determined codes to verify the rationality of the commodity codes. The similarity comparison requires calculation. When calculating the commodity similarity, the feature vectors of the commodity to be declared and the known commodity are substituted into the formula: in, and are the characteristic vectors of the two commodities, and is their modulus. If the similarity exceeds a certain threshold (e.g., 0.8) and the customs declaration code of the known product is accurate, the rationality of the code of the product to be declared is further verified. If the similarity is low, the product information and code determination process are re-examined and necessary adjustments are made. When encountering products that are difficult to accurately classify, fuzzy matching technology is used to fuzzily match the product characteristics with those in the historical customs declaration case library. At the same time, the case learning mechanism continuously updates the case library and model based on new customs declaration cases.

[0039] S500, the adjusted and verified customs declaration code is output as the final result; the output code format is strictly in accordance with the format specified by the customs, including chapter, item, sub-item and other levels of coding to ensure the accuracy and standardization of the coding.

[0040] S600: The customs declaration code, related commodity information, and the basis for determining the code are generated into an electronic document for storage. The stored electronic document is then used for subsequent query and audit. The stored electronic document is also used as a new training sample to feed back to the data analysis module to further optimize the classification model.

[0041] S700 also integrates an expert system when determining commodity codes. When complex and difficult commodity codes are encountered during product code output, the expert system intervenes to assist in determining the commodity code. The expert system, comprised of expert knowledge in the field of customs codes, uses rule-based and case-based reasoning to review the customs declaration codes of difficult commodities.

[0042] The above is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for determining a customs declaration code, characterized in that: The specific steps of this method are as follows: S100, collecting information on goods to be declared; S200: Processing the collected product information; S300. Establish a rule engine based on customs coding rules; S400: Compare the commodity to be declared with similar commodities with determined codes to verify the rationality of the commodity codes; S500: The adjusted and verified customs declaration code is output as the final result; S600: Generate an electronic document with the customs declaration code, related commodity information, and the basis for determining the code for storage.

2. A method for determining a customs declaration code according to claim 1, characterized in that: In step S100, the specific steps of collecting information are as follows: S110, configure the data collection interface to connect the trade order system and database; S120. Obtain basic information of the product from the trade order system, including product name, model, specifications, quantity, and price; S130. Extract more in-depth technical parameters and functional descriptions from product specifications or technical documents; S140. Obtain the trade mode, import and export countries or regions, and mode of transportation from the trade contract; S150. Clean the collected information, remove duplicate and useless information, and standardize the information.

3. A method for determining a customs declaration code according to claim 1, characterized in that: In step S200, the specific steps for processing product information are as follows: S210, extracting keywords from the product information using a method based on a term frequency-inverse document frequency (TF-IDF) algorithm combined with part-of-speech tagging; S220: Construct product feature vectors based on keywords, and each keyword corresponds to a dimension in the feature vector, whose value can be binary or assigned a certain weight value based on the importance of the keyword; S230, using a support vector machine learning algorithm, and using a large amount of accurately classified commodity data to train a classification model.

4. A method for determining a customs declaration code according to claim 1, characterized in that: In step S300, after the rule engine is established, when the intelligent classification module outputs the preliminary customs declaration code, the rule engine checks and adjusts the code according to the commodity information and customs coding rules.

5. A method for determining a customs declaration code according to claim 4, characterized in that: In step S400, similarity comparison requires calculation. When calculating the similarity of commodities, the feature vectors of the commodity to be declared and the known commodity are substituted into the formula: in, and are the characteristic vectors of the two commodities, and It is their model.

6. A method for determining a customs declaration code according to claim 1, characterized in that: In step S400, when encountering goods that are difficult to accurately classify, fuzzy matching technology is used to fuzzy match the product features with the product features in the historical customs declaration case library. At the same time, the case learning mechanism will continuously update the case library and model based on new customs declaration cases.

7. A method for determining a customs declaration code according to claim 1, characterized in that: In step S500, the output coding format is strictly in accordance with the format prescribed by the customs, including coding at various levels such as chapter, item, and sub-item, to ensure the accuracy and standardization of the coding.

8. A method for determining a customs declaration code according to claim 1, characterized in that: In step S600, the stored electronic document is subsequently queried and audited in English. The stored electronic document is also used as a new training sample to be fed back to the data analysis module to further optimize the classification model.

9. A method for determining a customs declaration code according to any one of claims 1 to 8, characterized in that: The following steps are also included: S700. When determining the commodity code, an expert system is also connected. When a complex and difficult-to-determine commodity code appears during the output of the product code, the expert system will intervene to assist in determining the commodity code.

10. A method for determining a customs declaration code according to claim 9, characterized in that: In step S700, the expert system is composed of expert knowledge in the field of customs coding, and reviews the customs declaration codes of difficult commodities through rule reasoning and case reasoning.

Images