A commodity name search prompting method, device, equipment, medium and product

Abstract

The application provides a commodity name search prompting method, device, equipment, medium and product, and relates to the technical field of data processing. The method receives a target input word input by a user and matches to generate an initial candidate commodity name set in a preset commodity name mapping table, quickly completing commodity candidate screening at a semantic level. By obtaining commodity data associated with each candidate commodity name in the initial candidate commodity name set and performing supply chain feature extraction and standardization processing on the commodity data to generate a corresponding supply chain feature vector, real-time integration of supply chain related information such as commodity inventory status, logistics timeliness, and promotion benefits can be achieved. By calculating a negotiable score for each candidate commodity name based on the supply chain feature vector and outputting a preset number of target commodity names in order of the negotiable score, the recommendation of commodities that are out of stock, lack of stock, or do not participate in the current promotion is avoided, and user invalid clicks are reduced. By outputting target commodity names that match the actual negotiable state, the user search query and order operation experience is optimized.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, device, medium and product for providing product name search suggestions. Background Technology

[0002] Currently, in e-commerce, logistics ordering, and various online transaction scenarios, product name search suggestion technology is widely used to improve user input efficiency. When a user enters part of the text in the search box, the system will automatically recommend a list of relevant complete product names to help the user quickly complete the query or place an order.

[0003] Existing search suggestion technologies primarily rely on static lexicon matching or historical language models for simple semantic matching to generate an initial candidate list. However, this method focuses on the semantic relevance of words and fails to perceive the current commercial availability of products. This leads to the system recommending products that, while semantically matching, are actually sold out, out of stock, or not participating in current promotions. Consequently, this not only results in invalid clicks and reduced search conversion rates but also negatively impacts the user's shopping experience. Summary of the Invention

[0004] This application provides a product name search suggestion method, apparatus, device, medium, and product to solve existing problems.

[0005] Firstly, this application provides a product name search suggestion method, including:

[0006] Receive the target input word from the user;

[0007] The target input word is matched against a preset product name mapping table to generate an initial set of candidate product names;

[0008] Obtain the product data associated with each candidate product name in the initial candidate product name set;

[0009] The commodity data is subjected to supply chain feature extraction and standardization processing to generate a supply chain feature vector corresponding to each candidate product name;

[0010] Each candidate product name is scored based on the supply chain feature vector to obtain a transaction score for each candidate product name.

[0011] The initial set of candidate product names is sorted based on the transaction score, and a preset number of target product names are output and displayed.

[0012] Secondly, this application provides a product name search suggestion device, comprising:

[0013] The receiving module is used to receive the target input word from the user.

[0014] The matching module is used to match the target input word in a preset product name mapping table to generate an initial set of candidate product names;

[0015] The acquisition module is used to acquire the product data associated with each candidate product name in the initial candidate product name set;

[0016] The generation module is used to extract and standardize the supply chain features of the commodity data to generate supply chain feature vectors corresponding to each candidate product name.

[0017] The scoring module is used to score each of the candidate product names based on the supply chain feature vector to obtain the transaction score of each candidate product name.

[0018] The output module is used to sort the initial candidate product name set based on the transactionable score and output a preset number of target product names.

[0019] Thirdly, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0020] The memory stores computer-executed instructions;

[0021] The processor executes computer execution instructions stored in the memory to implement the method as described in any of the first aspects.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any of the first aspects.

[0023] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the method described in any of the first aspects.

[0024] This application provides a product name search suggestion method, apparatus, device, medium, and product. The method receives target input words from a user and matches them in a preset product name mapping table to generate an initial set of candidate product names. This quickly completes semantic-level product candidate filtering, avoiding the problems of relying solely on static dictionaries or simple semantic matching, and the limited selection methods for candidate products. By acquiring product data associated with each candidate product name in the initial set of candidate product names, and performing supply chain feature extraction and standardization on the product data to generate corresponding supply chain feature vectors, it integrates supply chain-related information such as product inventory status, logistics timeliness, and promotional benefits in real time, avoiding the problem of existing technologies being unable to perceive the real-time commercial availability of products. By calculating a tradability score for each candidate product name based on the supply chain feature vectors and outputting a preset number of target product names according to the tradability score, it avoids recommending semantically matched but actually sold out, out of stock, or not participating in current promotions, reducing invalid user clicks and improving search conversion rates. By outputting target product names that accurately reflect actual tradability status, it optimizes the user's search query and order placement experience, improving the user shopping experience in e-commerce, logistics ordering, and various online transaction scenarios. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0026] Figure 1 An application scenario diagram corresponding to a product name search suggestion method provided in an embodiment of this application;

[0027] Figure 2 A flowchart illustrating a product name search suggestion method provided in an embodiment of this application;

[0028] Figure 3 A flowchart illustrating a product name search suggestion method provided in another embodiment of this application;

[0029] Figure 4 A flowchart illustrating a product name search suggestion method provided in another embodiment of this application;

[0030] Figure 5 This is a schematic diagram of the structure of a product name search suggestion device provided in an embodiment of this application;

[0031] Figure 6 This is a structural example diagram of an electronic device provided in an embodiment of this application.

[0032] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0033] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0034] To clearly understand the technical solution of this application, the solutions of the prior art will be described in detail first.

[0035] In current e-commerce, logistics ordering, and various online transaction scenarios, product name search suggestion technology is widely used to improve user input efficiency. When a user enters part of the text in the search box, the system automatically recommends a list of relevant complete product names to help the user quickly complete the search or place an order.

[0036] However, existing search suggestion technologies primarily rely on static lexicon matching or historical language models for simple semantic matching to generate an initial candidate list. This method focuses on the semantic relevance of words but fails to consider the current commercial availability of goods. Specifically, in real-world business, the inventory status (adequacy), logistics timeliness (fast shipping / delivery), and promotional benefits (discounts) are key factors determining whether a user ultimately places an order. However, existing search suggestion systems are often disconnected from backend enterprise resource planning (ERP) systems, warehouse management systems, and marketing systems. When generating search suggestion terms, this dynamic supply chain data is not acquired and integrated in real time. This leads to the system frequently recommending products that, while semantically matching, are actually sold out, out of stock, or not participating in current promotions. Users clicking on these recommendations often face the inability to place an order or seeing an out-of-stock message, resulting in invalid clicks, reduced search conversion rates, and a severe impact on the user's shopping experience.

[0037] Figure 1 An application scenario diagram corresponding to a product name search suggestion method provided in an embodiment of this application is shown, such as... Figure 1As shown, the application scenario provided in this embodiment includes: a user terminal 10, a search suggestion device 11, and a backend system 12. The user terminal and the search suggestion device communicate in real time via a network, and the search suggestion device establishes a linkage with the backend system through an API interface. The user terminal can be a mobile phone, computer, tablet, etc., equipped with an e-commerce APP, a logistics ordering mini-program, or an online transaction platform frontend. The backend system can be an enterprise resource planning system, a warehouse management system, or a marketing system, storing supply chain data such as product inventory, logistics, and promotions.

[0038] The product name search suggestion method provided in this application can be applied to scenarios such as e-commerce platform product search, logistics order product entry, and online retail transactions. In such scenarios, users need to complete the query and order placement by entering the product name, which requires high search input efficiency and recommendation accuracy.

[0039] Specifically, the search suggestion process is as follows:

[0040] First, when a user enters a target word in the search box of user terminal 10, user terminal 10 sends the target word to search suggestion device 11. Search suggestion device 11 receives the target word in real time and matches it against a preset product name mapping table to generate an initial set of candidate product names, ensuring that the candidate product names are semantically relevant to the user's input. Next, search suggestion device 11 obtains real-time product data such as inventory status, logistics timeliness, and promotional benefits for each candidate product name in the initial set of candidate product names from the backend system 12 via an API interface. Then, search suggestion device 11 performs supply chain feature extraction and standardization on the product data to generate a supply chain feature vector for each candidate product name. Based on the supply chain feature vector, it calculates the tradability score for each candidate product name to quantify the probability of a transaction. Finally, search suggestion device 11 sorts the target product names from high to low according to their tradability scores, selects a preset number of target product names, and sends them to user terminal 10. User terminal 10 displays the target product names below the search box for the user to quickly select, thereby achieving a dual match of semantic relevance and commercial availability.

[0041] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0042] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0043] Figure 2 This is a flowchart illustrating a product name search suggestion method provided in one embodiment of this application, as shown below. Figure 2 As shown, the execution subject of this embodiment is a product name search suggestion device. This device can be implemented by a computer program, or by a medium storing the relevant computer program, such as a USB flash drive and / or optical disc; alternatively, it can be implemented by a physical device integrating or installing the relevant computer program, such as a chip or electronic device. The electronic device may be a computer or a server, etc. The product name search suggestion method provided in this embodiment includes the following steps:

[0044] S201, Receive the target input word input by the user.

[0045] Optionally, partial text information entered by the user can be captured in real time through user interaction interfaces (such as search boxes, input panels, etc.), i.e., target input words. Target input words can be keyword fragments, abbreviations, synonyms, or fuzzy inputs of product names. For example, users entering "pen," "laundry detergent," or "mobile phone" all fall under the target input words referred to in this step.

[0046] Optionally, the method of receiving user input can be adapted to various online transaction scenarios, including but not limited to web search boxes, mobile app search bars, and mini-program input boxes. Furthermore, it supports real-time input and real-time reception; that is, during the user's input process (each time a character or word is entered), the input is captured in real time and subsequent steps are triggered, ensuring the real-time nature of search suggestions and avoiding user waiting.

[0047] Optionally, simple preprocessing can be performed on the target input words entered by the user, such as removing spaces and special characters, and unifying capitalization, to ensure the accuracy of subsequent matching and improve matching efficiency.

[0048] S202. Match the target input word in the preset product name mapping table to generate an initial set of candidate product names.

[0049] Specifically, this step is based on a preset product name mapping table. Through a keyword matching algorithm, the target input word is associated with and matched with the product names in the mapping table to filter out semantically related complete product names and form an initial candidate product name set.

[0050] The preset product name mapping table is a structured mapping relationship that integrates the complete names of all products sold on the platform and associates them with the core information of the products. The core information of the products includes, but is not limited to, product ID, product category, common synonyms, abbreviations, and fuzzy matching keywords.

[0051] Optionally, the matching algorithm used in the matching process can combine character matching and semantic matching. Character matching is used to accurately match the user's input keywords with character fragments of the product name, while semantic matching, based on a simple semantic analysis model, identifies the semantic association between the user's input words and the product name, such as the semantic association between "fast charging head" and "fast charger," avoiding matching omissions due to non-standard input. Through these matching methods, all complete product names semantically related to the target input word are filtered out, forming an initial candidate product name set.

[0052] It should be noted that this initial set of candidate product names is the basis for further screening and sorting. The number of candidate product names in the initial set can be flexibly set according to the platform's business needs to ensure that all relevant products are covered while avoiding an excessively large set that would reduce the efficiency of subsequent processing.

[0053] S203. Obtain the product data associated with each candidate product name in the initial candidate product name set.

[0054] Specifically, this step uses a data interface to obtain real-time dynamic product data for each candidate product name in the initial candidate product name set, thereby synchronizing search suggestions with supply chain data and marketing data.

[0055] Specifically, each candidate product name is associated with the backend warehouse management system and marketing system through the product ID. This step uses standardized data interfaces, such as API interfaces, to adapt to the data formats of different systems and achieve lightweight integration, retrieving product data corresponding to the candidate product name from the above systems in real time.

[0056] Optionally, product data may include inventory data, logistics data, and promotional data, covering key factors that make a product tradable.

[0057] Inventory data can be obtained from the warehouse management system, including the current inventory quantity, inventory status (sufficient, scarce, out of stock, sold out), and inventory location. The warehouse management system is responsible for the refined management of goods in the warehouse, and its inventory data reflects the actual supply capacity of goods in real time, which is the basis for determining whether a product can be ordered.

[0058] Logistics data can be obtained from enterprise resource planning or logistics management systems, including whether the product supports same-day delivery, estimated delivery time, estimated arrival time, logistics coverage, and delivery method (such as express delivery or regular delivery). Logistics timeliness directly affects users' willingness to place orders and is an important indicator of the product's transactability.

[0059] Promotional data can be obtained from the marketing system, including whether the product participates in the current promotion, the type of promotion (discount, spending threshold reduction, coupon, buy-one-get-one-free, etc.), the promotion strength, and the promotion validity period. Promotional benefits are key factors in increasing users' willingness to place orders and promoting transactions.

[0060] Specifically, data acquisition can adopt a real-time retrieval mode to ensure that the acquired product data is consistent with the backend system, avoiding invalid recommendations due to data delays. Furthermore, for multi-source, heterogeneous system data (different systems have different data formats and statistical standards), preliminary data cleaning is performed to remove invalid and duplicate data, ensuring the accuracy of subsequent processing.

[0061] S204. Extract and standardize the supply chain features of the commodity data to generate supply chain feature vectors for each candidate product name.

[0062] This step transforms the acquired unstructured, multi-dimensional product data (inventory, logistics, and promotional data) into quantifiable and computable feature vectors, providing data support for subsequent scoring and ranking, and avoiding the problem of multi-source heterogeneous data being unable to be directly compared and calculated.

[0063] Optionally, supply chain feature extraction can be performed by extracting core features reflecting the transactability of goods from inventory data, logistics data, and promotional data respectively, forming a feature set. For example, extracting features such as inventory adequacy (e.g., the proportion of inventory to safety stock), inventory status indicators (sufficient with high weight, scarce with medium weight, out of stock / sold out with low weight or 0 weight), shipping timeliness (same-day shipping with high weight, next-day shipping with medium weight, shipping after 3 days with low weight), and delivery timeliness (delivery within 24 hours with high weight, delivery within 48 hours with medium weight, delivery after 48 hours with low weight). Extracting features such as promotional strength (discount percentage, percentage of minimum purchase amount reduction to unit price), and promotional effectiveness (whether the promotion is within its validity period).

[0064] It should be noted that the extraction of the above-mentioned supply chain features can be flexibly adjusted according to the platform's business needs to ensure that the extracted key indicators can directly affect users' ordering decisions and reflect the transactability of goods, thus ensuring the relevance and effectiveness of the features.

[0065] It should be noted that since the extracted features belong to different dimensions, their numerical ranges and units of measurement differ. For example, inventory quantity is an integer, discount percentage is a decimal, and timeliness is a time length, making direct fusion calculation impossible. Therefore, it is necessary to standardize all extracted features, transforming features of different dimensions into standardized values ​​within a unified range (such as the [0,1] interval).

[0066] Optionally, standardization can employ commonly used standardization algorithms in this field, such as min-max standardization and z-score standardization. For example, the min-max standardization algorithm can map the values ​​of features such as inventory adequacy, delivery timeliness, and promotional intensity to the [0,1] interval, where the closer the value is to 1, the stronger the scalability of the feature. Standardization ensures that various features are comparable and computable, providing a unified standard for subsequent calculation of scalability scores.

[0067] Optionally, all standardized features corresponding to each candidate product name are combined in a preset order (such as the order of inventory features, logistics features, and promotion features) to form a supply chain feature vector corresponding to the candidate product name. Each feature vector is a multi-dimensional numerical vector that intuitively reflects the transaction capability of the product in terms of inventory, logistics, promotion, and other dimensions.

[0068] S205. Based on the supply chain feature vector, score each candidate product name to obtain the transaction score for each candidate product name.

[0069] This step uses supply chain feature vectors to quantify the scalability of each candidate product name, resulting in a scalability score. This scalability score measures a product's scalability; a higher score indicates a higher probability that the product will be ordered by a user.

[0070] Optionally, based on the influence weight of each supply chain feature on the tradability of the product, a weighted sum is performed on each standardized feature in the feature vector to obtain a tradability score. The weight allocation can be dynamically adjusted according to platform business needs and user behavior data; features with a greater impact on user order decisions are assigned higher weights.

[0071] For example, in e-commerce scenarios, inventory status has the greatest impact on user orders, so the weight of inventory features can be allocated to 40%; logistics timeliness is the second most important factor, with a weight allocation of 30%; and promotional features have a weight allocation of 30%. The scoring formula can be expressed as: Score for Transaction Potential = (Standardized Value of Inventory Features × 40%) + (Standardized Value of Logistics Features × 30%) + (Standardized Value of Promotional Features × 30%). Of course, the weight allocation is not fixed and can be dynamically optimized based on platform type (e.g., fresh food e-commerce can increase the weight of logistics timeliness, while discount e-commerce can increase the weight of promotional features) and user group characteristics to ensure the rationality and relevance of the scoring.

[0072] Optionally, points can be deducted for products with obvious reasons why they cannot be sold. For example, for products that are sold out, the potential sale score can be set to 0. For products that cannot be shipped on the same day and whose delivery time exceeds 7 days, 50% of the score can be deducted. For products whose promotional activities have expired, all scores corresponding to the promotional features can be deducted. Deducting points can further improve the accuracy of the potential sale score and prevent invalid products from entering the final recommendation list.

[0073] It should be noted that each candidate product name will receive a specific transaction score (e.g., 0-100 points). The higher the score, the more abundant the product's inventory, the faster the logistics, and the more attractive the promotion, the higher the probability of users placing an order.

[0074] S206. Sort the initial candidate product name set based on the transaction score, and output the target product name with a preset number of items.

[0075] This step sorts the initial set of candidate product names based on their convertibility scores, filters out product names with higher convertibility, and outputs them as target product names to achieve the goal of prioritizing the recommendation of convertible products.

[0076] Optionally, the sorting logic adopts a "descending order" sorting, that is, all candidate product names in the initial candidate product name set are arranged in descending order of their tradable scores. Furthermore, for candidate product names with the same score, a secondary sorting can be performed by combining auxiliary factors such as semantic relevance (the matching degree between the target input word and the product name) and historical order frequency to ensure the rationality of the sorting.

[0077] Optionally, the number of target product names displayed in the output is a preset number, which can be flexibly set according to the platform's interaction needs, such as 5, 8, or 10. This ensures that users have enough choices while avoiding overwhelming them with too many recommendations. During the output display, product availability information such as "Sufficient Stock," "Same-Day Shipping," and "20% Discount" can be shown simultaneously to help users quickly determine the product's purchasability and improve the user experience.

[0078] Optionally, when product data (such as inventory, logistics, and promotional information) in the backend system changes, steps S203 to S206 are re-executed to update the sorting and display content of the target product names in real time, ensuring the timeliness and accuracy of the recommendation information.

[0079] This application provides a product name search suggestion method. It receives target input words from the user and matches them in a preset product name mapping table to generate an initial set of candidate product names. This quickly completes semantic-level product candidate filtering, avoiding the problems of relying solely on static dictionaries or simple semantic matching, and the limited selection methods for candidate products. By acquiring product data associated with each candidate product name in the initial set of candidate product names, and performing supply chain feature extraction and standardization on the product data to generate corresponding supply chain feature vectors, it integrates supply chain-related information such as product inventory status, logistics timeliness, and promotional benefits in real time, avoiding the problem of existing technologies being unable to perceive the real-time commercial availability of products. By calculating a tradability score for each candidate product name based on the supply chain feature vectors and sorting them according to the tradability score, it outputs a preset number of target product names, avoiding recommending semantically matched but actually sold out, out of stock, or not participating in current promotions, reducing invalid user clicks and improving search conversion rates. By outputting target product names that accurately reflect actual tradability status, it optimizes the user's search query and order placement experience, improving the user shopping experience in e-commerce, logistics ordering, and various online transaction scenarios.

[0080] Figure 3 A flowchart illustrating a product name search suggestion method provided in another embodiment of this application; as shown Figure 3 As shown, as an optional implementation, based on any of the above embodiments, the product data includes real-time inventory data, estimated delivery time data, promotional activity data, and historical click data.

[0081] Specifically, real-time inventory data is mainly obtained from the warehouse management system and includes the current inventory quantity and inventory warning threshold of each candidate product name. Optional data includes inventory status (sufficient, tight, out of stock, sold out). Real-time inventory data is the core data for evaluating the sellability of products and directly determines whether users can successfully place an order.

[0082] Specifically, the estimated delivery time data can be obtained in real time from the warehouse management system. The core data includes the estimated delivery hours of each candidate product name and the matching relationship between the shipping location and the receiving location. The estimated delivery time data can more accurately reflect the level of logistics services.

[0083] Specifically, promotional activity data can be obtained in real time from the marketing system. This data can include whether the products associated with each candidate product name are participating in the promotion, the type of promotion (discount, full reduction, coupon, etc.), the discount rate, and the remaining time of the promotion. This data is an important incentive to attract users to place orders and increase the probability of a transaction.

[0084] Specifically, historical click data can be obtained from the platform's user behavior log system. It can be the total number of clicks by users on products associated with each candidate product name in the past 24 hours, which reflects the user attention and interest preferences of the category represented by the candidate product name.

[0085] Optionally, to ensure the timeliness and availability of various data, full-scale relevant data can be retrieved from e-commerce platforms, enterprise resource planning systems, warehouse management systems, marketing systems, and user behavior log systems via API interfaces at preset intervals (e.g., every 10 minutes), ensuring data consistency with each backend system. Furthermore, the retrieved product data is cached for a short period (e.g., 10 minutes) to avoid increasing system load due to multiple API calls within a short timeframe. The cached data will be automatically updated with the full-scale synchronization every 10 minutes. Additionally, if any core product data cannot be retrieved for a candidate product name (e.g., product is no longer available, API is malfunctioning, no historical click records), the candidate product name is marked as "unavailable" and will be excluded from subsequent sorting.

[0086] Specifically, the commodity data undergoes supply chain feature extraction and standardization to generate supply chain feature vectors corresponding to each candidate product name, including the following steps:

[0087] S301. Determine inventory adequacy characteristics based on real-time inventory data; determine logistics timeliness characteristics based on estimated delivery time data; determine promotional attractiveness characteristics based on promotional activity data; determine category popularity characteristics based on historical click data.

[0088] The inventory adequacy feature is used to quantitatively assess the overall sellability of products associated with a candidate product name. It can be calculated as the ratio of the average inventory of associated products to the maximum inventory across all product categories. The formula is: Inventory Adequacy Feature Value = Average Inventory of Associated Products / Maximum Inventory Across All Product Categories. Here, the average inventory of associated products is the arithmetic mean of the inventory quantities of all product sets mapped to the candidate product name, and the maximum inventory across all product sets is the maximum inventory value among all product sets within the platform. The inventory adequacy feature ranges from [0,1]. A higher value indicates a more sufficient overall inventory and stronger sellability of the products associated with the candidate product name. For example, if a candidate product name "Perfume" is associated with two product sets with inventory quantities of 80 and 20 units respectively, and the average inventory of associated products is 50 units, and the maximum inventory across all product categories is 1000 units, then the inventory adequacy feature of this candidate product name is 50 / 1000 = 0.05.

[0089] The logistics timeliness feature is used to quantitatively evaluate the logistics convenience of associated products for a candidate product name. Specifically, the logistics timeliness feature value = 1 - average delivery time of associated products / maximum delivery time of all product categories. The average delivery time of associated products is the arithmetic mean of the estimated delivery hours for all product sets mapped to the candidate product name, and the maximum delivery time of all product categories is the maximum estimated delivery hours for all product sets within the platform. The value range of the logistics timeliness feature is [0,1]. A larger value indicates a shorter average delivery time for the associated products of the candidate product name, faster logistics, and lower waiting costs for users. For example, if a candidate product name "perfume" is associated with two product sets with estimated delivery times of 24 hours and 48 hours respectively, and the average delivery time of the associated products is 36 hours, and the maximum delivery time of all product categories is 72 hours, then the logistics timeliness feature of this candidate product name is 1 - 36 / 72 = 0.5.

[0090] The promotional attractiveness feature is used to quantitatively evaluate the marketing attractiveness of associated products for a candidate product name. Specifically, the promotional attractiveness feature value is equal to the maximum discount rate of the associated products. The discount rate is calculated as (original price - promotional price) / original price. If a product does not participate in any promotional activities, the discount rate is 0. The maximum discount rate among associated products is prioritized as the promotional attractiveness feature value to capture the most attractive marketing information for the candidate product name and reflect its appeal to users to the greatest extent. For example, if a candidate product name "perfume" is associated with two product sets, one with a discount rate of 0.5 (50% off) and the other with a discount rate of 0.3 (70% off), then the promotional attractiveness feature of this candidate product name is 0.5.

[0091] The category popularity feature is used to quantitatively evaluate user interest and attention to the category represented by the candidate product name. The calculation formula is: Category Popularity Feature Value = log(Sum of Clicks of Related Products + 1) / log(Maximum Clicks Across the Site + 1). Here, the sum of clicks of related products is the total number of user clicks on all product sets mapped to the candidate product name within the past 24 hours; adding 1 is to avoid the logarithm being meaningless when the click count is 0. The maximum clicks across the site is the maximum number of clicks on all product sets on the platform within the past 24 hours. The category popularity feature ranges from [0,1]. The larger the value, the higher the user attention to the category represented by the candidate product name, and the greater the probability of a transaction after a user clicks. For example, a candidate product name "perfume" is associated with two product sets, with 5,000 and 1,000 clicks in the past 24 hours respectively. The sum of clicks on the associated products is 6,000. If the maximum number of clicks on the entire site is 10,000, then the category popularity characteristic of this candidate product name is log(6000+1) / log(10000+1)≈3.77 / 4≈0.9425.

[0092] It should be noted that inventory adequacy and logistics timeliness are aggregated using average values ​​to assess the overall level of the category represented by the candidate product name; promotional attractiveness is aggregated using maximum values ​​to capture the most competitive marketing highlights; and category popularity is aggregated using summation and normalization to reflect the overall user attention of the category, ensuring that each feature accurately serves the goal of recommending products that can be sold.

[0093] S302. Normalize the inventory adequacy characteristics, logistics timeliness characteristics, promotional attractiveness characteristics, and category popularity characteristics to generate supply chain feature vectors corresponding to each candidate product name.

[0094] Optionally, a min-max normalization method is used to uniformly map the values ​​of the four types of features to the [0,1] interval, eliminating the dimensional differences between different features and ensuring that each type of feature has equal weight in the subsequent calculation of the tradable score. Furthermore, for features that already meet the [0,1] value range, the calculation results can be directly used without additional normalization processing; if the feature value exceeds the [0,1] interval due to data fluctuations, it is adjusted to the target interval using the min-max normalization formula.

[0095] Specifically, after normalization, the four types of features are combined in a preset order, such as inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity, to form a supply chain feature vector for each candidate product name. This supply chain feature vector integrates factors such as "product sellability, logistics service, marketing attractiveness, and user interest." For example, if the four feature values ​​of a candidate product name are 0.05, 0.5, 0.5, and 0.9425, then its supply chain feature vector is [0.05, 0.5, 0.5, 0.9425]. This supply chain feature vector can more comprehensively and accurately reflect the transaction potential of the candidate product name.

[0096] This application provides a product name search suggestion method that comprehensively considers multi-dimensional information from product data, such as real-time inventory, estimated delivery time, promotional activities, and historical clicks. It extracts features such as inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity, and generates supply chain feature vectors through normalization processing. This allows the search suggestions to not only match semantics but also accurately reflect the actual status of the product at each stage of the supply chain and user preferences. As a result, it effectively selects products with high transaction potential for priority recommendation, improves the usability and conversion rate of search results, and optimizes the efficiency and experience of users' shopping decisions.

[0097] As an optional implementation, based on any of the above embodiments, determining promotional attractiveness characteristics based on promotional activity data includes the following steps:

[0098] It should be noted that if only the maximum discount rate is used as the criterion for judging the promotion attractiveness, the timeliness of the promotion activity is not considered. For example, if a commodity has a large discount but the promotion is about to end, users may miss the preferential treatment and cannot enjoy the rights and interests, resulting in a reduction in the actual value of the promotion attractiveness; on the contrary, if the discount rate is moderate but the promotion period is sufficient, users have enough time to make a decision to place an order, and its actual attractiveness is higher. Therefore, in this embodiment, the promotion attractiveness is quantified through the dual dimensions of the maximum discount rate and the remaining promotion time, taking into account both the preferential strength and the timeliness, making the features more in line with the actual order placement decisions of users and improving the accuracy of feature quantification.

[0099] First, extract the maximum discount rate among all the commodities associated with any candidate product name, and determine the commodity providing the maximum discount rate as the target promotion commodity.

[0100] Since a candidate product name may be associated with multiple commodity sets and there are differences in the promotion strength of different commodity sets, the maximum discount rate is the promotion signal that can most attract users. Therefore, the maximum discount rate is preferentially extracted as the basis for basic judgment. Determining the commodity providing the maximum discount rate as the target promotion commodity facilitates the subsequent accurate acquisition of the promotion timeliness information of this commodity and ensures the data correspondence and accuracy.

[0101] For example, if the candidate product name "perfume" is associated with commodity set 101 (discount 0.5, i.e., 50% off) and commodity set 102 (discount 0.7, i.e., 70% off), then extract the maximum discount rate 0.5 and determine commodity set 101 as the target promotion commodity. If all the commodities associated with the candidate product name have no promotion activities, the maximum discount rate takes the value of 0, and there is no need for subsequent steps. The promotion attractiveness feature is directly determined to be 0.

[0102] Second, obtain the remaining promotion time of the target promotion commodity.

[0103] Among them, the remaining promotion time refers to the remaining duration from the current moment to the end moment of the promotion activity of the target promotion commodity, and this data is sourced from the promotion activity data synchronized by the marketing system. For example, if the promotion activity of the target promotion commodity starts at 0:00 on the current day and ends 72 hours later, and there are still 48 hours until the end of the activity at the current moment, then the remaining promotion time of this commodity is 48 hours. If the promotion activity of the target promotion commodity has expired, the remaining promotion time takes the value of 0, and the promotion attractiveness of this commodity will be greatly reduced after subsequent coefficient correction, avoiding ineffective promotion recommendations.

[0104] Finally, perform coefficient correction on the maximum discount rate according to the remaining promotion time, and determine the corrected value as the promotion attractiveness feature.

[0105] Specifically, by using a correction coefficient, the impact of the remaining time of the promotion on the attractiveness of the promotion is quantified, so that the attractiveness characteristics of the promotion can reflect both the discount strength and the timeliness of the promotion, avoiding characteristic bias caused by ignoring timeliness.

[0106] Optionally, the adjustment coefficient can be set according to the logic of "the longer the remaining time of the promotion, the larger the adjustment coefficient; the shorter the remaining time, the smaller the adjustment coefficient; and when the remaining time is 0, the adjustment coefficient is 0," which aligns with users' decision-making habits. When there is ample remaining time, users have enough time to consider placing an order, making the promotion more attractive; when there is insufficient remaining time, users may hesitate for fear of missing out on the discount or be unable to place an order in time, reducing the promotion's attractiveness; and when the promotion expires, it has no appeal whatsoever.

[0107] Optionally, the correction formula can be: k = t1 / t2, where k is the correction coefficient (ranging from [0,1]), t1 is the remaining promotional time for the target promotional product, and t2 is the total duration of the promotional activity for the target promotional product. Therefore, in this embodiment, the calculation formula for the promotional attractiveness feature is: Promotional attractiveness feature value = k × the maximum discount ratio of the candidate product name associated with the product.

[0108] For example, if the maximum discount ratio of the target promotional product is 0.5, the total promotion duration is t2 = 72 hours, and the remaining time of the current promotion is t1 = 48 hours, then the correction coefficient k = 48 / 72 ≈ 0.667, and the promotion attractiveness characteristic value = 0.667 × 0.5 ≈ 0.333. This correction logic can accurately quantify the impact of promotional timeliness on attractiveness, making the promotion attractiveness characteristics more closely match real-world scenarios.

[0109] It should be noted that the calculation logic of the correction coefficient can be flexibly adjusted according to the platform's business needs. For example, setting a minimum correction coefficient threshold (such as setting the correction coefficient to 1 when the remaining time is ≥24 hours) is within the protection scope of this embodiment.

[0110] This application provides a product name search suggestion method that extracts the maximum discount ratio from candidate product name-related products to lock in the target promotional product, and combines the remaining time of the promotion to adjust the discount ratio with a coefficient to determine the promotional attractiveness characteristics. This method not only highlights the attractiveness of high-discount products, but also dynamically adjusts the weight of promotional value through the time dimension, so that the search suggestion can more accurately reflect the current urgency of the product's promotion and the actual discount strength, thereby effectively recommending users to prioritize products with high cost performance and strong promotional timeliness.

[0111] As an optional implementation, based on any of the above embodiments, each candidate product name is scored based on the supply chain feature vector to obtain a transactional score for each candidate product name, including the following steps:

[0112] First, obtain the weights of the following characteristics: inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity.

[0113] It should be noted that the supply chain feature vector includes four dimensions: inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity. Each feature has a different impact on a product's scalability; for example, inventory adequacy typically has a higher impact on sales than category popularity. Therefore, by assigning corresponding weights to each feature, multiplying each feature by its corresponding weight, and then summing the results, a quantitative assessment of a product's overall scalability is achieved. The weight allocation aligns with business logic; a higher score indicates stronger overall scalability and makes the product more worthy of priority recommendation. This approach balances the impact of each dimension while also adapting to the business needs of different platforms through weight adjustments.

[0114] Specifically, the weights of inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity are the core parameters for quantifying the influence of each feature. Their values ​​are all in the range of [0,1], and the sum of the weights of inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity is 1, ensuring that the scoring results are regular and comparable, and avoiding score distortion due to weight imbalance.

[0115] Optionally, the system can preset a set of default weights. These default weights are set to align with general e-commerce transaction logic and are determined based on industry experience and user behavior data. For example, the default weight configuration could be: inventory adequacy feature weight w1=0.3, logistics timeliness feature weight w2=0.2, promotional attractiveness feature weight w3=0.3, and category popularity feature weight w4=0.2. This default configuration can meet the needs of most comprehensive e-commerce platforms. Furthermore, platforms can flexibly adjust the weight allocation according to their own business type. For example, discount e-commerce platforms can increase the promotional attractiveness feature weight (e.g., w3=0.4), fresh food e-commerce platforms can increase the logistics timeliness feature weight (e.g., w2=0.3), and platforms with tight inventory can increase the inventory adequacy feature weight (e.g., w1=0.4). After adjusting the weights, it must be ensured that the sum of the weights remains 1.

[0116] Optionally, the storage and retrieval of weights are associated with real-time synchronized product data, and can be configured and updated through the system backend. The updates take effect immediately, ensuring that the scoring logic can quickly adapt to platform business adjustments and improving the flexibility and scalability of the solution.

[0117] Secondly, the products of inventory adequacy feature and its weight, logistics timeliness feature and its weight, promotional attractiveness feature and its weight, and category popularity feature and its weight are summed. The summation result is used as the tradable score for the candidate product name.

[0118] It should be noted that by assigning different influence weights to each feature, the score can comprehensively reflect the overall performance of the product in four dimensions: inventory, logistics, promotion, and user attention. This avoids a single feature dominating the scoring result and ensures the comprehensiveness and rationality of the score.

[0119] Specifically, this can be expressed by a formula: Let the inventory adequacy characteristic be finv with a weight of w1, the logistics timeliness characteristic be flog with a weight of w2, the promotional attractiveness characteristic be fpro with a weight of w3, and the category popularity characteristic be fpop with a weight of w4. Then the formula for summing up the results is: .

[0120] The convertibility score is a quantitative indicator of the overall convertibility of the candidate product name. The closer the score is to 1, the more sufficient the product's inventory, the faster the logistics, the stronger the promotional appeal, and the higher the user attention. The stronger the overall convertibility, the more suitable it is to be recommended to users. The closer the score is to 0, the weaker the overall convertibility of the product. The recommendation priority should be reduced, or it should even be removed from the recommendation list.

[0121] For example, the normalized feature values ​​of a candidate product name "Perfume" are as follows: inventory adequacy feature finv=0.05, logistics timeliness feature flog=0.5, promotional attractiveness feature fpop=0.333, and category popularity feature fpop=0.9425. Using the default weights (w1=0.3, w2=0.2, w3=0.3, w4=0.2), the summation result is: 0.05×0.3 +0.5×0.2 +0.333×0.3 +0.9425×0.2=0.015+0.1+0.0999+0.1885=0.4034. Therefore, the transaction score of the candidate product name "Perfume" is 0.4034, which directly reflects its comprehensive transaction capability.

[0122] This application provides a product name search suggestion method that sets reasonable weights for supply chain characteristics such as inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity. The method then multiplies each characteristic by its corresponding weight and sums the results to obtain a transaction score. This achieves a quantitative processing of multi-dimensional comprehensive evaluation of candidate product names. It takes into account factors that affect transactions, such as actual product supply capacity, logistics efficiency, and promotional discounts, and also incorporates category popularity reflected by users' historical preferences. This allows the scoring results to more comprehensively and accurately reflect the transaction potential of the product.

[0123] Figure 4 A flowchart illustrating a product name search suggestion method provided in another embodiment of this application; as shown Figure 4As shown, as an optional implementation, based on any of the above embodiments, after obtaining the weights of inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity, the following steps are also included:

[0124] S401. Obtain the current operating scenario.

[0125] It should be noted that the platform has different operational goals at different stages of operation, and the corresponding user needs and product recommendation priorities will also change. For example, the goal of the inventory clearance stage is to quickly reduce inventory, the goal of the time-sensitive promotion stage is to highlight logistics advantages to attract users who want to place orders immediately, and the goal of the discount activity stage is to improve conversion rates through discounts. Therefore, by combining the current operational scenario and dynamically adjusting the weight allocation of the four types of features, the calculation of the conversion score can be more aligned with the current operational goals, ensuring that the recommendation results accurately serve the platform's operational needs.

[0126] The current operational scenario refers to the platform's current operational stage or key promotional scenario. This scenario information can be manually configured through the platform's operations backend or automatically identified by the system based on preset operational strategies, ensuring the accuracy and real-time nature of the scenario information. Specifically, common operational scenarios include, but are not limited to: inventory clearance scenarios, time-limited promotion scenarios, discount activity scenarios, and regular operational scenarios. Among these, the regular operational scenario is the default scenario, using preset default weight configurations, and does not require weight adjustment.

[0127] Optionally, the method for obtaining scenario information can be flexibly selected according to the platform's operation mode. For the manual configuration mode, operators can set the current operation scenario and the scenario's effective time in the backend according to the actual operation plan, and the system will read the scenario information on time and perform weight adjustments. Alternatively, the system can automatically determine the current operation scenario by analyzing data such as the platform's current promotional activity type, inventory status, and logistics strategy. For example, when the platform launches an "inventory clearance special" event, it will automatically be identified as an inventory clearance scenario.

[0128] S402. When the current operating scenario is an inventory clearance scenario, increase the weight of the inventory adequacy feature; when the current operating scenario is a time-sensitive promotion scenario, increase the weight of the logistics timeliness feature; when the current operating scenario is a discount activity scenario, increase the weight of the promotion attractiveness feature.

[0129] The operational goal of inventory clearance scenarios is to quickly reduce existing inventory, minimize inventory backlog, and improve inventory turnover efficiency. Therefore, it is necessary to increase the weight of the inventory adequacy feature. By increasing the tradability score of well-stocked products, products with sufficient inventory are prioritized for recommendation, thus facilitating inventory clearance. For example, the default weight of the inventory adequacy feature is adjusted from 0.3 to 0.45, and the weights of the other three features are correspondingly reduced, ensuring that the sum of the weights remains 1.

[0130] Among them, the operation goal of the time-limited promotion scenario is to highlight the logistics fulfillment advantages of the products, attract users with high demand for delivery timeliness (such as users who purchase urgently needed supplies), and improve the instant order conversion rate. Therefore, it is necessary to increase the weight of the logistics timeliness feature. By increasing the成交 score of products with fast logistics speed, products that are shipped on the same day and delivered quickly are preferentially recommended to meet the needs of users in the scenario. Exemplarily, the default weight of the logistics timeliness feature is adjusted from 0.2 to 0.4, and the weights of the other three types of features are correspondingly reduced to ensure that the sum of the weights is still 1.

[0131] Among them, the operation goal of the discount activity scenario is to attract users through the preferential intensity, improve the product click-through rate and the deal conversion rate, and expand the promotion effect. Therefore, it is necessary to increase the weight of the promotion attractiveness feature. By increasing the成交 score of products with strong promotion intensity, products with large preferential intensity and high cost performance are preferentially recommended to maximize the drainage and conversion effects of the promotion activity. Exemplarily, the default weight of the promotion attractiveness feature is adjusted from 0.3 to 0.45, and the weights of the other three types of features are correspondingly reduced to ensure that the sum of the weights is still 1.

[0132] It should be noted that the specific values of the weight adjustment can be flexibly set according to the specific needs of the platform operation scenario. For example, in the inventory clearance scenario, if the inventory backlog is serious, the weight of the inventory sufficiency feature can be further increased to 0.5; in the discount activity scenario, if the promotion intensity is large, the adjustment range of the weight of the promotion attractiveness feature can be appropriately increased, which all fall within the protection scope of this optional implementation manner.

[0133] A product name search hint method provided by this application can dynamically adjust the weight distribution according to different operation scenarios based on obtaining the weights of each supply chain feature, so as to be able to flexibly adapt to diverse business goals.

[0134] As an optional implementation manner, based on any of the above embodiments, sorting the initial candidate product name set according to the成交 score includes the following steps:

[0135] Weightedly fuse the成交 score with the preset basic semantic score to obtain the target sorting score of each candidate product name in the initial candidate product name set.

[0136] Among them, the preset basic semantic score is preset and is an index used to quantitatively evaluate the semantic matching degree between the candidate product name and the user's target input word. Its value range is uniformly mapped to the [0,1] interval. The higher the preset basic semantic score, the higher the semantic matching degree between the candidate product name and the user's input, and the more in line with the user's search intention.

[0137] Optionally, for each candidate product name in the initial candidate product name set, the matching degree is calculated by combining the semantic features and character features of the target input word. A combination of cosine similarity algorithm and edit distance normalization algorithm is used to ensure the comprehensiveness and accuracy of the calculation results. Specifically, the target input word and candidate product names are converted into word vectors respectively, and the cosine value of the two word vectors is calculated, with a value ranging from [0,1]. The closer the value is to 1, the higher the semantic similarity. The edit distance between the target input word and the candidate product name is calculated, which is the minimum number of editing operations required to transform one string into the other, including replacement, insertion, and deletion. Then, the character matching degree score is obtained by normalizing the formula semantic character score = 1 - edit distance / candidate product name character length, with a value ranging from [0,1].

[0138] Optionally, the cosine similarity result and the edit distance score are weighted and averaged (the weights can be adjusted according to the business scenario, such as semantic vector weight 0.6 and edit distance weight 0.4) to finally obtain the basic semantic score of the candidate product name.

[0139] Optionally, if a candidate product name is a newly listed product, has no corresponding semantic vector, or has no historical character matching record, the basic semantic score is preset to a fixed value (such as 0.5) to avoid sorting deviation due to missing data.

[0140] Optionally, based on the actual business scenarios of the e-commerce platform, a transactional score weight w5 and a basic semantic score weight w6 are preset, with a total weight of 1.0. For example, the transactional score weight w5 can range from 0.6 to 0.8, prioritizing the transaction feasibility of recommended products and serving as the primary basis for ranking; the basic semantic score weight w6 can range from 0.2 to 0.4, ensuring the semantic matching degree between recommended products and user search intent, and serving as a secondary basis for ranking.

[0141] Specifically, a weighted summation formula is used to calculate the target ranking score for each candidate product name. The formula is: Target Ranking Score = w5 × Potential for Sale Score + w6 × Basic Semantic Score. The higher the target ranking score, the more likely the candidate product name is to both match the user's search intent and have a high probability of being sold.

[0142] Finally, the candidate product names are sorted in descending order based on the target ranking score to generate a list of target product names.

[0143] Specifically, all candidate product names in the initial candidate product name set are sorted in descending order according to the target sorting score. The higher the score of the candidate product name, the higher its sorting position and the more likely it is to be displayed to the user.

[0144] Optionally, for candidate product names with the same target ranking score, they are further sorted in descending order based on the basic semantic score; if the basic semantic scores are still the same, they are sorted in descending order based on the category popularity feature to ensure the uniqueness and rationality of the ranking results.

[0145] Optionally, according to a preset display quantity (e.g., 3 or 5), the top N candidate product names are selected from the candidate product name list in descending order to form the final target product name list, and the results are displayed in the prompt area of ​​the search box.

[0146] This application provides a product name search suggestion method that generates a target ranking score by weightedly fusing a tradable score with a preset basic semantic score, and then sorts candidate product names in descending order based on this score. This method retains the ability of semantic matching to capture user input intent, while also incorporating multi-dimensional features of the supply chain to quantitatively assess the actual tradability of products, making the search results ranking more closely match the user's real needs.

[0147] As an optional implementation, based on any of the above embodiments, before obtaining the product data associated with each candidate product name in the initial candidate product name set, the following steps are further included:

[0148] Based on a preset time period, product data is retrieved from the target system via an application programming interface.

[0149] The purpose of this step is to pre-fetch and filter invalid product data before obtaining the product data associated with the initial candidate product names. This avoids unnecessary data retrieval, feature extraction, and scoring of products that are delisted, out of stock, or unreachable by logistics. This improves system processing efficiency, reduces resource consumption, and ensures that all subsequently obtained product data meets the basic criteria for sale.

[0150] The preset time period is a pre-set time for data retrieval, which can be flexibly set according to the update frequency of the platform's product data. Optionally, the preset time period can be set to 5-15 minutes. For example, if the time period is set to 10 minutes, the system will automatically trigger an API interface call every 10 minutes to retrieve all product data from the target system, ensuring that the pre-retrieved data is synchronized with the target system.

[0151] The target system refers to the backend system that stores various types of product data. This target system can be an enterprise resource planning system, warehouse management system, marketing system, or user behavior log system, and must be consistent with the system acquiring the product data to ensure the completeness and consistency of the retrieved data dimensions. When retrieving data via API, an interface protocol adapted to the data formats of each target system is used to ensure data retrieval compatibility and stability. Furthermore, the retrieved product data includes basic information and dynamic data of all products within the platform, such as product ID, full product name, inventory data, logistics data, promotional data, historical click data, and product status (listed / delisted), providing comprehensive data support for subsequent filtering steps.

[0152] Secondly, filter the product data and delete product data that is unavailable, has zero inventory, or is not covered by logistics.

[0153] Optionally, the "product status" field in the product data can be read. If the field is marked as "unlisted", "discontinued", "expired" or other non-listed status, all data corresponding to the product can be deleted directly. Such products cannot be searched or ordered by users and do not need to be included in the subsequent data processing.

[0154] Optionally, the "Current Inventory Quantity" field in the product data is read. If the value of this field is 0 and there is no subsequent replenishment plan (which can be determined by the replenishment warning field in the enterprise resource planning system), the product data is deleted. If the inventory is 0 but there is a clear replenishment plan (such as replenishment within 24 hours), the product data is temporarily retained and marked as "Inventory to be replenished", and further verification is performed in conjunction with real-time inventory data.

[0155] Optionally, the "delivery range" field in the product data can be read and combined with the commonly used delivery areas of all users on the platform. If the delivery range of the product does not cover any user's commonly used delivery area, or the delivery range is empty or invalid, the product data will be deleted. If the delivery range only covers some users' delivery areas, the product data will be retained and further matched with the user's actual delivery address later.

[0156] Optionally, after the filtering operation is completed, a valid product data set is generated. This valid product data set only includes product data that is in stock, has an inventory greater than 0 (or has a clear replenishment plan), and whose logistics cover at least part of the user's delivery area. Furthermore, when obtaining product data associated with the initial candidate product names, it is retrieved directly from this valid product data set, without having to call the API interfaces of each target system again to pull the full data, thereby improving data acquisition efficiency.

[0157] This application provides a product name search suggestion method that automatically retrieves product data from the target system through a preset time period and filters out invalid data such as products that are no longer available, have zero inventory, or are not covered by logistics in advance. This ensures that subsequent processing is based only on real-time and valid product information, avoids interference from invalid data on supply chain feature extraction and sorting results, and improves the accuracy and reliability of search suggestions.

[0158] Figure 5 This is a schematic diagram of the structure of a product name search suggestion device provided in an embodiment of this application, as shown below. Figure 5 As shown, the product name search suggestion device provided in this embodiment is located in an electronic device. The product name search suggestion device 50 provided in this embodiment includes: a receiving module 51, a matching module 52, an acquisition module 53, a generation module 54, a scoring module 55, and an output module 56.

[0159] Specifically, the receiving module 51 is used to receive the target input word input by the user; the matching module 52 is used to match the target input word in a preset product name mapping table to generate an initial candidate product name set; the obtaining module 53 is used to obtain the product data associated with each candidate product name in the initial candidate product name set; the generating module 54 is used to extract and standardize the product data according to the supply chain features to generate the supply chain feature vector corresponding to each candidate product name; the scoring module 55 is used to score each candidate product name based on the supply chain feature vector to obtain the transaction score of each candidate product name; and the output module 56 is used to sort the initial candidate product name set according to the transaction score and output and display a preset number of target product names.

[0160] Optionally, product data includes real-time inventory data, estimated delivery time data, promotional activity data, and historical click data.

[0161] Optionally, the generation module 54, when performing supply chain feature extraction and standardization on the product data to generate supply chain feature vectors corresponding to each candidate product name, is specifically used for: determining inventory adequacy features based on real-time inventory data; determining logistics timeliness features based on estimated delivery time data; determining promotional attractiveness features based on promotional activity data; determining category popularity features based on historical click data; and normalizing the inventory adequacy features, logistics timeliness features, promotional attractiveness features, and category popularity features to generate supply chain feature vectors corresponding to each candidate product name.

[0162] Optionally, the generation module 54, when determining the promotional attractiveness features based on promotional activity data, is specifically used to: extract the maximum discount ratio among all products associated with any candidate product name, and determine the product providing the maximum discount ratio as the target promotional product; obtain the remaining promotion time of the target promotional product; adjust the maximum discount ratio according to the remaining promotion time, and determine the adjusted value as the promotional attractiveness feature.

[0163] Optionally, the scoring module 55, when scoring each candidate product name based on the supply chain feature vector to obtain the tradable score of each candidate product name, is specifically used to: obtain the weights of inventory adequacy features, logistics timeliness features, promotional attractiveness features, and category popularity features.

[0164] The products of inventory adequacy feature and its weight, logistics timeliness feature and its weight, promotional attractiveness feature and its weight, and category popularity feature and its weight are summed. The summation result is used as the transaction score for the candidate product name.

[0165] Optionally, the product name search suggestion device provided in this embodiment further includes a weight adjustment module and a filtering module.

[0166] Optionally, after obtaining the weights of inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity, the weight adjustment module is used to: obtain the current operating scenario; increase the weight of inventory adequacy if the current operating scenario is an inventory clearance scenario; increase the weight of logistics timeliness if the current operating scenario is a time-sensitive promotion scenario; and increase the weight of promotional attractiveness if the current operating scenario is a discount activity scenario.

[0167] Optionally, the output module 56, when sorting the initial candidate product name set according to the tradable score, is specifically used to: perform weighted fusion of the tradable score and the preset basic semantic score to obtain the target ranking score of each candidate product name in the initial candidate product name set; and sort each candidate product name in descending order based on the target ranking score to generate a target product name list.

[0168] Optionally, before obtaining the product data associated with each candidate product name in the initial candidate product name set, the filtering module is used to: pull product data from the target system through the application programming interface based on a preset time period; filter the product data and delete product data that is out of stock, has zero inventory, or is not covered by logistics.

[0169] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, as shown below. Figure 6As shown, the electronic device 60 provided in this embodiment includes a processor 61 and a memory 62 communicatively connected to the processor 61.

[0170] The memory 62 stores computer execution instructions; the processor 61 executes the computer execution instructions stored in the memory 62 to implement the method provided in any of the above embodiments.

[0171] The program may include program code, which includes computer-executable instructions. Memory 62 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk storage device.

[0172] In this embodiment, the memory 62 and the processor 61 are connected via a bus. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single straight line, but this does not mean that there is only one bus or one type of bus.

[0173] This application also provides a computer-readable storage medium, including computer-executable instructions stored in the computer-readable storage medium, which, when executed by a processor, are used to implement the method provided in any of the above embodiments.

[0174] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the method provided in any of the above embodiments.

[0175] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to implement the solution of this embodiment according to actual needs.

[0176] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The unit composed of the above modules can be implemented in hardware or in the form of hardware plus software functional units.

[0177] The integrated modules described above, implemented as software functional modules, can be stored in a computer-readable storage medium. These software functional modules, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of the various embodiments of this application.

[0178] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0179] The memory may include high-speed RAM, and may also include non-volatile storage (NVM), such as at least one disk storage device, and may also be a USB flash drive, external hard drive, read-only memory, disk or optical disc, etc.

[0180] The aforementioned storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0181] An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Alternatively, the storage medium can be an integral part of the processor. The processor and storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and storage medium can exist as discrete components in an electronic control unit or main control device.

[0182] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0183] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A product name search suggestion method, characterized in that, include: Receive the target input word from the user; The target input word is matched against a preset product name mapping table to generate an initial set of candidate product names; Obtain the product data associated with each candidate product name in the initial candidate product name set; The commodity data is subjected to supply chain feature extraction and standardization processing to generate a supply chain feature vector corresponding to each candidate product name; Each candidate product name is scored based on the supply chain feature vector to obtain a transaction score for each candidate product name. The initial set of candidate product names is sorted based on the transaction score, and a preset number of target product names are output and displayed.

2. The method according to claim 1, characterized in that, The product data includes real-time inventory data, estimated delivery time data, promotional activity data, and historical click data; The step of extracting and standardizing supply chain features from the product data to generate supply chain feature vectors for each candidate product name includes: Based on the real-time inventory data, inventory adequacy characteristics are determined; based on the estimated delivery time data, logistics timeliness characteristics are determined; based on the promotional activity data, promotional attractiveness characteristics are determined; based on the historical click data, category popularity characteristics are determined. The inventory adequacy feature, logistics timeliness feature, promotional attractiveness feature, and category popularity feature are normalized to generate the supply chain feature vector corresponding to each candidate product name.

3. The method according to claim 2, characterized in that, The determination of promotional attractiveness characteristics based on the promotional activity data includes: Extract the maximum discount percentage from all products associated with any candidate product name, and identify the product that provides the maximum discount percentage as the target promotional product; Obtain the remaining promotional time for the target promotional product; The maximum discount percentage is adjusted by a coefficient based on the remaining time of the promotion, and the adjusted value is determined as the promotion attractiveness feature.

4. The method according to claim 2, characterized in that, The step of scoring each candidate product name based on the supply chain feature vector to obtain a transactional score for each candidate product name includes: Obtain the weights of features such as inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity. The products of the inventory adequacy feature and its weight, the logistics timeliness feature and its weight, the promotional attractiveness feature and its weight, and the category popularity feature and its weight are summed together. The summation result is determined as the transaction score for the candidate product name.

5. The method according to claim 4, characterized in that, After obtaining the weights for inventory adequacy, logistics timeliness, promotional attractiveness, and category popularity, the process also includes: Obtain the current operational scenario; When the current operating scenario is an inventory clearance scenario, increase the weight of the inventory adequacy feature; when the current operating scenario is a time-sensitive promotion scenario, increase the weight of the logistics timeliness feature; when the current operating scenario is a discount activity scenario, increase the weight of the promotion attractiveness feature.

6. The method according to claim 1, characterized in that, The step of sorting the initial candidate product name set according to the tradable score includes: The transaction score is weighted and fused with the preset basic semantic score to obtain the target ranking score of each candidate product name in the initial candidate product name set. Based on the target ranking score, each candidate product name is sorted in descending order to generate a target product name list.

7. The method according to any one of claims 1-6, characterized in that, Before obtaining the product data associated with each candidate product name in the initial candidate product name set, the process also includes: Based on a preset time period, product data is retrieved from the target system via an application programming interface; The product data is filtered to remove products that are unavailable, have zero inventory, or are not covered by logistics.

8. A product name search suggestion device, characterized in that, include: The receiving module is used to receive the target input word from the user. The matching module is used to match the target input word in a preset product name mapping table to generate an initial set of candidate product names; The acquisition module is used to acquire the product data associated with each candidate product name in the initial candidate product name set; The generation module is used to extract and standardize the supply chain features of the commodity data to generate supply chain feature vectors corresponding to each candidate product name. The scoring module is used to score each of the candidate product names based on the supply chain feature vector to obtain the transaction score of each candidate product name. The output module is used to sort the initial candidate product name set based on the transactionable score and output a preset number of target product names.

9. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

11. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1-7.

Images