A commodity description information generation method and device

Abstract

This disclosure provides a method and apparatus for generating product description information, relating to computer technology and data processing, particularly big data. The specific implementation involves: extracting product keywords from historical product search information; searching for target products based on these keywords; and generating product description information for the target product based on its product information and historical search information. This approach reduces the complexity of user operations, improves the efficiency of generating product description information, and the product description information determined jointly by product information and historical search information is more unique, thus improving its relevance to user needs.

Description

Technical Field

[0001] This disclosure relates to the fields of computer technology and data processing, and in particular to the field of big data. Background Technology

[0002] In existing technologies, if a user wants a description of a product, they can give precise instructions to the large model using prompts, such as telling the model what the product is, who the target customer is, and what the core advantages are, to guide the model in generating product description information in the desired style. However, this method suffers from serious homogenization problems, resulting in low-quality product description information and low efficiency in generating such information. Summary of the Invention

[0003] This disclosure provides a method and apparatus for generating product description information, so as to improve the efficiency of generating product description information.

[0004] According to one aspect of this disclosure, a method for providing product description information is provided, comprising: Extract product keywords from historical search information; Search for target products based on the product keywords; Based on the product information of the target product and the product's historical search information, product description information of the target product is generated.

[0005] According to another aspect of this disclosure, a product description information generation apparatus is provided, comprising: The extraction unit is configured to extract product keywords from historical product search information; The search unit is configured to search for target products based on the product keywords; The generation unit is configured to generate product description information for the target product based on the product information of the target product and the product's historical search information.

[0006] According to another aspect of this disclosure, an electronic device is provided, comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the methods described in the embodiments of this disclosure.

[0007] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the methods described in embodiments of this disclosure.

[0008] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the methods described in the embodiments of this disclosure.

[0009] This disclosure can extract product keywords from historical search information, then search for target products based on these keywords, and generate product descriptions based on the target product's information and historical search information. On one hand, the user does not need to input any suggestions for the target product during the entire generation process, reducing the complexity of user operations and improving the efficiency of generating product descriptions. On the other hand, the product descriptions determined based on both product information and historical search information are more unique, meaning they better match user needs, thus improving the quality of the product descriptions. Furthermore, this disclosure generates product descriptions based on historical search information rather than a specific category of target products specified by the user, broadening the scope and enriching the range of target products.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0011] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure.

[0012] Figure 1 This is the system architecture diagram to which this disclosure applies.

[0013] Figure 2 This is a flowchart of the product description information generation method disclosed herein.

[0014] Figure 3 This is a diagram illustrating the extraction of product keywords provided in this public document.

[0015] Figure 4 This is a schematic diagram of obtaining candidate products provided in this disclosure.

[0016] Figure 5 This is a schematic diagram of the process of selecting target products from candidate products, as provided in this publication.

[0017] Figure 6 This is a schematic diagram of generating product description information provided in this disclosure.

[0018] Figure 7 This is a schematic block diagram of the product description information generation device provided in this disclosure.

[0019] Figure 8This is a block diagram of an electronic device used to implement the product description information generation method of the present disclosure embodiments. Detailed Implementation

[0020] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0021] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0023] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0024] Technology for generating product descriptions already exists. Users can leverage general-purpose models (such as ChatGPT, qwen, and deepseek) to directly generate product descriptions for a specific product. Specifically, users guide the general-purpose model to generate copy in the desired style by providing precise instructions. First, prompts can be generated to indicate what the product is, who the target customer is, what the core advantages are, and to suggest a specific user identity, such as a "senior beauty blogger," a "professional fitness coach," or a "helpful mom." These prompts can also indicate which parts the generated product description should include (such as an "attractive title," "pain point introduction," "product demonstration," "user experience," "effect comparison," "promotional information," and "call to action"), and specify the platform where the product description will be published to generate a description matching the style of that platform.

[0025] It can be seen that for the same product, the product description information generated in this way is quite similar, that is, there is a serious problem of homogenization. When the prompt words are not complete, the general model affects the quality of the generated product description information due to the lack of guidance. In addition, this method requires users to spend a lot of time writing prompt words, which reduces the generation efficiency.

[0026] In view of this, this disclosure provides a new approach. To facilitate understanding of this disclosure, the system architecture on which this disclosure is based will first be described. Figure 1 Exemplary system architectures that can be applied to embodiments of this disclosure are shown, such as Figure 1 As shown, the system architecture may include: a client and a server.

[0027] The server side and the client side are the two main components of an application service. The server side uses a server as its primary hardware infrastructure and may include one or more software service modules. The server side and the client side form a collaborative front-end and back-end.

[0028] The client can be set on the terminal device. In this embodiment of the disclosure, the client can be a local application, a mini-program, or a web application running through a browser on the terminal device.

[0029] Terminal devices can include, but are not limited to, smart mobile terminals, wearable devices, PCs (Personal Computers), and smart home devices. Smart mobile devices can include devices such as mobile phones, tablets, laptops, PDAs (Personal Digital Assistants), and connected car terminals. Wearable devices can include devices such as smartwatches, smart glasses, smart bracelets, VR (Virtual Reality) devices, AR (Augmented Reality) devices, and mixed reality devices (devices that support both virtual and augmented reality). Smart home devices can include devices such as smart TVs and smart refrigerators with displays.

[0030] A server can be a single server, a server cluster consisting of multiple servers, or a cloud server. A cloud server, also known as a cloud computing server or cloud host, is a hosting product in the cloud computing service system, designed to address the shortcomings of traditional physical hosts and Virtual Private Servers (VPS) services, such as high management difficulty and weak service scalability.

[0031] It should be understood that Figure 1 The number of client and server components shown is merely illustrative. Depending on implementation needs, there can be any number of client and server components.

[0032] As one embodiment, a user can enter search information on a search page displayed on the user's device. The user sends the search information to the server, which uses the search information as search criteria to obtain search results and returns them to the user. During this process, the server can collect and store search information, obtaining historical search information for the product (i.e., historical search information related to the product). Then, the server extracts product keywords from the historical search information, searches for target products based on these keywords, and generates a product description based on the target product's information and the historical search information. This product description can then be sent to the user for display.

[0033] Figure 2 This is a flowchart of a product description information generation method provided in this disclosure embodiment. This product description information generation method can be... Figure 1 The server-side execution in the system shown. For example... Figure 2 As shown, the method may include the following steps: Step 201: Extract product keywords from historical product search information.

[0034] Step 202: Search for the target product based on product keywords.

[0035] Step 203: Generate product description information for the target product based on the product information and historical search information of the target product.

[0036] As can be seen from the above process, this disclosure can extract product keywords from historical product search information, then search for target products based on these keywords, and generate product description information for the target products based on their product information and historical search information. On the one hand, the user does not need to input any prompts for the target product during the entire generation process, reducing the complexity of user operations and improving the efficiency of generating product description information. On the other hand, the product description information determined jointly by product information and historical search information is more unique, meaning it improves the matching degree with user needs, thereby improving the quality of the product description information. Furthermore, this disclosure generates product description information based on historical search information rather than a specific category of target products specified by the user, broadening the scope of target products and increasing their richness.

[0037] The following describes in detail each step of the above process and the effects that can be further produced, with reference to the embodiments.

[0038] First, the above step 201, namely "extracting product keywords from historical product search information", will be described in detail with reference to the embodiments.

[0039] Search information refers to the text string that a user enters to request specific information or content from a search engine or website. Product search history information can refer to search content entered by the user that is directly or indirectly related to product information, purchase, comparison, or evaluation. Product keywords extracted from product search history information can refer to information such as the product's name, brand, and model.

[0040] The server can collect and store search information entered through the search page provided by the user. When the stored search information reaches a certain amount or the collection time reaches a preset time, the server can obtain the product's historical search information based on the collected search information or the search information collected within the preset time. The user can refer to shopping platforms, search software, social media applications, etc., and the obtained product historical search information can refer to one or multiple items.

[0041] As a preferred approach, product search history information can be obtained based on the classification of previously collected search information or search information collected within a preset time period. Specifically, the server can classify the previously collected search information or search information collected within a preset time period to obtain search information under each search category. Here, search categories can refer to product categories, manufacturing categories, book categories, animal science categories, etc. Search information under the product category can be directly used as product search history information. Of course, search information under search categories related to the product (such as book categories) can also be used as product search history information.

[0042] Then, the server can directly extract product keywords from the product's historical search information. For example, the product's historical search information can refer to "women's long winter down jackets", "high-performance mobile phone recommendations", "fast food recommendations", etc. The product keyword extracted by the server from "women's long winter down jackets" can be "down jacket", the product keyword extracted from "high-performance mobile phone recommendations" can be "mobile phone", and the product keyword extracted from "fast food recommendations" can be "fast food".

[0043] In practical applications, there is often a large amount of historical search information for products. In order to extract product keywords from the historical search information of more popular products in the future, the server can filter the historical search information of products based on search volume to improve the quality of the historical search information of products.

[0044] Specifically, the search volume of each product's historical search information is obtained. For product historical search information whose search volume meets preset conditions, the step of extracting product keywords is performed. One feasible approach is to use a search volume threshold as the preset condition; product keywords can only be extracted when the search volume of a product's historical search information exceeds the threshold. Another feasible approach is to use the top N preset conditions, where N is a positive integer. That is, all product historical search information is sorted from highest to lowest search volume, and product keywords are extracted from the top N products.

[0045] For example, consider two historical search results for a product: "Women's Long Down Jacket in Winter" and "Summer Short-Sleeve Jacket." "Women's Long Down Jacket in Winter" has been searched 10,000 times, while "Summer Short-Sleeve Jacket in Summer" has been searched 100 times. If the search volume threshold is 1,000 times, then the historical search result that meets the preset criteria is "Women's Long Down Jacket in Winter." Subsequent steps only require extracting product keywords from "Women's Long Down Jacket in Winter." It can be seen that this method can filter historical search results for products with a wider impact and greater user demand. For instance, if the current time is winter, the search volume for "Women's Long Down Jacket in Winter" is higher, and subsequent steps can generate product descriptions that match the target product at the current time or the product descriptions of currently popular target products.

[0046] Furthermore, while the above methods can filter historical search information for products based on search volume, they also contain a large amount of semantically similar historical search information. For example, "cost-effective mobile phones in 2025" and "What cost-effective mobile phones are recommended in 2025" actually express similar meanings. Extracting historical search information for semantically similar products separately may lead to a waste of resources.

[0047] As a preferred embodiment, the server can determine the semantic features of each historical search result for a product. Based on these features, the historical search results are clustered (using algorithms such as K-Means or DBSCAN) to obtain clustering results. Then, target search information is determined from the historical search results corresponding to each cluster category. Subsequently, product keywords can be extracted based on the target search information. In short, semantically similar historical search results are grouped together, and the most representative historical search results from each cluster category are selected as target search information. Product keywords are then extracted from the target search information. This avoids redundant processing of similar historical search results, improving resource utilization. Furthermore, deduplication of historical search results before keyword extraction avoids generating homogenized product descriptions, further enhancing the differentiation of product descriptions and their relevance to user needs.

[0048] When extracting product keywords, one feasible method is to use a pre-defined keyword dictionary (such as brand, model, category, etc.) to directly extract product keywords from historical search information through string matching or regular expressions.

[0049] As another possible approach, such as Figure 3As shown, historical search information for products can be input into the primary search engine (such as deepseek-v3), and prompts can be used to guide the primary search engine's search intent based on this historical information, thereby extracting product keywords from the historical search data. This method of extracting product keywords using the primary search engine can significantly improve the accuracy of extraction. Specifically, the primary search engine can fully understand the search intent behind the historical search information to extract accurate product keywords. Furthermore, compared to direct extraction via string matching or regular expressions, using the primary search engine can extract more diverse product keywords, avoiding omissions.

[0050] The aforementioned extraction prompts can also be used to prompt the first major model to determine whether the product's historical search information is suitable for extracting product keywords, and to extract product keywords from the product's historical search information that is suitable for extracting product keywords.

[0051] Next, this disclosure provides a specific example of extracting prompt words. That is, the prompt words extracted for the first model can be: "You are a product creator who excels at using historical product search information to decide whether a product is suitable for recommendation. If it is suitable, you generate appropriate product keywords for recommendation."

[0052] The requirements are as follows: First, determine whether the product's historical search information is relevant to the product. If not, output "Do not recommend the product". Then, determine whether the product's historical search information is a positive description. If it is a positive description, extract the brand, model, or product name of the recommended product based on the historical search information; if it is a negative description, directly output "Do not recommend the product".

[0053] Additionally, if there are multiple recommended products in the product search history, separate the brand, model, or product name of each product with a comma. Among the extraction prompts provided above, the first model can be positioned as a product creator, prompting it to output product keywords from historical search information. Specifically, "determining whether the historical search information is a positive recommendation for the product" involves analyzing the search intent behind the historical search information. For example, the historical search information "winter down jacket recommendations" can be considered a positive description of the product, making it suitable for extracting product keywords. Conversely, the historical search information "cookies to avoid" can be considered a negative description of the product, making it unsuitable for extracting product keywords.

[0054] Additionally, you can add "If the product's historical search information involves comparisons of two or more products, it can also be considered a positive description of the product" to the extraction prompts. For example, the product's historical search information "Which is better, brand 1's mobile phone or brand 2's mobile phone?" is suitable for extracting product keywords.

[0055] In practical applications, positive and negative examples can be added at the end of the extracted prompt words to indicate that the first major model provides accurate extraction results based on the examples. For example, a positive example could be that the product keyword extracted from "winter down jacket recommendations" is "down jacket", and a negative example could be that the product keyword extracted from "winter down jacket recommendations" is "winter".

[0056] The following describes step 202, namely "searching for target products based on product keywords", in detail with reference to the embodiments.

[0057] In this embodiment of the disclosure, the server uses the product keywords extracted in step 201 as search terms to search for target products from a preset product library. For example, based on the product keywords, it matches the product names of products included in the preset product library and selects products whose product names include the product keywords as target products.

[0058] To obtain high-quality target products, you can first filter products whose names include product keywords. Specifically, you can first search for candidate products based on product keywords, as follows: Figure 4 As shown, the system matches product names with product keywords from a pre-defined product database. Products whose names contain the product keywords are selected as candidate products. The server can also limit the number of candidate products, for example, matching a preset number of candidate products. Then, the target product is selected from the candidate products based on at least one of the following: the similarity between the candidate product name and the product keywords, and the product's sales data.

[0059] When filtering candidate products based on the similarity between the product name and the product keywords, it is actually to determine the similarity between the product keywords and the product name of each candidate product. The candidate products with a similarity greater than the similarity threshold are selected as the target products. Alternatively, the candidate products can be sorted from high to low according to the similarity, and the top M candidate products are selected as the target products, where M is a positive integer.

[0060] The similarity can be obtained using edit distance, and its calculation process can be expressed by the following formula: (Formula 1) For similarity, For product keywords, The product name of the candidate product. The edit distance between product keywords and candidate product names. The length of the product keywords, The length of the candidate product name.

[0061] When filtering candidate products based on sales data, the process essentially involves ranking the candidates based on their sales data (such as sales volume, price, commission, etc.) and selecting the top K candidates as the target products, where K is a positive integer. For example, the weights of different sales data can be adjusted for different scenarios and needs. Based on different sales data and their corresponding weights, a score is determined for each candidate product. The candidate products are then ranked based on their scores. For instance, during promotional events, the weight of sales volume can be higher, and if users want a higher commission rate, the weight of commission can be higher.

[0062] As a preferred embodiment, such as Figure 5 As shown, candidate products and their corresponding product information (such as product name, product ID, sales volume, price, commission, platform, cover image, etc.) can be obtained by searching based on product keywords. First, candidate products are filtered based on the similarity between the product name and product keywords to obtain the filtered candidate products. Then, the filtered candidate products are sorted based on sales data, and the top K candidate products are retained as target products.

[0063] In addition, the purpose of this disclosure is actually to realize the process of automatically generating product description information of target products. After obtaining the product description information of the target products, users can publish the product description information of the target products on the corresponding platforms for sales. Therefore, when obtaining candidate products from the preset product library, it is also possible to search based on product keywords and the platforms that the user has opened to obtain candidate products and their corresponding product information. In this way, the obtained candidate products are the products that users can sell. Different users have different candidate products, which further meets the personalized needs of users.

[0064] This approach can improve the accuracy of the target products obtained, reduce invalid searches, ensure a high degree of matching between the target products and user needs, and reduce the subsequent computational burden by screening candidate products, thereby improving resource utilization.

[0065] It should be noted that if a product search history includes one product keyword, the target product corresponding to that product keyword is obtained based on that product keyword. If a product search history includes multiple product keywords, the target product corresponding to each product keyword is obtained based on each product keyword.

[0066] The following describes in detail step 203, namely "generating product description information of the target product based on the product information of the target product and the product historical search information", with reference to the embodiments.

[0067] In the embodiments disclosed herein, such as Figure 6 As shown, the product information of the target product and the product's historical search information can be input into a pre-trained model for generating product description information to obtain the product description information of the target product output by the model.

[0068] As a preferred embodiment, historical search information for products and the corresponding product information of target products are input into a second large model (such as deepseek-r1). The second large model obtains and outputs the product description information of the target products based on preset prompts. It should be noted that if several target products are obtained in step 202, the second large model can output the product description information corresponding to each of the several target products.

[0069] This method of generating product description information using the second major model can significantly improve the accuracy of product description information and its matching degree with user needs. That is, the second major model can combine product historical search information to generate relevant product description information, which improves the differentiation and personalization of product description information, thereby improving the quality of product description information. Moreover, the whole process does not require manual editing by users, thus improving generation efficiency.

[0070] To enrich product descriptions and facilitate faster access to key information for other users (i.e., users who view product descriptions posted by other users), the prompts in this disclosure can be used to guide the second model in outputting the title and body text of the product description based on the first preset requirement. For example, given the historical search information "Recommended 2025 household washing machines, high cost-performance ratio, shock absorption to minimize disturbance to neighbors," and the target product obtained from the historical search information is "XX Silent Treadmill," the title output by the second model could be "2025 Treadmill Review! Price Reduced by 300 Yuan!", and the body text could be "XX Silent Treadmill Designed for Small Apartments, Featuring a Dual Shock Absorption System, Noise Level ≤ 45 Decibels (Equivalent to Quiet Conversation), No Complaints from Downstairs Neighbors." It can be seen that the title and body text of the product description are highly relevant to the historical search content, improving the differentiation of the product description and its matching with other users' needs. Furthermore, outputting the product description in a title and body text format allows other users to quickly grasp the key points of the product description, thus increasing its appeal to them.

[0071] It should be noted that if multiple target products are obtained in step 202, the prompt words can be used to prompt the second model to output the title and body text corresponding to each target product, or to prompt the second model to output the title that is common to multiple target products and the body text corresponding to each target product.

[0072] In the above content, the product description information output by the second major model generally refers to the text description information generated based on the product information of the target product (such as product name, product ID, sales volume, price, commission, etc.). In order to further improve the richness of the product description information, if the product information includes a product image, the aforementioned prompt words can also be used to prompt the second major model to determine the output position of the product image in the product description information and insert the product image into the corresponding output position in the product description information. For example, the product image corresponding to the target product can be inserted after the text description information corresponding to the target product, or the product image corresponding to the target product can be inserted after the description related to appearance in the text description information corresponding to the target product.

[0073] Furthermore, to facilitate other users quickly jumping to the resource transfer interface of the corresponding target product, if the product information includes a product card, the product card must at least include resource transfer prompt information. This prompt information is used to trigger the resource transfer interface. Additionally, in this disclosure, the product card may also include other product information, such as product name, thumbnail, inventory, price changes, and reviews. In this case, the prompt can also be used to guide the second major model to determine the output position of the product card in the product description information and insert the product card into the corresponding output position. For example, the product card corresponding to the target product can be inserted after the product image corresponding to the target product, or it can be inserted before the text description information corresponding to the target product.

[0074] This method of inserting product images and product cards into product descriptions enriches the content of product descriptions, reduces the understanding cost for other users by using visual information, and allows other users to quickly jump to the resource transfer interface when they are interested in a target product through resource transfer prompts, shortening their decision-making path, saving them time, and reducing their operational complexity.

[0075] It should be noted that if multiple target products are identified in step 202, the prompt words can also be used to guide the second model to determine a preset number of products to be displayed from among the multiple target products that meet the similarity criteria, and to generate product description information for the products to be displayed. Here, target products that meet the similarity criteria can be understood as any two target products among multiple target products having a similarity greater than a threshold. In other words, for relatively similar target products, a preset number of target products can be determined as products to be displayed, and the second model only needs to output the product description information for the products to be displayed.

[0076] Alternatively, the second model can output product descriptions for all target products, but for target products that meet the similarity criteria, it determines a preset number of products to be displayed, and only inserts the product cards of these products into the corresponding output positions in their respective product descriptions. For example, if mobile phone 1 and mobile phone 2 have the same configuration parameters and models, but differ only in color, then product cards can be inserted into the corresponding output positions in the product descriptions for mobile phone 1, or only into the corresponding output positions in the product descriptions for mobile phone 2.

[0077] Of course, the prompt words can also prompt the second major model to compare product images of multiple target products. If the product images are completely identical, select one of the target products with completely identical product images to insert a product image, without having to insert a product image for every target product.

[0078] This approach avoids excessive repetition in product descriptions, which lowers the quality of the descriptions. Furthermore, it allows for the display of higher-quality product descriptions to other users, enhancing their experience.

[0079] It should also be noted that this disclosure can utilize a second model to further filter target products. Specifically, the product information, historical search information, and keywords of the target product can be input into the second model, and the second model can be prompted to output the product description information of the target product based on preset prompt words. These prompt words can also be used to prompt the second model to judge whether the product information of the target product is correct based on the product keywords. For target products with correct product information, the step of outputting the product description information of the target product is executed.

[0080] In other words, the second model uses prompt words to further filter target products based on semantic features. For example, if the product keyword is "mobile phone", but the product name of the target product is mobile phone case, mobile phone charging cable or headphones, then the product information of the target product is incorrect, that is, the target product is incorrect, and the second model will not output the product description information of the target product.

[0081] In this way, when using the second model to further filter target products, the correctness of the target products can be judged at the semantic level, reducing the generation of low-quality content and improving the accuracy of the output content.

[0082] Next, this disclosure provides a specific example of a prompt word input into the second large model, that is, the prompt word for the second large model can be: "You're a top-performing livestreamer, your content boasts the highest readership and conversion rate across the entire internet. Now you want to promote target products, which may include multiple items. Based on the product information, generate information on product selling points, target audience, user pain points, and applicable scenarios. Using these points, create a product description specifically for the target product. Note that the outline should not completely copy these points, and should even avoid including the specific points listed above. The generated product description includes a title and body text. The title and body text need to incorporate historical search information, analyzing user search intent from this data and highlighting the core elements of that search history."

[0083] Next, the requirements for the title and body text, and the total character count, are outlined. The title should be no more than 25 characters, and the full text no more than 800 characters. The insertion positions for the product image and product card are also specified. The similarity of the target products' content must also be considered. If the input target products are too similar, for example, identical configurations just with different colors, such as "XX Mobile Phone 6000mAh Large Battery Obsidian Black 8G+128G" or "XX Mobile Phone 6000mAh Large Battery Snow White 8G+128G," then only the product card for one of the target products needs to be inserted.

[0084] While generating product descriptions, it's also necessary to determine if the target product is correct. The product name is checked against product keywords. If the product keyword is "mobile phone," but the target product name is something like "phone case," "phone charging cable," or "earphones," it's considered incorrect, and no product description is generated for an incorrect target product. If all target products are incorrect, only the semantically closest target product is used to generate its description. In addition to the above, this disclosure may also add some specific examples to the prompt words to prompt the second model to generate a complete product description for the target product based on the writing format and behavioral style of the examples.

[0085] Based on this, the server can obtain the product description information of the target product generated by the second model and send the product description information of the target product to the user terminal. The user terminal responds by receiving the product description information of the target product, displaying the product description information of the target product, and responding to the user's publishing operation on the product description, publishing the product description information of the target product for other users to browse.

[0086] The technical solutions disclosed herein involve the collection, storage, use, processing, transmission, provision, and disclosure of information such as user personal information, all of which comply with relevant laws and regulations and do not violate public order and good morals.

[0087] The foregoing has described specific embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0088] According to another embodiment, a product description information generation apparatus is provided. Figure 7 A schematic block diagram of a product description information generation apparatus according to one embodiment is shown. The product description information generation apparatus is disposed in... Figure 1 The server side in the illustrated architecture. For example... Figure 7 As shown, the product description information generation device 700 includes an extraction unit 701, a search unit 702, and a generation unit 703. The main functions of each component are as follows: Extraction unit 701 is configured to extract product keywords from historical product search information; Search unit 702 is configured to search for target products based on product keywords; The generation unit 703 is configured to generate product description information for the target product based on the product information and historical search information of the target product.

[0089] As one possible implementation method, when extracting product keywords from historical product search information, the extraction unit 701 can be specifically configured to: perform clustering processing on historical product search information based on similarity to obtain clustering result information; extract target search information from historical product search information corresponding to each cluster category based on the clustering result information; and extract product keywords based on the target search information.

[0090] As one possible approach, when extracting product keywords from historical product search information, the extraction unit 701 can be specifically configured to: input historical product search information into the first model, and extract product keywords from the historical product search information based on the search intent of the historical product search information through the first model.

[0091] As one possible implementation method, when the search unit 702 searches for target products based on product keywords, it can be specifically configured to: obtain candidate products based on product keywords; and select the target product from the candidate products based on at least one of the following: the similarity between the product name and product keywords of the candidate products, and the sales data of the products.

[0092] As one possible approach, when generating product description information of a target product based on its product information and historical search information, the generation unit 703 can be specifically configured to: input the product information and historical search information into the second model, and obtain and output the product description information based on preset prompts through the second model.

[0093] As one possible approach, the product description information includes a title and body text, with prompts used to prompt the second model to output the title based on the first preset requirement and the body text based on the second preset requirement.

[0094] As one possible approach, product information includes product images and product cards. Each product card includes at least resource transfer prompts, which trigger the resource transfer interface. Prompts are used to prompt the second model to determine the output positions of the product images and product cards in the product description information and to insert the product images and product cards into the output positions of the product description information.

[0095] As one possible approach, there are multiple target products. The prompt words are also used to prompt the second model to determine a preset number of products to be displayed from multiple target products that meet the similarity conditions, and to generate product description information for the products to be displayed.

[0096] As one possible implementation method, when the generation unit 703 inputs the product information and historical search information of the target product into the second model, and the second model obtains and outputs the product description information based on preset prompt words, it can be specifically configured as follows: inputting the product information, historical search information, and product keywords of the target product into the second model, and the second model obtains and outputs the product description information of the target product based on preset prompt words; wherein, the prompt words are used to prompt the second model to judge whether the product information of the target product is correct based on the product keywords, and for the target product with correct product information, the step of outputting the product description information of the target product is executed.

[0097] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0098] Figure 8A schematic block diagram of an example electronic device 800 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0099] like Figure 8 As shown, device 800 includes a computing unit 801, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 802 or a computer program loaded from storage unit 808 into random access memory (RAM) 803. RAM 803 may also store various programs and data required for the operation of device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via bus 804. Input / output (I / O) interface 805 is also connected to bus 804.

[0100] Multiple components in device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of monitors, speakers, etc.; storage unit 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0101] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as the product description information generation method. For example, in some embodiments, the product description information generation method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the product description information generation method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the product description information generation method by any other suitable means (e.g., by means of firmware).

[0102] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0103] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0104] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0105] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0106] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0107] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0108] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0109] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for generating product description information, comprising: Extract product keywords from historical search information; Search for target products based on the product keywords; Based on the product information of the target product and the product's historical search information, product description information of the target product is generated.

2. The method according to claim 1, wherein, The extraction of product keywords from historical product search information includes: The historical search information of the products is clustered based on similarity to obtain clustering results. Based on the clustering results, target search information is extracted from the historical search information of the products corresponding to each cluster category; Extract the product keywords based on the target search information.

3. The method according to claim 1, wherein, The extraction of product keywords from historical product search information includes: The product's historical search information is input into the first model, and the first model extracts the product keywords from the product's historical search information based on the search intent of the product's historical search information.

4. The method according to claim 1, wherein, The process of searching for target products based on the product keywords includes: Candidate products are obtained by searching based on the product keywords; The target product is selected from the candidate products based on at least one of the following: the similarity between the product name and the product keywords of the candidate products, and the sales data of the products.

5. The method according to claim 1, wherein, The step of generating product description information for the target product based on the product information and the product's historical search information includes: The product information and the product's historical search information are input into the second model. The second model then obtains and outputs the product description information based on preset prompts.

6. The method according to claim 5, wherein, The product description information includes a title and a body text. The prompt words are used to prompt the second model to output the title based on the first preset requirement and to output the body text based on the second preset requirement.

7. The method according to claim 5, wherein, The product information includes a product image and a product card. The product card includes at least resource transfer prompt information, which is used to trigger the resource transfer interface. The prompt word is used to prompt the second model to determine the output position of the product image and the product card in the product description information, and to insert the product image and the product card into the output position in the product description information.

8. The method according to claim 5, wherein, There are multiple target products, and the prompt words are also used to prompt the second model to determine a preset number of products to be displayed from multiple target products that meet the similarity conditions, and generate product description information for the products to be displayed.

9. The method according to claim 5, wherein, The step of inputting the product information of the target product and the product's historical search information into the second model, and obtaining and outputting the product description information based on preset prompts through the second model, includes: The product information of the target product, the product's historical search information, and the product's keywords are input into the second model. The second model then obtains and outputs the product description information based on preset prompts. The prompt word is used to prompt the second model to determine whether the product information of the target product is correct based on the product keywords. For target products with correct product information, the step of outputting the product description information of the target product is executed.

10. A product description information generation device, comprising: The extraction unit is configured to extract product keywords from historical product search information; The search unit is configured to search for target products based on the product keywords; The generation unit is configured to generate product description information for the target product based on the product information of the target product and the product's historical search information.

11. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-9.

12. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-9.

13. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-9.

Images