Commodity category determination method and device, computer equipment, storage medium and program product

Abstract

The invention relates to a commodity category determination method and device, computer equipment, a storage medium and a program product. The method comprises the steps of determining a first similarity between a target semantic vector of a to-be-classified commodity and a first semantic vector corresponding to a category of each commodity in a category system of a target platform under the condition that text information of the to-be-classified commodity is matched with an image; the target semantic vector comprises a second semantic vector corresponding to the text information and a third semantic vector corresponding to the image, and generating an interpretable classification report of the to-be-classified commodity by using a classification decision engine according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report at least comprises the category of the to-be-classified commodity. The probability of category determination errors caused by mismatching of the text information and the images is reduced, and the accuracy of the obtained categories of the to-be-classified commodities is improved.

Description

Technical Field

[0001] This application relates to the field of data analysis technology, and particularly to a method, apparatus, computer device, storage medium, and program product for determining a product category. Background Art

[0002] In terms of product management, product categories provide great convenience for the organizational management of products. In the product category management scenario, generally, an e-commerce platform pre-defines a product category system. When merchants manage products, they need to set the product categories of the products and store them on the e-commerce platform to complete the association between the products and the product categories.

[0003] Currently, keywords are used to match the product categories of products. However, this method has the problem of low accuracy in the matched product categories. Summary of the Invention

[0004] Based on this, in view of the above technical problems, it is necessary to provide a product category determination method, apparatus, computer device, computer-readable storage medium, and computer program product that can improve the accuracy of the determined product category.

[0005] In a first aspect, this application provides a method for determining a product category, including:

[0006] When the text information of the product to be classified matches the image, determining a first similarity between the target semantic vector of the product to be classified and the first semantic vectors corresponding to the categories of each product in the category system of the target platform; the target semantic vector includes a second semantic vector corresponding to the text information and a third semantic vector corresponding to the image;

[0007] Using a classification decision engine, generating an interpretable classification report for the product to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report at least includes the category of the product to be classified.

[0008] In one embodiment, a large language model is used to parse the text information of the product to be classified to obtain a first semantic vector;

[0009] Based on a vision large model, extracting image features in the image of the product to be classified, and using a convolutional network to extract the image features to obtain the attribute information of the product to be classified, and obtaining a third semantic vector based on the attribute information;

[0010] When the second similarity between the first semantic vector and the third semantic vector is greater than a preset similarity, it is determined that the text information of the product to be classified matches the image.

[0011] In one embodiment, the method further includes:

[0012] When the second similarity between the first semantic vector and the third semantic vector is not greater than a preset similarity, a prompt message is output; the prompt message is used to prompt that the text information of the commodity to be classified does not match the image.

[0013] In one embodiment, using a classification decision engine, according to the first similarity between the target semantic vector and each first semantic vector, an interpretable classification report of the commodity to be classified is generated, including:

[0014] Using the classification decision engine, sort the first similarities to obtain a sorting result, and starting from the largest first similarity in the sorting result, select a preset number of first similarities;

[0015] Using the classification decision engine, according to the constraint conditions, Prompt information, and the categories of the commodities corresponding to the preset number of first similarities, generate the interpretable classification report of the commodity to be classified; the Prompt information is determined based on the target semantic vector and the categories of the commodities corresponding to the preset number of first similarities.

[0016] In one embodiment, the method further includes:

[0017] Obtain feedback information on the interpretable classification report of the commodity to be classified, and optimize the classification decision engine based on the feedback information.

[0018] In one embodiment, the method further includes:

[0019] Periodically update the category system corresponding to the target platform with the categories of new commodities to obtain a new category system of the target platform.

[0020] In a second aspect, the present application further provides a commodity category determination device, including:

[0021] A first determination module, configured to determine the first similarity between the target semantic vector of the commodity to be classified and the first semantic vectors corresponding to the categories of each commodity in the category system of the target platform when the text information of the commodity to be classified matches the image; the target semantic vector includes a second semantic vector corresponding to the text information and a third semantic vector corresponding to the image;

[0022] A generation module, configured to use a classification decision engine to generate an interpretable classification report of the commodity to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report at least includes the category of the commodity to be classified.

[0023] In a third aspect, the present application further provides a computer device, including a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0024] In the case where the text information of the commodity to be classified matches the image, determine the first similarity between the target semantic vector of the commodity to be classified and the first semantic vectors corresponding to the categories of each commodity in the category system of the target platform; the target semantic vector includes the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image.

[0025] Using a classification decision engine, generate an interpretable classification report for the commodity to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report includes at least the category of the commodity to be classified.

[0026] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0027] In the case where the text information of the commodity to be classified matches the image, determine the first similarity between the target semantic vector of the commodity to be classified and the first semantic vectors corresponding to the categories of each commodity in the category system of the target platform; the target semantic vector includes the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image.

[0028] Using a classification decision engine, generate an interpretable classification report for the commodity to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report includes at least the category of the commodity to be classified.

[0029] In a fifth aspect, the present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0030] In the case where the text information of the commodity to be classified matches the image, determine the first similarity between the target semantic vector of the commodity to be classified and the first semantic vectors corresponding to the categories of each commodity in the category system of the target platform; the target semantic vector includes the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image.

[0031] Using a classification decision engine, generate an interpretable classification report for the commodity to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report includes at least the category of the commodity to be classified.

[0032] The above-mentioned method, device, computer device, storage medium, and program product for determining a product category determine a first similarity between a target semantic vector of a product to be classified and a first semantic vector corresponding to the category of each product in the category system of a target platform when the text information of the product to be classified matches the image; the target semantic vector includes a second semantic vector corresponding to the text information and a third semantic vector corresponding to the image. Using a classification decision engine, an interpretable classification report for the product to be classified is generated according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report at least includes the category of the product to be classified. Since the first similarity between the target semantic vector and the first semantic vector is further determined only when the text information of the product to be classified matches the image, the probability of incorrect category determination due to non-matching of the text information and the image is reduced, and the first similarity is determined using a target semantic vector that includes a second semantic vector corresponding to the text information and a third semantic vector corresponding to the image, thereby considering the multimodal information of the text information and the image to generate an interpretable classification report, improving the accuracy of the category of the product to be classified obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following will briefly introduce the drawings required for use in the description of the embodiments or the related art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0034] Figure 1 is a flowchart showing a method for determining a product category provided in this embodiment;

[0035] Figure 2 is a flowchart showing a method for matching text information and an image of a product to be classified provided in this embodiment;

[0036] Figure 3 is a flowchart showing a method for generating an interpretable classification report provided in this embodiment;

[0037] Figure 4 is a flowchart showing another method for determining a product category provided in this embodiment;

[0038] Figure 5 is a block diagram showing the structure of a device for determining a product category provided in an embodiment of the present application;

[0039] Figure 6 is an internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] To make the objectives, technical solutions, and advantages of this application clearer, the following further details this application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for explaining this application and are not intended to limit this application.

[0041] In an exemplary embodiment, as Figure 1 shown, Figure 1 is a schematic flowchart of a method for determining a product category provided in this embodiment. Taking the application of this method to a computer device as an example, it includes the following S101 to S102:

[0042] S101. When the text information of the product to be classified matches the image, determine the first similarity between the target semantic vector of the product to be classified and the first semantic vectors corresponding to the categories of each product in the category system of the target platform; the target semantic vector includes the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image.

[0043] Among them, the text information of the product to be classified may include a title and description information. The title of the product refers to the name displayed on the e-commerce platform, which is usually used to concisely and accurately convey the core information of the product. The title may include, for example, the brand, model, and keywords of the product to be classified. The description information may include, for example, color, size, and specification information. The category system of the target platform can be crawled in real time, such as the three-level category tree of a certain e-commerce platform.

[0044] A large language model such as (Generative Pre-trained Transformer 4, GPT-4) or other models capable of parsing text information can be used to parse the text information of the product to be classified to obtain the first semantic vector. A vision large model such as the Contrastive Language-Image Pretraining (CLIP) model or other models capable of extracting image features is used to extract the image features of the image of the product to be classified. A traditional Convolutional Neural Network (CNN) or other network is used to detect the attribute information of the image features, and the third semantic vector is obtained by analyzing the attribute information. Determine the second similarity between the first semantic vector and the third semantic vector. If the second similarity is greater than the preset similarity, it means that the text information matches the image. In this case, the first similarity between the target semantic vector of the product to be classified and each first semantic vector can be determined.

[0045] In this embodiment, since the target semantic vector includes the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image, the target semantic vector includes multimodal information, that is, the text information includes the image information, and the formats of the text information and the image information are unified in the form of the semantic vector, and the information of different modalities is converted into the information of the same modality.

[0046] The target platform may refer to the e-commerce platform where the commodity to be classified needs to be listed. For example, when a user lists a commodity to be classified on a certain e-commerce platform, the category of the commodity to be classified needs to be filled in. Therefore, in this embodiment, the first similarity between the target semantic vector of the commodity to be classified and the first semantic vector corresponding to the category of each commodity in the category system of the target platform is calculated, so that based on the first similarity between the target semantic vector and the first semantic vector corresponding to the category of each commodity in the category system of the target platform, the category corresponding to the higher first similarity can be determined from the first similarities, and an interpretable classification report of the commodity to be classified can be generated based on the category corresponding to the higher first similarity.

[0047] S102. Using a classification decision engine, generate an interpretable classification report for the commodity to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report at least includes the category of the commodity to be classified.

[0048] The categories of the commodity to be classified are, for example, floor washers, brooms, mobile phones, etc., and will not be listed one by one here. The classification decision engine can be used to sort the first similarities to obtain a sorting result, starting from the largest first similarity in the sorting result, select a preset number of first similarities, and determine the category of the commodity to be classified from the categories of the commodities corresponding to the preset number of first similarities.

[0049] The interpretable classification report at least includes the category of the commodity to be classified, and the interpretable report may also include the reasons for determining the category of the commodity to be classified, that is, based on which reasons the category of the commodity to be classified is determined.

[0050] The reasons for determining the category of the commodity to be classified may include text features showing a preset number of keywords affecting classification and their attention scores, image features highlighting the detected brand logo area and key components, conflict records, the similarity between the commodity to be classified in the image and the category of the commodity to be classified, etc. Conflict records are, for example, listing all detected data contradictions (such as the title claiming 'waterproof', but the IP rating is not marked in the parameters). The preset number of keywords is the preset number of keywords selected starting from the most important keyword after sorting the keywords according to their importance.

[0051] The interpretable classification report can be a classification report in natural language form. The interpretable report is, for example:

[0052] "Category of the recommended product to be classified: Smart phone (XX_123)

[0053] Reason:

[0054] 1) The title contains "Brand XX Model", matching the target knowledge graph template;

[0055] 2) "Chip Model = XXX" in the target parameter information belongs to the exclusive configuration of mobile phones;

[0056] 3) The similarity between the image recognized by CLIP and the "mobile phone" category reaches 89.2%."

[0057] In this embodiment, when the text information of the product to be classified matches the image, the first similarity between the target semantic vector of the product to be classified and the first semantic vector corresponding to the category of each product in the category system of the target platform is determined, and the classification decision engine is used to generate an interpretable classification report for the product to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report at least includes the category of the product to be classified. Since the first similarity between the target semantic vector and the first semantic vector is further determined only when the text information of the product to be classified matches the image, the probability of incorrect category determination due to non - matching of text information and image is reduced, and the first similarity is determined by using the target semantic vector including the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image, so as to consider the multi - modal information of text information and image to generate an interpretable classification report, improving the accuracy of the category of the product to be classified obtained.

[0058] In one embodiment, the classification decision engine can be used to sort each first similarity to obtain a sorting result, and starting from the largest first similarity in the sorting result, a preset number of first similarities are selected; the classification decision engine is used to generate an interpretable classification report for the product to be classified according to the constraint conditions and the categories of the products corresponding to the preset number of first similarities. The constraint condition is, for example, that if there is ambiguity, the category with higher sales volume within a certain period of time is preferentially selected as the category of the product to be classified. For example, when the classification decision engine classifies the product to be classified, in the case of being unable to determine which category to use as the category of the product to be classified from the categories of the products corresponding to the preset number of first similarities, the category with higher sales volume within a certain period of time can be determined from the categories corresponding to the preset number of first similarities as the category of the product to be classified. The certain period of time is, for example, one month, or one quarter, etc.

[0059] In an exemplary embodiment, as Figure 2 shown Figure 2It is a schematic flowchart of a method for matching text information and images of a commodity to be classified provided in this embodiment. The method includes the following steps:

[0060] S201, Parse the text information of the commodity to be classified using a large language model to obtain a first semantic vector.

[0061] S202, Extract image features from the image of the commodity to be classified based on a vision large model, and use a convolutional network to extract the image features to obtain the attribute information of the commodity to be classified, and obtain a third semantic vector based on the attribute information.

[0062] S203, When the second similarity between the first semantic vector and the third semantic vector is greater than a preset similarity, determine that the text information and the image of the commodity to be classified match.

[0063] In this embodiment, when the second similarity between the first semantic vector and the third semantic vector is greater than a preset similarity, it means that the text information and the image of the commodity to be classified match. In this case, further determine the first similarity between the target semantic vector of the commodity to be classified and the first semantic vector corresponding to the category of each commodity in the category system of the target platform, so that the finally obtained interpretability report is more accurate.

[0064] In an exemplary embodiment, the method further includes:

[0065] When the similarity between the first semantic vector and the third semantic vector is not greater than a preset similarity, output a prompt message; the prompt message is used to prompt that the text information and the image of the commodity to be classified do not match.

[0066] When the similarity between the first semantic vector and the third semantic vector is not greater than a preset similarity, it means that the text information of the commodity to be classified conflicts with the image of the commodity to be classified, and conflict detection is realized. For example, the title is "Apple mobile phone", but the image of the commodity to be classified is an image of a fruit.

[0067] In this embodiment, by outputting a prompt message for prompting that the text information and the image of the commodity to be classified do not match when the similarity between the first semantic vector and the third semantic vector is not greater than a preset similarity, conflict detection between the text information and the image of the commodity to be classified is realized. In the case of conflicts or contradictions between the text information and the image, manual review can be triggered through the prompt message.

[0068] In an exemplary embodiment, as Figure 3 shown, Figure 3 It is a schematic flowchart of a method for generating an interpretability classification report provided in this embodiment. The above S102 may include the following steps:

[0069] S301. Use the classification decision engine to sort the first similarities to obtain a sorting result, and starting from the largest first similarity in the sorting result, select a preset number of first similarities.

[0070] Since there are many categories of goods in the category system, the classification decision engine may not be able to determine the category of the goods to be classified from a large number of categories. Therefore, the first similarities can be sorted to obtain a sorting result, and starting from the largest first similarity in the sorting result, a preset number of first similarities are selected. The preset number is, for example, 10.

[0071] S302. Use the classification decision engine to generate an interpretable classification report for the goods to be classified according to the constraint conditions, Prompt information, and the categories of the goods corresponding to the preset number of first similarities; the Prompt information is determined based on the target semantic vector and the categories of the goods corresponding to the preset number of first similarities.

[0072] The Prompt information can be determined based on the target semantic vector and the categories of the goods corresponding to the preset number of first similarities by using Prompt engineering. The Prompt information refers to the natural language instructions input by the user to the classification decision engine. The Prompt information may include the target semantic vector of the goods to be classified and the categories of the goods corresponding to the preset number of first similarities, so that the classification decision engine can be prompted to more accurately determine the category of the goods to be classified through the Prompt information.

[0073] In this embodiment, by using the classification decision engine to generate an interpretable classification report for the goods to be classified according to the constraint conditions, Prompt information, and the categories of the goods corresponding to the preset number of first similarities, the accuracy of obtaining the interpretable classification report can be further improved.

[0074] In one embodiment, feedback information on the interpretable classification report for the goods to be classified is obtained, and the classification decision engine is optimized based on the feedback information. If there are problems with the interpretable classification report of the goods to be classified obtained by the user, for example, the category of the goods to be classified in the interpretable report is incorrect, then feedback information indicating that the category of the goods to be classified in the interpretable report is misclassified can be provided. The error cases (such as misclassification) can be added to the training set based on the feedback information to drive the optimization of the classification decision engine and improve the classification accuracy of the classification decision engine. Among them, the feedback information may expose data missing or logical defects in the knowledge graph.

[0075] In one embodiment, the category system corresponding to the target platform can be updated periodically with new product categories, so as to improve the comprehensiveness of the category system of the target platform adopted, and reduce the probability of incorrect determination of the categories of new products due to the lack of new product categories in the category system. The new category system is used to determine the first similarity between the target semantic vector of the subsequent product to be classified and the first semantic vectors corresponding to the categories of each product in the new category system, so as to determine the category of the subsequent product to be classified based on the first similarity. For example, the category system is updated once a week or once a month. After the category system is updated, it becomes richer. A richer category system can improve the accuracy of classification by the classification decision engine, reduce the number of error cases feedback by users in the future, and form a positive cycle of data update → classification decision engine optimization → reduction of feedback information.

[0076] In an exemplary embodiment, as Figure 4 shown, Figure 4 is a schematic flow chart of another method for determining product categories provided in this embodiment. The method may include the following steps:

[0077] S401, Parse the text information of the product to be classified using a large language model to obtain a first semantic vector.

[0078] S402, Extract image features from the image of the product to be classified based on a visual large model, and use a convolutional network to extract the image features to obtain the attribute information of the product to be classified, and obtain a third semantic vector based on the attribute information.

[0079] S403, When the second similarity between the first semantic vector and the third semantic vector is greater than a preset similarity, it is determined that the text information and the image of the product to be classified match.

[0080] S404, When the text information and the image of the product to be classified match, determine the first similarity between the target semantic vector of the product to be classified and the first semantic vectors corresponding to the categories of each product in the category system of the target platform.

[0081] S405, Use the classification decision engine to sort the first similarities to obtain a sorting result, and start from the largest first similarity in the sorting result, and select a preset number of first similarities.

[0082] S406, Use the classification decision engine to generate an interpretable classification report for the product to be classified according to the constraint conditions, Prompt information, and the categories of the products corresponding to the preset number of first similarities.

[0083] It should be understood that although the steps in the flowcharts involved in the above embodiments are sequentially shown according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0084] Based on the same inventive concept, an embodiment of the present application further provides a commodity category determination device for implementing the commodity category determination method involved above. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the commodity category determination device provided below can refer to the limitations on the commodity category determination method in the above text, and will not be repeated here.

[0085] In an exemplary embodiment, as Figure 5 shown, Figure 5 is a structural block diagram of a commodity category determination device provided by an embodiment of the present application. The device 500 includes:

[0086] A first determination module 501, configured to determine a first similarity between a target semantic vector of a commodity to be classified and a first semantic vector corresponding to each commodity category in the category system of a target platform when the text information of the commodity to be classified matches the image; the target semantic vector includes a second semantic vector corresponding to the text information and a third semantic vector corresponding to the image;

[0087] A generation module 502, configured to use a classification decision engine to generate an interpretable classification report for the commodity to be classified according to the first similarity between the target semantic vector and each first semantic vector; the interpretable classification report at least includes the category of the commodity to be classified;

[0088] In an embodiment, the device may further include:

[0089] A second determination module, configured to parse the text information of the commodity to be classified using a large language model to obtain a first semantic vector; extract image features from the image of the commodity to be classified based on a visual large model, and use a convolutional network to extract the image features to obtain attribute information of the commodity to be classified, and obtain a third semantic vector based on the attribute information; determine that the text information of the commodity to be classified matches the image when the similarity between the first semantic vector and the third semantic vector is greater than a preset similarity.

[0090] In one embodiment, the second determination module is further configured to output a prompt message when the similarity between the first semantic vector and the third semantic vector is not greater than a preset similarity; the prompt message is used to prompt that the text information of the commodity to be classified does not match the image.

[0091] In one embodiment, the generation module 502 is specifically configured to use a classification decision engine to sort each first similarity to obtain a sorting result, and start from the largest first similarity in the sorting result to select a preset number of first similarities; use the classification decision engine to generate an interpretable classification report for the commodity to be classified according to the constraint conditions, Prompt information, and the categories of the commodities corresponding to the preset number of first similarities; the Prompt information is determined based on the target semantic vector and the categories of the commodities corresponding to the preset number of first similarities.

[0092] In one embodiment, the apparatus may further include:

[0093] An acquisition module, configured to acquire feedback information on the interpretable classification report for the commodity to be classified, and optimize the classification decision engine based on the feedback information.

[0094] In one embodiment, the apparatus may further include:

[0095] An update module, configured to periodically update the category system corresponding to the target platform with the categories of new commodities to obtain a new category system of the target platform.

[0096] Each module in the above commodity category determination apparatus can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above respective modules.

[0097] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as Figure 6As shown in the figure. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. The computer program, when executed by the processor, implements a method for determining a product category. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the computer device housing, or an external keyboard, touchpad, or mouse, etc.

[0098] Those skilled in the art can understand that Figure 6 the structure shown in the figure is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0099] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the steps of any of the above method embodiments are implemented. The technical principle and technical effect are similar and will not be elaborated here.

[0100] In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, the steps of any of the above method embodiments are implemented. The technical principle and technical effect are similar and will not be elaborated here.

[0101] In an embodiment, a computer program product is provided, including a computer program. When the computer program is executed by the processor, the steps of any of the above method embodiments are implemented. The technical principle and technical effect are similar and will not be elaborated here.

[0102] 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 for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0103] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. 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), magnetoresistive 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 be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0104] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, 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, it should be considered to be within the scope described in this specification.

[0105] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method for determining a commodity category, characterized in that, The method includes: When the text information of the commodity to be classified matches the image, determining the first similarity between the target semantic vector of the commodity to be classified and the first semantic vectors corresponding to the categories of each commodity in the category system of the target platform; the target semantic vector includes the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image; Using a classification decision engine, generating an interpretable classification report for the commodity to be classified according to the first similarities between the target semantic vector and the first semantic vectors; the interpretable classification report at least includes the category of the commodity to be classified.

2. The method according to claim 1, wherein The method further includes: Using a large language model to parse the text information of the commodity to be classified to obtain the first semantic vector; Based on a vision large model, extracting image features in the image of the commodity to be classified, and using a convolutional network to extract the image features to obtain the attribute information of the commodity to be classified, and obtaining the third semantic vector based on the attribute information; When the second similarity between the first semantic vector and the third semantic vector is greater than a preset similarity, determining that the text information of the commodity to be classified matches the image.

3. The method according to claim 2, wherein The method further includes: When the second similarity between the first semantic vector and the third semantic vector is not greater than the preset similarity, outputting a prompt message; the prompt message is used to prompt that the text information of the commodity to be classified does not match the image.

4. The method according to any one of claims 1-3, characterized in that, The step of using a classification decision engine to generate an interpretable classification report for the commodity to be classified according to the first similarities between the target semantic vector and the first semantic vectors includes: Using the classification decision engine to sort the first similarities to obtain a sorting result, and starting from the largest first similarity in the sorting result, selecting a preset number of first similarities; Using the classification decision engine to generate an interpretable classification report for the commodity to be classified according to the constraint conditions, Prompt information, and the categories of the commodities corresponding to the preset number of first similarities; the Prompt information is determined based on the target semantic vector and the categories of the commodities corresponding to the preset number of first similarities.

5. The method according to claim 1, characterized in that, The method further includes: Obtaining feedback information on the interpretable classification report for the commodity to be classified, and optimizing the classification decision engine based on the feedback information.

6. The method according to claim 1, characterized in that, The method further includes: Periodically updating the category system corresponding to the target platform with the categories of new commodities to obtain a new category system of the target platform.

7. A commodity category determination device, characterized in that, The apparatus includes: A first determination module, configured to determine the first similarity between the target semantic vector of the commodity to be classified and the first semantic vectors corresponding to the categories of each commodity in the category system of the target platform when the text information of the commodity to be classified matches the image; the target semantic vector includes the second semantic vector corresponding to the text information and the third semantic vector corresponding to the image; A generation module, configured to use a classification decision engine to generate an interpretable classification report for the commodity to be classified according to the first similarities between the target semantic vector and the first semantic vectors; the interpretable classification report at least includes the category of the commodity to be classified.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, When this computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 6.

Images