Page information processing method, system and electronic device

Abstract

The embodiment of the application discloses a page information processing method, system and electronic equipment, comprising: in the process of displaying a page in a first network site through the web browser, providing an operation option for viewing commodity demand trend information for a target picture in the page in an operation focus state; after receiving an operation request of a user through the operation option, obtaining information of the same commodity based on the target picture, and generating a commodity demand trend analysis result after calling and analyzing a generative artificial intelligence (AI) model; after receiving an operation request of finding a source through the AI interaction area, initiating a request for commodity search based on the target picture to a second network site, and after determining a recommended purchase commodity from the commodity search result through the generative AI model, displaying the recommended result in the AI interaction area. Through the embodiment of the application, the user can be assisted to quickly make a product selection decision.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to page information processing methods, systems and electronic devices. Background Technology

[0002] In cross-border commodity information service systems (also known as "cross-border e-commerce platforms"), "product selection" is a crucial core strategic element. It refers to a systematic process that, based on market data, consumer insights, competitive analysis, and the company's own resources, selects products with the greatest sales potential or profit margin from a vast array of goods for listing and sale in the company's online store. Product selection is not simply a matter of "picking goods," but rather a "decision-making process" aimed at finding the "right products."

[0003] The specific product selection process typically includes the following steps: 1. Find inspiration: Discover the product demand trends of overseas users by browsing e-commerce platforms (mainly overseas e-commerce platforms) and social media (overseas social media); 2. Data Analysis: For products with good demand trends, it is necessary to analyze the search volume of the product, the intensity of competition, the sales volume of top sellers, and whether there is profit margin. 3. Supply chain assessment: Can suppliers be found (by sourcing goods from domestic B2B (business-to-business) e-commerce platforms)? What are the costs and quality like? 4. Risk assessment: Are there any risks of patent or trademark infringement? Is logistics convenient? 5. Final decision: Based on all the information, decide whether to invest in purchasing the product and then list it in your store.

[0004] In existing technologies, completing the product selection process typically requires manual browsing of products, recording of product information, analysis of reviews, comparison of multiple products, and a final decision on whether to purchase a particular product. This consumes a significant amount of time and manpower for merchants, resulting in low efficiency. Therefore, how to help merchants improve product selection efficiency and reduce costs has become a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0005] This application provides page information processing methods, systems, and electronic devices that can assist users in making quick product selection decisions.

[0006] This application provides the following solution: A method for processing page information, the method being applied to a web browser plugin, comprising: During the process of displaying a page from the first website through the web browser, an operation option is provided for viewing product demand trend information for the target image that is in the focus state on the page; After receiving the user's operation request through the operation options, the system obtains information about the same product based on the target image, calls the generative artificial intelligence (AI) model to analyze it, generates a product demand trend analysis result, and then displays the trend analysis result in the AI ​​interaction area of ​​the web browser window. After receiving a request to find sources of goods through the AI ​​interaction area, the system initiates a request to the second website to search for goods based on the target image, displays the search results for goods in the second website, and determines the recommended goods for purchase from the search results through the generative AI model, and then displays the recommendation results in the AI ​​interaction area.

[0007] Wherein, the first website is a website related to overseas product information services, a social network website targeting overseas users, or a content system targeting overseas users, and "overseas" is relative to the country / region to which the user belongs; The second website is a site related to product information services where both parties to the transaction are located within the user's country / region.

[0008] The provision of operation options for viewing product demand trend information for the target image that is in the focus state on the first page includes: When the mouse hovers over an image, that image is designated as the target image in the operation focus state, and the operation option for viewing product demand trend information is provided at the location of the target image.

[0009] This also includes: After receiving an operation request through the operation option for viewing product demand trend information, the target image is uploaded to the server to be converted into a standardized image link address, so as to obtain the same product and search for products based on the target image using the standardized image link address.

[0010] The step of initiating a request to the second network site to perform a product search based on the target image includes: Send an instruction to the web browser to open the target page of the second website; After displaying the target page, the location information of the search input control and search operation options on the target page is obtained, and the link address of the target image is filled into the search input control. Then, the operation of triggering the search operation option is executed to initiate the product search based on the target image.

[0011] The generative AI model includes an AI language model and an AI vision model. The step of obtaining the location information of the search input control and search operation options on the target page includes: After displaying the target page, obtain the Document Object Model (DOM) information of the target page, as well as a screenshot of the page; The DOM information and prompt word information are provided to the AI ​​language model so that the AI ​​language model can locate the position information of the search input control by analyzing the DOM information and return it. The page screenshot and prompt information are provided to the AI ​​visual model so that the AI ​​visual model can locate the position information of the search operation option by analyzing the page screenshot and return it.

[0012] Wherein, determining the recommended purchase product from the product search results using the generative AI model includes: After displaying the product search results page in the browser, the DOM information of the product search results page is obtained from the browser, and the DOM information and prompt word information are provided to the generative AI model so that the generative AI model can extract recommended products for purchase based on the analysis of the DOM information.

[0013] Specifically, when determining recommended products from the product search results using a generative AI model, the prompts provided to the AI ​​model include priority rules for various ranking indicators. This allows the AI ​​model to sort the products in the search results according to the priority ranking indicators and then determine the recommended products.

[0014] This also includes: After displaying the recommended products, the system receives a request to perform a detailed analysis of one of the target products and sends an instruction to the browser to open the product details page corresponding to the target product. After displaying the product details page, a generative AI model is used to analyze the details of the target product, and the analysis results are displayed in the AI ​​interaction area.

[0015] Specifically, when analyzing the detailed information of the target product using a generative AI model, the information input into the generative AI model includes: Valid information extracted from the DOM information of the product details page, and / or a screenshot of the product details page.

[0016] The analysis of the target product's details using a generative AI model includes: The browser is requested to obtain the DOM information of the product details page and the location information of the rating operation option, which is used to expand the user rating content of the target product; The DOM information and prompt word information are provided to the generative AI model so that the generative AI model can analyze the key value points of the product based on the DOM information. Based on the location information of the evaluation operation option, a command is sent to the browser to scroll the page to the location of the evaluation operation option, and the operation of triggering the evaluation operation option is executed to obtain the user's evaluation content. User reviews and prompts are provided to a generative AI model so that the model can summarize the user reviews.

[0017] This includes providing user reviews and prompts to the generative AI model, specifically: The system provides the default positive reviews displayed in the user review content display area to the generative AI model, locates the negative review tags in the user review content display area, executes the operation to trigger the negative review tags, obtains the negative review content of the target product, and provides the negative review content to the generative AI model so that the generative AI model can analyze and summarize the positive and negative review content.

[0018] Specifically, a screenshot of the user review content area is provided to the generative AI model so that the generative AI model can perform user review content analysis based on the screenshot.

[0019] This also includes: During the analysis of the generative AI model based on page screenshots, instructions are sent to the web browser to scroll the page or click on the target coordinates on the page, based on the information fed back by the generative AI model. The page screenshot is then taken again and input into the generative AI model for analysis. In this way, page information is gradually provided to the AI ​​model through multiple rounds of interaction with the generative AI model.

[0020] This also includes: During the process of displaying the product details page of the target product on the second website, options are provided for publishing the same product on a third website; After receiving the user's operation request through this operation option, the product details information of the target product is provided to the generative AI model. The generative AI model then extracts information related to product release from the product details information and fills it into the product release form of the third network site.

[0021] A page information processing system, comprising: Web browser plugins and generative AI models; The web browser plugin is used to receive requests from users for product selection assistance during the process of users browsing pages through web browsers and performing automated operations by interacting with the browser to obtain relevant page data and generate prompt words. It then calls the generative AI model based on the page data and prompt words. The generative AI model is used to process the page data and prompt information to generate content to assist in product selection decisions. The web browser plugin is also used to display the content generated by the generative AI model through the AI ​​interaction area within the browser window.

[0022] The content of the product selection decision-making assistance generated by the generative AI model includes one or more of the following: the analysis results of demand trend analysis of the same product based on the target image, the product search results of finding sources of goods in the second network site based on the target image in the first network site, the recommended purchase product information analyzed from the found sources of goods, and the results of detailed analysis and summary of the recommended purchase products.

[0023] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of any of the preceding methods.

[0024] An electronic device, comprising: One or more processors; and A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of any of the preceding methods.

[0025] A computer program product includes a computer program / computer executable instructions that, when executed by a processor in an electronic device, implement the steps of any of the preceding methods.

[0026] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a web browser plugin that, while displaying a page from a first website through the web browser, offers an option to view product demand trend information for a target image that is the focus of the page. Upon receiving a user's request via this option, the plugin can retrieve information on the same product from the first website and / or other cross-border product information service-related websites based on the target image. After analyzing this information using a generative AI model, a product demand trend analysis result is generated and displayed in the AI ​​interaction area of ​​the web browser window. Similarly, upon receiving a request to find suppliers via the AI ​​interaction area, the plugin can initiate a product search request based on the target image from a second website, displaying the search results from that website. The generative AI model then identifies recommended products from the search results and displays the recommendations in the AI ​​interaction area. This approach, using a plugin and generative AI model, enables automated product demand trend analysis, sourcing, and product recommendations, thus improving the efficiency of product selection decisions.

[0027] Alternatively, upon receiving a request to perform a detailed analysis of a target product, an instruction can be sent to the browser to open the product details page corresponding to the target product. A generative AI model can then analyze the product's details and display the results in the AI ​​interaction area. This allows users to make further product selection decisions from multiple AI-recommended products based on the specific product details analysis results.

[0028] The plugin program provides page information to the generative AI model, including page screenshots. Based on feedback from the generative AI model, it can send instructions to the web browser to scroll the page or click on target coordinates on the page, and then re-capture the page and input it into the generative AI model for analysis. This multi-round interaction gradually provides page information to the generative AI model, enabling on-demand information provision and reducing resource waste. Furthermore, this method is more effective for analyzing pages rendered using Canvas or image stitching.

[0029] Of course, any product implementing this application does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is a schematic diagram of the system architecture and interaction process provided in the embodiments of this application; Figure 2 This is a flowchart of the method provided in the embodiments of this application; Figure 3 This is a schematic diagram of the first interface provided in an embodiment of this application; Figure 4 This is a schematic diagram of the second interface provided in an embodiment of this application; Figure 5 This is a schematic diagram of the interactive process of searching for goods on a second website based on images, provided in an embodiment of this application. Figure 6 This is a schematic diagram of the interactive process for recommending the top 3 best-selling products from product search results, provided in an embodiment of this application. Figure 7 This is a schematic diagram of the third interface provided in an embodiment of this application; Figure 8 This is a schematic diagram of the interactive process for analyzing detailed product information provided in an embodiment of this application; Figure 9 This is a schematic diagram of the fourth interface provided in an embodiment of this application; Figure 10 This is a schematic diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0033] In this embodiment, to facilitate product selection for merchants, a web browser plugin can be provided. Merchants can install this plugin in their web browser to complete the product selection process more conveniently. Specifically, after installing the plugin in the user's browser, while browsing a page on a website (referred to as the first website for easy distinction), a first operation option for viewing product demand trend information can be provided for images that are currently in the focus state (e.g., when the mouse hovers over an image, that image is currently in the focus state). Users can click on this first operation option to request to view the demand trend information of that product. Afterwards, information on similar products related to that image can be obtained. Specifically, similar products can be searched on the current first website, or on other cross-border product information service-related websites, or separately on the first website and other cross-border product information service-related websites. Additionally, information on similar products can be queried through social networking systems, content platforms (mainly service systems such as short video publishing), and other systems, etc. After finding similar products, product demand trend analysis can be performed. Of course, if the same product is found on different websites, the demand trends for that product on each website can be analyzed separately, yielding different trend analysis results. These results can also be compared. Specifically, when searching for the same product, methods such as image comparison can be used. For example, if the image in the focus state is known, related products can then be found in the product databases of the first website and / or other websites.

[0034] In practice, the first online site may be considered an "external system" relative to the product information service system associated with the plugin provider. In this case, data authorization can be obtained from the first online site before querying the corresponding product information database. Alternatively, if the same product is not found on the first online site, a search for the same product can be performed through other cross-border product information service-related online sites, providing demand trend analysis results from those other sites.

[0035] The specific analysis results can be displayed within the AI ​​interactive area provided in the web browser window. This AI interactive area can also be created by a plugin. During the display of the demand trend analysis results, a second operation option for finding suppliers can be provided within this AI interactive area. Subsequently, if the user clicks this second operation option, a request to search for goods based on the target image can be initiated to a second website. The search results from the second website will be displayed, and a generative AI model will determine the recommended products from the search results, displaying the recommendations in the AI ​​interactive area. This second website can be pre-configured; that is, regardless of which first website the user is browsing, after initiating a supplier search request, they can perform a product search through the pre-configured second website.

[0036] The first type of website typically involves overseas-related sites, such as product information service systems (e.g., "overseas e-commerce platforms"), social networking systems (e.g., "overseas social platforms"), or content systems (e.g., "overseas content platforms") targeting overseas users. These websites can be used to discover products that are of general interest to overseas users. These products can then be purchased, usually from domestic B2B product information service systems (e.g., "domestic B2B e-commerce platforms") where the merchants are located. The purchased products are then listed and sold on a cross-border product information service system, targeting overseas users. In other words, in current technology, merchants in cross-border product information service systems need to switch between three types of websites during the product selection and sales process. The first type consists of the aforementioned "overseas e-commerce platforms," ​​"overseas social platforms," ​​and "overseas content platforms"; the second type consists of domestic B2B e-commerce platforms used for product procurement; and the third type consists of cross-border e-commerce platforms. The terms "overseas," "domestic," and "overseas" mentioned above are all relative to the country of origin of the "merchants in the cross-border commodity information service system." For example, if a merchant's country is A, then everything within country A is considered domestic, and other countries are considered overseas or abroad. Specifically, for the first type of overseas e-commerce platform, buyers are typically overseas users, while sellers (or merchants) can be from country A or other overseas sellers. The second type, domestic e-commerce platforms, refers to platforms where both buyers and sellers are within the same country / region. Since procurement is involved, this is usually a B2B model, meaning both buyers and sellers are corporate users, not individuals. The third type, cross-border e-commerce platforms, typically have sellers (or merchants) located within country A, but their target users are overseas.

[0037] In the solution provided in this application embodiment, merchants only need to browse a first website. If they find an image of interest, they can hover the mouse over it. Correspondingly, the browser's plugin can display a first operation option, such as "Check Trends," near the image (e.g., the upper left corner). Then, the user simply clicks this first operation option to view the AI-generated product demand trend analysis results in the AI ​​interactive area, such as the sidebar on the right. Afterward, clicking a second operation option, such as "Find Suppliers," will automatically initiate a "Search Products Based on Images" search request on a second website and display the search results page of that second website in the browser. Furthermore, since search results typically include multiple products, but merchants usually need to select the best ones during procurement, AI-recommended preferred product information can also be provided in the AI ​​interactive area. Through this process, users can quickly complete product selection.

[0038] The AI ​​may recommend more than one product; typically, it might recommend three (or other numbers). However, merchants may need to further refine their selection. In this case, a third option can be provided for analyzing the AI-recommended products. Users can click this option to request analysis of a specific recommended product. The plugin then sends a command to the browser to open the product details page, analyzes the details using an AI model, and displays the analysis results in the AI ​​interaction area. These results can include analysis of the product's main value propositions and user reviews. This allows users to make their final product selection decision based on the detailed analysis results.

[0039] In addition, the specific plugin program can also support the "product listing" function. For example, after completing product selection and subsequent procurement, when it is necessary to list the product to one's own store on the aforementioned third online site, if the product was selected through the aforementioned process, it can be found again on the aforementioned second online site (listed by other merchants on the second online site). The plugin program can then provide options such as "List the same product." Clicking this option will extract key product information from the second online site, such as product images, titles, specifications, prices, etc. The plugin program can then automatically open the product listing form page on the third online site, filling in the aforementioned key information to help users quickly list products. Of course, the information entered into the form can be modified. For example, merchants can modify the specific price, or if product details change, the product images can also be modified, and so on.

[0040] In this embodiment, the plug-in program involves interaction with the browser. Specifically, from a technical perspective, this interaction can be accomplished using the browser's underlying debugging protocols. For example, for the Chrome browser, CDP (Chrome DevTools Protocol) is a low-level debugging protocol that allows external programs to connect and control the browser via WebSocket (a protocol for full-duplex communication over a single TCP (Transmission Control Protocol) connection). Therefore, this embodiment can utilize this CDP protocol to achieve interaction between the plug-in program and the browser. Specifically, the plug-in program can create automated proxy tasks for the CDP engine. Then, the CDP engine interacts with the browser to execute automated actions. The CDP engine then returns the processing results to the plug-in program, which displays the results in a specific AI interaction area.

[0041] Based on the above, from a technical architecture perspective, such as Figure 1 As shown, this involves browser plugins, plugin backends (server-side), a CDP engine, AI model services, and interactions between various websites (including the aforementioned first, second, and third websites). First, users can click on product selection-related options (e.g., "Check Trends," "Check Sources," "Find TOP 3 Products," "Send Similar Products," etc.), and the plugin sends product selection messages to the backend. Next, automated agent tasks can be created through the CDP engine, establishing a CDP connection between the CDP engine and the target website. Then, various automated operations can be performed on the target website through the CDP engine, and the target website can return page data. Afterward, the CDP engine can send data analysis requests to the generative AI model, which can then stream analysis results. The CDP engine can push intermediate results to the backend, and the backend sends interface update messages to the plugin, which are then displayed through the AI ​​interaction area.

[0042] The specific implementation schemes provided in the embodiments of this application will be described in detail below.

[0043] Example 1 First, this first embodiment provides a page information processing method from the perspective of the aforementioned plugin program. This method is applied to web browser plugins. (See [link to relevant documentation]). Figure 2 The method may specifically include: S201: During the process of displaying the first page of the first website through the web browser, provide the target image in the first page that is in the focus state with an operation option for viewing product demand trend information.

[0044] After installing the plugin program provided in this application embodiment in their web browser, merchants can browse web pages normally. Specifically, when selecting products from external websites (relative to the third website where products will be shipped later), they can browse web pages on the first website through the browser with the plugin program installed. During browsing, if the user hovers the mouse over an image on the page (or performs other operations), the plugin program can detect the image and display a first operation option for viewing product demand trend information in the upper left corner of the image. For example, such as... Figure 3 As shown, assuming it is part of a page on the first website the user is browsing, at a certain moment the mouse hovers over it... Figure 3 Above the image shown at point 31, you can... Figure 3 The location shown at point 32 displays the operation entry provided by the plug-in program. For example, it may display the "Check Trend" option, which is the operation option described in the embodiments of this application.

[0045] S202: After receiving the user's operation request through the operation options, obtain information about the same product based on the target image, and generate a product demand trend analysis result by calling a generative artificial intelligence (AI) model. Then, display the trend analysis result in the AI ​​interaction area of ​​the web browser window.

[0046] Upon receiving a user's "check trends" request, the system can first identify the corresponding product information for a specific image. This information can then be analyzed by a generative AI model to generate demand trend analysis results. This product information can be obtained in several ways. One method is if the plugin provider has a pre-existing partnership with the first website being accessed, granting the plugin access to its product database. In this case, the plugin can send a query to the first website, using the specific image as the search term, and the website can return results for the corresponding product. Alternatively, the plugin provider may also have a connection to a cross-border e-commerce information service system. Therefore, the plugin can also use the specific image as the search term to retrieve product information from the database of this system. Both methods can also be used simultaneously. Furthermore, information about the same product in the image can be obtained from social networking systems, content platforms, and other systems. The process of searching for similar products based on images can be completed in the background. That is, it is not necessary to show the user the specific results of the similar product search. These results can be provided to the generative AI model, which will then analyze the product demand trend.

[0047] Specifically, the information obtained regarding the same product can include sales volume / sales revenue of the same product in multiple different countries / regions, buyer user information, and related notes, videos, and topic discussions on relevant social networks or content platforms. When the generative AI model performs product demand trend analysis, prompts can guide the model to analyze from the required aspects, such as the increase in product demand trend (mainly analyzed from sales volume, sales revenue, etc., combined with data from the most recent 12 months to determine whether the demand trend is continuously increasing, stable, fluctuating, or declining), buyer country preferences (obtained by analyzing the country distribution of buyer users), product type demand, social media buzz trends (determined by analyzing changes in related notes, videos, and topic discussions on specific social networks and content platforms), buyer demand (including whether there are emotional or social needs, and whether new usage scenarios have emerged), purchase paths (whether more buyers purchase directly through live streams or search and compare prices), etc. Furthermore, the generative AI model can provide conclusive analysis results, such as, "Since the increase in demand trend for the same product is negative, it is recommended that merchants continue to observe and not list it for the time being." Alternatively, if the demand trend for a product is showing a strong upward trend, you can advise merchants to "find sources immediately and put the product on the shelves as soon as possible," and so on.

[0048] It should be noted that, in the optional implementation, after receiving the operation request through the first operation option, the target image can also be uploaded to the server corresponding to the plugin program to be converted into a standardized image link address. This standardized image link address can then be used to retrieve similar products and for subsequent product searches based on the target image. In other words, the target image is first transferred to the server corresponding to the plugin program to regenerate the link address, ensuring the stability of subsequent processes.

[0049] After obtaining product demand trend analysis results through generative AI models, the plugin can create an AI interactive area in a web browser window and display the trend analysis results within this area, allowing users to view the results. For example, ... Figure 3As shown at point 33, this is the AI ​​interaction area mentioned above. It displays the trend analysis results provided by the generative AI model, allowing users to make decisions based on these results. However, since the primary website typically targets overseas buyers, in practice, merchants in cross-border e-commerce systems usually need to source goods through domestic B2B e-commerce platforms, comparing multiple suppliers before deciding whether to purchase and from which supplier. Therefore, after the generative AI model provides the product demand trend analysis results, the decision at this point mainly revolves around whether to continue searching for relevant sources of goods. Therefore, the process of displaying the product demand trend analysis results can also provide operational options for finding sources of goods. For example, it can be specifically as follows... Figure 3 As shown at point 34, after displaying the product demand trend analysis results, options such as "Find sources for this product outside of a certain website" can be provided below the analysis results. Alternatively, users can submit specific sourcing requests by entering dialogue content such as "Help me find sources for this product outside of a certain website" in the dialog box of the AI ​​interaction area.

[0050] In practice, the specific second website used for sourcing can be pre-configured in the plugin program, typically an online site associated with the plugin provider, or another system offering domestic B2B product information services. Alternatively, it can be a second website temporarily determined based on the specific product; for example, it could be a service site related to a factory or trade show for that product. The key is to provide information services for large-scale procurement transactions where both parties are within the same country / region.

[0051] S203: After receiving the operation request to find sources of goods through the AI ​​interaction area, a request to search for goods based on the target image is initiated to the second website, and the search results of goods in the second website are displayed through the web browser. After determining the recommended goods to be purchased from the search results through the generative AI model, the recommendation results are displayed in the AI ​​interaction area.

[0052] If a user, after viewing the demand trend analysis results for a specific product, decides to list it in their store for sale, they can initiate a sourcing request using the aforementioned options or by entering dialogue in the AI ​​interaction area's dialog box. For the plugin program, upon receiving the user's sourcing request, it can send a request to a second website to search for the product based on the target image. If a second website is pre-configured, the search request can be sent directly to it; alternatively, the second website can be dynamically determined based on the current image or previous searches for similar products, and then a search request can be sent to that website.

[0053] In one specific implementation, the plugin program can first send an instruction to the web browser to open the target page (e.g., the homepage) of a second website. After displaying the target page, the plugin program can locate the search input control on the target page, then fill in the link address of the target image into the search input control and trigger the "Start Search" button to initiate a request to search for products based on the target image. The link address of the image can be the link address obtained after the aforementioned standardization process. Specifically, the location of the search input control can also be assisted by a generative AI model. For example, the plugin program can provide a screenshot of the target page on the second website to the generative AI model, which can then identify the search input control and the "Search" button, locate their positions, and provide this information to the plugin program. The plugin program then executes the specific action of filling in the link address and triggering the search button. In this approach, the specific generative model can be a multimodal AI model, that is, an AI model capable of simultaneously processing multiple different modalities of data (e.g., text, images, etc.). Alternatively, generative AI models can be divided into single-modal models such as AI language models and AI vision models. The plugin can provide the DOM information of the target page to the AI ​​language model, which analyzes the DOM information to locate the search input control and returns the result to the plugin. On the other hand, a screenshot of the page can be provided to the AI ​​vision model, which locates the search button and returns the result to the plugin. The plugin can then use this location information to input image links and trigger the search button.

[0054] After obtaining product search results from a second website, a product search results page can be displayed in a browser, allowing users to view the product information found on the second website. Since there are usually multiple products found on the second website, merchants typically need to select the best ones for purchase. Therefore, in this embodiment, after obtaining the product search results from the second website, a generative AI model can be used to analyze the search results to determine recommended products for purchase, which are then displayed in the AI ​​interaction area. Alternatively, in a practical implementation, the step of having the generative AI model provide recommended product information can be performed after the user initiates a request. For example, after providing the product search results from the second website, corresponding operation options can be offered in the AI ​​interaction area. Alternatively, the merchant can input a message such as "Help me find the best-selling products" in the dialog box of the AI ​​interaction area, and then the generative AI model can analyze the product search results to determine recommended products for purchase.

[0055] For example, in Figure 3 Based on the example shown, after a user clicks the "Help me find sources for this product outside of this site" option, the interface displayed can be as follows: Figure 4 As shown, it displays information based on... Figure 3 The images shown at point 31 are part of the search results page from a product search on a second website. Among them, such as... Figure 4 As shown at point 41, when a user inputs "Help me select the best-selling items from the product results," an AI model can analyze the search results and select three (or a different number) of the best-selling items as recommended purchases. For example, as... Figure 4 As shown in point 42, information on these recommended purchase items can be displayed in the AI ​​interaction area.

[0056] The plugin can provide the DOM (Document Object Model) information of the search results page to the generative AI model, allowing the model to analyze product information based on this DOM information and select recommended products. Alternatively, it can provide screenshots of the search results page to the AI ​​model for analysis. Determining whether a specific product is a recommended purchase source typically involves analyzing data on certain metrics for that product and comparing it with data on the same metrics for other products. However, multiple metrics are often present. Therefore, to facilitate recommendations from the generative AI model, the prompt word information can include priority rules for various ranking metrics. This allows the model to sort the products in the search results according to the most frequently used ranking metrics and determine the recommended products. For example, a prompt word might include the following information: "Analyze the top 3 best-selling products on the page."

[0057] Priority rules: 1. If trading volume exists, sort by trading volume. 2. Otherwise, sort by transaction amount. 3. Otherwise, sort by repurchase rate. 4. Otherwise, the AI ​​will make its own judgment. Through the aforementioned methods, it is possible to search for similar products based on the target images that users initially browse on the first website, and to analyze product demand trends. Furthermore, it can help users source products on the second website and provide recommended sourcing information, thus helping users narrow down their sourcing scope and quickly make product selection decisions.

[0058] The following is combined Figure 5 , Figure 6 This paper introduces the interactive process described above, which involves searching for products based on images on a second website and recommending the top 3 best-selling products from the search results, from a technical implementation perspective.

[0059] Firstly, as Figure 5 As shown, when a user hovers their mouse over an image and clicks the "Check Trends" option, the plugin program provides trend analysis results through a generative AI model. If the user then selects "Find Suppliers," the following steps can be included: 1. Provide product images to the plugin program based on the images corresponding to the user's clicked operation options. The plugin program can recognize the image formats. 2. The plugin program identifies the image format and requests the image to be uploaded to the backend; 3. The backend converts the image into a File object; 4. The backend returns the image link; 5. The plug-in program creates automated agent tasks for the CDP engine: searching for similar products by image; 6. The CDP engine sends a command to the browser to open the second web site page; 7. The browser returns a page loading completion event; afterwards, AI-driven intelligent element positioning can be used to wait for the search box to load; 8. The CDP engine executes JavaScript to inspect DOM information; 9. The browser returns DOM element state information; if the search box is not loaded, it will wait 100ms (or other time) and then retry until the search box is loaded. 10. The CDP engine retrieves a screenshot of the page from the browser; 11. Screenshot of the page returned by the browser; 12. The CDP engine sends the page content represented by DOM information to the AI ​​language model and instructs the AI ​​model to locate the search box in the Prompt. 13. The AI ​​language model analyzes the DOM structure, understands the semantics of the "search box", and returns the coordinates of the search box and element features; 14. The CDP engine allows you to enter the URL of the target image in the search box, and the specific page can be filled in with the image link in the search box; 15. The CDP engine sends a screenshot of the page to the visual model and instructs the visual model to locate the search button in the Prompt. 16. The AI ​​vision model understands the page layout through page screenshots, identifies the search icon button, and returns the location of the search button; 17. The CDP engine executes the action of clicking the search button; 18. Trigger a request to search for products using images; 19. The browser will redirect to the search results page for display; 20. Display product search results to the user.

[0060] In practice, the same multimodal generative model can be used to understand and analyze DOM content and page screenshots. In this case, the AI ​​language model and the AI ​​vision model can be merged into the same model.

[0061] Afterwards, users can continue to use features such as "help me select the best-selling items from the product results," specifically, for example... Figure 6 As shown: 1. After viewing the product search results on the second website, users can initiate a request to analyze best-selling products; 2. After receiving the request, the plug-in program creates an automated agent for the CDP engine; 3. The CDP engine executes JavaScript checks on the browser to obtain the current URL; 4. The browser returns URL information; 5. The CDP engine verifies whether the page is a search results page; 6. The CDP engine scrolls the page to the top; 7. The CDP engine refreshes the page; 8. The browser returns that the page has finished loading; 9. The CDP engine retrieves the page's DOM content; 10. The browser returns the page's DOM structure; 11. The CDP engine sends the page's DOM content and prompts to the generative AI model; 12. Generative AI model analyzes DOM content to extract the top 3 products; 13. Generative AI models return structured product data to the CDP engine; 14. The CDP engine pushes analysis results to plug-in programs; 15. The plugin renders a list of product cards, displaying the top 3 products.

[0062] Since generative AI models typically recommend multiple sources of goods for purchase, but merchants usually need to specify a particular source when making actual purchases, this system can, as an option, further assist users in identifying the source. Specifically, while providing recommended source information, the AI ​​interaction area can offer options for detailed analysis of specific products, such as... Figure 4 As shown at point 43, when displaying recommended sources of goods, an "Analyze this product" option can be provided for each recommended source. Users can initiate specific analysis requests by clicking this option. Of course, in practice, users can also input specific analysis requirements in the dialog box in the AI ​​interaction area, and so on.

[0063] After a user requests analysis of a product's details, the plugin can instruct the browser to open the target product's details page. For example, it can obtain the link to the target product's details page from the product search results page and provide this link to the browser, enabling the browser to open the page. The plugin can then provide the information from the details page to a generative AI model, which will then analyze the product details.

[0064] The analysis results of generative AI models on product details information can include key value propositions (e.g., what might be called "selling points") and / or summaries of user reviews. These analysis results can also be displayed in the AI ​​interaction area, for example... Figure 7 As shown at point 71, this is an example demonstrating the product details analysis results. The summarized value points include: customized services, preferential policies, fast delivery, etc. The summarized review information includes the ratio of positive to negative reviews, the reasons for positive and negative reviews, and so on. In the specific process of analyzing the details, the AI ​​model can not only extract explicit data but also understand implicit information. For example, in the review analysis, the AI ​​model can identify the sentiment tendencies of reviews such as "fast logistics" and "average quality," and summarize them as the core reasons for "positive" or "negative" reviews, etc.

[0065] In one specific implementation, the plugin can request the DOM information of the product details page and the location information of the rating option (rating button) from the browser. The rating button is used to expand user reviews of the target product. Then, the DOM information and prompt words can be provided to a generative AI model, enabling the model to analyze the product's key value proposition based on the DOM information. Simultaneously, based on the location information of the rating option, a command can be sent to the browser to scroll the page to the location of the rating option and execute the operation to trigger the rating option, thereby obtaining user review content. Finally, the user review content and prompt words are provided to the generative AI model, allowing it to summarize the user review content.

[0066] In some pages, positive user reviews are typically displayed by default, while negative reviews require clicking an additional negative review tag to be shown. Therefore, to obtain more comprehensive review content, the plugin can also locate the negative review tag, click it to expand the negative review content, and provide it to a generative AI model for analysis and summarization. This ensures that the final summarized user reviews more comprehensively and accurately reflect the actual review situation.

[0067] There are several ways to provide user review content to the generative AI model. For example, one method is to provide a screenshot of the user review section to the generative AI model, allowing it to analyze the content based on this screenshot. However, in this screenshot-based analysis, the content may be insufficient to obtain the necessary information, as the screenshot typically only captures the content within the first screen's visible area. Furthermore, the page may contain action options that could provide more information about the product, leading to more detailed and accurate analysis. Therefore, during the screenshot-based analysis, the generative AI model can interact with the plugin program in multiple rounds. For instance, after analyzing the first screen's screenshot, the model can instruct the plugin program to scroll down and take a new screenshot. The plugin program can then send corresponding scrolling instructions to the browser, including the scrolling direction and distance. After scrolling, a new screenshot is taken and input into the generative AI model. Alternatively, if the generative AI model detects an option in a screenshot of a page, such as the "View All" option in a user review section, it can instruct the plugin to click on that option at a specific coordinate. The plugin then sends a corresponding click command to the browser, obtaining the response after the option is clicked. This response is then provided to the generative AI model for further analysis. Through multiple rounds of interaction between the plugin and the generative AI model, the model can obtain more complete information about the product details page. Furthermore, the specific page scrolling and option selection can be determined by the generative AI model based on its analysis and understanding of the received screenshots, achieving on-demand provision and avoiding resource waste.

[0068] Through the AI ​​analysis and summarization of product details described above, merchants can quickly understand the core selling points, positive and negative user reviews of specific products, thus helping them make the final product selection decision. After making the product selection decision, if a specific product needs to be purchased, the subsequent procurement process can be completed by communicating with the corresponding supplier.

[0069] The following is combined Figure 8 From a technical implementation perspective, the interactive process for analyzing detailed product information described above will be introduced. Specifically, it may include the following steps: 1. The user initiates a request to analyze product details; 2. The plug-in program creates an automated agent for the CDP engine; 3. Perform parallel tasks: retrieve selling points and check the review button. Specifically, 3a: the CDP engine requests the page DOM from the browser; 3b: the CDP engine requests the browser to locate the review button. 4. 4a: The browser returns the DOM content; 4b: The browser returns the location of the rating button; 5. 5a: The CDP engine requests analysis of core selling points from the generative AI model; 5b: The CDP engine sends a request to the browser to check the login status. 6. 6a: The generative AI model returns a list of selling points; 6b: The browser returns login information; 7. The CDP engine sends a command to the browser to scroll to the button position; 8. The CDP engine sends a command to the browser to click the button to expand the rating; 9. The browser returns that the evaluation content has finished loading; 10, 10a: The CDP engine requests the generative AI model to analyze positive reviews; 10b: The CDP engine clicks on negative review tags; 10c: The CDP engine requests the generative AI model to analyze negative review content. 11, 11a: Generative AI model returns positive review analysis results; 11c: Generative AI model returns negative review analysis results; 12. The CDP engine requests a summary of the generative AI model, which is available in the evaluation content. 13. The generative AI model returns a summary result; 14. The CDP engine returns all summarized analysis results to the plug-in program; 15. The plug-in program displays a complete analysis report to the user through the AI ​​interactive area.

[0070] After completing the procurement, the merchant can publish the specific product information on the third website. At this time, the plug-in program in this embodiment can continue to help the user complete the quick product posting process. Specifically, since the products to be published were procured through sourcing on the second website, the specific product images, specifications, and other information are usually consistent with the product information already published on the second website. In this case, when the merchant needs to post products, they can revisit the product details page on the second website through a browser. At this time, the plug-in program can be displayed at a certain location on the product details page (e.g., ...). Figure 9 (near the product title shown in 91), or in Figure 9In the AI ​​interaction area shown at point 92, options such as "Post the same style" are provided. Clicking this option allows you to provide key information from the current product details page to the generative AI model. The model then extracts information relevant to product posting, including product images and specifications, and fills it into the product posting form on a third-party website. Information about this processing by the generative AI model can also be displayed in the AI ​​interaction area, for example... Figure 9 As shown at point 93. Users can also view specific product listing records by clicking on options such as "View Generation Records". Since merchants may customize product graphics and logos during the procurement process, the product images may not be entirely consistent with the original images published on the second website. Therefore, users can also modify the information in the product listing form before proceeding with the listing process. In this way, at least some information in the product listing process can be automatically filled out by AI to help merchants, thereby improving listing efficiency.

[0071] By combining the above product distribution process, specific plugin programs can help users improve efficiency in multiple stages of the process, from product selection and sourcing to product distribution.

[0072] The following points will be explained.

[0073] First, the process from product selection and sourcing to product delivery involves various types of pages. Users' states of awareness differ across these page types, and correspondingly, different operational options can be provided to facilitate easier interaction with the plugin or generative AI model. To achieve this, the plugin can identify the type of the currently viewed page. Alternatively, it can provide a screenshot of the currently viewed page to the generative AI model, which then identifies the page type and returns the information to the plugin. This allows the plugin to offer different operational options for different page types. For example, for pages like the client homepage or store page, where users may not yet have a clear intention or goal, the plugin can primarily offer the aforementioned "trend checking" related options. However, for search results pages or product detail pages, where users already have a clear brand, category, or even product preference, other operational options can be provided. For instance, search results pages could offer options related to finding suppliers, while product detail pages could offer options related to product analysis, "send the same product," and so on.

[0074] Secondly, in this embodiment, the entire process of interacting with the generative AI model can utilize streaming response technology, allowing merchants to see the progress of each step in real time, thus enhancing the user experience. All data can be saved with a single click, facilitating subsequent processes such as product launches.

[0075] Third, the interaction between the plugin program and the browser based on protocols such as CDP can specifically include: (1) Executing arbitrary JavaScript code in the target page context through the Runtime.evaluate method in the CDP protocol to achieve functions such as page state acquisition, DOM manipulation, and event triggering. For example, it includes checking the current page URL (UniformResource Locator), scrolling the page to a specified position, and obtaining DOM element information. (2) Page screenshot capture. Specifically, the Page.captureScreenshot method in the CDP protocol can be used to obtain a screenshot of the current view of the page for visual analysis of generative AI models. (3) Network request information acquisition. Monitor the network requests and responses of the page to obtain dynamic content such as Ajax (Asynchronous JavaScript and XML) data and API (Application Programming Interface) call results, breaking through the limitation of traditional crawlers that can only obtain static HTML (Hyper Text Markup Language). (4) Page lifecycle management. Monitor events such as page loading status, DOM changes, and resource loading completion to ensure that automated operations are performed at the correct time.

[0076] Fourth, the generative AI model used in this application embodiment can be pre-trained with massive amounts of web page data, possessing capabilities such as understanding HTML structure, recognizing semantic information, and extracting key data. Furthermore, when understanding page screenshots is involved, the generative AI model also needs image understanding capabilities. This can be achieved using a multimodal generative AI model, meaning the same model can understand multiple modalities such as text and images. To improve the generative AI model's ability to recognize product-related page information or screenshots, it can be fine-tuned based on existing open-source generative AI models; specific training methods will not be detailed here. Additionally, a dedicated Prompt project can be pre-designed to transform page analysis tasks into queries understandable to the model. For example, if the generative AI model needs to perform DOM structure analysis, the page's DOM tree and text content can be sent to the model, and a Prompt expressed in natural language can guide the model to extract data from specified fields. If the generative AI model needs to analyze recommended products on the page, the priority rules of various ranking indicators can be expressed in natural language within the Prompt. Furthermore, the Prompt can prompt the generative AI model to understand and reason about the page content semantically, not only extracting explicit data but also understanding implicit information. In this way, the generalization ability of the generative AI model can be leveraged; a single Prompt can adapt to various different websites without needing to maintain separate XPath (nested path) rules, thus reducing maintenance costs. Moreover, when the page structure changes, the generative AI model can find the corresponding data fields through semantic understanding, without requiring manual rule updates. For example, product prices might change... or In this context, generative AI models can correctly identify all data. Furthermore, when extracting data, generative AI models can automatically filter invalid information, standardize unit formats, and complete missing fields. For example, the price "￥99.00" and "99 yuan" can be unified as "￥99.00". Moreover, generative AI models can also perform page context analysis. For instance, on a product list page, a generative AI model can understand the boundaries of each product card, correctly associating information such as titles, prices, and images belonging to the same product, avoiding data confusion.

[0077] For complex pages, as mentioned earlier, a multi-turn dialogue approach can be used to gradually obtain information. For example, in the first round, the plugin provides a screenshot of the first screen to the generative AI model, which identifies the page type (search page / details page / store page, etc.); in the second round, the generative AI model extracts structured fields from the page; in the third round, if the generative AI model finds missing data, it reports it to the plugin, which then instructs the browser to scroll the page or click an option to get more content via the CDP protocol, takes a new screenshot, and provides it to the generative AI model, and so on.

[0078] Fifth, when using a generative AI model for page screenshot analysis, the plugin sends a screenshot of the webpage to the generative AI model. The model can then identify the page layout, locate product cards, extract product images, and so on. This allows for page analysis without relying on the DOM structure, which is more effective for pages using Canvas rendering (drawing 2D images on a webpage using JavaScript) or image stitching. Additionally, it can perform product image recognition, such as analyzing the main product image to extract product features like color, shape, and category, for image-based product search functions. When product information is displayed as images (such as price tags or promotional information), OCR (Optical Character Recognition) capabilities can be used to extract text from the image, ensuring data integrity. Furthermore, multimodal capabilities can be utilized to combine DOM information, text, and image information for comprehensive analysis. For example, when analyzing product selling points, it can analyze both text descriptions in the DOM information and product close-ups and parameter annotations in the page screenshot.

[0079] In summary, this application provides a web browser plugin that, during the display of a page on a first website through a web browser, offers an option to view product demand trend information for a target image that is the focus of the operation. Upon receiving a user's operation request via these options, the plugin can retrieve information on the same product from the first website and / or other cross-border product information service-related websites based on the target image. After analyzing this information using a generative artificial intelligence (AI) model, a product demand trend analysis result is generated and displayed in the AI ​​interaction area of ​​the web browser window. Similarly, upon receiving a request to find suppliers via the AI ​​interaction area, the plugin can initiate a product search request based on the target image to a second website, displaying the search results from that website. The generative AI model then identifies recommended products from the search results and displays the recommendations in the AI ​​interaction area. This approach, through the plugin and generative AI model, enables automated product demand trend analysis, sourcing, and product recommendations, thereby improving the efficiency of product selection decisions.

[0080] Alternatively, upon receiving a request to perform a detailed analysis of a target product, an instruction can be sent to the browser to open the product details page corresponding to the target product. A generative AI model can then analyze the product's details and display the results in the AI ​​interaction area. This allows users to make further product selection decisions from multiple AI-recommended products based on the specific product details analysis results.

[0081] The plugin program provides page information to the generative AI model, including page screenshots. Based on feedback from the generative AI model, it can send instructions to the web browser to scroll the page or click on target coordinates on the page, and then re-capture the page and input it into the generative AI model for analysis. This multi-round interaction gradually provides page information to the generative AI model, enabling on-demand information provision and reducing resource waste. Furthermore, this method is more effective for analyzing pages rendered using Canvas or image stitching.

[0082] Example 2 This second embodiment mainly provides a page information processing system from a system perspective, which may specifically include: Web browser plugins and generative AI models; The web browser plugin is used to receive requests from users for product selection assistance during the process of users browsing pages through web browsers and performing automated operations by interacting with the browser to obtain relevant page data and generate prompt words. It then calls the generative AI model based on the page data and prompt words. The generative AI model is used to process the page data and prompt information to generate content to assist in product selection decisions. The web browser plugin is also used to display the content generated by the generative AI model through the AI ​​interaction area within the browser window.

[0083] The content of the product selection decision-making assistance generated by the generative AI model includes one or more of the following: the analysis results of demand trend analysis of the same product based on the target image, the product search results of finding sources of goods in the second network site based on the target image in the first network site, the recommended purchase product information analyzed from the found sources of goods, and the results of detailed analysis and summary of the recommended purchase products.

[0084] Through the above-described second embodiment, by combining browser plugins and generative AI models, content can be generated to help users make product selection decisions during the process of selecting products outside the website, thereby assisting users in quickly completing product selection decisions.

[0085] For details regarding the parts of this embodiment that are not described in detail, please refer to the description in embodiment one and other parts of this specification. They will not be repeated here.

[0086] It should be noted that the embodiments of this application may involve the use of user data. In practical applications, user-specific personal data may be used in the scheme described herein within the scope permitted by applicable laws and regulations, provided that it complies with the applicable laws and regulations of the country (e.g., with the user's explicit consent, with the user being properly notified, etc.).

[0087] Corresponding to Embodiment 1, this application also provides a page information processing device, which is applied to a web browser plugin program and includes: An operation option providing unit is used to provide operation options for viewing product demand trend information to a target image in the operation focus state on the page during the process of displaying a page in the first website through the web browser; The trend analysis result display unit is used to receive the user's operation request through the operation options, obtain information about the same product based on the target image, and generate product demand trend analysis results by calling the generative artificial intelligence AI model. The trend analysis results are then displayed in the AI ​​interaction area of ​​the web browser window. The recommendation result display unit is used to receive a request to find goods through the AI ​​interaction area, initiate a request to the second network site to search for goods based on the target image, display the product search results in the second network site, and determine the recommended products to purchase from the product search results through the generative AI model, and then display the recommendation results in the AI ​​interaction area.

[0088] Wherein, the first website is a website related to overseas product information services, a social network website targeting overseas users, or a content system targeting overseas users, and "overseas" is relative to the country / region to which the user belongs; The second website is a site related to product information services where both parties to the transaction are located within the user's country / region.

[0089] Specifically, the operation option providing unit can be used for: When the mouse hovers over an image, that image is designated as the target image in the operation focus state, and the operation option for viewing product demand trend information is provided at the location of the target image.

[0090] Additionally, the device may also include: The link address standardization conversion unit is used to upload the target image to the server after receiving an operation request through the operation option for viewing product demand trend information, so as to convert it into a standardized image link address, so as to use the standardized image link address to obtain the same product and search for products based on the target image.

[0091] Specifically, the recommendation result display unit may include: The instruction sending subunit is used to send an instruction to the web browser to open the target page of the second web site; The location information acquisition subunit is used to acquire the location information of the search input control and search operation options on the target page after the target page is displayed, fill the link address of the target image into the search input control, and then execute the operation of triggering the search operation option to initiate the product search based on the target image.

[0092] In one specific implementation, the generative AI model includes an AI language model and an AI vision model; At this time, the location information acquisition subunit may specifically include: The page information acquisition subunit is used to acquire the document object model (DOM) information of the target page and a page screenshot after the target page is displayed. The search input control location information acquisition subunit is used to provide the DOM information and prompt word information to the AI ​​language model, so that the AI ​​language model can locate the location information of the search input control by analyzing the DOM information and return it; The search operation option location information acquisition subunit is used to provide the page screenshot and prompt word information to the AI ​​vision model, so that the AI ​​vision model can locate the location information of the search operation option by analyzing the page screenshot and return it.

[0093] Specifically, when the recommendation result display unit determines the recommended purchase products from the product search results using the generative AI model, it can be used to: After displaying the product search results page in the browser, the DOM information of the product search results page is obtained from the browser, and the DOM information and prompt word information are provided to the generative AI model so that the generative AI model can extract recommended products for purchase based on the analysis of the DOM information.

[0094] Specifically, when determining recommended products from the product search results using a generative AI model, the prompts provided to the AI ​​model include priority rules for various ranking indicators. This allows the AI ​​model to sort the products in the search results according to the priority ranking indicators and then determine the recommended products.

[0095] Additionally, the device may also include: The detail analysis request receiving unit is used to receive an operation request for detail analysis of one of the target products after displaying the recommended purchase products, and send an instruction to the browser to open the product details page corresponding to the target product; The detailed analysis results display unit is used to analyze the detailed information of the target product through a generative AI model after the product details page is displayed, and to display the analysis results in the AI ​​interaction area.

[0096] Specifically, when the detail analysis result display unit analyzes the detail information of the target product using a generative AI model, the information input to the generative AI model includes: Valid information extracted from the DOM information of the product details page, and / or a screenshot of the product details page.

[0097] Specifically, the detailed analysis results display unit may include: The evaluation operation option location acquisition sub-unit is used to request the DOM information of the product details page and the location information of the evaluation operation option from the browser. The evaluation operation option is used to expand the user evaluation content of the target product. The key value point information analysis subunit is used to provide the DOM information and prompt word information to the generative AI model, so that the generative AI model can analyze the key value point information of the product based on the DOM information. The user review content acquisition subunit is used to send an instruction to the browser to scroll the page to the position of the review operation option based on the position information of the review operation option, and then execute the operation to trigger the review operation option to acquire user review content. The user review summary subunit is used to provide user review content and prompt word information to the generative AI model so that the generative AI model can summarize the user review content.

[0098] Specifically, when providing user review content and prompt information to the generative AI model, the positive review content displayed by default in the user review content display area can be provided to the generative AI model, and the location information of the negative review tag in the user review content display area can be located. After the operation of triggering the negative review tag is executed, the negative review content of the target product can be obtained, and the negative review content can be provided to the generative AI model so that the generative AI model can analyze and summarize the positive and negative review content.

[0099] Alternatively, a screenshot of the user review content area can be provided to the generative AI model, enabling the model to analyze the user review content based on the screenshot.

[0100] Furthermore, the device may also include: The multi-turn interaction unit is used to send instructions to the web browser to scroll the page or click on the operation option at the target coordinate position on the page, based on the information fed back by the generative AI model during the analysis of the page screenshot. The unit then takes a new screenshot of the page and inputs it into the generative AI model for analysis. In this way, page information is gradually provided to the AI ​​model through multi-turn interaction with the generative AI model.

[0101] To facilitate product launch, the device may also include: The product publishing operation option providing unit is used to provide operation options for publishing the same product to a third website during the process of displaying the product details page of the target product in the second website; The product release information generation unit is used to receive the user's operation request through the operation option, provide the product details information of the target product to the generative AI model, and the generative AI model extracts information related to product release from the product details information and fills it into the product release form of the third network site.

[0102] In addition, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in any of the foregoing method embodiments.

[0103] And an electronic device, comprising: One or more processors; and A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method described in any of the foregoing method embodiments.

[0104] A computer program product includes a computer program / computer executable instructions that, when executed by a processor in an electronic device, implement the steps of the method described in the foregoing method embodiments.

[0105] in, Figure 10 An exemplary architecture of an electronic device is shown, which may include a processor 1010, a video display adapter 1011, a disk drive 1012, an input / output interface 1013, a network interface 1014, and a memory 1020. The processor 1010, video display adapter 1011, disk drive 1012, input / output interface 1013, network interface 1014, and memory 1020 can communicate with each other via a communication bus 1030.

[0106] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solution provided in this application.

[0107] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system 1021 for controlling the operation of the electronic device 1000, and the basic input / output system (BIOS) for controlling the low-level operations of the electronic device 1000. Additionally, it can store a web browser 1023, a data storage management system 1024, and a page information processing system 1025, etc. The aforementioned page information processing system 1025 can be the application program that specifically implements the aforementioned steps in this embodiment. In summary, when the technical solution provided in this application is implemented through software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0108] Input / output interface 1013 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0109] The network interface 1014 is used to connect the communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0110] Bus 1030 includes a pathway for transmitting information between various components of the device, such as processor 1010, video display adapter 1011, disk drive 1012, input / output interface 1013, network interface 1014, and memory 1020.

[0111] It should be noted that although the above-described device only shows the processor 1010, video display adapter 1011, disk drive 1012, input / output interface 1013, network interface 1014, memory 1020, bus 1030, etc., in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the solution of this application, and does not necessarily include all the components shown in the figures.

[0112] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0113] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0114] The page information processing method, system, and electronic device provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and its core ideas. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for processing page information, characterized in that, The method is applied to web browser plugins and includes: During the process of displaying a page from the first website through the web browser, an operation option is provided for viewing product demand trend information for the target image that is in the focus state on the page; After receiving the user's operation request through the operation options, the system obtains information about the same product based on the target image, calls the generative artificial intelligence (AI) model to analyze it, generates a product demand trend analysis result, and then displays the trend analysis result in the AI ​​interaction area of ​​the web browser window. After receiving a request to find sources of goods through the AI ​​interaction area, the system initiates a request to the second website to search for goods based on the target image, displays the search results for goods in the second website, and determines the recommended goods for purchase from the search results through the generative AI model, and then displays the recommendation results in the AI ​​interaction area.

2. The method according to claim 1, characterized in that, The first website is a website related to overseas product information services, a social network website targeting overseas users, or a content system targeting overseas users, where "overseas" is relative to the country / region to which the user belongs; The second website is a site related to product information services where both parties to the transaction are located within the user's country / region.

3. The method according to claim 1, characterized in that, The provision of operation options for viewing product demand trend information for the target image that is in the focus state on the first page includes: When the mouse hovers over an image, that image is designated as the target image in the operation focus state, and the operation option for viewing product demand trend information is provided at the location of the target image.

4. The method according to claim 1, characterized in that, Also includes: After receiving an operation request through the operation option for viewing product demand trend information, the target image is uploaded to the server to be converted into a standardized image link address, so as to obtain the same product and search for products based on the target image using the standardized image link address.

5. The method according to claim 1, characterized in that, The step of initiating a request to the second website to perform a product search based on the target image includes: Send an instruction to the web browser to open the target page of the second website; After displaying the target page, the location information of the search input control and search operation options on the target page is obtained, and the link address of the target image is filled into the search input control. Then, the operation of triggering the search operation option is executed to initiate the product search based on the target image.

6. The method according to claim 5, characterized in that, The generative AI model includes an AI language model and an AI vision model; The step of obtaining the location information of the search input control and search operation options on the target page includes: After displaying the target page, obtain the Document Object Model (DOM) information of the target page, as well as a screenshot of the page; The DOM information and prompt word information are provided to the AI ​​language model so that the AI ​​language model can locate the position information of the search input control by analyzing the DOM information and return it. The page screenshot and prompt information are provided to the AI ​​visual model so that the AI ​​visual model can locate the position information of the search operation option by analyzing the page screenshot and return it.

7. The method according to claim 1, characterized in that, When determining recommended purchase items from the product search results using the generative AI model, the process includes: After displaying the product search results page in the browser, the DOM information of the product search results page is obtained from the browser, and the DOM information and prompt word information are provided to the generative AI model so that the generative AI model can extract recommended products for purchase based on the analysis of the DOM information.

8. The method according to claim 7, characterized in that, When determining recommended products from the product search results using a generative AI model, the prompts provided to the AI ​​model include priority rules for various ranking indicators. This allows the AI ​​model to sort the products in the search results according to the priority ranking indicators and then determine the recommended products.

9. The method according to claim 1, characterized in that, Also includes: After displaying the recommended products, the system receives a request to perform a detailed analysis of one of the target products and sends an instruction to the browser to open the product details page corresponding to the target product. After displaying the product details page, a generative AI model is used to analyze the details of the target product, and the analysis results are displayed in the AI ​​interaction area.

10. The method according to claim 9, characterized in that, When analyzing the detailed information of the target product using a generative AI model, the information input into the generative AI model includes: Valid information extracted from the DOM information of the product details page, and / or a screenshot of the product details page.

11. The method according to claim 9, characterized in that, The analysis of the target product's details using a generative AI model includes: The browser is requested to obtain the DOM information of the product details page and the location information of the rating operation option, which is used to expand the user rating content of the target product; The DOM information and prompt word information are provided to the generative AI model so that the generative AI model can analyze the key value points of the product based on the DOM information. Based on the location information of the evaluation operation option, a command is sent to the browser to scroll the page to the location of the evaluation operation option, and the operation of triggering the evaluation operation option is executed to obtain the user's evaluation content. User reviews and prompts are provided to a generative AI model so that the model can summarize the user reviews.

12. The method according to claim 11, characterized in that, When providing user reviews and prompts to generative AI models, the following should be included: The system provides the default positive reviews displayed in the user review content display area to the generative AI model, locates the negative review tags in the user review content display area, executes the operation to trigger the negative review tags, obtains the negative review content of the target product, and provides the negative review content to the generative AI model so that the generative AI model can analyze and summarize the positive and negative review content.

13. The method according to claim 11, characterized in that, Provide the generative AI model with a screenshot of the user review content area so that the generative AI model can perform user review content analysis based on the screenshot.

14. The method according to claim 13, characterized in that, Also includes: During the analysis of the generative AI model based on page screenshots, instructions are sent to the web browser to scroll the page or click on the target coordinates on the page, based on the information fed back by the generative AI model. The page screenshot is then taken again and input into the generative AI model for analysis. In this way, page information is gradually provided to the AI ​​model through multiple rounds of interaction with the generative AI model.

15. The method according to claim 1, characterized in that, Also includes: During the process of displaying the product details page of the target product on the second website, options are provided for publishing the same product on a third website; After receiving the user's operation request through this operation option, the product details information of the target product is provided to the generative AI model. The generative AI model then extracts information related to product release from the product details information and fills it into the product release form of the third network site.

16. A page information processing system, characterized in that, include: Web browser plugins and generative AI models; The web browser plugin is used to receive requests from users for product selection assistance during the process of users browsing pages through web browsers and performing automated operations by interacting with the browser to obtain relevant page data and generate prompt words. It then calls the generative AI model based on the page data and prompt words. The generative AI model is used to process the page data and prompt information to generate content to assist in product selection decisions. The web browser plugin is also used to display the content generated by the generative AI model through the AI ​​interaction area within the browser window.

17. The system according to claim 16, characterized in that, The content of the product selection decision-making assistance generated by the generative AI model includes one or more of the following: the analysis results of demand trend analysis of the same product based on the target image, the product search results of finding sources of goods in the second website based on the target image in the first website, the recommended purchase information analyzed from the found sources of goods, and the results of detailed analysis and summary of the recommended purchase products.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program performs the steps of the method described in any one of claims 1 to 15.

19. An electronic device, characterized in that, include: One or more processors; as well as A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method according to any one of claims 1 to 15.

20. A computer program product comprising a computer program / computer-executable instructions, characterized in that, When the computer program / computer-executable instructions are executed by a processor in an electronic device, they implement the steps of the method according to any one of claims 1 to 15.

Images