Imported goods value dynamic risk control method and system based on multi-source heterogeneous data fusion

Abstract

The invention discloses an imported goods value dynamic risk control method and system based on multi-source heterogeneous data fusion, and relates to the technical field of dynamic risk control. Comprising the steps that customs data information is acquired, a customs information set is obtained, and the customs information set at least comprises position data and information data of one customs; based on the customs information set, commodity information in the customs is obtained, a commodity information set is obtained, and the commodity information set at least comprises the number information of one kind of commodities and customs declaration data information of the commodities. Multi-source data fusion enables information to be richer and more comprehensive, basic information of a single commodity is included, mutual influence among different commodities is also considered, the prediction capability of imported commodity values is improved, the complexity and the dynamism of the market are fully considered in the model level, the interaction model and the price trend model are established, and the prediction efficiency is improved. The method can reflect the actual condition of the imported commodity more accurately, and achieves the dynamic monitoring of the imported commodity value.

Description

Technical Field

[0001] The present invention relates to the field of dynamic risk control technology, and specifically to a method and system for dynamic risk control of imported goods value based on multi-source heterogeneous data fusion. Background Art

[0002] The value of imported goods refers to the value of imported goods, that is, the price of goods used in customs declaration upon import. It usually includes tariffs, freight, insurance and other related fees of the goods, reflecting the total value of the goods upon entry. It is often used to calculate tariffs, import taxes and other taxes. The risk control of the value of goods refers to a systematic measure to identify, predict and control the value declaration, assessment and associated risks of imported goods through technology, policies and process design. Its core goal is to avoid tariff losses, legal risks and market fluctuations caused by overvaluation or undervaluation of goods. Therefore, it is very necessary to conduct risk control on the value of imported goods.

[0003] Patent publication number CN115860747A is a customs import and export commodity risk management method and system based on big data. By standardizing and integrating the multi-dimensional customs commodity data information obtained from big data, and then desensitizing the multi-dimensional standard customs commodity data information and uploading it to the risk feature identification module for analysis, customs commodity risk clue information is obtained. Based on the customs commodity risk clue information, a customs commodity risk analysis model is constructed. Based on the customs commodity risk analysis model, commodity hazard analysis information, commodity quality risk analysis information, and commodity transaction risk analysis information are output to determine the customs import and export commodity risk information. When the customs import and export commodity risk information exceeds the preset commodity risk threshold, a technical solution for risk early warning management of customs import and export commodities is implemented.

[0004] When the above-mentioned and similar technical solutions are used for dynamic risk control of the value of imported goods, one of the data that needs to be referred to is the domestic market price. By continuously obtaining real price data from the domestic market, the value declaration of imported goods can be identified. However, when there are different types of goods in the same batch of goods, as the goods flood into the market, the prices of different types of goods in the goods will produce certain fluctuations and impacts on each other, such as promoting each other, impacting each other, or one growing while the other shrinks, which will lead to inaccurate identification of the value declaration of the batch of goods and further affect the accuracy of risk control. Summary of the Invention

[0005] The purpose of the present invention is to provide a dynamic risk control method and system for import cargo value based on multi-source heterogeneous data fusion to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a dynamic risk control method for import cargo value based on multi-source heterogeneous data fusion, comprising: Obtaining customs data information to obtain a customs information set, where the customs information set includes at least location data and information data of one customs; Based on the customs information set, obtain commodity information within the customs to obtain a commodity information set, wherein the commodity information set includes at least quantity information of a type of commodity and customs declaration data information of the commodity; Perform feature extraction based on the product information set to extract product category information and obtain a target product set, where the target product set includes at least one category of imported products; By acquiring data, based on the target product set, real-time price information of the target product set is obtained to obtain a product price set, wherein the product price set corresponds to the target product set respectively; Create an impact model for different types of imported goods and obtain an interaction model. The interaction model is used to represent the quantity and price impact information of different types of imported goods. Based on the actual price fluctuation trend of the commodity price set, a price trend model is created, and based on the combination of the interaction model and the price trend model, a data matching model is obtained; Based on the interaction results between the product information set and the data matching model, the predicted price corresponding to the product information set is obtained to obtain a predicted price set, and the predicted price set is compared with the product information set; Set an early warning threshold. When the comparison result between the customs declaration data corresponding to the commodity information set and the predicted price set exceeds the early warning threshold, an early warning prompt is triggered. This combines multi-source data such as commodity type, commodity market price, commodity import quantity, and the influence relationship between different types of commodities to improve the accuracy of dynamic risk control of imported goods value.

[0007] Furthermore, the method for obtaining the target product set includes: Based on the customs information set, using the customs information set as a marking feature point, respectively obtain the real-time commodity category information of the marking feature point to obtain a first category set; Setting an acquisition threshold, where the acquisition threshold is a state threshold, and acquiring previous commodity category information of the marked feature points based on the acquisition threshold to obtain a second category set; The repeated parts of the first category set and the second category set are filtered out to obtain the target product set.

[0008] Furthermore, the method for obtaining the second type set includes: A maximum time limit value is set, which is a time threshold. Based on the maximum time limit value, the previous commodity status of the marked feature point is obtained to determine whether it has entered the market. When the previous commodity status is the market entry status, it is determined to be a rejection status. When the previous commodity status is the market non-entry status, it is determined to be the status threshold, and then the previous commodity category information of the marked feature point that meets the status threshold is obtained to obtain the second category set.

[0009] Furthermore, the method for creating the interaction model includes: At least two judgment deadlines are set, where the judgment deadlines are time deadlines. Based on the judgment deadlines, corresponding prices of the target product set are obtained to obtain a corresponding price set. Based on the judgment period, the quantity change data of the target product set is obtained to obtain a quantity change set, and the quantity change set corresponds to the target product set respectively; Splitting the target product set and the quantity change set into at least two target product items and two quantity change items, where the target product items correspond to the quantity change items; The quantity change item corresponding to a single target commodity item is used as a variable, and the quantity change items corresponding to the remaining target commodity items are used as quantitative quantities. The corresponding data of the variables in the price corresponding set are obtained to obtain the first target training set. Using the quantity change item corresponding to a single target commodity item as a variable and the quantity change items corresponding to the remaining target commodity items as quantitative, obtain the corresponding data of the remaining target commodity items in the price corresponding set to obtain the second target training set; The first target training set and the second target training set are used as training data for training, and an interaction model is output.

[0010] Furthermore, the method for creating the price trend model includes: Based on the target product set, a search period is set, where the search period is a time period, and actual price information of the target product set is obtained; Split the search period into at least two search time periods, obtain actual price information for each search time period, and obtain price segment items; The price segment items are used as training data for training, and the price trend model is output.

[0011] Furthermore, the product information set includes the remaining transportation time of the product, and the method for obtaining the data matching model includes: Based on the product information set, obtain the remaining transportation time of the target product set to obtain the transportation time set; Based on the combined results of the transportation time set and the price trend model, the predicted price of the product information set is obtained to obtain the predicted price item; Based on the combination of the predicted price item and the product information set and the interaction model as training data, a data matching model is output.

[0012] Furthermore, the method for setting the warning threshold includes: Set the first fluctuation value, the second fluctuation value, and the fluctuation judgment value, and also set the fluctuation period, which is a time period. Obtain the difference between the starting point and the end point of the price of the commodity information set within the fluctuation period to obtain the benchmark difference item.

[0013] Determine whether the benchmark difference item exceeds the fluctuation judgment value. When it exceeds the fluctuation judgment value, the first warning threshold is obtained by combining the first fluctuation value and the price endpoint difference. When it does not exceed the fluctuation judgment value, the second warning threshold is obtained by combining the second fluctuation value and the price endpoint difference. The first warning threshold and the second warning threshold are combined to obtain the warning threshold.

[0014] Furthermore, the method for obtaining the product information set includes: Obtain commodity quantity information within the target customs to obtain commodity quantity items, and simultaneously obtain customs declaration data information corresponding to the commodity quantity items to obtain declaration information items; Based on the customs information set, a combination result of the commodity quantity item and the declaration information item is obtained to obtain the commodity information set.

[0015] Furthermore, the dynamic risk control system for import cargo value based on multi-source heterogeneous data fusion uses the above-mentioned dynamic risk control method for import cargo value based on multi-source heterogeneous data fusion, including: Acquisition module: obtains customs data information, obtains customs information set, obtains commodity information within the customs based on the customs information set, and obtains commodity information set; Extraction module: performs feature extraction based on the product information set, extracts product category information, and obtains the target product set; Creation module: Based on the target commodity set, obtain the real-time price information of the target commodity set to obtain the commodity price set. The commodity price set corresponds to the target commodity set, and create impact models for different types of imported commodities to obtain an interaction model. The interaction model is used to represent the quantity and price impact information of different types of imported commodities. Based on the actual price fluctuation trend of the commodity price set, a price trend model is created. Based on the combination of the interaction model and the price trend model, a data matching model is obtained. Feedback module: Based on the interaction results between the commodity information set and the data matching model, the predicted price corresponding to the commodity information set is obtained to obtain the predicted price set. The predicted price set is compared with the commodity information set, and an early warning threshold is set. When the comparison result between the customs declaration data corresponding to the commodity information set and the predicted price set exceeds the early warning threshold, an early warning prompt is triggered.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This dynamic risk control method and system for import goods value based on multi-source heterogeneous data fusion creates interaction models and price trend models respectively. At the data level, multi-source data fusion makes the information richer and more comprehensive, including not only the basic information of a single commodity, but also considering the mutual influence between different commodities. This comprehensive data support helps to build a more accurate model and improve the ability to predict the value of import goods. At the model level, the establishment of the interaction model and the price trend model fully considers the complexity and dynamics of the market, and can more accurately reflect the actual situation of imported goods. The data matching model organically combines the two to provide a more reasonable predicted price for the commodity information set, thereby achieving a dynamic monitoring effect on the value of import goods. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram of the customs information set of the present invention; Figure 3 This is a schematic diagram of the composition of the commodity information set of the present invention; Figure 4 This is a schematic diagram of the price correspondence set of the present invention; Figure 5 It is a schematic diagram of the quantity change set of the present invention. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] In order to achieve dynamic risk control of the value of imported goods, regulatory authorities usually refer to domestic market prices and identify possible false reporting, concealment, etc. by real-time monitoring and comparing domestic and foreign price differences. By continuously obtaining real price data from the domestic market, it is possible to more accurately judge whether the declared value of imported goods is reasonable. If the declared value is significantly lower than the domestic market price, there may be a risk of under-reporting, which requires further investigation and verification. However, when a batch of goods containing multiple commodities enters the market, the price relationship between these commodities does not exist in isolation, but will affect each other, forming a complex dynamic relationship. There may be a complementary relationship between certain commodities, such as a certain raw material and its related downstream products. When the demand for a certain raw material increases, its price increase may drive up the price of downstream products, thereby forming a situation where prices promote each other. If two commodities are highly substitutable in use, then a drop in the price of one of the commodities may lead to a decrease in demand for the other commodity, thereby lowering its price. These complex factors are intertwined. Together, the domestic market price fluctuations show a high degree of uncertainty and complexity. If we only rely on simple price comparisons without considering the influence of these factors, it may lead to incorrect identification of the import value declaration, thereby affecting the accuracy of risk control. The dynamic risk control method for import value based on multi-source heterogeneous data fusion provided by this application creates an interaction model and a price trend model according to the type information of the goods through the obtained customs information and the commodity information in the customs. The interaction model can represent the price impact of different types of imported goods on each other when the quantity changes, and the price trend model can represent the natural change of the price of different types of goods over time. By combining the interaction model with the price trend model, the market price of the goods in the customs is estimated, and then the corresponding warning threshold and warning prompt are set, thereby achieving the effect of combining multi-source data such as commodity type, commodity market price, commodity import quantity and the influence relationship between different types of commodities, thereby improving the accuracy of dynamic risk control of import value. Figure 1 As shown, steps S100-S700 are included.

[0020] Step S100: Obtain customs data information to obtain a customs information set.

[0021] It should be noted that the customs information set includes the location data and information data of at least one customs office. Since there is more than one customs office entering the domestic market, it is necessary to obtain the location data and corresponding information data of each customs office separately, such as scale, traffic volume, etc. These data are all public data and can be obtained through data query.

[0022] Step S200: Based on the customs information set, obtain commodity information within the customs to obtain a commodity information set.

[0023] It should be noted that the commodity information set includes at least the quantity information of one type of commodity and the customs declaration data information of the commodity. The method for obtaining the commodity information set includes: obtaining the commodity quantity information within the target customs to obtain the commodity quantity item, and at the same time obtaining the customs declaration data information corresponding to the commodity quantity item to obtain the declaration information item; based on the customs information set, obtaining the combination result of the commodity quantity item and the declaration information item to obtain the commodity information set.

[0024] Example 1

[0025] like Figure 2-Figure 3 As shown, the customs information set currently obtained includes the first target customs, the second target customs, and the third target customs. Through data acquisition, the commodity quantity information and corresponding customs declaration data information of the first target customs are queried, which are: 10,000 voltage-stabilized power supply devices with a unit price of US$1,200 per unit, 250,000 integrated circuits with a unit price of US$8.2 per unit, and 10,000 smart terminal devices with a unit price of US$800 per unit; the commodity quantity information and corresponding customs declaration data information of the second target customs are: 10,000 sets of clothing with a unit price of US$100 per set, and 10,000 sets of gloves with a unit price of US$1 per set; the commodity quantity information and corresponding customs declaration data information of the third target customs are: 180 tons of phthalate plasticizers with a unit price of US$10,000 per ton. At this time, based on the combination of the commodity quantity items and the declaration information items, the commodity information set is obtained.

[0026] Step S300: performing feature extraction based on the product information set, extracting product category information, and obtaining a target product set.

[0027] It should be noted that the target commodity set includes at least one type of imported commodity. The method for obtaining the target commodity set includes: based on the customs information set, using the customs information set as the marking feature point, respectively obtaining the real-time commodity category information of the marking feature point to obtain the first category set; setting an acquisition threshold, the acquisition threshold being the state threshold, obtaining the previous commodity category information of the marking feature point based on the acquisition threshold to obtain the second category set; screening out the repeated parts of the first category set and the second category set to obtain the target commodity set.

[0028] It should be noted that the method for obtaining the second category set includes: setting a maximum term value, the maximum term value is a time threshold, the maximum term value is one year, and based on the maximum term value, obtaining the previous commodity status of the marking feature point to determine whether it has entered the market. When the previous commodity status is the market entry status, it is determined to be a rejection status. When the previous commodity status is the non-market entry status, it is determined to be the status threshold, and then obtaining the previous commodity category information of the marking feature point that meets the status threshold to obtain the second category set.

[0029] Example 2

[0030] During the specific implementation process, the customs information set obtained includes the first target customs and the second target customs. At this time, the first target customs and the second target customs are used as the first marking feature point and the second marking feature point respectively, where the real-time commodity category information of the first marking feature point and the second marking feature point is: voltage-stabilized power supply equipment, integrated circuits, smart terminal equipment, clothing, and gloves, and the first category set is obtained. At the same time, according to the set maximum time limit value, the previous commodity status of the first target customs and the second target customs within the maximum time limit value is obtained, and it is obtained that among the commodities entering through the first target customs and the second target customs within one year, voltage-stabilized power supply equipment and smart terminal equipment are both entering the market. At this time, the voltage-stabilized power supply equipment and smart terminal equipment are judged to be eliminated, and integrated circuits, clothing, gloves, electronic products, and cosmetics are status thresholds, and the second category set is obtained. The repeated parts of the first category set and the second category set are screened out to obtain the target commodity set, namely integrated circuits, clothing, and gloves.

[0031] Step S400: acquiring real-time price information of the target product set based on the target product set by means of data acquisition to obtain a product price set.

[0032] It should be noted that the commodity price sets correspond to the target commodity sets respectively. By setting the acquisition interval, the price trend information of the target commodity set in the domestic market is repeatedly obtained continuously. Since the target commodity set contains multiple commodities, the obtained commodity price sets are also multiple.

[0033] Step S500: Create an impact model of different types of imported goods to obtain an interaction model.

[0034] It should be noted that the interaction model is used to represent the quantity and price impact information of different types of imported goods. The method for creating the interaction model includes: setting at least two judgment periods, the judgment period is a time period, and the judgment period is 10 days. Based on the judgment period, the corresponding prices of the target commodity set are obtained to obtain a price correspondence set; based on the judgment period, the quantity change data of the target commodity set are obtained to obtain a quantity change set, and the quantity change sets correspond to the target commodity sets respectively; the target commodity set and the quantity change set are split into at least two target commodity items and two quantity change items, and the target commodity items correspond to the quantity change items; using the quantity change item corresponding to a single target commodity item as a variable and the quantity change items corresponding to the remaining target commodity items as a quantitative, obtaining the corresponding data of the variable in the price correspondence set to obtain the first target training set; using the quantity change item corresponding to a single target commodity item as a variable and the quantity change items corresponding to the remaining target commodity items as a quantitative, obtaining the corresponding data of the remaining target commodity items in the price correspondence set to obtain the second target training set.

[0035] Example 3

[0036] In the specific implementation process, Figure 4-Figure 5 As shown, the target product set is integrated circuits, clothing, and gloves. Two judgment periods are set, both of which are 10 days. It is found that within the first judgment period, the price trend of integrated circuits is: 56-57-58-57-56-55-56-54-55-56, in yuan; the price trend of clothing is: 700-701-702-701-700-701-702-703-705-707, in yuan; the price trend of gloves is: 7-8-7-8-7-8-9-10-11-12, in yuan. During the second judgment period, the price trend of integrated circuits is: 56-57-58-57-56-55-56-54-55-56, in RMB; the price trend of clothing is: 700-701-702-701-700-699-698-699-700-701, in RMB; the price trend of gloves is: 7-8-7-8-7-8-7-8-7-6, in RMB. The corresponding price set is obtained. At the same time, based on the judgment period, the quantity change data of the target product set is obtained. During the first judgment period, the quantity change of integrated circuits is: 110224-110021-110001-110001-110001-109124-108754-108453-108110-107985, the unit is piece, the number of clothing changes is: 9856-9760-9654-9654-9654-8765-8564-8435-8354-8254, the unit is piece. The quantity of gloves changes to: 4358-4038-3865-3865-3865-3721-3545-3215-3320-3021, the unit is set. During the second judgment period, the quantity of integrated circuits changes to: 98546-96587-94214-93258-92145-91254-91102-90354-90001-89924, the unit is piece, the number of clothing changes is: 6878-6758-6721-6654-6612-6545-6512-6423-6401-6325, the unit is piece, the quantity change of gloves is: 2548-2354-2302-2258-2275-2238-2210-2185-2135-2103, the unit is set, and the quantity change set is obtained; the target product set and the quantity change set are split into three target product items and three quantity change items. The three target product items are integrated circuits, clothing and gloves. At this time, the quantity change item corresponding to the integrated circuit is used as the variable, and the quantity change item corresponding to clothing and gloves is used as the quantitative, that is, the first judgment is used. During the judgment period, the number of clothing is 9654, and the number of gloves is 9865 as the quantitative value. The corresponding data of the integrated circuits in the price corresponding concentration are obtained to obtain the first target training item. The change in the number of clothing is used as a variable, and the change in the number of integrated circuits and gloves is used as the quantitative value. That is, during the first judgment period, the number of integrated circuits is 110001, and the number of gloves is 9865 as the quantitative value. The corresponding data of clothing in the price corresponding concentration are obtained to obtain the second target training item. The change in the number of gloves is used as a variable, and the change in the number of integrated circuits and clothing is used as the quantitative value. That is, during the first judgment period, the number of integrated circuits is 110001, and the number of clothing is 9654 as the quantitative value. For quantification, obtain the corresponding data of gloves in the price corresponding set to obtain the third target training item, and the first target training item, the second target training item and the third target training item are combined to obtain the first target training set; then use the quantity change item corresponding to a single target commodity item as a variable, and the quantity change items corresponding to the remaining target commodity items as the quantification, that is, within the first judgment period, the number of clothing is 9654 and the number of gloves is 9865 as the quantification, obtain the price corresponding data of clothing and gloves in the price corresponding set to obtain the fourth target training item, and then use the quantity change corresponding to clothing as a variable, and the quantity change corresponding to integrated circuits and gloves as the quantification, that is, within the first judgment period, the number of integrated circuits is 1 10001, the number of gloves is 9865 as the quantitative value, and the corresponding data of the integrated circuits and gloves in the price corresponding set are obtained to obtain the fifth target training item. Then, the change in the number of gloves is used as the variable, and the change in the number of integrated circuits and clothing is used as the quantitative value. That is, within the first judgment period, the number of integrated circuits is 110001, and the number of clothing is 9654 as the quantitative value. The corresponding data of the integrated circuits and clothing in the price corresponding set are obtained to obtain the sixth target training item. The fourth target training item, the fifth target training item and the sixth target training item are combined to obtain the second target training set. At this time, the first target training set and the second target training set are used as training data for training, and the interaction model is output.

[0037] Step S600: creating a price trend model based on the actual price fluctuation trend of the commodity price set, and obtaining a data matching model based on the combination result of the interaction model and the price trend model.

[0038] It should be noted that the method for creating a price trend model includes: setting a retrieval period based on the target product set, where the retrieval period is a time period and the retrieval cycle is set to 10 days, and obtaining the actual price information of the target product set; splitting the retrieval period into ten retrieval time periods, obtaining the actual price information of each retrieval time period, and obtaining price segmentation items; training with the price segmentation items as training data, and outputting the price trend model.

[0039] It should be noted that the product information set includes the remaining transportation time of the product. The method for obtaining the data matching model includes: based on the product information set, obtaining the remaining transportation time of the target product set to obtain the transportation time set; based on the combination set result of the transportation time set and the price trend model, obtaining the predicted price of the product information set to obtain the predicted price item; based on the predicted price item and the combination result of the product information set and the interaction model as training data, outputting the data matching model.

[0040] Specifically, since the price trend model can roughly predict the price of goods after different time nodes, after obtaining the remaining transportation time of the target product set, the price trend model can be used to output the preliminary predicted price of the target product set, and the interactive model can be used to further output the re-predicted price of the target product set.

[0041] Step S700: Based on the interaction result between the product information set and the data matching model, the predicted price corresponding to the product information set is obtained to obtain a predicted price set.

[0042] It should be noted that it is necessary to compare the predicted price set with the commodity information set and set an early warning threshold. When the comparison result between the corresponding customs declaration data in the commodity information set and the predicted price set exceeds the early warning threshold, an early warning prompt is triggered. This combines multi-source data such as commodity types, commodity market prices, commodity import quantities, and the influence relationship between different types of commodities to improve the accuracy of dynamic risk control of imported goods value.

[0043] It should be noted that the method for setting the warning threshold includes: setting a first fluctuation value, a second fluctuation value and a fluctuation judgment value, wherein the first fluctuation value is ±15%, the second fluctuation value is 10±%, and the fluctuation judgment value is ±5%, and at the same time setting a fluctuation period, which is a time period, and the fluctuation period is 10 days. The difference between the starting point and the price end point of the commodity information set within the fluctuation period is obtained to obtain a benchmark difference item; determine whether the benchmark difference item exceeds the fluctuation judgment value. When it exceeds the fluctuation judgment value, the first warning threshold is obtained by combining the first fluctuation value and the price end point difference. When it does not exceed the fluctuation judgment value, the second warning threshold is obtained by combining the second fluctuation value and the price end point difference. The first warning threshold and the second warning threshold are combined to obtain the warning threshold.

[0044] Example 4

[0045] During the specific implementation process, for a batch of goods at a certain customs, the unit price of the declared data information is 80 yuan. At this time, according to the set fluctuation period, the starting price of the commodity within the fluctuation period is 94, the end price is 100, and the benchmark difference item is 6.4%. Since it exceeds the fluctuation judgment value, the first warning threshold is obtained by combining the first fluctuation value and the price end point difference, that is, the first warning threshold is 85-115, and the unit price of the declared data information is not within the first warning threshold, thereby triggering a warning prompt.

[0046] The dynamic risk control system for import value based on multi-source heterogeneous data fusion uses the above-mentioned dynamic risk control method for import value based on multi-source heterogeneous data fusion, including: an acquisition module: acquiring customs data information to obtain a customs information set; based on the customs information set, acquiring commodity information within the customs to obtain a commodity information set; an extraction module: performing feature extraction based on the commodity information set to extract commodity category information to obtain a target commodity set; a creation module: acquiring real-time price information of the target commodity set based on the target commodity set to obtain a commodity price set, wherein the commodity price set corresponds to the target commodity set respectively; creating an impact model for different types of imported commodities to obtain an interaction model, which is used to represent the quantity and price impact information of different types of imported commodities; creating a price trend model based on the actual price fluctuation trend of the commodity price set; and obtaining a data matching model based on the combination of the interaction model and the price trend model; and a feedback module: acquiring a predicted price corresponding to the commodity information set based on the interaction result between the commodity information set and the data matching model to obtain a predicted price set; comparing the predicted price set with the commodity information set; setting an early warning threshold; and triggering an early warning prompt when the comparison result between the customs declaration data corresponding to the commodity information set and the predicted price set exceeds the early warning threshold.

[0047] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is limited by the accompanying embodiments and their equivalents.

Claims

1. A dynamic risk control method for import cargo value based on multi-source heterogeneous data fusion, including: Obtaining customs data information to obtain a customs information set, where the customs information set includes at least location data and information data of one customs; Based on the customs information set, obtain commodity information within the customs to obtain a commodity information set, wherein the commodity information set includes at least quantity information of a type of commodity and customs declaration data information of the commodity; Its characteristics are: Perform feature extraction based on the product information set to extract product category information and obtain a target product set, where the target product set includes at least one category of imported products; By acquiring data, based on the target product set, real-time price information of the target product set is obtained to obtain a product price set, wherein the product price set corresponds to the target product set respectively; Create an impact model for different types of imported goods and obtain an interaction model. The interaction model is used to represent the quantity and price impact information of different types of imported goods. Based on the actual price fluctuation trend of the commodity price set, a price trend model is created, and based on the combination of the interaction model and the price trend model, a data matching model is obtained; Based on the interaction results between the product information set and the data matching model, the predicted price corresponding to the product information set is obtained to obtain a predicted price set, and the predicted price set is compared with the product information set; Set an early warning threshold. When the comparison result between the customs declaration data corresponding to the commodity information set and the predicted price set exceeds the early warning threshold, an early warning prompt is triggered. This combines multi-source data such as commodity type, commodity market price, commodity import quantity, and the influence relationship between different types of commodities to improve the accuracy of dynamic risk control of imported goods value.

2. The method for dynamic risk control of import cargo value based on multi-source heterogeneous data fusion according to claim 1 is characterized by: The method for obtaining the target product set includes: Based on the customs information set, using the customs information set as a marking feature point, respectively obtain the real-time commodity category information of the marking feature point to obtain a first category set; Setting an acquisition threshold, where the acquisition threshold is a state threshold, and acquiring previous commodity category information of the marked feature points based on the acquisition threshold to obtain a second category set; The repeated parts of the first category set and the second category set are filtered out to obtain the target product set.

3. The method for dynamic risk control of import cargo value based on multi-source heterogeneous data fusion according to claim 2 is characterized by: The method for obtaining the second type collection includes: A maximum time limit value is set, which is a time threshold. Based on the maximum time limit value, the previous commodity status of the marked feature point is obtained to determine whether it has entered the market. When the previous commodity status is the market entry status, it is determined to be a rejection status. When the previous commodity status is the market non-entry status, it is determined to be the status threshold, and then the previous commodity category information of the marked feature point that meets the status threshold is obtained to obtain the second category set.

4. The method for dynamic risk control of import cargo value based on multi-source heterogeneous data fusion according to claim 1 is characterized by: The method for creating the interaction model includes: At least two judgment deadlines are set, where the judgment deadlines are time deadlines. Based on the judgment deadlines, corresponding prices of the target product set are obtained to obtain a corresponding price set. Based on the judgment period, the quantity change data of the target product set is obtained to obtain a quantity change set, and the quantity change set corresponds to the target product set respectively; Splitting the target product set and the quantity change set into at least two target product items and two quantity change items, where the target product items correspond to the quantity change items; The quantity change item corresponding to a single target commodity item is used as a variable, and the quantity change items corresponding to the remaining target commodity items are used as quantitative quantities. The corresponding data of the variables in the price corresponding set are obtained to obtain the first target training set. Using the quantity change item corresponding to a single target commodity item as a variable and the quantity change items corresponding to the remaining target commodity items as quantitative, obtain the corresponding data of the remaining target commodity items in the price corresponding set to obtain the second target training set; The first target training set and the second target training set are used as training data for training, and an interaction model is output.

5. The import goods value dynamic risk control method based on multi-source heterogeneous data fusion according to claim 1 is characterized by: The method for creating the price trend model includes: Based on the target product set, a search period is set, where the search period is a time period, and actual price information of the target product set is obtained; Split the search period into at least two search time periods, obtain actual price information for each search time period, and obtain price segment items; The price segment items are used as training data for training, and the price trend model is output.

6. The import goods value dynamic risk control method based on multi-source heterogeneous data fusion according to claim 1 is characterized by: The product information set includes the remaining transportation time of the product, and the method for obtaining the data matching model includes: Based on the product information set, obtain the remaining transportation time of the target product set to obtain the transportation time set; Based on the combined results of the transportation time set and the price trend model, the predicted price of the product information set is obtained to obtain the predicted price item; Based on the combination of the predicted price item and the product information set and the interaction model as training data, a data matching model is output.

7. The import goods value dynamic risk control method based on multi-source heterogeneous data fusion according to claim 1 is characterized by: The method for setting the warning threshold includes: Set the first fluctuation value, the second fluctuation value, and the fluctuation judgment value, and also set the fluctuation period, which is a time period. Obtain the difference between the starting point and the end point of the price of the commodity information set within the fluctuation period to obtain the benchmark difference item. Determine whether the benchmark difference item exceeds the fluctuation judgment value. When it exceeds the fluctuation judgment value, the first warning threshold is obtained by combining the first fluctuation value and the price endpoint difference. When it does not exceed the fluctuation judgment value, the second warning threshold is obtained by combining the second fluctuation value and the price endpoint difference. The first warning threshold and the second warning threshold are combined to obtain the warning threshold.

8. The method for dynamic risk control of import cargo value based on multi-source heterogeneous data fusion according to claim 1 is characterized by: The method for obtaining the product information set includes: Obtain commodity quantity information within the target customs to obtain commodity quantity items, and simultaneously obtain customs declaration data information corresponding to the commodity quantity items to obtain declaration information items; Based on the customs information set, a combination result of the commodity quantity item and the declaration information item is obtained to obtain the commodity information set.

9. The dynamic risk control system for import cargo value based on multi-source heterogeneous data fusion is characterized by: The method for dynamic risk control of import cargo value based on multi-source heterogeneous data fusion according to any one of claims 1 to 8 is used, comprising: Acquisition module: obtains customs data information, obtains customs information set, obtains commodity information within the customs based on the customs information set, and obtains commodity information set; Extraction module: performs feature extraction based on the product information set, extracts product category information, and obtains the target product set; Creation module: Based on the target commodity set, obtain the real-time price information of the target commodity set to obtain the commodity price set. The commodity price set corresponds to the target commodity set, and create impact models for different types of imported commodities to obtain an interaction model. The interaction model is used to represent the quantity and price impact information of different types of imported commodities. Based on the actual price fluctuation trend of the commodity price set, a price trend model is created. Based on the combination of the interaction model and the price trend model, a data matching model is obtained. Feedback module: Based on the interaction results between the commodity information set and the data matching model, the predicted price corresponding to the commodity information set is obtained to obtain the predicted price set. The predicted price set is compared with the commodity information set, and an early warning threshold is set. When the comparison result between the customs declaration data corresponding to the commodity information set and the predicted price set exceeds the early warning threshold, an early warning prompt is triggered.

Images