Information generation method and system, information release method and system, equipment, storage medium and program product

Abstract

The embodiment of the invention provides an information generation method and system, an information release method and system, equipment, a storage medium and a program product. The method comprises the following steps that object information of a first object is obtained, the first object is a published object on a first website, and the object information of the first object comes from the first website; according to the object information of the first object, screening a similar object similar to the first object from second objects published by a second website; according to the category to which the similar object belongs in the second website, determining a publishing category to which the first object belongs in the second website; and generating release information required for releasing the first object on the second website according to the release category and the object information of the first object. According to the technical scheme provided by the embodiment of the invention, the automation degree of information release can be improved, and the labor cost of information release can be greatly reduced.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information generation and publishing method, system, device, storage medium and program product. Background Art

[0002] With the continuous development of Internet technology, more and more merchants choose to sell goods on e-commerce websites (or e-commerce platforms). If merchants want to sell goods on e-commerce websites, they need to publish the goods on the e-commerce websites first. Product publishing is an important part of achieving online sales.

[0003] Currently, more and more merchants choose to open stores on multiple different websites to broaden sales channels and reach more consumers. While this strategy increases sales opportunities, it also brings operational challenges.

[0004] Although most of the products sold by merchants on different websites are the same, due to the significant differences in product delivery between the websites, merchants have to repeat multiple release operations (for example, multiple category selection operations), which incurs extremely high product delivery costs. Repeated operations not only consume merchants' time and energy, but also increase the risk of errors (for example, selecting the wrong category), greatly affecting merchants' operational efficiency. Summary of the invention

[0005] In view of the above problems, the present application is proposed to provide an information generation and publishing method, system, device, storage medium and program product that solve the above problems or at least partially solve the above problems.

[0006] In a first aspect of the present application, there is provided an information generation method, comprising:

[0007] Acquire object information of a first object, where the first object is a published object on a first website, and the object information of the first object comes from the first website;

[0008] According to the object information of the first object, selecting similar objects similar to the first object from second objects published on the second website;

[0009] Determining, according to the category to which the similar object belongs on the second website, the publishing category to which the first object belongs on the second website;

[0010] Publishing information required for publishing the first object on the second website is generated according to the publishing category and the object information of the first object.

[0011] The second aspect of the present application provides an information publishing method, comprising:

[0012] Acquire object information of a first object, where the first object is a published object on a first website, and the object information of the first object comes from the first website;

[0013] According to the object information of the first object, selecting similar objects similar to the first object from second objects published on the second website;

[0014] Determining, according to the category to which the similar object belongs on the second website, the publishing category to which the first object belongs on the second website;

[0015] generating, according to the publishing category and the object information of the first object, publishing information required for publishing the first object on the second website;

[0016] Based on the publishing information, the first object is published on the second website.

[0017] A third aspect of the present application provides a method for determining an information category, comprising:

[0018] Acquire object information of a first object, where the first object is a published object on a first website, and the object information of the first object comes from the first website;

[0019] According to the object information of the first object, selecting similar objects similar to the first object from second objects published on the second website;

[0020] According to the category to which the similar object belongs on the second website, the publishing category to which the first object belongs on the second website is determined.

[0021] A fourth aspect of the present application provides an information display method, comprising:

[0022] Display a first input box and a second input box;

[0023] Receive a target URL of the first object on the first website input in the first input box and identity information of the user registered on the first website input in the second input box;

[0024] When the first object belongs to the user, displaying publishing information required for publishing the first object on the second website, the publishing information being generated based on object information of the first object, the object information of the first object being determined based on web page data of a target web page, the target web page being a web page identified by the target URL;

[0025] When the first object does not belong to the user, first prompt information is displayed, where the first prompt information is used to prompt that the first object does not belong to the user.

[0026] A fifth aspect of the present application provides an information processing system, including:

[0027] The terminal device is used to send an information generation request to the server device, wherein the information generation request is used to request generation of publishing information required for publishing a first object on a second website, wherein the first object is a published object on the first website;

[0028] The server-side device is used to obtain object information of a first object in response to the information generation request, where the object information of the first object comes from the first website; based on the object information of the first object, filter out similar objects similar to the first object from second objects published on the second website; and determine the publishing category to which the first object belongs on the second website based on the category to which the similar objects belong on the second website.

[0029] In a sixth aspect of the present application, an electronic device is provided. The electronic device comprises: a memory and a processor, wherein:

[0030] The memory is used to store programs;

[0031] The processor is coupled to the memory, and is used to execute the program stored in the memory to implement any of the methods described above.

[0032] In a seventh aspect of the present application, a computer-readable storage medium storing a computer program is provided, and when the computer program is executed by a computer, it can implement any of the methods described above.

[0033] In an eighth aspect of the present application, a computer program product is provided, comprising a computer program, which implements any of the methods described above when executed by a processor.

[0034] In the technical solution provided by the embodiment of the present application, the object information of the first object is obtained from the first website, and based on the object information, similar objects similar to the first object are screened from the second objects published on the second website, and based on the category to which the similar objects belong on the second website, the first object is classified, and then the publishing category to which the first object belongs on the second website is determined. It can be seen that the adoption of this solution can automatically determine the appropriate publishing category for the object to be published, improve the degree of automation, and greatly reduce the labor cost of information publishing.

[0035] In addition, based on the publishing category and the object information of the first object, the publishing information required for publishing the first object on the second website can be automatically generated, which can further improve the degree of automation and further reduce the labor cost of information publishing. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 A schematic diagram of an information processing system provided by an embodiment of the present application;

[0038] Figure 2a This is a flowchart of an information generation method provided in one embodiment of the present application;

[0039] Figure 2b This is a flowchart of a method for predicting a release category provided in one embodiment of the present application;

[0040] Figure 3 This is an example diagram of a first interface provided by an embodiment of the present application;

[0041] Figure 4 This is an example diagram of a second interface provided in an embodiment of the present application;

[0042] Figure 5 An example flow chart of an attribute information mapping method provided in one embodiment of the present application;

[0043] Figure 6a An example flow chart of an information publishing method provided in an embodiment of the present application;

[0044] Figure 6b An example flow chart of an information publishing method provided in an embodiment of the present application;

[0045] Figure 7 A service system architecture diagram of a server device side provided in an embodiment of the present application;

[0046] Figure 8 This is a flowchart of an information category determination method provided in an embodiment of the present application;

[0047] Fig. 9 This is a flowchart of an information display method provided in one embodiment of the present application;

[0048] Fig.10 A structural block diagram of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0049] In actual applications, the Category-Property-Value (CPV) systems of different websites are different. The differences are reflected in the different category structures (or category trees), different attribute names, and different attribute values. For example, the category structure is composed of multiple levels of categories, and the multiple levels include first-level categories and leaf categories. In other words, the same product may belong to different categories on different websites. The attribute names and attribute values ​​of different websites are also different.

[0050] Therefore, in actual applications, when users want to publish the same product on multiple websites, they need to manually select the category to which the product belongs on each website to complete the product publishing, which is time-consuming and easy to make mistakes. These problems are more prominent when publishing large quantities of products. Once the wrong category is selected for a product, it will not only affect the subsequent search and push flow of the product, reducing the sales volume of the product, but may also recommend the wrong product to consumers.

[0051] In order to solve or partially solve the above technical problems, in the technical solution provided by the embodiment of the present application, the same-item object (i.e., similar object) of the off-site object (i.e., the first object) is obtained through the same-item matching, and the category to which the off-site object belongs in the site is determined based on the category to which the same-item object belongs, thereby realizing the conversion of the off-site category expression into the on-site category expression. The adoption of this solution can automatically determine the appropriate publishing category for the object to be published, improve the degree of automation, and greatly reduce the labor cost of information publishing, especially when a large number of information needs to be published, the labor cost reduction effect is more significant.

[0052] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below according to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0053] In addition, some of the processes described in the specification, claims and the above-mentioned figures of the present application include multiple operations that appear in a specific order, and these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0054] 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, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0055] First, the vocabulary involved in the embodiments of the present application is explained. It can be understood that this explanation is for a clearer understanding of the embodiments of the present application, and does not necessarily constitute a limitation on the embodiments of the present application.

[0056] Product launch: refers to the process of creating and displaying product information on e-commerce platforms, including product name, description, price, image, etc., to attract consumers to buy.

[0057] Category: an object, such as the classification of goods, which refers to a collection of goods with certain common characteristics. It can be divided into multiple categories according to the granularity of division, for example: clothing-men's clothing-jackets, where clothing is the first-level category, men's clothing is the second-level category, and jackets are the leaf category.

[0058] Attribute: refers to the characteristics of an object, such as a product. The attribute value is the specific content of the attribute. For example, the attributes of a product include color, size, collar type, sleeve type, etc. The attribute values ​​of the color attribute include red, pink, brown, etc.

[0059] Large Language Model (LLM): refers to natural language processing models based on deep learning. They are trained using neural networks and can automatically learn the laws and features of natural languages, thereby achieving tasks such as language models, text generation, machine translation, and speech recognition. These models usually require large amounts of data and computing resources to train, so they are called "large" language models.

[0060] Multimodal large model: a model that is trained by combining multimodal information such as text, images, video, and audio. In other words, it refers to a large model that integrates multiple types of data processing capabilities under a unified framework. It can process data in multiple modalities at the same time, making the machine more comprehensive and in-depth in understanding the world.

[0061] Data collection: refers to the use of information acquisition technology to automatically collect product information from e-commerce websites, such as product name, description, price, pictures, comments, etc., to facilitate data analysis, comparison or market research.

[0062] Information release: refers to the process of creating and displaying information on a website. Taking product release as an example, product release refers to the process of creating and displaying product information on an e-commerce platform, including product name, description, price, pictures, etc., to attract consumers to buy.

[0063] Same-style matching: refers to identifying products of the same style as a product by analyzing objects such as the product title, image, attributes, etc.

[0064] Before introducing the information category determination method, information generation method and information publishing method provided in the embodiments of the present application, an information processing system for implementing these methods is introduced. Figure 1 As shown, the information processing system may include: a terminal device 100 and a server device 200.

[0065] The terminal device 100 is used to send an information generation request to the server device 200, wherein the information generation request is used to request the generation of publishing information required for publishing a first object on the second website, wherein the first object is a published object on the first website;

[0066] The server device 200 is used to obtain object information of the first object in response to the information generation request, where the object information of the first object comes from the first website; based on the object information of the first object, filter out similar objects similar to the first object from second objects published on the second website; and determine the publishing category to which the first object belongs on the second website based on the category to which the similar objects belong on the second website.

[0067] The terminal device 100 may send an information generation request to the server device 200 in response to the user's operation. Exemplarily, the terminal device 100 may send an information generation request to the server device 200 in response to the user's operation on the interface.

[0068] In the technical solution provided by the embodiment of the present application, the object information of the first object is obtained from the first website, and based on the object information, similar objects similar to the first object are screened from the second objects published on the second website, and based on the category to which the similar objects belong on the second website, the first object is classified, and then the publishing category to which the first object belongs on the second website is determined. It can be seen that the adoption of this solution can automatically determine the appropriate publishing category for the object to be published, improve the degree of automation, and greatly reduce the labor cost of information publishing.

[0069] In some embodiments, the server device 200 is further used to generate publishing information required to publish the first object on the second website according to the publishing category and the object information of the first object.

[0070] In the technical solution provided in the embodiment of the present application, the publishing information required for publishing the first object on the second website is automatically generated based on the publishing category and the object information of the first object, which can further improve the degree of automation and further reduce the labor cost of information publishing.

[0071] In some embodiments, the server device 200 is further used to publish the first object on the second website based on the publishing information.

[0072] The technical solution provided in the embodiments of the present application helps to improve the efficiency of information release.

[0073] Exemplarily, the above information processing system can be applied in the field of e-commerce. Accordingly, the above objects can be commodities, specifically but not limited to: clothing, home furnishings, shoes and boots, bags (such as handbags, backpacks, bags, etc.), hats, gloves, accessories (such as bracelets, necklaces, bracelets, earrings, headwear, etc.), electronic products (such as watches, mobile phones, laptops, tablet computers, etc.), bedding, etc.

[0074] Exemplarily, the above information processing system can be applied in the field of video on demand. Accordingly, each of the above objects can be a video, which can be but not limited to: variety show video, animation video, movie, TV series, documentary, etc.

[0075] The terminal device may be a smart phone, a tablet computer, a desktop computer, a smart wearable device, etc. The server device may be a single server, a service cluster, a virtual server deployed on a server or a cloud, etc. Exemplarily, the server device may be a server device of the second website.

[0076] The functional implementation contents of the various hardware such as the terminal device and the server device in the information processing system will be described in detail in the following method embodiments.

[0077] Figure 2aA flowchart of an information generation method provided in one embodiment of the present application. The execution subject of all steps in the method described in this embodiment may be a server device in the above-mentioned system. Of course, if the terminal device has strong computing power, the execution subject of the method in this embodiment may also be a client, or the client and the server may execute the method in collaboration (i.e., some steps are executed by the client, and other steps are executed by the server). For ease of understanding, the information generation method provided in the embodiment of the present application will be introduced below by taking the execution subject as the server device as an example. Figure 2a As shown, the method may include the following steps:

[0078] 201. Obtain object information of a first object.

[0079] The first object is a published object on the first website.

[0080] The object information of the first object may include description information for describing the first object. Exemplarily, the description information may include multimodal description information, and the multimodal description information may include but is not limited to: text information, image information, and video information.

[0081] Among them, the object information of the first object comes from the first website. The first object is a published object on the first website, and the object information of the first object comes from the publishing information published by the first website for the first object. The publishing information published by the first website for the first object may include but is not limited to: the category to which the first object belongs on the first website, the original attribute information configured by the first object on the first website, and the description information used to describe the first object. The description information may include multimodal description information, and the multimodal description information may include but is not limited to: text information, image information and video information. It should be noted that the publishing information in this article is used to describe the corresponding object.

[0082] In some embodiments, the object information of the first object may be derived from web page data of a web page used to display the first object on the first website, and the web page is used to carry the published information of the first object. The web page data may include: the category to which the first object belongs on the first website, the original attribute information configured by the first object on the first website (including the original attribute name and the attribute value of the original attribute name), and description information used to describe the first object.

[0083] Taking the first website as an e-commerce platform and the first object as a product as an example, the product information of the product can come from a web page for displaying the product in the e-commerce platform, such as a product details page. The web page data can be obtained by acquiring data from the web page, and the object information of the first object can be determined based on the web page data.

[0084] In some other embodiments, the object information of the first object may come from a server device of the first website. Figure 1 The server device 200 shown (which may be referred to as the server device of the second website) sends a data acquisition request to the server device of the first website. The data acquisition request may carry the object identifier of the first object. The server device of the first website responds to the data acquisition request and sends the publishing information about the first object on the first website to the server device of the second website. The server device of the second website determines the object information of the first object based on the publishing information.

[0085] 202. Filter similar objects that are similar to the first object from second objects published on a second website according to the object information of the first object.

[0086] The number of the second objects published by the second website is one or more.

[0087] The first website and the second website are two different websites.

[0088] In an optional implementation, the publishing information published by the second website for the second object may be obtained, and based on the similarity between the object information of the first object and the publishing information published by the second website for the second object, similar objects similar to the first object may be screened out from one or more second objects published by the second website.

[0089] The information published by the second website for the second object may include, but is not limited to: the category to which the second object belongs on the second website, the attribute information configured on the second website for the second object, and description information for describing the second object. The description information may include multimodal description information, which may include, but is not limited to: text information, image information, and video information.

[0090] In some embodiments, the object information of the first object includes multimodal description information of the first object. In the above 202, "according to the object information of the first object, selecting similar objects similar to the first object from the second objects published on the second website" can be implemented by the following steps:

[0091] 2021. Vectorize the multimodal description information of the first object to obtain a vector for characterizing the first object.

[0092] The multimodal description information of the first object includes: first text information and first image information.

[0093] Taking the first object as a product as an example, the first text information may include: product title, product details description text, etc., and the first image information may include: product pictures, such as: product main picture, product inventory unit (SKU) picture and product details picture.

[0094] The first text information is vectorized to obtain a first text vector, and the first image information is vectorized to obtain a first image vector.

[0095] In some embodiments, the vector used to characterize the first object includes: a first text vector and a first image vector.

[0096] 2022. Obtain a vector for characterizing the second object.

[0097] The multimodal description information of the second object includes: second text information and second image information.

[0098] In some embodiments, the vector used to represent the second object may include: a second text information vector obtained by vectorizing the second text information and a second image vector obtained by vectorizing the second image information.

[0099] The vector used to represent the second object can be obtained in advance through vectorization operations and stored in the database. When it is needed later, it can be directly obtained, which can save the time consumed by vectorization and help shorten the time required for information generation. Of course, if the delay is not considered, the vector used to represent the second object can also be directly obtained through vectorization operations when it is needed later.

[0100] 2023. Determine a similar object that is similar to the first object based on the vector used to characterize the first object and the vector used to characterize the second object.

[0101] Based on the first text vector, the first image vector, the second text vector and the second image vector, the similarity between the first object and the second object is determined, and similar objects are determined according to the similarity.

[0102] The similarity between the first object and the second object may be calculated by one or more of the following methods:

[0103] Method 1:

[0104] A first similarity between the first object and the second object is determined based on the similarity between the first text vector and the second text vector. A second similarity between the first object and the second object is determined based on the similarity between the first image vector and the second image vector. A third similarity between the first object and the second object is determined based on the similarity between the first text vector and the second image vector. A fourth similarity between the first object and the second object is determined based on the similarity between the first image vector and the second text vector.

[0105] Exemplarily, the similarity between two vectors may be cosine similarity.

[0106] Exemplarily, the sum of the first similarity and the second similarity can be determined as the comprehensive similarity between the first object and the second object, and then similar objects are screened out according to the comprehensive similarity. For example, objects whose comprehensive similarity is greater than a preset threshold are determined as similar objects. Of course, the top N similar objects can also be screened out in descending order, where N is an integer greater than or equal to 1.

[0107] Exemplarily, the first similar object is screened out according to the first similarity, the second similar object is screened out according to the second similarity, and the first similar object and the second similar object are determined as similar objects. The first similar object and the second similar object can be screened in the same screening manner as in the above embodiment.

[0108] In practical applications, the first text information can be input into a text search engine, so that the text search engine determines the first similarity, and the information is sorted in descending order according to the first similarity, thereby screening out N1 first similar objects; the first image information can be input into an image search engine, so that the image search engine determines the second similarity, and the information is sorted in descending order according to the second similarity, thereby screening out N2 second similar objects. Among them, N1 and N2 are integers greater than or equal to 1, and their sizes can be equal, for example: 10. The N1 first similar objects and the N2 second similar objects can be deduplicated to obtain the final similar objects. Taking the first object as a product as an example, Figure 2b As shown, the product information of the product includes a title and an image. The image is input into the image search engine 21, and the image search engine 21 outputs the image search result. The text, that is, the title, is input into the text search engine 22 to obtain the text search result. The text search result and the image search result are deduplicated to obtain the final similar objects.

[0109] For example, the sum of the first similarity, the second similarity, the third similarity and the fourth similarity can be determined as the comprehensive similarity between the first object and the second object, and then similar objects are screened out according to the comprehensive similarity. The specific screening method can refer to the corresponding content in the above embodiments.

[0110] Method 2:

[0111] The first text vector and the first image vector are concatenated to obtain a first concatenated vector; the second text vector and the second image vector are concatenated to obtain a second concatenated vector, and the similarity between the first object and the second object is determined based on the similarity between the first concatenated vector and the second concatenated vector, and then similar objects are screened out based on the similarity. The object screening method can refer to the corresponding contents in the above embodiments.

[0112] In an embodiment of the present application, by vectorizing the multimodal description information of the first object, a vector that can accurately represent the first object can be obtained. In this way, subsequently, based on the vector similarity, similar objects similar to the first object can be found more accurately, which helps to accurately determine the publication category subsequently.

[0113] 203. Determine the publishing category to which the first object belongs on the second website according to the category to which the similar object belongs on the second website.

[0114] When there is only one similar object, the category to which the similar object belongs on the second website may be determined as the publishing category to which the first object belongs on the second website.

[0115] When there are multiple similar objects, one or more of the following methods can be used to determine them:

[0116] Method 1: The category to which the similar object most similar to the first object belongs on the second website may be determined as the publishing category to which the first object belongs on the second website.

[0117] Method 2: Count the categories to which multiple similar objects belong on the second website, and determine the category with the most appearances as the publishing category to which the first object belongs on the second website. For example: multiple similar objects are: A, B, C, D, E, among which A belongs to a1 on the second website, B belongs to a1 on the second website, C belongs to a1 on the second website, D belongs to a2 on the second website, and E belongs to a3 on the second website. Among them, a1 appears 3 times, a2 appears 1 time, and a3 appears 1 time. Therefore, a1 with the most appearances is determined as the publishing category to which the first object belongs on the second website.

[0118] Method 3: Use LLM to screen.

[0119] Method 3 can be implemented by the following steps:

[0120] 2031. Determine the categories to which the multiple similar objects belong on the second website as multiple candidate categories.

[0121] Using the above example, a1, a2, and a3 can be determined as multiple candidate categories.

[0122] That is to say, the categories to which the multiple similar objects belong on the second website are not the same, and the categories to which the multiple similar objects belong on the second website need to be taken as multiple candidate categories, that is, multiple candidate categories.

[0123] 2032. Generate a first prompt word according to the object information of the first object and the multiple candidate categories.

[0124] The first prompt word is used to prompt the large language model to select the target category to which the first object belongs from the multiple candidate categories.

[0125] In practical applications, in order to improve the efficiency of prompt word construction, a first prompt word template can be constructed in advance. The prompt word template is a pre-designed prompt word framework with a certain structure and format, which includes a fixed part and a placeholder. The position occupied by the placeholder in the template is the position to be filled. By filling the placeholder, different prompt words can be generated. The prompt word template helps to improve the efficiency and standardization of prompt word generation, making the interaction with LLM more convenient and effective.

[0126] In this way, the object information of the first object and a plurality of candidate categories are subsequently filled into the first prompt template to obtain the first prompt word.

[0127] In the embodiment of the present application, using the reasoning ability of LLM to predict the release category helps to improve the prediction accuracy of the release category, thereby improving the rationality and availability of the automatically generated release information, thereby reducing the operating cost of users releasing information.

[0128] In order to improve the accuracy of LLM output, you can add some additional examples in the first prompt word to help LLM better understand the data content of this input, the operations / tasks to be performed, and the output format. Specifically, the following steps can be used to achieve this:

[0129] S11. Generate a plurality of first examples according to the object information of the plurality of similar objects and the categories to which the plurality of similar objects belong on the second website.

[0130] The first example may include multimodal description information of similar objects and the categories to which the similar objects belong. Exemplarily, the format of the first example may be [multimodal description information of similar objects, categories to which the similar objects belong].

[0131] S12: Generate a first prompt word according to the object information of the first object, the multiple candidate categories, and the multiple first examples.

[0132] A first prompt word is generated according to the multimodal description information of the first object, the multiple candidate categories, and the multiple first examples.

[0133] Exemplarily, the multimodal description information of the first object, the multiple candidate categories, and the multiple first examples may be filled into a first prompt word template to obtain a first prompt word.

[0134] 2033. Input the first prompt word into the large language model to obtain the target category output by the large language model.

[0135] Among them, the large language model can process multimodal information, so it can be called a multimodal large model.

[0136] Take the first object as a commodity as an example. Figure 2b As shown, a first prompt word is constructed based on a first prompt word template, multiple candidate categories and product information, and the first prompt word is input into the multimodal large model 23, and the large model 23 outputs category 1 and category 2.

[0137] It should be noted that the large language model can output one or more categories. When the large language model outputs multiple categories, the multiple categories can be sent to the user so that the user can select the final category. The number of multiple categories here is small, and the user can easily select from these small categories, which helps to improve the accuracy of category division.

[0138] 2034. Determine the target category as the publishing category to which the first object belongs on the second website.

[0139] When the target category output by the large language model is one, the target category output by the large language model may be determined as the publishing category.

[0140] When the large language model outputs multiple target categories, the target category selected by the user may be determined as the publishing category.

[0141] In some embodiments, since the CPV of different websites is different, the published category and the category configured by the first website for the first object are likely to be different. For example, the published category is: Apparel-Men's Clothing-Coats, and the category configured by the first website for the first object is: Men's Clothing-Tops-Jackets. The categories in this application are multi-level categories, that is, they are composed of multiple categories of different levels.

[0142] 204. Generate publishing information required for publishing the first object on the second website according to the publishing category and the object information of the first object.

[0143] Optionally, the publishing information required for publishing the first object on the second website may include: a publishing category and multimodal description information for describing the first object, such as: a picture, a title, etc.

[0144] In the technical solution provided by the embodiment of the present application, the object information of the first object is obtained from the first website, and based on the object information, similar objects similar to the first object are screened out from the second objects that have been published on the second website, and based on the category to which the similar objects belong on the second website, the first object is classified, and then the publishing category to which the first object belongs on the second website is determined. It can be seen that the adoption of this solution can automatically determine the appropriate publishing category for the object to be published, improve the degree of automation, and greatly reduce the labor cost of information publishing. In addition, based on the publishing category and the object information of the first object, the publishing information required to publish the first object on the second website can be automatically generated, which can further improve the degree of automation and further reduce the labor cost of information publishing.

[0145] The following describes a process for obtaining object information of a first object, which may include the following steps:

[0146] 205. Obtain a target uniform resource identifier of the first object.

[0147] A Uniform Resource Locator (URL) is a string used to identify the name of an Internet resource.

[0148] In an optional manner, the target URL is provided by the user. For example, Figure 3 As shown, the terminal device displays a first interface 300, which displays a first input box 302 and a control "one-click product transfer" 303. In response to the user triggering the control 303, the target URL entered by the user in the first input box is sent to the server device.

[0149] In another optional manner, the target URL is generated based on the website information of the first website provided by the user and the object ID of the first object provided by the user, wherein the object ID of the first object is assigned to the first object by the first website and is used to uniquely identify the first object on the first website. The website information of the first website may include the URL of the first website and / or the website name of the first website. It should be noted that based on the website name of the first website, the URL of the first website can be obtained through the network or the correspondence between the pre-set website name and the URL. The website information of the first website and the object ID of the first object can be provided by the user through the interface of the terminal device.

[0150] Taking the e-commerce platform as an example, the URL of the product is usually generated based on the URL of the e-commerce platform and the ID number of the product according to certain rules. Therefore, in the embodiment of the present application, the URL of the product can be generated based on the URL of the e-commerce platform and the product ID of the product according to the same rules.

[0151] 206. Acquire webpage data of the target webpage according to the target URL.

[0152] The target webpage is identified by the target URL, the target webpage belongs to the first website, and the target webpage is used to display the first object.

[0153] According to the target URL of the first object on the first website, the target webpage identified by the target URL can be obtained, and the webpage data can be obtained from the target webpage. Exemplarily, the webpage data can be obtained from the webpage based on information acquisition technology and multi-terminal information acquisition technology.

[0154] Among them, the web page data may include publishing information published by the first website for the first object, such as: the category to which the first object belongs on the first website, the original attribute information configured by the first object on the first website, and description information used to describe the first object. The description information may include multimodal description information, and the multimodal description information may include but is not limited to: text information, image information and video information.

[0155] Taking the first object as a product as an example, based on information acquisition technology and multi-terminal information acquisition technology, the open product information such as price, title, picture, details, category, attribute information, etc. of the product can be obtained from the web page to finally obtain the original product information of the product.

[0156] 207. Determine object information of the first object according to the webpage data.

[0157] Optionally, description information (such as multimodal description information) in the webpage data may be directly used as object information of the first object.

[0158] In the embodiment of the present application, the technical solution of obtaining the object information of the first object through a web page has the advantage of being easy to implement.

[0159] In practical applications, when the languages ​​used by the first website and the second website are different, the web page data may be preliminarily processed, and the object information of the first object may be determined based on the preliminarily processed web page data. Exemplarily, the description information used to describe the first object in the preliminarily processed web page data is determined as the object information of the first object.

[0160] The preliminary processing may include: text translation and image translation. Among them, text translation refers to translating text into a target language, which is the language used by the second website, and image translation refers to identifying characters in an image, translating the characters into a target language, and replacing them with the original characters in the image. Taking the first object as a product as an example, the text may include information such as title, category, attribute information, and detailed description text.

[0161] When the first website and the second website are e-commerce platforms, the preliminary processing may also include: performing price conversion on the commodity prices in the web page data. For example, the first website uses RMB as the price unit, and the second website uses USD as the price unit. The RMB price in the commodity information can be converted into USD price based on the current exchange rate.

[0162] Optionally, the image in the first object information can also be stored in the content delivery network (CDN) of the second website to facilitate subsequent information release. Taking products as an example, the main image, detail image and other images of the product can be stored in the content delivery network CDN of the second website to facilitate subsequent product release.

[0163] The above steps 205-207 are performed before the above step 201, so that in the subsequent steps, corresponding processing is performed based on the initially processed web page data (eg, object information, original attribute information).

[0164] Optionally, when the target URL is provided by the user, before executing step 206, the target URL may be verified for legitimacy based on the configuration file; when the legitimacy verification passes, the object information of the target object is obtained from the first website based on the target URL.

[0165] The legitimacy check may include but is not limited to: whether it is supported to obtain the web page of the website to which the URL belongs (obtaining the web page means obtaining web page data from the web page), whether the format of the target URL is correct. In actual application, the URL formats of different websites can be written into a configuration file, and then, based on the configuration file, it is determined whether it is supported to obtain the web page of the first website and whether the URL format entered by the user is correct.

[0166] When the legality check fails, a prompt message may be sent to the terminal device. The prompt message is used to prompt the user that the URL entered is illegal. After sending the prompt message to the terminal device, the server device may also end the current process.

[0167] The validity check can ensure that the user enters a valid URL.

[0168] In order to avoid infringement risks in data acquisition and information use, in some embodiments, ownership can be verified. Specifically, the above 206 "According to the target URL, obtaining the webpage data of the target webpage" can be implemented by the following steps:

[0169] 2061. Obtain identity information provided by the user for registration on the first website.

[0170] For example, Figure 4 As shown, the terminal device displays a second interface 400, which displays a second input box (for example, 401 and 402) and a confirmation control 403. In response to the user's triggering operation on the confirmation control 403, the identity information entered by the user in the second input box is sent to the server device.

[0171] Exemplarily, the second input box may include an input box 401 for inputting an account number and an input box 402 for inputting a password.

[0172] The first interface 300 and the second interface 400 may be the same interface or different interfaces. When the first interface 300 and the second interface 400 are the same interface, the first input box and the second input box are displayed on the same interface.

[0173] Exemplarily, when the first interface 300 and the second interface 400 are different interfaces, the terminal device displays (e.g., pops up) the second interface 400 in response to the user's triggering operation on the control 303 on the first interface 300, and the second interface 400 displays a second input box (e.g., 401 and 402).

[0174] 2062. Determine whether the first object belongs to the user according to the identity information and the target URL.

[0175] In an optional implementation, the step 2062 of "determining whether the first object belongs to the user according to the identity information and the target URL" can be implemented by the following steps:

[0176] S21. Acquire, according to the identity information, a URL of at least one object belonging to the user on the first website.

[0177] According to the identity information of the user registered on the first website, the server device of the first website may be requested to provide the URL of at least one object belonging to the user on the first website. The server device of the first website may first verify the identity information, and after the verification is passed, return the URL of at least one object belonging to the user on the first website to the server device of the second website.

[0178] It should be noted that when the identity information verification fails, the server device of the first website can notify the server device of the second website, and the server device of the second website can notify the terminal device. The terminal device displays a prompt message that the identity information verification fails and prompts the user to re-enter the identity information.

[0179] In actual applications, the server device of the second website can access the authorization component of the server device of the first website, and verify the user's identity information by calling the authorization component. After the verification is passed, the URL of at least one object belonging to the user on the first website can be obtained through the application programming interface (API) provided by the server device of the first website. Taking the e-commerce platform as an example, the URL of at least one product under the user's store can be obtained through the application programming interface.

[0180] S22: If the target URL exists in the URL of the at least one object, determine that the first object belongs to the user.

[0181] When the target URL exists in the URL of at least one object, it indicates that the first object belongs to the user.

[0182] S23: If the target URL does not exist in the URL of the at least one object, it is determined that the first object does not belong to the user.

[0183] When the target URL does not exist in the URL of at least one object, it indicates that the first object does not belong to the user.

[0184] 2063. When the first object belongs to the user, obtain web page data of the target web page according to the target URL.

[0185] When the first object belongs to the user, the server device of the second website obtains the object information of the first object from the first website according to the target URL.

[0186] When the first object does not belong to the user, the server device of the second website may send first prompt information to the user, where the first prompt information is used to prompt that the first object does not belong to the user.

[0187] In the embodiment of the present application, when obtaining the object information of the first object from the first website, it is necessary to verify whether the first object belongs to the user. When the first object belongs to the user, the object information of the first object is obtained from the first website. When the first object does not belong to the user, the object information of the first object is not obtained from the first website. This helps to avoid the risk of infringement in data acquisition and product information use.

[0188] In actual applications, taking the release of goods as an example, when releasing goods, it is necessary not only to determine the category to which the goods belong, but also to mount the corresponding attribute information for the goods under the category. However, different websites have different category structures, and the attribute names (or attribute names) mounted under the categories are also different. Therefore, a method for determining the attribute information that needs to be mounted for a first object will be introduced below. In some embodiments, the above 204 "generating the release information required to release the first object on the second website according to the release category and the object information of the first object" can be implemented by the following steps:

[0189] 2041. Determine a target attribute name that needs to be configured for the first object according to a plurality of preset attribute names pre-configured by the second website for the publishing category.

[0190] In actual application, the second website will assign an attribute ID to each preset attribute name to uniquely identify the corresponding preset attribute name. Therefore, in the above 2041, the target attribute name required to be configured for the first object and its corresponding attribute ID can be determined.

[0191] In an optional implementation, a plurality of preset attribute names pre-configured by the second website for the publishing category may be determined as the target attribute names required to be configured for the first object.

[0192] In actual applications, there are many preset attribute names pre-configured on the website for a certain category, some of which are required attribute names and some are optional attribute names.

[0193] In some other optional implementations, the required attribute name pre-configured by the second website for the publishing category can be determined as the target attribute name that needs to be configured for the first object.

[0194] In some other optional implementations, LLM can be used to filter out the target attribute name that the first object needs to configure from multiple preset attribute names. Specifically, in the above step 2041, "determining the target attribute name that the first object needs to configure according to multiple preset attribute names pre-configured by the second website for the publishing category" can be implemented by the following steps:

[0195] S31. Generate a second prompt word according to the object information of the first object, the original attribute name configured by the first website for the first object, and the plurality of preset attribute names.

[0196] The second prompt word is used to prompt the large language model to filter out a target attribute name required to be configured for the first object from the multiple preset attribute names.

[0197] A second prompt word template may be obtained, and the object information of the first object, the original attribute name configured by the first website for the first object, and a plurality of preset attribute names may be filled into the second prompt word template to obtain a second prompt word.

[0198] In order to improve the accuracy of LLM output, you can add some additional examples in the second prompt word to help LLM better understand the data content of this input, the operations / tasks to be performed, and the output format. Specifically, you can use the following steps to achieve this:

[0199] A. Determine a reference object from the similar objects.

[0200] Wherein, the reference object belongs to the target category.

[0201] Optionally, objects belonging to the target category among similar objects may be determined as reference objects.

[0202] Optionally, the reference object may be an object belonging to the target category in the search results obtained based on the above image search engine.

[0203] B. Generate a second instance according to the object information of the reference object and the attribute name configured by the second website for the reference object.

[0204] The second example may include multimodal description information of the reference object and the attribute name configured by the second website for the reference object. Exemplarily, the format of the second example may be [multimodal description information of the reference object, original attribute name configured by the reference object].

[0205] C. Generate a second prompt word according to the object information of the first object, the multiple preset attribute names and the second example.

[0206] The object information of the first object, the original attribute name configured by the first website for the first object, the second example and a plurality of preset attribute names are filled into the second prompt word template to obtain a second prompt word.

[0207] S32: Input the second prompt word into the large language model to obtain the target attribute name output by the large language model.

[0208] The number of target attribute names can be one or more.

[0209] In the embodiment of the present application, the reasoning ability of LLM can be used to complete a more accurate attribute name mapping (that is, mapping from the original attribute name to the target attribute name).

[0210] 2042. Determine an attribute value of the target attribute name according to original attribute information configured by the first website for the first object.

[0211] The original attribute information includes an original attribute name and an original attribute value corresponding to the original attribute name. The original attribute name and the original attribute value corresponding to the original attribute name are configured by the first website for the first object. Exemplarily, the original attribute name and the original attribute value corresponding to the original attribute name may be configured by the first website in response to a configuration operation of a user (e.g., a merchant) to whom the first object belongs.

[0212] In an optional implementation, the step 2042 of “determining the attribute value of the target attribute name according to the original attribute information configured by the first website for the first object” can be implemented by the following steps:

[0213] S41. Obtain multiple optional values ​​pre-configured by the second website for the target attribute name.

[0214] Taking the e-commerce platform as an example, suppose that the first website pre-configures the attribute names for the category of men's clothing - tops - jackets: brand, color, fabric, among which the multiple optional values ​​of fabric are: combed cotton, mercerized cotton, wool, cashmere; the second website pre-configures the attribute names for the category of clothing - men's clothing - coats: brand, color, material, etc., among which the multiple optional values ​​of material include: cotton, wool, spandex.

[0215] S42: Determine a reference attribute name that matches the target attribute name from the original attribute name.

[0216] Wherein, there are multiple original attribute names. In an optional manner, multiple original attribute names may be vectorized to obtain text vectors of one or more original attribute names; a target attribute name may be vectorized to obtain a text vector of the target attribute name; vector similarities between the text vector of the target attribute name and the text vectors of each of the multiple original attribute names are calculated, and the original attribute name with the greatest vector similarity is determined as the reference attribute name.

[0217] Using the above example, product A belongs to the category of men's clothing - tops - jackets on the first website. The first website configures multiple original attribute names for product A, namely: brand, color, fabric (main fabric). The publishing category of product A on the second website is: clothing - men's clothing - coats. The target attribute name that needs to be configured for the first object to be published on the second website is material. Among the three brand, color, and fabric, the original attribute name that best matches material is fabric. Therefore, fabric is determined as the reference attribute name.

[0218] S43: Determine a target attribute value corresponding to the target attribute name from the multiple optional values ​​according to the similarity between the original attribute value corresponding to the reference attribute name and the multiple optional values.

[0219] The original attribute value and multiple optional values ​​may be vectorized, and then based on vector similarity, the optional value with the highest similarity is selected from the multiple optional values ​​as the target attribute value corresponding to the target attribute name.

[0220] Using the above example, on the first website, the original attribute value corresponding to the attribute name "Fabric" of Product A is mercerized cotton. Mercerized cotton is most similar to "cotton" among the multiple optional values ​​of cotton, wool, and spandex corresponding to the attribute name "Material" on the second website. Therefore, cotton can be determined as the target attribute value of "Material".

[0221] That is, the original attribute value corresponding to the reference attribute name is different from the target attribute value corresponding to the target attribute name.

[0222] In actual applications, attributes can be divided into mandatory items and optional items. Therefore, when the target attribute name is a mandatory attribute name, the optional value that is most similar to the original attribute value is used as the target attribute value corresponding to the target attribute name. When the target attribute name is an optional item, the optional value that is most similar to the original attribute value is determined. When the similarity between the optional value and the original attribute value is greater than a preset threshold, the optional value is used as the target attribute value corresponding to the target attribute name. Otherwise, the target attribute value corresponding to the target attribute name is set to empty, that is, no attribute value is mounted for the target attribute name.

[0223] In actual applications, the second website will assign a value ID to each optional value. Therefore, the above step S43 can determine the target attribute value corresponding to the target attribute name and the value ID of the target attribute value.

[0224] In the embodiment of the present application, the original attribute value of the original attribute name is not simply directly used as the attribute value of the target attribute name. Instead, the optional value that best matches the original attribute value among the multiple optional values ​​configured by the second website for the target attribute name is used as the attribute value of the target attribute name, which helps to improve the mapping accuracy of the attribute value.

[0225] 2043. Generate publishing information required for publishing the first object on the second website according to the publishing category, the target attribute name, and the target attribute value corresponding to the target attribute name.

[0226] The release information may include: the release category, the target attribute name and the target attribute value corresponding to the target attribute name. Optionally, the release information may also include: description information for describing the first object, such as title, image and the like.

[0227] For example, Figure 5 As shown, for the above Figure 2bThe image search result of the image search engine 21 performs category filtering, filters out the products belonging to the publishing category in the image search result, and generates a second example based on the product information of the filtered products and the attribute name configured by the second network for the products; according to the publishing category, query the multiple preset attribute names configured by the second website for the publishing category, and these multiple preset attribute names can be called category attributes of the publishing category, wherein the multiple preset attribute names are divided into two types: required attribute names and optional attribute names (such as sales attributes); query the merchant information according to the merchant ID of the merchant (that is, the merchant who needs to publish the product on the second website), and the merchant information can include the merchant's main content, name and other information. The second prompt word can be generated based on the product information (for example: title, image, description text, etc.), the category attribute of the publishing category, the second example, and the merchant information, and the second prompt word is input into the multimodal large model 501 to obtain the target attribute name output by the large model 501. According to the preset rules and text similarity, the target attribute value is mounted for the target attribute name, where the preset rules refer to: the attribute management rules of the second website, for example, if it is divided into required attribute names, then this attribute name must have an attribute value, and when the attribute value is mounted, it will be guaranteed that there is a value mounted under this attribute name, while sales attributes such as size may not need to be guaranteed to have a value. Among them, multiple target attribute names can be multiple. Finally, the result assembly / aggregation is executed to obtain the required attribute list [{attribute name, value, attribute ID}] and the sales attribute list [{attribute name, value, attribute ID}].

[0228] When the second website is an e-commerce website, after obtaining the target attribute information (i.e., the target attribute name and the target attribute value corresponding to the target attribute name), SKU information can be generated based on this. SKU information may include but is not limited to target attribute information, price information, weight information, etc. In addition, a logistics recommended route can also be generated based on information such as weight and size.

[0229] In summary, by using the same-item matching to obtain the same item on the site for off-site products, combined with the big model, the category prediction accuracy recall rate is improved, the attribute mapping function is realized, and the off-site category system expression is converted into the on-site category system expression, solving the problem of inaccurate categories and the inability to fill in the original product attributes. In addition, based on artificial intelligence algorithms, such as LLM, the original title and description of the product are optimized, the selling points, logistics information, price information, etc. are automatically generated, and the images are translated based on the visual model to generate images that meet the on-site standards, so that merchants can send products without modification.

[0230] Figure 6aA flowchart of an information publishing method provided for an embodiment of the present application. The execution subject of all steps in the method described in this embodiment may be a server device in the above system. Of course, if the terminal device has strong computing power, the execution subject of the method in this embodiment may also be a client, or the client and the server may execute the method in collaboration (i.e., some steps are executed by the client, and other steps are executed by the server). Figure 6a As shown, the method may include the following steps:

[0231] 61. Obtain object information of the first object.

[0232] The first object is a published object on a first website, and the object information of the first object comes from the first website.

[0233] 62. Filter similar objects that are similar to the first object from second objects published on a second website according to the object information of the first object.

[0234] 63. Determine the publishing category to which the first object belongs on the second website based on the category to which the similar object belongs on the second website.

[0235] 64. Generate publishing information required for publishing the first object on the second website according to the publishing category and the object information of the first object.

[0236] 65. Based on the publishing information, publish the first object on the second website.

[0237] In an optional implementation, the first object may be published directly based on the publishing information.

[0238] It should be noted here that: for the contents of each step not fully described in detail in the method provided in the embodiment of the present application, please refer to the corresponding contents in the above embodiment, which will not be repeated here. In addition, in addition to the above steps, the method provided in the embodiment of the present application may also include other parts or all of the steps in the above embodiments, which can be specifically referred to the corresponding contents in the above embodiments, which will not be repeated here.

[0239] Let's take the first object as a product as an example. Figure 6b The technical solutions provided in the embodiments of the present application are introduced by way of example. Figure 6b As shown, the method may include the following steps:

[0240] 601. The user enters a URL.

[0241] Users (ie merchants) can use the terminal device to provide Figure 3The URL of the product is input into the first input box 302 on the first interface 300 shown, and the "one-click product transfer" control 303 on the first interface 300 is clicked, and the terminal device sends the URL of the product to the server device of the second website.

[0242] 602. URL validity check.

[0243] The server device of the second website uses the URL format configuration file to verify the legitimacy of the product URL. After the verification is successful, the terminal device is notified so that the terminal device can display the following information: Figure 4 The second interface 400 is shown.

[0244] If the verification fails, the terminal device is notified so that the user can be prompted by the terminal device to input errors and re-enter the input.

[0245] 603. The user enters identity information.

[0246] The user can input the user identity information through the second input box 401 on the second interface 400 and click the confirmation control 403 on the second interface 400, and the terminal device sends the user identity information to the server device of the second website.

[0247] 604. Verification of commodity ownership.

[0248] The server device of the second website verifies whether the product belongs to the user by calling the authorization component provided by the first website.

[0249] 605. Determine whether the user is the owner of the product.

[0250] Based on the verification result, it is determined whether the user is the owner of the product. If the product is owned by the user, step 606 is executed, otherwise, the entire process ends.

[0251] 606. Collect product information in real time.

[0252] The web page of the product is obtained according to the URL of the product, and data of the web page is obtained to obtain web page data, which includes product information.

[0253] 607. Perform data processing on the collected data (also known as the preliminary processing mentioned above).

[0254] Data processing involves: picture-text translation, picture transfer and price conversion, etc. The specific implementation can refer to the corresponding content in the above embodiments, which will not be repeated here.

[0255] 608. Data conversion.

[0256] Data conversion includes: category prediction (i.e. determination of published categories), attribute mapping (i.e. determination of target attribute names and attribute values), currency conversion (e.g. conversion of RMB to USD), logistics information generation, and SKU information generation. For specific implementation, please refer to the corresponding contents in the above embodiments, which will not be described here.

[0257] Based on the data conversion result, the publishing information required for publishing the product on the second website is generated.

[0258] 609. Provide publishing information to users.

[0259] The published information can be sent to the terminal device in the form of a product form, and the terminal device displays the product form on the third interface for modification and confirmation by the user.

[0260] 610. User publishes product.

[0261] In response to the user's triggering operation on the publishing control of the third interface, the terminal device sends a publishing request to the server device of the second website, and the server device of the second website publishes the product on the second website based on the publishing information confirmed by the user.

[0262] In actual application, the user can enter the URL of the product to be released outside the site (the first website) in the input box displayed by the terminal device, and then click the "one-click product transfer" control displayed by the terminal device. In response to the user's click operation on the control, the terminal device sends the URL of the product to be released outside the site to the server device in the site (the second website). The server device in the site verifies the legitimacy of the URL. After the verification is successful, it notifies the terminal device to display an authorization interface (such as Figure 4 As shown in the figure), the user enters the account and password registered by the user outside the site in the authorization interface. After receiving the account and password sent by the terminal device, the server device in the site verifies whether the product to be released belongs to the user by calling the authorization component provided by the website outside the site. When the product to be released belongs to the user, the server device in the site obtains the web page data based on the URL, processes and converts the web page data, and generates the release information required for the release of the product in the site, and sends the release information to the user so that the user can confirm and release the product.

[0263] The technical solution provided by the embodiment of the present application solves the difficulty of how to convert the expression of the commodity system outside the website into the commodity system inside the website, among which category prediction and attribute mapping are the key. First, based on the commodity information, the category to which the commodity outside the website belongs in the website is predicted, and based on the commodity information and the predicted category, the original attributes of the commodity outside the website are mapped to the attributes under the attribute system inside the website.

[0264] The technical solution provided in the embodiment of the present application is based on multi-terminal collection and information acquisition technology to achieve the acquisition of complete original information of goods, which is beneficial to improving the success rate and completeness of the collection of original information of goods. The categories and attributes obtained based on official category attribute data, combined with same-item matching and large model prediction are more accurate and of better quality, and can parse and fill in complex sku information. Users can implement one-click product delivery based on the generated information; by accessing OAuth authorization and verifying the ownership of goods, users can only publish their own goods to avoid the risk of infringement of data acquisition and use of product information.

[0265] The technical solution provided in the embodiment of the present application is based on real-time acquisition of merchants' products on other platforms, and through category prediction, attribute mapping and other operations based on large language models, merchants can quickly move products from outside the website to the website with one click, thereby improving the efficiency and quality of merchants' product delivery, driving the growth of the website's high-quality supply scale, and having strong practical value. One-click product delivery refers to the operation of users entering the product URL and publishing the product published outside the website to the website with one click. For example, Figure 3 As shown, after the user enters the product URL, he clicks the "one-click product move" control 302 on the first interface 300 to complete the product release.

[0266] Still taking the first object as a commodity as an example, Figure 1 The interaction process between the devices is described in detail:

[0267] 1. The user inputs the URL of the product through the terminal device 100.

[0268] 2. The terminal device 100 sends the URL of the product to the server device 200 of the second website.

[0269] 3. The server device 200 performs a validity check on the URL.

[0270] After verification is passed, proceed to step 4.

[0271] 4. The server device 200 notifies the terminal device 100 to prompt the user to enter the identity information registered on the first website.

[0272] 5. The user enters identity information through the terminal device 100.

[0273] 6. The terminal device 100 sends the identity information to the server device 200 of the second website.

[0274] 7. The server device 200 of the second website performs a product ownership verification.

[0275] When the product is verified to belong to the user, proceed to step 8.

[0276] 8. The server device 200 of the second website requests the server device 300 of the first website to access the web page identified by the URL.

[0277] 9. The server device 300 of the first website returns the web page identified by the URL to the server device 200 of the second website.

[0278] 10. The server device 200 of the second website collects product information in real time by acquiring web pages.

[0279] 11. The server device 200 of the second website is initially processed.

[0280] 12. Data conversion of the server device 200 of the second website.

[0281] Through steps 11 and 12, the publishing information required for publishing the product on the second website is obtained.

[0282] 13. The server device 200 of the second website sends the publishing information to the terminal device 100.

[0283] 14. The user confirms to publish the information through the terminal device 100.

[0284] 15. The terminal device 100 requests the server device 200 of the second website to publish the product.

[0285] 16. The server device 200 of the second website publishes the product on the second website.

[0286] Taking the first object as a product as an example, Figure 7 A schematic diagram of the service architecture of the server device side provided in one embodiment of the present application. Figure 7As shown, the server device includes a data collection module 701, a data processing module 702 and a data conversion module 703. Among them, the data collection module 701 is used to realize: commodity ownership verification, real-time information acquisition, multi-terminal information acquisition and other functions, specifically including: responsible for the commodity data collection of the e-commerce website, commodity ownership verification is required before collection, and whether the collected commodity belongs to the user. Only after the verification is passed will the collection continue, and the data acquisition problem is solved through verification code recognition, proxy, cookie planting and other technologies, and multi-terminal collection is implemented on the APP (Application) side, mobile side and web side to collect target commodities and obtain the original information of commodities. The data processing module 702 is used to realize: multi-language translation, currency conversion, picture transfer, picture translation and other functions, specifically including: preliminary processing of commodity information, text translation: translation of non-English texts such as titles and descriptions in commodity information into English, currency conversion: conversion of currency into US dollars. Image transfer: Since the collected product images exist on the CDN of the first website, they cannot be used directly when the product is released. It is necessary to download the images and upload them to the CDN of the second website, including the main image of the product, the sku image and the product detail image. The data conversion module 703 is used to implement the following functions: category prediction, attribute mapping, detailed description text generation, logistics information generation, text optimization and SKU information generation, specifically including: converting the product model outside the site (i.e. the first website) into the product model inside the site (i.e. the second website). The product information of other e-commerce platforms cannot be used directly on this e-commerce platform, such as inconsistent price types, inconsistent category structures, and inability to map attributes. With the help of basic capabilities such as same-item matching and large models, category prediction, attribute mapping, detailed description generation and other operations are performed, converted into the product model inside the site, and the form is filled. The user can click to publish. Among them, the detailed description text generation can be generated using a large language model according to the multimodal information of the product, and the text optimization may include title optimization, for example: the title and multimodal information of the product are input into the large language model, so that the large language model can optimize the title in combination with the multimodal information.

[0287] Among them, some functions of the data processing module 702 can be realized by calling the text translation engine 704 and the image translation engine 705, and some functions of the data conversion module 703 can be realized by calling the large language model 706 and the same-item matching model 707.

[0288] Figure 7 The specific implementation of each module and each engine and the specific interaction process can be found in the corresponding contents of the above embodiments, which will not be described in detail here.

[0289] Figure 8A flowchart of a method for determining information categories provided in an embodiment of the present application. The execution subject of all steps in the method described in this embodiment may be a server device in the above system. Of course, if the terminal device has strong computing power, the execution subject of the method in this embodiment may also be a client, or the client and the server may execute the method in collaboration (i.e., some steps are executed by the client, and other steps are executed by the server). Figure 8 As shown, the method may include the following steps:

[0290] 801. Obtain object information of a first object.

[0291] The first object is a published object on a first website, and the object information of the first object comes from the first website.

[0292] 802. Filter similar objects similar to the first object from second objects published on a second website according to the object information of the first object;

[0293] 803. Determine the publishing category to which the first object belongs on the second website according to the category to which the similar object belongs on the second website.

[0294] It should be noted here that: for the contents of each step not fully described in detail in the method provided in the embodiment of the present application, please refer to the corresponding contents in the above embodiment, which will not be repeated here. In addition, in addition to the above steps, the method provided in the embodiment of the present application may also include other parts or all of the steps in the above embodiments, which can be specifically referred to the corresponding contents in the above embodiments, which will not be repeated here.

[0295] Fig. 9 The flowchart of the information display method provided in one embodiment of the present application. The execution subject of all the steps in the method described in this embodiment may be the server device in the above system. Of course, if the terminal device has strong computing power, the execution subject of the method in this embodiment may also be the client, or the client and the server may execute the method in collaboration (i.e., some steps are executed by the client, and other steps are executed by the server). Fig. 9 As shown, the method may include the following steps:

[0296] 901. Display a first input box and a second input box.

[0297] 902, receiving a target URL of a first object on a first website input in the first input box and identity information of a user registered on the first website input in the second input box;

[0298] 903. When the first object belongs to the user, publishing information required for publishing the first object on a second website is displayed.

[0299] The publishing information is generated based on the object information of the first object, the object information of the first object is determined based on the webpage data of the target webpage, and the target webpage is the webpage identified by the target URL;

[0300] 904. When the first object does not belong to the user, display a first prompt message.

[0301] The first prompt information is used to prompt that the first object does not belong to the user.

[0302] In the above 901, the first input box and the second input box can be displayed in the same interface or in different interfaces. In an optional implementation, the above 901 may include: displaying a first interface, the first interface including a first input box; in response to receiving the target URL input in the first input box, displaying a second interface, the second interface including the second input box.

[0303] It should be noted here that: for the contents of each step not fully described in detail in the method provided in the embodiment of the present application, please refer to the corresponding contents in the above embodiment, which will not be repeated here. In addition, in addition to the above steps, the method provided in the embodiment of the present application may also include other parts or all of the steps in the above embodiments, which can be specifically referred to the corresponding contents in the above embodiments, which will not be repeated here.

[0304] Fig.10 FIG. 1 is a schematic diagram showing the structure of an electronic device provided by an embodiment of the present application. Fig.10 As shown, the electronic device includes a memory 1101 and a processor 1102. The memory 1101 can be configured to store various other data to support operations on the electronic device. Examples of these data include instructions for any application or method for operating on the electronic device. The memory 1101 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read only memory (EEPROM), an erasable programmable read only memory (EPROM), a programmable read only memory (PROM), a read only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0305] The memory 1101 is used to store programs;

[0306] The processor 1102 is coupled to the memory 1101 and is used to execute the program stored in the memory 1101 to implement the methods provided by the above-mentioned method embodiments.

[0307] Further, if Fig.10 As shown, the electronic device also includes: a communication component 1103, a display 1104, a power component 1105, an audio component 1106 and other components. Fig.10 Only some components are shown schematically, which does not mean that the electronic device only includes Fig.10 Components shown.

[0308] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a computer, the steps or functions of the methods provided in the above-mentioned method embodiments can be implemented.

[0309] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, can implement the steps or functions of the methods provided in the above-mentioned method embodiments.

[0310] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0311] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM (Read Only Memory) / RAM (Random Access Memory), a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0312] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for generating information, characterized in that: include: Acquire object information of a first object, where the first object is a published object on a first website, and the object information of the first object comes from the first website; According to the object information of the first object, selecting similar objects similar to the first object from second objects published on the second website; Determining, according to the category to which the similar object belongs on the second website, the publishing category to which the first object belongs on the second website; Publishing information required for publishing the first object on the second website is generated according to the publishing category and the object information of the first object.

2. The method according to claim 1, characterized in that: The object information of the first object includes multimodal description information of the first object; The step of screening out similar objects similar to the first object from second objects published on a second website according to the object information of the first object includes: Vectorizing the multimodal description information of the first object to obtain a vector for representing the first object; Obtaining a vector for representing the second object; Based on the vector used to characterize the first object and the vector used to characterize the second object, a similar object similar to the first object is determined.

3. The method according to claim 2, characterized in that The multimodal description information of the first object includes: first text information and first image information; the vector used to represent the first object includes: a first text vector obtained by vectorizing the first text information and a first image vector obtained by vectorizing the first image information; the multimodal description information of the second object includes: second text information and second image information, and the vector used to represent the second object includes: a second text information vector obtained by vectorizing the second text information and a second image vector obtained by vectorizing the second image information; and Determining a similar object similar to the first object based on the vector used to characterize the first object and the vector used to characterize the second object includes: Determining a first similarity between the first object and the second object according to the similarity between the first text vector and the second text vector; determining a second similarity between the first object and the second object according to the similarity between the first image vector and the second image vector; The similar objects are determined according to the first similarity and the second similarity.

4. The method according to any one of claims 1 to 3, characterized in that The number of the similar objects is multiple; and determining the publishing category to which the first object belongs on the second website according to the category to which the similar objects belong on the second website includes: determining the categories to which the multiple similar objects belong on the second website as multiple candidate categories; generating a first prompt word according to the object information of the first object and the multiple candidate categories, wherein the first prompt word is used to prompt the large language model to select a target category to which the first object belongs from the multiple candidate categories; Inputting the first prompt word into the large language model to obtain the target category output by the large language model; The target category is determined to be the publishing category to which the first object belongs on the second website.

5. The method according to claim 4, characterized in that The generating a first prompt word according to the object information of the first object and the plurality of candidate categories includes: generating a plurality of first examples according to the object information of the plurality of similar objects and the categories to which the plurality of similar objects belong on the second website; A first prompt word is generated according to the object information of the first object, the multiple candidate categories, and the multiple first examples.

6. The method according to any one of claims 1 to 3, characterized in that Generating publishing information required for publishing the first object on the second website according to the publishing category and the object information of the first object, including: Determining a target attribute name to be configured for the first object according to a plurality of preset attribute names pre-configured by the second website for the publishing category; Determine, according to original attribute information configured by the first website for the first object, a target attribute value corresponding to the target attribute name, wherein the original attribute information includes an original attribute name and an original attribute value corresponding to the original attribute name; The publishing information required for publishing the first object on the second website is generated according to the publishing category, the target attribute name, and the target attribute value corresponding to the target attribute name.

7. The method according to claim 6, characterized in that The determining the target attribute value of the target attribute name according to the original attribute information configured by the first website for the first object includes: Obtaining multiple optional values ​​pre-configured by the second website for the target attribute name; Determine a reference attribute name matching the target attribute name from the original attribute name; According to the similarity between the attribute value corresponding to the reference attribute name and the multiple optional values, a target attribute value corresponding to the target attribute name is determined from the multiple optional values.

8. The method according to claim 6, characterized in that Determining a target attribute name to be configured for the first object according to a plurality of preset attribute names pre-configured by the second website for the publishing category includes: Generate a second prompt word according to the object information of the first object, the original attribute name configured by the first website for the first object, and the plurality of preset attribute names, wherein the second prompt word is used to prompt the large language model to filter out a target attribute name to be configured for the first object from the plurality of preset attribute names; The second prompt word is input into the large language model to obtain the target attribute name output by the large language model.

9. The method according to claim 8, characterized in that The step of generating a second prompt word according to the object information of the first object and the plurality of preset attribute names includes: Determine a reference object from the similar objects, wherein the reference object belongs to the target category; generating a second instance according to the object information of the reference object and the attribute name configured by the second website for the reference object; A second prompt word is generated according to the object information of the first object, the plurality of preset attribute names and the second example.

10. The method according to any one of claims 1 to 3, characterized in that Also includes: Get the target uniform resource identifier URL of the first object; According to the target URL, obtaining web page data of the target web page; The target webpage is identified by the target URL, the target webpage belongs to the first website, and the target webpage is used to display the first object; According to the webpage data, object information of the first object is determined.

11. The method according to any one of claims 1 to 3, characterized in that According to the target URL, the webpage data of the target webpage is obtained, including: Obtaining identity information provided by the user registered on the first website; Determining whether the first object belongs to the user according to the identity information and the target URL; When the first object belongs to the user, webpage data of the target webpage is acquired according to the target URL.

12. The method according to claim 11, characterized in that Also includes: When the first object does not belong to the user, first prompt information is sent to the user, where the first prompt information is used to prompt that the first object does not belong to the user.

13. The method according to claim 11, characterized in that Determining whether the first object belongs to the user according to the identity information and the target URL includes: Acquire, according to the identity information, a URL of at least one object belonging to the user on the first website; If the target URL exists in the URL of the at least one object, determining that the first object belongs to the user; If the target URL does not exist in the URL of the at least one object, it is determined that the first object does not belong to the user.

14. The method according to any one of claims 1 to 3, characterized in that The first object is a commodity.

15. An information publishing method, characterized in that: include: Acquire object information of a first object, where the first object is a published object on a first website, and the object information of the first object comes from the first website; According to the object information of the first object, selecting similar objects similar to the first object from second objects published on the second website; Determining, according to the category to which the similar object belongs on the second website, the publishing category to which the first object belongs on the second website; generating, according to the publishing category and the object information of the first object, publishing information required for publishing the first object on the second website; Based on the publishing information, the first object is published on the second website.

16. The method according to claim 15, characterized in that Publishing the first object on the second website based on the publishing information includes: Sending the published information to the user, A publishing request for the publishing information from the user is received, and the first object is published on the second website according to the publishing information.

17. The method according to claim 16, characterized in that Based on the publishing information, publishing the first object on the second website further includes: receiving a publishing request from the user for the modified publishing information, and publishing the first object on the second website according to the modified publishing information; The modified published information is obtained by the user modifying the published information.

18. A method for determining information categories, characterized in that: include: Acquire object information of a first object, where the first object is a published object on a first website, and the object information of the first object comes from the first website; According to the object information of the first object, selecting similar objects similar to the first object from second objects published on the second website; According to the category to which the similar object belongs on the second website, the publishing category to which the first object belongs on the second website is determined.

19. An information display method, characterized in that: include: Display a first input box and a second input box; Receive a target URL of the first object on the first website input in the first input box and identity information of the user registered on the first website input in the second input box; When the first object belongs to the user, displaying publishing information required for publishing the first object on the second website, the publishing information being generated based on object information of the first object, the object information of the first object being determined based on web page data of a target web page, the target web page being a web page identified by the target URL; When the first object does not belong to the user, first prompt information is displayed, where the first prompt information is used to prompt that the first object does not belong to the user.

20. The method according to claim 19, characterized in that Display the first input box and the second input box, including: Display a first interface, wherein the first interface includes a first input box; In response to receiving the target URL input in the first input box, a second interface is displayed, where the second interface includes the second input box.

21. An information processing system, characterized in that: include: The terminal device is used to send an information generation request to the server device, wherein the information generation request is used to request generation of publishing information required for publishing a first object on a second website, wherein the first object is a published object on the first website; The server device is configured to obtain object information of a first object in response to the information generation request, where the object information of the first object comes from the first website; According to the object information of the first object, similar objects similar to the first object are screened from the second objects published on the second website; according to the category to which the similar objects belong on the second website, the publishing category to which the first object belongs on the second website is determined.

22. The system according to claim 21, characterized in that The server-side device is further used to generate publishing information required for publishing the first object on the second website according to the publishing category and the object information of the first object.

23. The system according to claim 22, characterized in that The server device is further configured to publish the first object on the second website based on the publishing information.

24. An electronic device, characterized in that: include: A memory and a processor, wherein The memory is used to store programs; The processor, coupled to the memory, is configured to execute the program stored in the memory to implement the method according to any one of claims 1 to 20.

25. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a computer, the method according to any one of claims 1 to 20 can be implemented.

26. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 20 is implemented.

Images