Commodity category prediction method and device, medium and product
By acquiring and matching the source platform category name path of the target product, and combining the category prediction models of the same platform, related platforms and the target platform, the problem of low efficiency and accuracy of product category prediction in e-commerce multi-platform is solved, and more efficient category name path determination is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-17
- Publication Date
- 2026-03-27
AI Technical Summary
In the multi-platform e-commerce operation environment, the accurate classification of product categories is inefficient and inaccurate, especially when crossing platforms. Existing technologies have low accuracy when predicting product categories based on information such as product names, descriptions, and images.
By obtaining the category name paths of the source platforms where the target product has been listed, paths that meet the predetermined requirements in terms of similarity to the category name paths of the target platform are selected as the category name paths of the target product on the target platform. These paths are then matched with category prediction models from the same platform, related platforms, and the target platform itself.
It improves the efficiency and accuracy of product category prediction, reduces manual intervention, and enhances the efficiency of product listing and review, making it suitable for large-scale e-commerce scenarios.
Smart Images
Figure CN121745992A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and in particular to a method, device, medium and product for predicting commodity categories. Background Technology
[0002] In a multi-platform e-commerce environment, accurate product category classification is fundamental for product listing, traffic allocation, search recommendations, and data analysis. Therefore, product category prediction is crucial. Currently, category prediction is achieved either manually or based on product names, descriptions, and images. However, manual category selection is not only inefficient and prone to errors, but its problems become even more pronounced in multi-platform e-commerce environments where product category systems differ across platforms. When predicting categories based on product names, descriptions, and images, the information provided is limited, making predictions prone to inaccuracies and resulting in low accuracy. Summary of the Invention
[0003] The purpose of this application is to provide a method, device, medium, and product for predicting commodity categories, in order to solve the problems of low efficiency and accuracy in predicting commodity categories as described in the background art.
[0004] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a method for predicting product categories, including: Receive a category prediction instruction, the category prediction instruction carrying the product identifier of the target product, the target product being the product to be listed on the target platform; Based on the product identifier, the category name path used by the target product when it was listed on the source platform is obtained and denoted as the source platform category name path. The source platform is the platform on which the target product was listed before it was listed on the target platform. From the target category name paths included in the target platform, determine the target category name paths that have a predetermined similarity to the source platform category name paths, and denot them as the matching category name paths; The matching category name path is determined as the category name path of the target product when it is listed on the target platform.
[0005] Optionally, determining the target category name path from the target category name paths included in the target platform that has a predetermined similarity to the source platform category name path, denoted as the matched category name path, includes: Determine whether each target category name path includes a target category name path that is the same as the source platform category name path; If included, the target category name path that is the same as the source platform category name path is determined as the matching category name path; If not included, split each of the source platform category name paths to obtain an array containing multiple category name nodes for each source platform category name path; For each category name node in each array, check whether the category name node is included in each target category name path; The target category name path containing the most category name nodes is determined as the matching category name path.
[0006] Optionally, before obtaining the category name path used by the target product when it was listed on the source platform based on the product identifier, denoted as the source platform category name path, the method further includes: Determine whether there exists a platform in the source platform that is identical to the target platform, and denote it as the same platform; If the same platform exists, the category name path of the target product on the source platform of the same platform is determined as the matching category name path of the target product.
[0007] Optionally, after determining whether there exists a platform in the source platform that is identical to the target platform, denoted as the same platform, the method further includes: If no platform exists, determine whether there is a platform in the source platform that is associated with the target platform, and denot it as an associated platform. The associated platform is a platform that uses the same category name path. If an associated platform exists, the source platform category name path of the target product on the associated platform is determined as the matching category name path of the target product.
[0008] Optionally, before determining whether there exists a platform in the source platform that is identical to the target platform, denoted as the same platform, the method further includes: Determine whether the target platform allows the use of external category prediction methods to predict the matching category name path of the target product, wherein the external category prediction method is a category prediction method that is not the target platform's own; If not allowed, then obtain the target parameters, which are the parameters required when calling the target platform's own category prediction model; The target parameters are sent to the category prediction model so that the category prediction model outputs the matching category name path of the target product.
[0009] Optionally, after determining the target category name path from the target category name paths included in the target platform that has a predetermined similarity to the source platform category name path, and denoting it as the matched category name path, the method further includes: When no target category name path with a similarity meeting the predetermined requirement is determined from the target platform, target parameters are obtained. The target parameters are the parameters required when calling the target platform's own category prediction model. The target parameters include the basic information of the target product. The basic information includes product information and category name path, which are respectively denoted as basic product information and basic category name path. If the basic category name path is not obtained, information including the category name is obtained from the product publishing information mapping table of the source platform and recorded as source category name related information; Extract the category name from the source category name information; concatenate the extracted category name to obtain the temporary category name path of the target product; Based on the basic product information and the temporary category name path, the target platform's category prediction model is invoked so that the category prediction model outputs the matching category name path of the target product.
[0010] Optionally, after invoking the category prediction model of the target platform based on the basic product information and the temporary category name path, so that the category prediction model outputs the matching category name path of the target product, the method further includes: When the category prediction model does not output a valid matching category name path, the basic product information and the temporary category name path are used as search keywords to search for category name paths that are the same as or have a similarity to the basic product information and the temporary category name path in the pre-edited category name path document, and the matching category name path is obtained.
[0011] In a second aspect, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described in any one of the first aspects above.
[0012] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any one of the first aspects above.
[0013] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any one of the first aspects above.
[0014] According to the specific embodiments provided in this application, the following technical effects are disclosed: The product category prediction method provided in this application first obtains the category name path of the source platform when the target product was listed; then, it filters out the target category name paths that meet the predetermined requirements in similarity to the source platform category name paths, and records them as matching category name paths; and determines the filtered matching category name paths as the category name paths of the target product when it was listed on the target platform. Compared with the prior art, category prediction is performed by matching product names, which provides limited information and low accuracy. Therefore, compared with the prior art, this application improves the efficiency and accuracy of target product category prediction. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a diagram illustrating the application environment of the commodity category prediction method in one embodiment of this application; Figure 2 A flowchart illustrating a product category prediction method provided in an embodiment of this application; Figure 3 A flowchart illustrating a method for determining the matching category name path provided in an embodiment of this application; Figure 4 A flowchart illustrating yet another product category prediction method provided in another embodiment of this application; Figure 5 A flowchart illustrating another product category prediction method provided in an embodiment of this application; Figure 6 A flowchart illustrating another product category prediction method provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation
[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0018] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the contents of this application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0019] This application provides an environment in which the product category prediction method can be applied. See [link to relevant documentation]. Figure 1 The application environment includes terminals and servers.
[0020] The data storage system stores the data that the server needs to process. This system can be set up independently, integrated into the server, or located in the cloud or on other servers. It stores the relevant data required for executing the product category prediction method. For example, the system may contain a product pool containing complete information about all products. This pool can include products that are currently scheduled to be listed on the target platform and those that are not. The information includes basic information, further subdivided into product information (referred to as basic product information) and category information (referred to as basic category information). It can also store a publishing information table, which records all information corresponding to products that have been listed on e-commerce platforms, such as relevant information about the source platform and its category information. Furthermore, the system also stores intermediate data generated during the category prediction process, allowing for timely retrieval when needed.
[0021] In this application, the basic information is collected by the developer and is independent of the product information on the source platform and also independent of the product information on the target platform.
[0022] In this system, the terminal communicates with the server via a network. The terminal can receive the product identifier of the target product whose category to be predicted, input by the user. This product identifier is then sent to the server. Upon receiving the product identifier, the server executes a product category prediction method based on it to obtain a prediction result. The server can then feed back the prediction result to the terminal. Furthermore, in some embodiments, the product category prediction method can be implemented independently by the server or the terminal. For example, the terminal can directly perform category prediction based on the target product's product identifier, or the server can perform category prediction based on data from the data storage system and the target product's product identifier.
[0023] The terminals can be, but are not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. Servers can be implemented using independent servers, server clusters composed of multiple servers, or cloud servers.
[0024] In one exemplary embodiment, see Figure 2 As shown, a product category prediction method is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking the server in the example, the explanation includes the following steps 100 to 103: Step 100: Receive a category prediction instruction, which carries the product identifier of the target product, and the target product is the product to be listed on the target platform; The category prediction instruction can be sent by the user to the server through the user interface, or it can be an instruction generated by other intermediate processes.
[0025] The target platform is the platform where the target product will be listed.
[0026] In addition, in this application, category prediction refers to the prediction of category name path.
[0027] Step 101: Based on the product identifier, obtain the category name path used by the target product when it was listed on the source platform, and record it as the source platform category name path. The source platform is the platform on which the target product was listed before it was listed on the target platform. Furthermore, the server stores a release information table, which records all historical listing information for all products in the product pool, including product ID, source platform ID, source platform category name path, etc.
[0028] Furthermore, for products to be listed on the target platform, the historical listing information of the target product is retrieved from the publishing information table based on the product ID, and then summarized together with the product's basic information into the source platform's product publishing information mapping table (source Goods Publishing Bo Map) so that it can be retrieved in a timely manner when needed, without having to retrieve it from the product pool and publishing information table again, thus improving the retrieval speed.
[0029] In addition, in this application, "uploading" and "publishing" have the same meaning.
[0030] There can be multiple source platforms. Each time a product is listed on a source platform, it is assigned a category name path. Therefore, when a target product has been listed on multiple source platforms, it will have multiple source platform category name paths. Thus, the source platform category name path here can be one or multiple, consistent with the number of source platforms on which the target product has been listed.
[0031] Furthermore, in addition to the category information such as the source platform category name and path, the target product also includes other category information, such as the source platform category ID and the source platform category ID path.
[0032] Furthermore, the category ID path represents a range of products from the initial level category ID to the final level category ID, decreasing in scope. For example, the category ID path for category ID 50012040 is 0 / 50024099 / 50012040. Furthermore, the category name path includes multi-level category names, and the range of products that the category name can include gradually decreases from the initial level category name to the final level category name. Therefore, the category name path describes a process of gradually finding the final level category name from a broad to a narrow scope. For example, the category name path for the final level category name "Smartphone" is: Digital / Mobile Communication / Smartphone.
[0033] Step 102: Determine the target category name path that meets the predetermined requirement of similarity with the category name path of the source platform from the target category name paths included in the target platform, and record it as the matching category name path; The target platform also has its own category information table, which records the category information of each product on the target platform, such as the category name path, which is described in this application as the target category name path.
[0034] Furthermore, achieving a predetermined similarity requirement can be defined as a similarity greater than a predetermined value.
[0035] The implementation process of step 102 can be as follows: compare each source platform category name path with each target category name path included in the target platform in turn, so as to determine the target category name path that meets the predetermined similarity requirement from each target category name path.
[0036] When a target product has multiple source platforms, there will be multiple source platform category name paths. In this case, each target category name path will still be compared with each source platform category name path. For example, when there are three source platforms, there will be three source platform category name paths, and each target category name path will be compared with these three source platform category name paths.
[0037] Step 103: Determine the matching category name path as the category name path of the target product when it is listed on the target platform.
[0038] The product category prediction method provided in this application first obtains the category name path of the source platform when the target product was listed; then, it filters out the target category name paths that meet the predetermined requirements in similarity to the source platform category name paths, and records them as matching category name paths; and determines the filtered matching category name paths as the category name paths of the target product when it was listed on the target platform. Compared with the prior art, category prediction is performed by matching product names, which provides limited information and low accuracy. Therefore, compared with the prior art, this application improves the accuracy of target product category prediction.
[0039] In addition, the server contains a product pool that stores complete information about all products. This pool includes target products to be listed on the target platform at the current time and products not currently listed. All information includes basic product information, which, for ease of description, is divided into product information and category information. Furthermore, for convenience, product information can also be referred to as basic product information, and category information as basic category information. Basic product information includes product name, price, specifications, etc. Basic category information includes basic category ID, basic category name path, etc. This basic information stored on the server can be collected and uploaded to the server by developers, or it can be directly retrieved from other storage locations.
[0040] Information about the target product can be converted into a format suitable for uploading, so that it can be retrieved in a timely manner when uploading.
[0041] Optionally, see Figure 3 In another exemplary embodiment of this application, step 102 includes steps 201 to 205: Step 201: Determine whether each target category name path includes a target category name path that is the same as the source platform category name path; The source platform category name path can be obtained by querying the source platform's product publishing information mapping table using the product ID. For example, assuming the target product's product ID is A, the source platform ID retrieved from the source platform's product publishing information mapping table using this product ID is 1. Assuming the source platform is Taobao (Platform Id=1), the category ID is 50012040 (representing the "Smartphone" category), and the source platform category name path is "Digital / Mobile Communication / Smartphone", the category ID path "0 / 50024099 / 50012040" can also be obtained.
[0042] Step 202: If included, determine the target category name path that is the same as the source platform category name path as the matching category name path; In other words, the similarity of the target category name path with the same category name path as the source platform meets the predetermined requirements.
[0043] Step 203: If not included, split each of the source platform category name paths to obtain an array containing multiple category name nodes for each of the source platform category name paths; For example, if the category name path of a certain source platform is "Digital / Mobile Communication / Smartphone", the array of the split category name nodes is ["Digital", "Mobile Communication", "Smartphone"].
[0044] Step 204: For each category name node in each array, check whether the target category name path includes that category name node; Step 205: Determine the target category name path containing the most category name nodes as the matching category name path.
[0045] During the processing, a predetermined score can be added to the target category name path for each matched category name node. The higher the score, the greater the similarity. Finally, the target category name path with the highest score is determined as the target category name path with the predetermined similarity requirement, i.e., the matched category name path. The predetermined score can be a human-made empirical value, such as 1 point.
[0046] When there are multiple source platforms, each target category name path will have multiple scores, but regardless of how many scores there are, the target category name path with the highest score will still be used as the matching category name path.
[0047] In addition, special handling is provided for specific platforms. For example, if the source platform is Tmall and the target platform is Taobao, queries can be performed using the category name and the Taobao category ID prefix (first-level category). For platforms with specific prefix patterns, such as Taobao, where category IDs have specific prefix patterns (e.g., those starting with "500" are usually 3C digital products), filtering by prefix can significantly narrow down the search scope and improve query efficiency and accuracy.
[0048] Optionally, see Figure 4 In another exemplary embodiment of this application, before step 101, the method further includes steps 301 to 304: Step 301: Determine whether there is a platform in the source platform that is the same as the target platform, and denot it as the same platform; The scenario in step 301 could be: the target product was previously listed on the same platform, but was removed from the platform for some reason, and now needs to be relisted on the same platform.
[0049] Step 302: If a product exists on the same platform, determine the category name path of the target product on the source platform of the same platform as the matching category name path of the target product. Reusing the category name path from the same platform can not only improve the speed of determining the matching category name path, but also improve the accuracy.
[0050] Step 303: If no platform exists, determine whether there is a platform in the source platform that is associated with the target platform, and denot it as an associated platform. The associated platform is a platform that uses the same category name path. Furthermore, for two related platforms, they can share a single category name path information table, which records the category name paths of all products on both platforms.
[0051] Step 304: If an associated platform exists, determine the source platform category name path of the target product on the associated platform as the matching category name path of the target product; For example, if the source platform of the target product is Tmall and the target platform is Taobao, since Tmall and Taobao are highly related platforms, the category name path of Tmall can be directly used as the matching category name path.
[0052] Mapping the source platform category name path of the associated platform can not only improve the speed of determining the matching category name path, but also improve the accuracy.
[0053] If no associated platform exists, proceed to step 101.
[0054] Optionally, see Figure 5 In another exemplary embodiment of this application, before step 301, the method further includes steps 401 to 403: Step 401: Determine whether the target platform allows the use of external category prediction methods to predict the matching category name path of the target product. The external category prediction method is a category prediction method that is not the target platform's own. Step 402: If not allowed, obtain the target parameters, which are the parameters required when calling the target platform's own category prediction model; Optionally, the target parameters may vary depending on the target platform. For example, the target parameters of a certain platform are the basic information of the target product, which includes product information and category name path, referred to as basic product information and basic category name path, respectively. Basic product information includes product name, product details, and product images.
[0055] The basic information consists of all the basic information about the target product stored on the server, such as name, price, and specifications.
[0056] Furthermore, the obtained basic category name paths need to be converted into a format that the category prediction model can recognize. For example, if the basic category name path is "Digital > Mobile Phone > Smartphone", the separator in the path needs to be unified to a forward slash (` / `), and the converted path will be "Digital / Mobile Phone / Smartphone".
[0057] Step 403: Send the target parameters to the category prediction model so that the category prediction model outputs the matching category name path of the target product.
[0058] The category prediction model of the target platform is the category prediction method of the target platform itself. This application does not focus on the category prediction method of the target platform, but only on the parameters that need to be input into the model when calling the category prediction model, that is, the target parameters.
[0059] Of course, the category prediction model may not always return a valid matching category name path. When a valid matching category name path is returned, the prediction process ends; otherwise, step 301 is executed.
[0060] Of course, if the target platform allows the use of external category prediction methods, steps 401 to 403 can be skipped, and step 301 can be executed directly.
[0061] Optionally, see Figure 6 In another exemplary embodiment of this application, after step 102, the method further includes steps 501 to 504: Step 501: When no target category name path with a similarity meeting the predetermined requirement is determined from the target platform, target parameters are obtained. The target parameters are the parameters required when calling the target platform's own category prediction model. The target parameters include basic information of the target product. The basic information includes product information and category name path, which are respectively denoted as basic product information and basic category name path. The basic product information includes the product name, product details, and product images.
[0062] The target parameters can be obtained from the product release information mapping table of the source platform.
[0063] Step 502: If the basic category name path is not obtained, obtain information including the category name from the product publishing information mapping table of the source platform, and record it as source category name related information; Furthermore, relevant information used when listing on the source platform can be collected from the product release information mapping table of the source platform. In addition, as much category information as possible can be obtained, especially information including the category name and the path of the category name on the source platform.
[0064] Step 503: Extract the category name from the source category name information; concatenate the extracted category name to obtain the temporary category name path of the target product; For example, the obtained category names are "Digital" and "Mobile Phone", and the combined temporary category name path is "Digital / Mobile Phone".
[0065] The concatenation process can be performed manually, by inputting the concatenated category name path into the server through the user interface.
[0066] Step 504: Based on the basic product information and the temporary category name path, call the category prediction model of the target platform so that the category prediction model outputs the matching category name path of the target product.
[0067] In step 504, the target parameters sent to the category prediction model are the basic product information and the temporary category name path.
[0068] When the basic category name path is obtained, step 502 is not required. The basic information and basic category name path are directly sent to the category prediction model to output the matching category name path of the target product.
[0069] Optionally, in another exemplary embodiment of this application, after step 504, the method further includes step 601: Step 601: When the category prediction model does not output a valid matching category name path, the basic product information and the temporary category name path are used as search keywords to search for category name paths that are the same as or have a similarity to the basic product information and the temporary category name path in the pre-edited category name path document, so as to obtain the matching category name path.
[0070] Among them, the category name path document is a pre-edited document that records all the category name paths that can be collected.
[0071] The method described in step 601 can be implemented using Elasticsearch (ES), an open-source distributed search engine and analytics engine built on Apache Lucene. It is written in Java and provides full-text search, real-time data analysis, and other functionalities through a RESTful API, while hiding the complexity of Lucene.
[0072] Furthermore, step 601 can be implemented by any of the following different methods: 1) Category Name Path Matching Method Objective: To search based on complete category name paths (e.g., "home appliances / large appliances / refrigerators"), prioritizing fully matching category name paths, and then returning partially matching category name paths.
[0073] logic: First, perform exact phrase matching and assign it a high weight (boost=3), so that the category name path that matches exactly will be ranked first in the results.
[0074] Next, tokenization matching is performed. ES will tokenize the input string (e.g., into "home appliances", "large appliances", "refrigerator") and then search for category name paths containing these words. This provides a wider scope but lower precision, hence a lower weight (boost=1).
[0075] 2) Product Name Matching Method Objective: To infer the possible category of a product from its name.
[0076] Logic: The user enters a product name (such as "iPhone 15 Pro Max") through the user interface, and the server uses this name to match the category path.
[0077] Use a word segmenter to segment the product name into different words. For example, you can specify the Ik_smart mode of the IK word segmenter. This mode will most likely segment "iPhone 15 Pro Max" into ["iPhone", "15", "pro", "max"], and then search for categories containing these words in the category name path (for example, "Mobile Communications > Mobile Phones" may be matched because it contains "Mobile Phones").
[0078] Of course, this is an expanded search with lower weight.
[0079] 3) Exact matching method for last-level category names Objective: To search directly based on the name of the lowest-level category (e.g., searching for "refrigerator").
[0080] Logic: Use the last-level category name field for word segmentation and matching. This method is the most direct way to search for categories, so it is given a high weight (boost=3) to ensure that the results appear at the top.
[0081] 4) Target platform site filtering method Objective: To filter sites targeting specific cross-border platforms (e.g., eBay, Lazada).
[0082] Logic: Platforms like eBay and Lazada have different sites in different countries / regions (e.g., eBay US, eBay UK), and each site may have different categories. This condition mandates that the query results must strictly match the specified siteCode. This is a filtering condition.
[0083] 5) Special handling method for target platform Purpose: To provide special handling for certain platforms, such as those where categories may be assigned to a specific scope. For example, the Xiaohongshu platform.
[0084] Logic: Because categories on this special platform may be assigned within a specific ID range (e.g., categoryId>=20000), a range query is used to ensure that only categories within this range are retrieved.
[0085] 6) Platform prefix filtering method Objective: To ensure that all query results belong to the specified platform.
[0086] Logic: In order to distinguish different platforms, each category will have a prefix field that identifies the platform.
[0087] 7) Authorized Store Filtering Method Objective: To filter categories based on authorization information. There are two modes: Platform Mode (authId=0): Queries the standard category library provided by the platform. Because standard categories are shared and may be duplicated, the collapse function is used to remove duplicates based on categoryId, ensuring that only one instance of each standard category is returned.
[0088] Shop mode (authId≠0): Queries the category of a specific authorized shop. Different shops may have custom categories; here, we directly match precisely using the shopId field.
[0089] 8) Node ID prefix filtering method Objective: To filter based on the ID path of the category path.
[0090] Logic: In addition to the name path (nodeNamePath), categories usually also have a path composed of IDs (nodeIdPath), such as "10001>10022>10035".
[0091] The input is a list of prefixes (e.g., ["10001>10022"]), and prefixQuery is used to match all categories that begin with these ID sequences.
[0092] This is often used to quickly locate all subcategories of a category. For example, passing in "10001>10022" will retrieve all third- and fourth-level categories that belong to the second-level category "10022".
[0093] Using filters (hierarchical filtering) does not contribute to the scoring.
[0094] In addition, when a matching category name path is predicted using the above-mentioned category prediction methods, it is possible to verify whether this matching category name path actually exists on the target platform and obtain its complete information. If the verification is successful, detailed information about the matching category name path is returned; if the verification fails (i.e., the matching category name path does not exist on the target platform), a log is logged, and other prediction methods will be tried or an empty result will be returned.
[0095] If any of the aforementioned prediction steps encounters an exception, it will be captured and detailed error information will be logged. Regardless of whether an exception occurs, as long as the prediction fails (does not return successfully in the previous steps), execution will eventually reach this point, where a failure log will be logged and null will be returned. This ensures that the method always has a return value, guaranteeing the robustness of the category prediction function. Even in the event of unexpected errors, the system can continue to run and provide useful error information.
[0096] This application integrates multiple category name path prediction methods, including direct reuse within the same platform, mapping to related platforms, the target platform's own category prediction model, ES full-text search category prediction method, and category path matching, thereby improving the success rate of matching category name path prediction for target products.
[0097] In addition, this application also supports multi-target platform differentiation processing to avoid mismatches caused by differences in platform category systems.
[0098] This application also reduces manual intervention and improves the efficiency of product listing and review; it supports batch product processing and is suitable for large-scale e-commerce scenarios.
[0099] This application can be used to predict the matching category name path for different target platforms, such as supporting multiple mainstream e-commerce platforms such as Douyin Store, Tmall, Taobao, Pinduoduo, Xiaohongshu, 1688, and Youzan; the prediction strategy can be dynamically adjusted according to the characteristics of the platform.
[0100] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram can be found in [reference needed]. Figure 7As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data related to a product category prediction method. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When the computer program is executed by the processor, it can implement a product category prediction method.
[0101] Those skilled in the art will understand, see Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0102] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.
[0103] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0104] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0105] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data that have been agreed to by the user or have been fully agreed to by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0106] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. In the embodiments provided in this application, any reference to memory, database, or other media can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).
[0107] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units, etc., and are not limited to these.
[0108] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0109] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for predicting product categories, characterized in that, include: Receive a category prediction instruction, the category prediction instruction carrying the product identifier of the target product, the target product being the product to be listed on the target platform; Based on the product identifier, the category name path used by the target product when it was listed on the source platform is obtained and denoted as the source platform category name path. The source platform is the platform on which the target product was listed before it was listed on the target platform. From the target category name paths included in the target platform, determine the target category name paths that have a predetermined similarity to the source platform category name paths, and denot them as the matching category name paths; The matching category name path is determined as the category name path of the target product when it is listed on the target platform.
2. The product category prediction method according to claim 1, characterized in that, The step of determining the target category name path that meets the predetermined similarity requirement with the category name path of the source platform from the target category name paths included in the target platform, denoted as the matching category name path, includes: Determine whether each target category name path includes a target category name path that is the same as the source platform category name path; If included, the target category name path that is the same as the source platform category name path is determined as the matching category name path; If not included, split each of the source platform category name paths to obtain an array containing multiple category name nodes for each source platform category name path; For each category name node in each array, check whether the category name node is included in each target category name path; The target category name path containing the most category name nodes is determined as the matching category name path.
3. The product category prediction method according to claim 1, characterized in that, Before obtaining the category name path used by the target product when it was listed on the source platform based on the product identifier, denoted as the source platform category name path, the method further includes: Determine whether there exists a platform in the source platform that is identical to the target platform, and denote it as the same platform; If the same platform exists, the category name path of the target product on the source platform of the same platform is determined as the matching category name path of the target product.
4. The product category prediction method according to claim 3, characterized in that, After determining whether there exists a platform in the source platform that is identical to the target platform, denoted as the same platform, the method further includes: If no platform exists, determine whether there is a platform in the source platform that is associated with the target platform, and denot it as an associated platform. The associated platform is a platform that uses the same category name path. If an associated platform exists, the source platform category name path of the target product on the associated platform is determined as the matching category name path of the target product.
5. The product category prediction method according to claim 3, characterized in that, Before determining whether there exists a platform in the source platform that is identical to the target platform, denoted as the same platform, the method further includes: Determine whether the target platform allows the use of external category prediction methods to predict the matching category name path of the target product, wherein the external category prediction method is a category prediction method that is not the target platform's own; If not allowed, then obtain the target parameters, which are the parameters required when calling the target platform's own category prediction model; The target parameters are sent to the category prediction model so that the category prediction model outputs the matching category name path of the target product.
6. The product category prediction method according to claim 1, characterized in that, After determining, from the target category name paths included in the target platform, a target category name path that meets a predetermined similarity requirement with the source platform category name path, and denoted as the matched category name path, the method further includes: When no target category name path with a similarity meeting the predetermined requirement is determined from the target platform, target parameters are obtained. The target parameters are the parameters required when calling the target platform's own category prediction model. The target parameters include the basic information of the target product. The basic information includes product information and category name path, which are respectively denoted as basic product information and basic category name path. If the basic category name path is not obtained, information including the category name is obtained from the product publishing information mapping table of the source platform and recorded as source category name related information; Extract the category name from the source category name information; concatenate the extracted category name to obtain the temporary category name path of the target product; Based on the basic product information and the temporary category name path, the target platform's category prediction model is invoked so that the category prediction model outputs the matching category name path of the target product.
7. The product category prediction method according to claim 6, characterized in that, After invoking the category prediction model of the target platform based on the basic product information and the temporary category name path, so that the category prediction model outputs the matching category name path of the target product, the method further includes: When the category prediction model does not output a valid matching category name path, the basic product information and the temporary category name path are used as search keywords to search for category name paths that are the same as or have a similarity to the basic product information and the temporary category name path in the pre-edited category name path document, and the matching category name path is obtained.
8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the commodity category prediction method according to any one of claims 1-7.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the commodity category prediction method according to any one of claims 1-7.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the commodity category prediction method according to any one of claims 1-7.