Block chain-based cross-border e-commerce commodity data real-time pushing system and method

Abstract

The invention relates to the technical field of business data processing, in particular to a blockchain-based cross-border e-commerce commodity data real-time pushing system and method. In response to retrieval and purchase behaviors of the target user on the new commodity category, constructing a target purchase behavior chain of the target user in the block chain network based on a commodity purchase time sequence of the target user; performing purchase behavior chain matching in the block chain network based on the target purchase behavior chain, and obtaining purchase probabilities of a plurality of subsequent commodity sub-categories; obtaining a plurality of specific commodity sets of the plurality of commodity sub-categories, and determining the quality matching degree of each specific commodity in the plurality of specific commodity sets in combination with the target purchase behavior chain to obtain a plurality of specific commodity matching sequences; and obtaining a commodity recommendation list for the target user according to the purchase probabilities of the plurality of commodity sub-categories and the plurality of specific commodity matching sequences. According to the invention, the user data security can be guaranteed, and the user shopping experience is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of commercial data processing, and particularly relates to a cross-border e-commerce commodity data real-time pushing system and method based on a block chain. BACKGROUND

[0002] In the field of traditional cross-border e-commerce commodity recommendation, mainly relying on user historical purchase records, browsing records and other data, a commodity recommendation list is generated through simple statistical analysis or basic algorithms. Data storage is mainly centralized database, and data management and maintenance are managed by the e-commerce platform. The existing technology has a shallow depth in the mining of user behavior data in the recommendation process, and is mainly focused on the user's past general purchase and browsing behavior, without fully considering the new product category scenario, and lacks effective protection of data security and traceability. SUMMARY

[0003] The present application provides a cross-border e-commerce commodity data real-time pushing system and method based on a block chain to solve the technical problems that the existing technology does not fully consider the new product category scenario and lacks effective protection of data security and traceability.

[0004] The technical solution of the present application to solve the above technical problems is as follows: In a first aspect, the present application provides a cross-border e-commerce commodity data real-time pushing system based on a block chain, comprising: a target purchase behavior chain construction module, configured to respond to the search and purchase behavior of a target user for a new commodity category, and to construct a target purchase behavior chain of the target user in a block chain network based on the time sequence of the purchase commodity of the target user; a purchase probability acquisition module, configured to perform purchase behavior chain matching in the block chain network based on the target purchase behavior chain, and to acquire the purchase probability of a plurality of commodity subcategories; a specific commodity matching sequence acquisition module, configured to acquire a plurality of specific commodity sets of the plurality of commodity subcategories, and to determine the quality matching degree of each specific commodity in the plurality of specific commodity sets in combination with the target purchase behavior chain, to obtain a plurality of specific commodity matching sequences; and a commodity recommendation list acquisition module, configured to obtain a commodity recommendation list for the target user according to the purchase probability of the plurality of commodity subcategories and the plurality of specific commodity matching sequences.

[0005] Optionally, in the target purchase behavior chain construction module, in response to the target user searching for a new commodity category and purchase behavior, a target purchase behavior chain of the target user is constructed based on the purchase time sequence of the target user in the blockchain network, including the following execution steps: detecting the first search for a new commodity category and purchase behavior of the target user, and judging whether the new commodity category is a new purchase scenario of the target user; when the new commodity category is a new purchase scenario of the target user, obtaining the purchase commodity information of the target user under the new commodity category, the purchase commodity information including commodity identifier, purchase time, commodity price and quality level; sorting the purchase commodity information in the order of purchase time to form a purchase time sequence; creating a purchase behavior chain with the target user as the main body in the blockchain network, writing the purchase time sequence into the purchase behavior chain in the form of a node, and constructing a target purchase behavior chain.

[0006] Optionally, in the purchase probability acquisition module, the target purchase behavior chain is matched in the blockchain network based on the target purchase behavior chain to obtain the purchase probability of the subsequent multiple commodity subcategories, including the following execution steps: retrieving multiple reference purchase behavior chains with similar purchase paths from the target purchase behavior chain in the blockchain network; based on the latest purchase node of the target purchase behavior chain, extracting subsequent purchase commodity data at the same purchase node from the multiple reference purchase behavior chains; aggregating the commodity subcategories in the subsequent purchase commodity data to obtain multiple commodity subcategories, and statistically obtaining the purchase probability of each commodity subcategory.

[0007] Optionally, in the purchase probability acquisition module, multiple reference purchase behavior chains with similar purchase paths are retrieved from the target purchase behavior chain in the blockchain network, including the following execution steps: extracting the commodity subcategories of each purchase node in the target purchase behavior chain to form a target commodity subcategory sequence; based on the new commodity category, retrieving and obtaining a set of historical purchase behavior chains in the blockchain network; traversing the set of historical purchase behavior chains to obtain a first historical purchase behavior chain and extracting the commodity subcategories of each purchase node in the first historical purchase behavior chain to form a first historical commodity subcategory sequence; according to the length of the target commodity subcategory sequence, a sequence segment of corresponding length is intercepted from the first historical commodity subcategory sequence to form a first comparison sequence; when each element in the target commodity subcategory sequence exists in the first comparison sequence, the first historical purchase behavior chain is added to the multiple reference purchase behavior chains.

[0008] Optionally, in the purchase probability acquisition module, the product subcategories in the subsequent purchase product data are summarized to obtain multiple product subcategories, and the purchase probability of each product subcategory is calculated. This includes the following steps: extracting the product subcategories of each purchased product from the subsequent purchase product data, removing duplicates to form a product subcategory set, the product subcategory set including multiple product subcategories; counting the number of times each product subcategory appears in the subsequent purchase product data in the product subcategory set; calculating the ratio of the number of times each product subcategory appears to the total number of subsequent purchase product data to obtain the purchase probability of each product subcategory.

[0009] Optionally, the specific product matching sequence acquisition module acquires multiple specific product sets of the multiple product subcategories, and determines the quality matching degree of each specific product in the multiple specific product sets in conjunction with the target purchase behavior chain to obtain multiple specific product matching sequences. This includes the following execution steps: Based on the multiple product subcategories, retrieve the specific products corresponding to each product subcategory from the cross-border product database to form multiple specific product sets; analyze the percentile ranking of each purchased product in the target purchase behavior chain within its corresponding product subcategory to obtain multiple percentile rankings of purchased prices, and calculate the average percentile ranking of the multiple percentile rankings of purchased prices; calculate the ranking deviation between the percentile ranking of each specific product in the multiple specific product sets and the average percentile ranking; calculate the quality matching degree of each specific product based on the ranking deviation value, where the quality matching degree and the ranking deviation value are negatively correlated; and sort the specific products in each specific product set in descending order according to the quality matching degree to obtain multiple specific product matching sequences.

[0010] Optionally, the product recommendation list acquisition module obtains a product recommendation list for the target user based on the purchase probabilities of the multiple product subcategories and the multiple specific product matching sequences, including the following execution steps: obtaining the total number of recommendations in the product recommendation list; calculating the number of recommended products corresponding to each product subcategory based on the purchase probabilities of each product subcategory and the total number of recommendations; selecting the corresponding number of recommended products from the corresponding specific product matching sequences in sequence according to the number of recommended products corresponding to each product subcategory to obtain the recommended product set for each product subcategory; and summarizing the recommended product sets of each product subcategory to form a product recommendation list for the target user.

[0011] Secondly, this invention provides a blockchain-based method for real-time push of cross-border e-commerce product data, comprising: responding to a target user's search and purchase behavior for a new product category; constructing a target user's target purchase behavior chain in a blockchain network based on the target user's purchase product time sequence; performing purchase behavior chain matching in the blockchain network based on the target purchase behavior chain to obtain the purchase probability of multiple subsequent product subcategories; obtaining multiple specific product sets of the multiple product subcategories, and determining the quality matching degree of each specific product in the multiple specific product sets in conjunction with the target purchase behavior chain to obtain multiple specific product matching sequences; and obtaining a product recommendation list for the target user based on the purchase probability of the multiple product subcategories and the multiple specific product matching sequences.

[0012] By implementing this invention, it is possible to respond to target users' retrieval and purchase behavior of new product categories, construct a target user's target purchase behavior chain in the blockchain network based on the purchase time sequence of the target user's products, focus on the purchase behavior of new product categories, obtain multi-dimensional information, provide accurate data for subsequent analysis, and rely on blockchain to achieve data traceability and tamper-proof, ensuring user data security; and form a clear time sequence, which is convenient for analyzing changes in user purchase habits and preferences. By implementing this invention, it is possible to match purchase behavior chains in a blockchain network based on the target purchase behavior chain, obtain the purchase probability of multiple subsequent product subcategories, and predict the product subcategories that users may purchase later by matching reference behavior chains with similar purchase paths and utilizing the similar purchase patterns of the group. This makes the acquisition of purchase probability more scientific and reasonable, and avoids subjective assumptions. By implementing this invention, it is possible to obtain multiple specific product sets of the multiple product subcategories, and determine the quality matching degree of each specific product in the multiple specific product sets in combination with the target purchase behavior chain, thereby obtaining multiple specific product matching sequences. The matching degree is calculated based on the average percentile ranking of the prices of the products purchased by the user, so that the price level of the recommended products matches the user's past spending power, and avoids recommending products that are too high or too low and do not meet the user's expectations. By implementing this invention, a product recommendation list for the target user can be obtained based on the purchase probability of the multiple product subcategories and the matching sequence of the multiple specific products. Based on the analysis of user purchase behavior and preferences by the aforementioned modules, the final recommendation list is highly tailored to the personalized needs of the target user, thereby increasing the user's interest in and willingness to purchase the recommended products.

[0013] In summary, by implementing this invention, the accuracy and personalization of cross-border e-commerce product recommendations can be improved while ensuring user data security, enhancing the user shopping experience, and helping to increase the transaction rate and user satisfaction of cross-border e-commerce platforms. Attached Figure Description

[0014] Figure 1 A schematic diagram of the structure of a blockchain-based real-time push system for cross-border e-commerce product data provided by the present invention; Figure 2 This is a flowchart illustrating a blockchain-based method for real-time push of cross-border e-commerce product data, as provided by the present invention.

[0015] In the attached diagram, the components represented by each number are as follows: The module includes: Target Purchase Behavior Chain Construction Module 11, Purchase Probability Acquisition Module 12, Specific Product Matching Sequence Acquisition Module 13, and Product Recommendation List Acquisition Module 14. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0018] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0019] Example 1, as Figure 1 As shown, this embodiment of the invention provides a blockchain-based real-time push system and method for cross-border e-commerce product data, including: The target purchase behavior chain construction module 11 is used to respond to the target user's search and purchase behavior for new product categories, and to construct the target user's target purchase behavior chain in the blockchain network based on the target user's purchase product time sequence; The purchase probability acquisition module 12 is used to match the purchase behavior chain in the blockchain network based on the target purchase behavior chain to obtain the purchase probability of multiple subsequent product subcategories. The specific product matching sequence acquisition module 13 is used to acquire multiple specific product sets of multiple product subcategories, and determine the quality matching degree of each specific product in the multiple specific product sets in combination with the target purchase behavior chain to obtain multiple specific product matching sequences. The product recommendation list acquisition module 14 is used to obtain a product recommendation list for the target user based on the purchase probability of multiple product subcategories and multiple specific product matching sequences.

[0020] In the target purchase behavior chain construction module 11 of this application embodiment, in response to the target user's search and purchase behavior for a new product category, the target user's target purchase behavior chain is constructed in the blockchain network based on the purchase time sequence of the target user's products, including the following execution steps: Detect the target user's first search and purchase behavior for a new product category to determine whether the new product category represents a new purchase scenario for the target user; When a new product category represents a new purchase scenario for the target user, obtain the target user's purchase information under the new product category. The purchase information includes product identifier, purchase time, product price, and quality level. The purchased goods information is sorted according to the purchase time to form a purchase time sequence; Create a purchase behavior chain centered on the target user in the blockchain network, and write the purchase sequence of goods into the purchase behavior chain in the form of nodes to construct the target purchase behavior chain.

[0021] In this embodiment, the target purchase behavior chain construction module 11 detects the user's first search and purchase behavior for a new product category to determine whether a new purchase scenario is triggered, accurately captures the user's previously unrepresented consumption preferences, and, based on the immutable and traceable characteristics of blockchain, transforms the user's purchase information in the new scenario into a behavior chain in chronological order, ensuring that the user behavior data used for subsequent recommendations is authentic and complete, and avoiding data tampering or forgery that could lead to recommendation bias.

[0022] To achieve the above objectives, firstly, it is necessary to detect the target user's initial search and purchase behavior for a new product category to determine whether the new product category represents a new purchase scenario for the target user. Specifically, the system first detects the target user's operational behavior, filtering out the two related actions: "initial search for a new product category" and "initial purchase." Both initial search and initial purchase must be satisfied; searches without purchases or purchases without prior search are not included in the filtering scope.

[0023] Next, compare the target user's historical behavior data: if the product category has no record in the user's past search or purchase history, it can be determined as a "new purchase scenario for the target user"; if the user has previously searched or purchased this product category, the subsequent target purchase behavior chain will not be triggered.

[0024] Then, when the new product category corresponds to a new purchase scenario for the target user, the system obtains the target user's purchase information within that new product category. This information includes the product identifier, purchase time, product price, and quality level. The product identifier is a unique code that identifies the product, such as SKU or product ID, ensuring accurate identification of the product's subcategory during subsequent matching. The purchase time needs to be accurate to the hour, minute, and second to provide a basis for subsequent time-series sorting. The product price is the actual transaction price, i.e., the actual payment amount excluding coupons, discounts, and other temporary offers. The quality level is a preset level based on the product's own attributes, such as A, B, C, or categories like "affordable luxury," "mass market," or "budget-friendly."

[0025] Furthermore, the purchased product information is sorted according to the purchase time to form a purchase product time sequence. That is, using "purchase time" as the sorting dimension, the collected multiple sets of purchased product information are arranged in order from morning to evening to form a "purchase product time sequence". For example, if a user buys a camping tent on January 5 and a camping sleeping bag on January 8, the purchase product time sequence is "Tent (1.5), Sleeping Bag (1.8)".

[0026] Finally, a unique "purchase behavior chain" is created for the target user in the blockchain network: using the user ID as the main identifier, each piece of product information in the above "purchase product sequence" is transformed into an independent "purchase node". Each node contains the product identifier, purchase time, product price and quality grade data.

[0027] Write nodes into the behavior chain in chronological order: Connect nodes sequentially according to the purchase time to form an immutable "target purchase behavior chain", and synchronize it to the distributed storage of the blockchain network to ensure that the data is traceable and cannot be modified unilaterally.

[0028] In the purchase probability acquisition module 12 of this application embodiment, purchase behavior chain matching is performed in the blockchain network based on the target purchase behavior chain to obtain the purchase probability of multiple subsequent product subcategories, including the following execution steps: Retrieve multiple reference purchase behavior chains from the blockchain network that have similar purchase paths to the target purchase behavior chain; Based on the latest purchase node of the target purchase behavior chain, extract subsequent purchase product data at the same purchase node from multiple reference purchase behavior chains; The product subcategories in subsequent purchase data are aggregated to obtain multiple product subcategories, and the purchase probability of each product subcategory is calculated.

[0029] In this embodiment of the application, the purchase probability acquisition module 12 is used to provide category priority for subsequent product recommendations. By mining the behavioral patterns of similar users in the blockchain network, it predicts the product sub-categories and probabilities that the target user may purchase next in the current purchase scenario.

[0030] To achieve the above objectives, it is first necessary to retrieve multiple reference purchase behavior chains from the blockchain network that have similar purchase paths to the target purchase behavior chain.

[0031] In the purchase probability acquisition module 12 of this application embodiment, retrieving multiple reference purchase behavior chains with similar purchase paths to the target purchase behavior chain from the blockchain network includes the following execution steps: Extract the product subcategories of each purchase node in the target purchase behavior chain to form a target product subcategory sequence; Based on the new product category, retrieve the historical purchase behavior chain set in the blockchain network; Traverse the set of historical purchase behavior chains to obtain the first historical purchase behavior chain, and extract the product sub-categories of each purchase node in the first historical purchase behavior chain to form the first historical product sub-category sequence; Based on the length of the target product sub-category sequence, a sequence segment of the corresponding length is extracted from the first historical product sub-category sequence to form the first comparison sequence; When all elements in the target product subcategory sequence exist in the first comparison sequence, the first historical purchase behavior chain is added to multiple reference purchase behavior chains.

[0032] Specifically, the first step is to extract the core features of the target purchase behavior chain. That is, from the target purchase behavior chain, extract the product subcategories corresponding to each purchase node to form a target product subcategory sequence. For example, the product of node 1 is "three-person tent", and its subcategory is "camping tent"; the product of node 2 is "down sleeping bag", and its subcategory is "camping sleeping bag".

[0033] The second step is to define the search scope. Based on the target user's current involvement with "new product categories," such as "outdoor camping equipment," all "historical purchase behavior chains" containing products of that category are filtered out in the blockchain network to exclude purchase behavior chains unrelated to that category, such as purchase behavior chains that only contain baby products or home appliances.

[0034] The third step requires precise matching of similar purchase behavior chains. First, traverse each chain in the historical purchase behavior chain set, namely the "first historical purchase behavior chain", extract the product sub-categories of each node in the first historical purchase behavior chain, and form the "first historical product sub-category sequence" in chronological order. For example, the first historical product sub-category sequence of the first historical purchase behavior chain is [camping tent, camping sleeping bag, folding table and chair, camping lamp].

[0035] Then, based on the length of the target product sub-category sequence, a segment of consecutive corresponding length is extracted from the first historical sequence as the first comparison sequence. For example, if the length of the first historical product sub-category sequence is greater than or equal to the length of the target product sub-category sequence, a continuous segment starting from the first node is extracted. Assuming the target sequence length is 2 and the historical sequence length is 4, the first two nodes are extracted to form [camping tent, camping sleeping bag]. If the length of the first historical product subcategory sequence is less than the length of the target product subcategory sequence, then the historical chain is directly excluded due to the length mismatch.

[0036] The fourth step is to compare the "target product sub-category sequence" with the "first comparison sequence".

[0037] Specifically, it is necessary to check whether each subcategory element in the target product subcategory sequence, such as "camping tent" and "camping sleeping bag", exists in the first comparison sequence. The standard for "existence" is that the order is not strictly required to be completely consistent; it is only necessary to include all elements.

[0038] If the above conditions are met, such as the target product sub-category sequence [camping tent, camping sleeping bag] being determined to be similar to the first comparison sequence [camping sleeping bag, camping tent] due to complete overlap of elements, then the first historical purchase behavior chain corresponding to the first comparison sequence will be included in the "multiple reference purchase behavior chains" set for subsequent extraction of subsequent purchase data.

[0039] Furthermore, based on the latest purchase node in the target purchase behavior chain, it is necessary to extract subsequent purchase data at the same purchase node from multiple reference purchase behavior chains.

[0040] First, we need to find the "latest purchase node" in the target purchase behavior chain, which is the product subcategory corresponding to the last purchase record in the purchase behavior chain. For example, the latest node of the target purchase behavior chain [camping tent, sleeping bag] is "sleeping bag".

[0041] Next, the reference purchase behavior chain data is filtered, and all reference purchase behavior chains are traversed to find the chain that also contains the "latest purchase node," i.e., the chain containing "sleeping bag." All purchase records after the latest purchase node are then extracted to form the "subsequent purchase product data." For example, in the above example, if a reference purchase behavior chain is [camping tent, sleeping bag, folding table and chair, camping lamp], then the purchase data corresponding to "folding table and chair" and "camping lamp" are extracted as the subsequent purchase product data.

[0042] Furthermore, it is necessary to summarize the product subcategories in the subsequent purchase data to obtain multiple product subcategories, and then calculate the purchase probability of each product subcategory.

[0043] In the purchase probability acquisition module 12 of this application embodiment, the product subcategories in the subsequent purchase product data are summarized to obtain multiple product subcategories, and the purchase probability of each product subcategory is calculated, including the following execution steps: Extract the product subcategories of each purchased product from the subsequent purchase data, and form a set of product subcategories after deduplication. The set of product subcategories includes multiple product subcategories. Count the number of times each product subcategory appears in subsequent purchase data within the product subcategory set; Calculate the ratio of the frequency of each product subcategory to the total number of subsequent purchase data to obtain the purchase probability of each product subcategory.

[0044] In this embodiment of the application, the function of the above-mentioned sub-execution steps in the purchase probability acquisition module 12 is to transform the subsequent purchase behavior data of similar users into quantifiable predictive indicators, so as to provide a clear priority basis for subsequent product recommendations.

[0045] To achieve the above, the first step is to extract the product subcategories of each purchased product from the subsequent purchase data. After deduplication, a set of product subcategories is formed, which includes multiple product subcategories. This involves iterating through all "subsequent purchase data" and extracting the corresponding "product subcategories" for each piece of data. For example, if a piece of data is "double folding table," its product subcategory is "folding table and chairs"; another is "LED camping light," and its product subcategory is "camping light." Then, all extracted product subcategories are deduplicated, such as removing duplicates of "folding table and chairs" and "camping light," ultimately forming a "product subcategory set" containing all unique subcategories, such as {folding table and chairs, camping light, portable stove, outdoor power supply}.

[0046] Next, it is necessary to count the number of times each product subcategory in the product subcategory set appears in subsequent purchase data. That is, for each product subcategory in the product subcategory set, such as "folding tables and chairs", backtrack through all "subsequent purchase data" and count the total number of times that product subcategory appears in the data. For example, "folding tables and chairs" appears 35 times in 100 subsequent data, and "camping lights" appears 28 times.

[0047] At the same time, the total number of subsequent purchase data is counted, that is, the sum of subsequent purchase data in all reference purchase behavior chains, such as the total data volume of 100 records in the example above.

[0048] Finally, for each product subcategory, the purchase probability of that subcategory is obtained by dividing the "number of occurrences" of the subcategory by the "total number of subsequent purchase data". The calculation formula is: Purchase probability = Number of occurrences of a product subcategory ÷ Total number of subsequent purchase data × 100%. For example, in the above example, the purchase probability of folding tables and chairs = 35 ÷ 100 = 35%, and the purchase probability of camping lights = 28 ÷ 100 = 28%.

[0049] The final output is a list of product subcategories with purchase probability values, such as [{Folding Tables and Chairs: 35%}, {Camping Lights: 28%}, ​​{Portable Stoves: 22%}, {Outdoor Power Supplies: 15%}], which serves as the core basis for subsequent recommendation ranking.

[0050] In the specific product matching sequence acquisition module 13 of this application embodiment, multiple specific product sets of multiple product subcategories are acquired, and the quality matching degree of each specific product in the multiple specific product sets is determined in combination with the target purchase behavior chain to obtain multiple specific product matching sequences, including the following execution steps: Based on multiple product subcategories, specific products corresponding to each product subcategory are retrieved from the cross-border product database to form multiple specific product sets; Analyze the percentile ranking of the purchased price of each purchased item in the target purchase behavior chain within the corresponding product subcategory to obtain multiple percentile rankings of the purchased price, and calculate the average percentile ranking of the multiple percentile rankings of the purchased price. Calculate the ranking deviation between the price percentile ranking of each specific product in multiple specific product sets and the average price percentile ranking; The quality matching degree of each specific product is calculated based on the ranking deviation value. The quality matching degree is negatively correlated with the ranking deviation value. Based on the quality matching degree, the specific products in each specific product set are sorted in descending order to obtain multiple specific product matching sequences.

[0051] In this embodiment, the specific product matching sequence acquisition module 13 is used to further filter out specific products that match the target user's spending power and quality preferences from the already determined high-probability product subcategories, so as to provide an accurate product list for the final recommendation.

[0052] To achieve the above objectives, firstly, it is necessary to retrieve the specific products corresponding to each product subcategory from the cross-border product database based on multiple product subcategories, forming multiple specific product sets.

[0053] Specifically, multiple product subcategories are needed as search criteria, such as folding tables and chairs, camping lights, and portable stoves. All specific products under each subcategory are selected from the cross-border product database, including product ID, price, brand, and specifications. Then, these are grouped by product subcategory to form multiple specific product sets. For example, the "folding tables and chairs product set" includes product A, product B, and product C; the "camping lights product set" includes product X, product Y, and product Z.

[0054] Next, it is necessary to analyze the percentile ranking of the purchased price of each purchased item in the target purchase behavior chain within the corresponding product subcategory, obtain multiple percentile rankings of the purchased price, and calculate the average percentile ranking of the multiple percentile rankings of the purchased price.

[0055] This involves analyzing the price tiers of products already purchased by the target customer. For each purchased product in the target customer's purchasing behavior chain, its subcategory is identified. For example, if the purchased product "Brand C Three-Person Tent" belongs to the "Camping Tent" subcategory, then the price range of all products under that subcategory is calculated, and the "percentile ranking" of the purchased product's price among similar products is calculated. For instance, if the "Camping Tent" subcategory has 100 products, and the purchased product's price is higher than 80 of those products, then its percentile ranking is 80%, indicating that its price is at a relatively high level within that subcategory.

[0056] Then, the average of the "purchased price percentile ranking" of all the products purchased by the target user is needed to obtain the "average price percentile ranking". For example, if the percentile rankings of the two products purchased by the user are 80% and 70% respectively, the average price percentile ranking is 75%, which means that the user prefers the higher-priced products in this category.

[0057] Furthermore, it is necessary to calculate the ranking deviation between the price percentile ranking of each specific product in multiple specific product sets and the average price percentile ranking.

[0058] Specifically, it is necessary to calculate the percentile ranking of the price of a specific product. That is, for each product in the specific product set, such as product A in the "folding tables and chairs product set", we count the prices of all products in its subcategory and calculate the percentile ranking of the price of that product. For example, product A ranks 60th in the "folding tables and chairs" subcategory.

[0059] Then, subtract the user's "average price percentile ranking" from the price percentile ranking of the specific product, and take the absolute value to obtain the "ranking deviation value". For example, if the price percentile ranking of product A is 60% - the average price percentile ranking is 75% = 15%, the ranking deviation value is 15%.

[0060] Furthermore, it is necessary to calculate the quality matching degree of each specific product based on the ranking deviation value. The quality matching degree is negatively correlated with the ranking deviation value. Optionally, the quality matching degree can be defined as 100% - ranking deviation value, in which case the quality matching degree of product A in the above example is 85%. Finally, all products within each "specific product set" need to be sorted from highest to lowest "quality matching degree". For example, in the "folding table and chair product set", product B with a quality matching degree of 90% is ranked first, product A with a quality matching degree of 85% is ranked second, and product C with a quality matching degree of 70% is ranked third.

[0061] The sorted list is the "specific product matching sequence". Each product subcategory corresponds to a specific product matching sequence. Finally, multiple specific product matching sequences are output, such as folding table and chair sequence, camping light sequence, etc., to provide a sorting basis for subsequent recommendation display.

[0062] In the product recommendation list acquisition module 14 of this application embodiment, a product recommendation list for the target user is obtained based on the purchase probability of multiple product subcategories and multiple specific product matching sequences, including the following execution steps: Get the total number of recommendations in the product recommendation list; Calculate the number of recommended products for each product subcategory based on the purchase probability and total number of recommendations for each product subcategory; Based on the number of recommended products corresponding to each product subcategory, the corresponding number of recommended products are selected sequentially from the corresponding specific product matching sequence to obtain the recommended product set for each product subcategory; The recommended product sets of each product subcategory are aggregated to form a product recommendation list tailored to the target user.

[0063] In this embodiment, the product recommendation list acquisition module 14 combines category priority with product quality matching to generate a final recommendation list that is structurally sound and highly accurate. To achieve the above objectives, the total number of recommended products in the product recommendation list needs to be obtained first. The total number of recommended products is preset by the system or dynamically set according to the business scenario. For example, the recommended area of ​​the product details page of an e-commerce APP will display 10 products, or the homepage recommendation bar will display 20 products. This number is the basic total for subsequent category quota allocation. Let's assume that the total number of recommended products is set to 10 here.

[0064] Next, based on the purchase probability and total number of recommendations for each product subcategory, the number of recommended products for each subcategory needs to be calculated. Specifically, the purchase probability of each product subcategory is used as a weight, combined with the total number of recommendations, and the number of recommendations allocated to each subcategory is calculated using the formula "purchase probability × total number of recommendations". If the result is a decimal, it is rounded to ensure that the number of recommendations is an integer, and the sum of the number of recommendations for all categories equals the total number of recommendations.

[0065] For example, if the total number of recommendations is 10, and the purchase probability of folding tables and chairs is 60%, camping lights 30%, and portable stoves 10%, then the number of recommendations for folding tables and chairs = 10 × 60% = 6; the number of recommendations for camping lights = 10 × 30% = 3; the number of recommendations for portable stoves = 10 × 10% = 1; and the total number of recommendations is 6 + 3 + 1 = 10, which is the same as the total number of recommendations.

[0066] Then, according to the number of recommended products corresponding to each product subcategory, the corresponding number of recommended products need to be selected sequentially from the corresponding specific product matching sequence to obtain the recommended product set for each product subcategory. Specifically, for each product subcategory, a corresponding "specific product matching sequence" is generated. This sequence is arranged in descending order of quality matching degree. For example, the specific product matching sequence for folding tables and chairs is "Product B (90%), Product A (85%), Product C (70%)...". Based on the allocated recommendation quota for that product subcategory, the corresponding number of products are selected sequentially starting from the head of the sequence. For example, if folding tables and chairs require 6 recommended products, the first 6 products in the sequence are selected; if camping lights require 3, the first 3 products in the sequence are selected, forming a unique "recommended product set" for each subcategory.

[0067] Finally, the "Recommended Product Sets" of all subcategories are integrated, and the products in each category are arranged from high to low according to the purchase probability of each product subcategory. For example, 6 products of folding tables and chairs are displayed first, followed by 3 products of camping lights, and finally 1 product of portable stoves. Alternatively, the products can be sorted in a mixed manner according to their quality matching degree.

[0068] The final product recommendation list, consisting of 10 items, is tailored to the target user and can be directly displayed in the recommendation area of ​​the cross-border e-commerce platform.

[0069] Example 2, as Figure 2 As shown, based on the same inventive concept as the blockchain-based real-time push system for cross-border e-commerce product data provided in Embodiment 1, this embodiment of the invention also provides a blockchain-based real-time push method for cross-border e-commerce product data, including: S100: Responding to the target user's search and purchase behavior for new product categories, constructing the target user's target purchase behavior chain in the blockchain network based on the target user's purchase product time sequence; S200: Based on the target purchase behavior chain, purchase behavior chain matching is performed in the blockchain network to obtain the purchase probability of multiple subsequent product subcategories; S300: Obtain multiple specific product sets from multiple product subcategories, and determine the quality matching degree of each specific product in the multiple specific product sets by combining the target purchase behavior chain, to obtain multiple specific product matching sequences; S400: Based on the purchase probability of multiple product subcategories and multiple specific product matching sequences, a product recommendation list is generated for the target user.

[0070] In step S100 of this embodiment, in response to the target user's retrieval and purchase behavior of a new product category, a target purchase behavior chain for the target user is constructed in the blockchain network based on the target user's purchase product time sequence. This includes: detecting the target user's first retrieval and purchase behavior of a new product category, and determining whether the new product category is a new purchase scenario for the target user; when the new product category is a new purchase scenario for the target user, obtaining the target user's purchase product information under the new product category, including product identifier, purchase time, product price, and quality grade; sorting the purchase product information according to the purchase time sequence to form a purchase product time sequence; creating a purchase behavior chain with the target user as the main body in the blockchain network, and writing the purchase product time sequence into the purchase behavior chain in the form of nodes to construct the target purchase behavior chain.

[0071] In step S200 of this application embodiment, the purchase behavior chain matching is performed in the blockchain network based on the target purchase behavior chain to obtain the purchase probability of multiple subsequent product subcategories. This includes: retrieving multiple reference purchase behavior chains from the blockchain network that have similar purchase paths to the target purchase behavior chain; extracting subsequent purchase product data at the same purchase node from the multiple reference purchase behavior chains based on the latest purchase node of the target purchase behavior chain; summarizing the product subcategories in the subsequent purchase product data to obtain multiple product subcategories, and statistically obtaining the purchase probability of each product subcategory.

[0072] In step S200 of this embodiment, retrieving multiple reference purchase behavior chains with similar purchase paths to the target purchase behavior chain from the blockchain network includes: extracting the product subcategories of each purchase node in the target purchase behavior chain to form a target product subcategory sequence; retrieving and obtaining a set of historical purchase behavior chains from the blockchain network based on the new product category; traversing the set of historical purchase behavior chains to obtain a first historical purchase behavior chain, and extracting the product subcategories of each purchase node in the first historical purchase behavior chain to form a first historical product subcategory sequence; extracting sequence segments of corresponding length from the first historical product subcategory sequence according to the length of the target product subcategory sequence to form a first comparison sequence; and adding the first historical purchase behavior chain to the multiple reference purchase behavior chains when all elements in the target product subcategory sequence exist in the first comparison sequence.

[0073] In step S200 of this embodiment, the product subcategories in the subsequent purchase product data are summarized to obtain multiple product subcategories, and the purchase probability of each product subcategory is calculated. This includes: extracting the product subcategories of each purchased product from the subsequent purchase product data, removing duplicates to form a product subcategory set, which includes multiple product subcategories; counting the number of times each product subcategory appears in the subsequent purchase product data; and calculating the ratio of the number of times each product subcategory appears to the total number of subsequent purchase product data to obtain the purchase probability of each product subcategory.

[0074] In step S300 of this embodiment, multiple specific product sets of multiple product subcategories are obtained, and the quality matching degree of each specific product in the multiple specific product sets is determined in combination with the target purchase behavior chain to obtain multiple specific product matching sequences. This includes: retrieving specific products corresponding to each product subcategory from the cross-border product database according to multiple product subcategories to form multiple specific product sets; analyzing the percentile ranking of each purchased product in the target purchase behavior chain within the corresponding product subcategory to obtain multiple percentile rankings of purchased prices, and calculating the average percentile ranking of the multiple percentile rankings of purchased prices; calculating the ranking deviation value between the percentile ranking of each specific product in the multiple specific product sets and the average percentile ranking; calculating the quality matching degree of each specific product based on the ranking deviation value, where the quality matching degree and the ranking deviation value are negatively correlated; and sorting the specific products in each specific product set in descending order according to the quality matching degree to obtain multiple specific product matching sequences.

[0075] In step S400 of this embodiment, a product recommendation list for the target user is obtained based on the purchase probability of multiple product subcategories and multiple specific product matching sequences. This includes: obtaining the total number of recommendations in the product recommendation list; calculating the number of recommended products corresponding to each product subcategory based on the purchase probability of each product subcategory and the total number of recommendations; selecting the corresponding number of recommended products from the corresponding specific product matching sequences in sequence according to the number of recommended products corresponding to each product subcategory to obtain a set of recommended products for each product subcategory; and summarizing the set of recommended products for each product subcategory to form a product recommendation list for the target user.

[0076] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0077] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0078] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0079] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0080] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0081] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0082] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A blockchain-based real-time push system for cross-border e-commerce product data, characterized in that: The system includes: The target purchase behavior chain construction module is used to respond to the target user's search and purchase behavior for new product categories, and to construct the target user's target purchase behavior chain in the blockchain network based on the time sequence of the target user's purchase of products. The purchase probability acquisition module is used to perform purchase behavior chain matching in the blockchain network based on the target purchase behavior chain to obtain the purchase probability of multiple subsequent product subcategories. The specific product matching sequence acquisition module is used to acquire multiple specific product sets of the multiple product subcategories, and determine the quality matching degree of each specific product in the multiple specific product sets in combination with the target purchase behavior chain to obtain multiple specific product matching sequences. The product recommendation list acquisition module is used to obtain a product recommendation list for the target user based on the purchase probability of the multiple product subcategories and the matching sequence of the multiple specific products.

2. The system according to claim 1, characterized in that, In response to target users' searches for and purchases of new product categories, a target user's target purchase behavior chain is constructed in the blockchain network based on the time sequence of the target user's purchases, including: Detect the target user's first search and purchase behavior for a new product category, and determine whether the new product category is a new purchase scenario for the target user; When the new product category is a new purchase scenario for the target user, the purchase product information of the target user under the new product category is obtained, and the purchase product information includes product identifier, purchase time, product price and quality level; The purchased product information is sorted according to the purchase time to form a purchase time sequence; In a blockchain network, a purchase behavior chain centered on the target user is created, and the purchase sequence of the goods is written into the purchase behavior chain in the form of nodes to construct the target purchase behavior chain.

3. The system according to claim 1, characterized in that, Based on the target purchase behavior chain, purchase behavior chain matching is performed in the blockchain network to obtain the purchase probability of multiple subsequent product subcategories, including: Retrieve from the blockchain network multiple reference purchase behavior chains that have similar purchase paths to the target purchase behavior chain; Based on the latest purchase node of the target purchase behavior chain, extract subsequent purchase product data at the same purchase node from the multiple reference purchase behavior chains; By summarizing the product subcategories in the subsequent purchase data, multiple product subcategories are obtained, and the purchase probability of each product subcategory is calculated.

4. The system according to claim 3, characterized in that, Retrieve multiple reference purchase behavior chains from the blockchain network that have similar purchase paths to the target purchase behavior chain, including: Extract the product subcategories of each purchase node in the target purchase behavior chain to form a target product subcategory sequence; Based on the new product category, retrieve the historical purchase behavior chain set in the blockchain network; Traverse the set of historical purchase behavior chains to obtain the first historical purchase behavior chain, and extract the product sub-category of each purchase node in the first historical purchase behavior chain to form the first historical product sub-category sequence; Based on the length of the target product sub-category sequence, a sequence segment of corresponding length is extracted from the first historical product sub-category sequence to form a first comparison sequence; When each element in the target product sub-category sequence exists in the first comparison sequence, the first historical purchase behavior chain is added to multiple reference purchase behavior chains.

5. The system according to claim 3, characterized in that, The product subcategories in the subsequent purchase data are summarized to obtain multiple product subcategories, and the purchase probability of each product subcategory is calculated, including: Extract the product subcategories of each purchased product from the subsequent purchase product data, and form a product subcategory set after deduplication. The product subcategory set includes multiple product subcategories. Count the number of times each product subcategory in the set of product subcategories appears in the subsequent purchase product data; Calculate the ratio of the frequency of each product subcategory to the total number of subsequent purchase data to obtain the purchase probability of each product subcategory.

6. The system according to claim 1, characterized in that, Obtain multiple specific product sets from the multiple product subcategories, and determine the quality matching degree of each specific product in the multiple specific product sets in conjunction with the target purchase behavior chain to obtain multiple specific product matching sequences, including: Based on the multiple product subcategories, specific products corresponding to each product subcategory are retrieved from the cross-border product database to form multiple specific product sets; Analyze the percentile ranking of the purchased price of each purchased item in the target purchase behavior chain within the corresponding product subcategory to obtain multiple percentile rankings of the purchased price, and calculate the average percentile ranking of the multiple percentile rankings of the purchased price. Calculate the ranking deviation between the price percentile ranking of each specific product in the multiple specific product sets and the average price percentile ranking; The quality matching degree of each specific product is calculated based on the ranking deviation value, and the quality matching degree is negatively correlated with the ranking deviation value. Based on the quality matching degree, the specific products in each specific product set are sorted in descending order to obtain multiple specific product matching sequences.

7. The system according to claim 1, characterized in that, Based on the purchase probabilities of the multiple product subcategories and the multiple specific product matching sequences, a product recommendation list is obtained for the target user, including: Get the total number of recommendations in the product recommendation list; Calculate the number of recommended products for each product subcategory based on the purchase probability of each product subcategory and the total number of recommendations. Based on the number of recommended products corresponding to each product subcategory, the corresponding number of recommended products are selected sequentially from the corresponding specific product matching sequence to obtain the recommended product set for each product subcategory; The recommended product sets of each product subcategory are aggregated to form a product recommendation list for the target user.

8. A method for real-time push of cross-border e-commerce product data based on blockchain, characterized in that: The method is implemented through the blockchain-based real-time push system for cross-border e-commerce commodity data as described in any one of claims 1-7, and the method includes: In response to the target user's search and purchase behavior for new product categories, a target user's target purchase behavior chain is constructed in the blockchain network based on the time sequence of the target user's purchases. Based on the target purchase behavior chain, purchase behavior chain matching is performed in the blockchain network to obtain the purchase probability of multiple subsequent product subcategories; Obtain multiple specific product sets from the multiple product subcategories, and determine the quality matching degree of each specific product in the multiple specific product sets in combination with the target purchase behavior chain to obtain multiple specific product matching sequences; A product recommendation list for the target user is obtained based on the purchase probability of the multiple product subcategories and the matching sequence of the multiple specific products.

Images