Cross-border e-commerce big data product selection system based on linear enhanced neural network
By integrating product information from cross-border e-commerce platforms through linear reinforced neural networks, the problem of high difficulty in product selection in cross-border e-commerce has been solved, enabling fast and accurate selection of cross-border products.
Patent Information
- Application Number
- CN202510787399.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-13
- Publication Date
- 2025-10-31
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In existing cross-border e-commerce, users need to repeatedly view and compare product information on different cross-border e-commerce websites, making product selection quite difficult.
The cross-border e-commerce big data product selection system, based on linear reinforced neural networks, connects with multiple e-commerce platforms, integrates information on similar products, compares prices, and ultimately generates the purchase path and link for the lowest-priced product, displaying the main image directly on the interface, thus simplifying the system setup process.
It enables rapid integration of cross-border goods and screening of high-quality purchasing paths, reduces the difficulty of system construction, and improves the efficiency and accuracy of users' product selection.
Smart Images
Figure CN120876004A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of cross-border e-commerce technology, and in particular to a cross-border e-commerce big data product selection system based on linear reinforced neural networks. Background Technology
[0002] Cross-border e-commerce refers to an international business activity in which trading entities belonging to different customs territories complete transactions through e-commerce platforms, conduct electronic payment settlements, and deliver goods through cross-border e-commerce logistics and off-site warehousing. However, most existing cross-border e-commerce relies on different cross-border e-commerce websites or public platforms to display products, and transactions are conducted by browsing these platforms.
[0003] However, since the products on different cross-border e-commerce websites or platforms are different, users need to repeatedly view product information on different websites and compare different product information in order to find their favorite products, which makes it difficult for users to select products. Summary of the Invention
[0004] This invention discloses a cross-border e-commerce big data product selection system based on a linear reinforced neural network, aiming to solve the technical problem that users need to repeatedly view product information from different websites and compare different product information in order to find their favorite products, which makes product selection difficult.
[0005] To achieve the above objectives, the present invention adopts the following technical solution: The cross-border e-commerce big data product selection system based on linear reinforcement neural networks includes an input layer, a hidden layer, and an output layer. Each of the input, hidden, and output layers includes multiple neurons, which are linearly reinforced. The input layer is connected to a category input unit, the hidden layer is connected to a data aggregation unit, and the output layer is connected to a data output unit and a backpropagation unit. The data aggregation unit includes an e-commerce platform connection module, a similar product integration module, a similar product comparison module, a link generation module, a link integration module, and a main image extraction module. The e-commerce platform connection module is used to establish connections with multiple existing cross-border e-commerce platforms and obtain all product information from those platforms. The similar product integration module is used to integrate identical products from all product information, mainly matching based on product category and supplier. The similar product comparison module is used to compare the prices of the same product on different cross-border e-commerce platforms and filter out the lowest price. The link generation module is used to generate links for the lowest-priced products after comparison. The link integration module is used to integrate all the compared product links. The main image extraction module is used to extract the main image from the product interface and present it in the product selection interface.
[0006] With a data aggregation unit, the entire product selection system primarily uses a linear reinforced network to match big data. During the data aggregation process, the system directly connects to multiple existing e-commerce platforms, integrates similar products, and then compares the lowest-priced products across different platforms using a product comparison module, directly generating links. Under this structure, there is no need to build separate connections between the system and all merchants. Through transfer, comparison, and filtering, cross-border products from all websites can be integrated, and the most suitable purchase path can be selected. This allows users to quickly purchase the high-quality cross-border products they need, and also reduces the difficulty of system construction.
[0007] In a preferred embodiment, the category input unit includes a category input module, a keyword extraction module, a keyword matching module, an old data analysis module, a precise filtering condition selection module, and a condition integration module. The category input module is used to input the required product category into the input layer. The keyword extraction module is used to extract keywords of the input product category. The keyword matching module is mainly used to establish the connection between the keyword extraction module and the precise filtering condition selection module. The old data analysis module is used to integrate and analyze the old data of the user's cross-border purchases, and analyze the condition options of all products in the input product category. If there is no purchase record of the product, it will be analyzed by big data. The precise filtering condition selection module presents condition options based on the analysis results, and these condition options are automatically displayed on the interface after the product category is entered. The condition integration module is used to match products based on the selected condition options and display them on the interface. The data output unit includes a precise matching output module, a broad matching output module, and a data sorting module. The precise matching output module matches precise products entirely based on the selection results in the precise filtering condition selection module. The broad match output module expands the matching range based on the matching results of the precise match output module, and the number of products matched by the broad match output module is greater than the number of products matched by the precise match output module. The precise match output module and the broad match output module can be manually switched. The data sorting module is used to sort the filtered product data.
[0008] By setting up a category input unit and a data output unit, the category input unit extracts keywords from the input text after the user inputs the product, and analyzes the user's purchase history and purchase records with the same keywords in big data. After extracting the keywords, it can match the corresponding filtering conditions. After presenting the conditions and filtering by the user, it can further refine the user's needs and optimize the accuracy of linear neural network matching.
[0009] In a preferred embodiment, the backpropagation unit includes a path identification module, a personal data aggregation module, a big data aggregation module, a data analysis module, and a backpropagation module. The path identification module is used to identify the matching path of the entire neural network, the personal data aggregation module is used to aggregate personal purchase data, and the big data aggregation module is used to aggregate all purchase data. The data analysis module is used to analyze the summarized data. The backpropagation module analyzes the user's final product selection under the same conditions based on the summarized data, and backpropagates the selected product to the matching path of the neural network, feeding it back to the data sorting module to change the data sorting result.
[0010] By incorporating a backpropagation unit, which primarily analyzes and aggregates individual data and big data separately, the analysis results reveal individual user selection habits and the selection habits of a majority of users under the same screening conditions. This information is then used to create new data sorting criteria. Consequently, when the final sorted results are presented to the user, they represent the results that best align with the user's or the general public's selection habits, making it easier for the user to choose a satisfactory and reliable product.
[0011] As shown above, the cross-border e-commerce big data product selection system based on linear reinforcement neural networks includes an input layer, a hidden layer, and an output layer. Each of the input, hidden, and output layers contains multiple neurons connected in a linear reinforcement manner. The input layer is connected to a category input unit, the hidden layer is connected to a data aggregation unit, and the output layer is connected to a data output unit and a backpropagation unit. The data aggregation unit includes an e-commerce platform connection module, a similar product integration module, a similar product comparison module, a link generation module, a link integration module, and a main image extraction module. The e-commerce platform connection module is used to establish connections with multiple existing cross-border e-commerce platforms and obtain all product information from these platforms. The similar product integration module is used to integrate identical products from all product information, mainly matching based on product category and supplier. The similar product comparison module is used to compare the prices of the same product on different cross-border e-commerce platforms and filter out the lowest price. The link generation module is used to generate links for the lowest-priced products after comparison. The link integration module is used to integrate all compared product links. The main image extraction module is used to extract the main image from the product interface and present it in the product selection interface. The cross-border e-commerce big data product selection system based on linear reinforced neural networks provided by this invention can integrate cross-border products from all websites and filter out the most suitable purchase path, enabling users to quickly purchase the high-quality cross-border products they need, and reducing the difficulty of system construction. Attached Figure Description
[0012] Figure 1This is a schematic diagram of the overall structure of the cross-border e-commerce big data product selection system based on linear reinforced neural networks proposed in this invention.
[0013] Figure 2 This is a schematic diagram of the data aggregation unit structure of the cross-border e-commerce big data product selection system based on linear reinforced neural networks proposed in this invention.
[0014] Figure 3 This is a schematic diagram of the category input unit structure of the cross-border e-commerce big data product selection system based on linear reinforced neural networks proposed in this invention.
[0015] Figure 4 This is a schematic diagram of the data output unit structure of the cross-border e-commerce big data product selection system based on linear reinforced neural networks proposed in this invention.
[0016] Figure 5 This is a schematic diagram of the backpropagation unit structure of the cross-border e-commerce big data product selection system based on linear reinforcement neural network proposed in this invention. Detailed Implementation
[0017] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0018] The cross-border e-commerce big data product selection system disclosed in this invention is mainly applied to cross-border product selection scenarios.
[0019] Reference Figure 1 and Figure 2 The cross-border e-commerce big data product selection system based on linear reinforcement neural networks includes an input layer, a hidden layer, and an output layer. Each of the input, hidden, and output layers contains multiple neurons, which are linearly reinforced. The input layer is connected to a category input unit, the hidden layer is connected to a data aggregation unit, and the output layer is connected to a data output unit and a backpropagation unit. The data aggregation unit includes an e-commerce platform connection module, a similar product integration module, a similar product comparison module, a link generation module, a link integration module, and a main image extraction module. The e-commerce platform connection module establishes connections with multiple existing cross-border e-commerce platforms and obtains all product information from those platforms. The similar product integration module integrates identical products from all product information, primarily matching by product category and supplier. The similar product comparison module compares the prices of the same product on different cross-border e-commerce platforms and filters out the lowest price. The link generation module generates links for the lowest-priced products after comparison. The link integration module integrates all compared product links. The main image extraction module extracts the main image from the product interface and displays it in the product selection interface. The entire product selection system... The system primarily uses a linear reinforced network to match big data. During data aggregation, it directly connects with multiple existing e-commerce platforms and integrates similar products, mainly by matching product categories and suppliers. This identifies identical products across different platforms. The product comparison module then compares the lowest-priced products across different platforms, generating direct links. These links are then integrated, and the main image is extracted and displayed on the interface for easy viewing and selection by users. Under this structure, there is no need to build separate connections between the system and all merchants. Through transfer, comparison, and filtering, cross-border products from all websites can be integrated, and the most suitable purchase path can be selected. This allows users to quickly purchase the high-quality cross-border products they need, while also reducing the difficulty of system construction.
[0020] Reference Figure 3 In a preferred embodiment, the category input unit includes a category input module, a keyword extraction module, a keyword matching module, an old data analysis module, a precise filtering condition selection module, and a condition integration module. The category input module is used to input the required product categories into the input layer.
[0021] Reference Figure 3 In a preferred embodiment, the keyword extraction module is used to extract keywords of the input product category, the keyword matching module is mainly used to establish the connection between the keyword extraction module and the precise filtering condition selection module, and the old data analysis module is used to integrate and analyze the old data of the user's cross-border purchases, analyze the condition options of all products in the input product category, and if there is no purchase record of the product, it is analyzed by big data.
[0022] Reference Figure 3 In a preferred embodiment, the precise filtering condition selection module presents condition options based on the analysis results, and these condition options are automatically displayed on the interface after the product category is entered. The condition integration module is used to match products based on the selected condition options and display them on the interface.
[0023] Reference Figure 4 In a preferred embodiment, the data output unit includes a precise matching output module, a broad matching output module, and a data sorting module. The precise matching output module matches precise products based entirely on the selection results in the precise filtering condition selection module.
[0024] Reference Figure 4 In a preferred embodiment, the broad match output module expands the matching range based on the matching results of the precise match output module, and the number of products matched by the broad match output module is greater than that of the precise match output module. The precise match output module and the broad match output module can be manually switched. The data sorting module is used to sort the filtered product data. After the user inputs a product, the category input unit extracts keywords from the input text and analyzes the user's purchase history and purchase records with the same keywords in big data. After extracting keywords, it can match corresponding filtering conditions, such as displaying the country, capacity, etc. of related products. After presenting these conditions and filtering by the user, the user's needs can be further refined, and the accuracy of linear neural network matching can be optimized. In addition, by switching between the precise match output module and the broad match output module interface with one click, the range of product recommendations can be quickly browsed for comparison and finding the desired product.
[0025] Reference Figure 5 In a preferred embodiment, the backpropagation unit includes a path identification module, a personal data aggregation module, a big data aggregation module, a data analysis module, and a backpropagation module. The path identification module is used to identify the matching path of the entire neural network, the personal data aggregation module is used to aggregate personal purchase data, and the big data aggregation module is used to aggregate all purchase data.
[0026] Reference Figure 5 In a preferred embodiment, the data analysis module analyzes the aggregated data, and the backpropagation module analyzes the user's final product selection under the same conditions based on the aggregated data. The selected product is then backpropagated to the matching path of the neural network and fed back to the data sorting module to change the data sorting results. The backpropagation unit primarily aggregates personal data and big data separately, and analyzes the aggregated data. The analysis results precisely indicate the user's personal selection habits under the same screening conditions, as well as the selection habits of a majority of users under big data. Personal data is the primary factor, and in the absence of personal data, big data analysis results are used as a supplement. The analysis results are backpropagated to the neural network matching path to create new data sorting conditions. Based on these conditions, a new sorting system is established in the data sorting module. Therefore, when the final sorted results are presented to the user, the results best match the user's selection habits or the general public's selection habits, making it easier for the user to choose a satisfactory and reliable product.
[0027] Working Principle: The entire product selection system primarily uses a linear reinforced network to match big data. During data aggregation, the system directly connects with multiple existing e-commerce platforms and integrates similar products, mainly matching product categories and suppliers. This identifies identical products across different platforms. The same product comparison module then compares the lowest-priced products across different platforms, generating direct links. These links are then integrated, and the main image is extracted and displayed on the interface for easy user viewing and selection. This structure eliminates the need to build separate connections between the system and all merchants. Through transfer, comparison, and filtering, it integrates cross-border products from all websites and selects the most suitable purchase path, allowing users to quickly purchase the high-quality cross-border products they need. This also reduces the difficulty of system setup. In the category input unit, after the user inputs a product, the system extracts keywords from the input text and analyzes the user's purchase history and purchase records with the same keywords in the big data. This allows for matching corresponding filtering conditions after keyword extraction. For example… The filtering criteria can display relevant product information such as country and capacity. After these criteria are presented and filtered by the user, user needs can be further refined, and the accuracy of linear neural network matching can be optimized. In addition, by switching between the precise matching output module and the broad matching output module interface with one click, users can quickly browse product recommendations within a range, making it easier to compare and find the desired products. In the backpropagation unit, personal data and big data are summarized separately and analyzed. The analysis results show the user's personal selection habits under the same filtering conditions, as well as the selection habits of more users under big data. Personal data is the primary factor, and in the absence of personal data, big data analysis results are used as a supplement. The analysis results are backpropagated to the neural network matching path to create new data sorting criteria. Based on these criteria, a new sorting system is established in the data sorting module. Therefore, when the final sorted results are presented to the user, the results are the ones that best match the user's selection habits or the selection habits of the general public, making it easier for users to choose satisfactory and reliable products.
[0028] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A cross-border e-commerce big data product selection system based on a linear reinforcement neural network, comprising an input layer, a hidden layer, and an output layer, characterized in that, The input layer, hidden layer, and output layer each include multiple neurons, which are linearly reinforced. The input layer is connected to a category input unit, the hidden layer is connected to a data aggregation unit, and the output layer is connected to a data output unit and a backpropagation unit. The data aggregation unit includes an e-commerce platform connection module, a similar product integration module, a similar product comparison module, a link generation module, a link integration module, and a main image extraction module. The e-commerce platform connection module is used to establish connections with multiple existing cross-border e-commerce platforms and obtain all product information from those platforms. The similar product integration module is used to integrate identical products from all product information, mainly matching based on product category and supplier. The similar product comparison module is used to compare the prices of the same product on different cross-border e-commerce platforms and filter out the lowest price. The link generation module is used to generate links for the lowest-priced products after comparison. The link integration module is used to integrate all the compared product links. The main image extraction module is used to extract the main image from the product interface and present it in the product selection interface.
2. The cross-border e-commerce big data product selection system based on a linear reinforcement neural network according to claim 1, characterized in that, The category input unit includes a category input module, a keyword extraction module, a keyword matching module, an old data analysis module, a precise filtering condition selection module, and a condition integration module. The category input module is used to input the required product category into the input layer.
3. The cross-border e-commerce big data product selection system based on a linear reinforcement neural network according to claim 2, characterized in that, The keyword extraction module is used to extract keywords for the input product category. The keyword matching module is mainly used to establish the connection between the keyword extraction module and the precise filtering condition selection module. The old data analysis module is used to integrate and analyze the user's old cross-border purchase data, analyze the condition options of all products in the input product category, and if there is no purchase record for the product, it will be analyzed from big data.
4. The cross-border e-commerce big data product selection system based on a linear reinforcement neural network according to claim 3, characterized in that, The precise filtering condition selection module presents condition options based on the analysis results, and these condition options are automatically displayed on the interface after the product category is entered. The condition integration module is used to match products based on the selected condition options and display them on the interface.
5. The cross-border e-commerce big data product selection system based on a linear reinforcement neural network according to claim 4, characterized in that, The data output unit includes a precise matching output module, a broad matching output module, and a data sorting module. The precise matching output module matches precise products based entirely on the selection results in the precise filtering condition selection module.
6. The cross-border e-commerce big data product selection system based on a linear reinforcement neural network according to claim 5, characterized in that, The broad match output module expands the matching range based on the matching results of the precise match output module, and the number of products matched by the broad match output module is greater than the number of products matched by the precise match output module. The precise match output module and the broad match output module can be manually switched. The data sorting module is used to sort the filtered product data.
7. The cross-border e-commerce big data product selection system based on a linear reinforcement neural network according to claim 5, characterized in that, The backpropagation unit includes a path identification module, a personal data aggregation module, a big data aggregation module, a data analysis module, and a backpropagation module. The path identification module is used to identify the matching path of the entire neural network, the personal data aggregation module is used to aggregate personal purchase data, and the big data aggregation module is used to aggregate all purchase data.
8. The cross-border e-commerce big data product selection system based on a linear reinforcement neural network according to claim 7, characterized in that, The data analysis module is used to analyze the summarized data. The backpropagation module analyzes the user's final product selection under the same conditions based on the summarized data, and backpropagates the selected product to the matching path of the neural network, feeding it back to the data sorting module to change the data sorting result.