Dynamic recommendation method, device and equipment for item information flow

Abstract

This application provides a method, apparatus, and device for dynamic recommendation of item information flow. The method divides the item information page into multiple areas, abstracting the content that can be displayed on the information page into UI elements, where each UI element is the smallest granularity for displaying information and styles. Each area can display one or more UI elements at a time. When generating the item information flow, for each item to be recommended, based on user characteristics, user behavior sequences, item characteristics, and candidate element characteristics, the correlation between candidate UI elements and the user and the item to be recommended is determined, and target UI elements with high correlation to the user and the item to be recommended are recommended. The item information flow is generated based on the target UI elements, display data, and layout information, enabling the dynamic assembly of UI elements for displaying page information for each item to be recommended for the current user, and providing more personalized and accurate recommendations of item information flow for different users.

Description

Technical Field

[0001] This application relates to data processing technology, and more particularly to a dynamic recommendation method, apparatus, and device for item information flow. Background Technology

[0002] With the development of internet technology, more and more users are choosing to browse, select, or purchase the items they need on online e-commerce platforms. However, as the number and variety of items offered by e-commerce platforms increase, users often need to spend a lot of time finding items that meet their needs.

[0003] Many existing e-commerce platforms personalize product recommendations using item summary cards. These cards are displayed sequentially, creating a waterfall-style layout. Each item summary card typically consists of an item thumbnail, a main title, and a price. Users can usually click on these cards. When a user clicks, the e-commerce platform selects that item as the target item, searches for similar items, generates an immersive page containing key details, and then displays this immersive page sequentially to form a feed of information about the recommended items.

[0004] For cross-regional and even cross-border e-commerce platforms, there are many B-type users such as suppliers and buyers. The key information and purchasing preferences of B-type users from different countries and industries are significantly different. However, the layout of the item information page (such as the information card of partial item details) in the information flow page of the recommended item is fixed. The details displayed and the display method of the item information page are the same for all categories of items. It is impossible to make personalized and accurate recommendations of the main details of recommended items for different B-type users. Summary of the Invention

[0005] This application provides a dynamic recommendation method, apparatus, and device for item information flow, which solves the problem that existing technologies cannot provide personalized and accurate recommendations of the main details of recommended items for different B-type users.

[0006] On one hand, this application provides a dynamic recommendation method for an item information stream, wherein the item information stream includes multiple information pages for recommended items, the information pages are divided into multiple areas, each area is used to display at least one UI element, and the method includes:

[0007] Receive an item information stream acquisition request sent when a user triggers an operation on a target item, acquire the user's user characteristics and user behavior sequence, and acquire the item characteristics of the target item;

[0008] Based on the characteristics of the target item, determine the item to be recommended;

[0009] For each item to be recommended, the correlation between each candidate UI element and the user and the item to be recommended is determined based on the user characteristics and user behavior sequence of the user, the item characteristics of the item to be recommended, and the element characteristics of the candidate UI elements in each area of ​​the information page. Based on the correlation between each candidate UI element and the user and the item to be recommended, the target UI element to be displayed in each area of ​​the information page of the item to be recommended is determined.

[0010] Obtain the display data for each of the target UI elements;

[0011] Based on the target UI elements displayed in each area of ​​the information page of each item to be recommended, the display data of the target UI elements, and the layout information of the information page, an item information stream containing the information page of each item to be recommended is rendered and generated.

[0012] On the other hand, this application provides a dynamic recommendation device for an item information stream, wherein the item information stream includes multiple information pages recommending items, the information pages are divided into multiple areas, each area is used to display at least one UI element, and the device includes:

[0013] The feature processing module is used to receive an item information stream acquisition request sent when a user triggers an operation on a target item, acquire the user's user features and user behavior sequence, and acquire the item features of the target item.

[0014] The item recommendation module is used to determine the items to be recommended based on the characteristics of the target item;

[0015] The intelligent UI recommendation module is used to determine the correlation between each candidate UI element and the user and the item to be recommended, based on the user characteristics and user behavior sequence of the user, the item characteristics of the item to be recommended, and the element characteristics of the candidate UI elements in each area of ​​the information page, and to determine the target UI element to be displayed in each area of ​​the information page of the item to be recommended based on the correlation between each candidate UI element and the user and the item to be recommended.

[0016] The display data acquisition module is used to acquire display data for each of the target UI elements.

[0017] The recommendation display module is used to render and generate an item information stream containing the information page of each item to be recommended, based on the target UI element displayed in each area of ​​the information page of each item to be recommended, the display data of the target UI element, and the layout information of the information page.

[0018] On the other hand, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0019] The memory stores computer-executed instructions;

[0020] The processor executes computer execution instructions stored in the memory to implement the method described above.

[0021] On the other hand, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method described above.

[0022] The dynamic recommendation method, apparatus, and device for item information streams provided in this application decompose the information pages of recommended items in the item information stream into very fine-grained units, abstracting the content that can be displayed on the information page into UI elements. Each UI element is the smallest granularity for displaying information and styles. The information page of recommended items is divided into multiple areas, and each area can display one or more UI elements at a time. When a user triggers an action on a target item on the waterfall flow page of the item summary card, a request to obtain the item information stream is sent to the server via the terminal. The server receives the request, obtains the user's user characteristics and user behavior sequence, and obtains the item characteristics of the target item. Based on the item characteristics of the target item, it determines the item to be recommended. When generating the information page for recommended items, for each item to be recommended, based on the user's characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the component characteristics of the candidate UI components in each area of ​​the information page, the correlation between each candidate UI component and the user and the item to be recommended is determined. Based on the correlation between each candidate UI component and the current user and the item to be recommended, one or more candidate UI components with a high correlation to the current user and the item to be recommended are selected from the candidate UI components in each area as the target UI components to be displayed in each area of ​​the information page for that item. The display data for each target UI component is then obtained, and based on the display data of each item in the information page for that item... The system displays target UI elements in a designated area, along with their display data and information page layout information. This data is then used to render and generate an item information stream containing information pages for each item to be recommended. For each item to be recommended to the current user, the system dynamically assembles the UI elements displayed on the page information for each item. The UI elements displayed on the page information for different items recommended to the same user may differ, and the UI elements displayed on the page information for the same item recommended to different users may also differ. This achieves intelligent recommendation of item information streams tailored to individual users (e.g., Category B users), providing more personalized and accurate recommendations. This helps users make decisions while effectively reducing the time cost associated with repeated clicks, comparisons, and communication. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0024] Figure 1 An example diagram of an information page in an existing item information flow provided for this application;

[0025] Figure 2 An example diagram of the system architecture applicable to the dynamic recommendation method for item information flow provided in this application;

[0026] Figure 3 A flowchart of a dynamic recommendation method for item information flow provided as an exemplary embodiment of this application;

[0027] Figure 4 An example diagram of the item information flow generated by the dynamic recommendation method for item information flow provided in this application;

[0028] Figure 5 A framework diagram of dynamic recommendation of UI elements for an item information flow provided in an exemplary embodiment of this application;

[0029] Figure 6 A flowchart illustrating the overall process of the dynamic recommendation method for the item information flow provided in this application;

[0030] Figure 7 A schematic diagram of the structure of a dynamic recommendation device for item information flow provided in an embodiment of this application;

[0031] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an example embodiment of this application.

[0032] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0033] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0034] First, let me explain the terms used in this application:

[0035] L-AB conversion rate: The conversion rate from the waterfall / information feed to the user's order preference behavior (such as communication, adding to cart, etc.).

[0036] User behavior sequence: This is the process of a series of events such as clicks, visits, and purchases generated by users in their daily operations. It can be represented as a time series of event sets. It contains the user's fine-grained habits and preferences and is one of the important feature sources of user-level machine learning models.

[0037] In the existing item information flow, the data items and display style displayed on each item's information page are fixed, for example, Figure 1 Example image of an information page in an existing item information flow, such as Figure 1 As shown, the data items (such as item main image, price, title, etc.) and action points (such as store, customer service, add to cart, purchase, etc.) displayed on all information pages are consistent. The data items and display style displayed in the same area of ​​the information page for all user-recommended item information streams are fixed.

[0038] To address the aforementioned technical problems, this application provides a dynamic recommendation method for item information streams. The method decomposes the information page for recommended items in the item information stream into a very fine-grained structure, abstracting the content that can be displayed on the information page into UI elements. Each UI element represents the smallest unit of information display and style. The information page for recommended items is divided into multiple areas, and each area can display one or more UI elements at a time. When a user triggers an action on a target item in the item summary card's waterfall layout (e.g., clicking the target item's summary card), the user sends an item information stream retrieval request to the server via the terminal. The server receives the item information stream retrieval request, obtains the user's user characteristics and user behavior sequence, and obtains the item characteristics of the target item. Based on the item characteristics of the target item, the server determines the item to be recommended. When generating the information page for recommended items, for each item to be recommended, based on the user's characteristics and user behavior sequence, the item's characteristics, and the component characteristics of the candidate UI components in each area of ​​the information page, the relevance of each candidate UI component to the user and the item to be recommended is determined. Then, based on the relevance of each candidate UI component to the current user and the item to be recommended, one or more candidate UI components with a high relevance to the current user and the item to be recommended are selected from the candidate UI components in each area and used as the target UI components to be displayed in each area of ​​the information page for that item. The information page for each target UI component is then obtained. The system can dynamically recommend the data and display style of each item's information page, allowing different items to display different data items and / or display styles within the item information flow. Furthermore, the display style of the same recommended item's information page can differ for different users, achieving a personalized item information flow with intelligent recommendations. This provides users with richer and more accurate recommendation reason tags and action points, helping them make decisions while effectively reducing the time cost of repeated clicks, comparisons, and communication.

[0039] For example, the dynamic recommendation method for item information flow provided in this application can be applied to Figure 2 The system architecture is shown below. Figure 2As shown, the system architecture includes: terminals and servers.

[0040] The server can be an e-commerce platform's server, specifically a server cluster deployed in the cloud. This server stores intelligent UI recommendation algorithms and item recommendation algorithms. Through preset computational logic, the server performs item recommendation to determine the items to be recommended, and performs intelligent UI recommendation to determine the target UI elements and display data for each item's information page, and renders and generates an item information stream containing the information pages for the recommended items.

[0041] The terminal can specifically be a hardware device with network communication, computing and information display functions, including but not limited to smartphones, tablets, desktop computers, and Internet of Things devices.

[0042] Through communication with the server, when a user triggers (e.g., clicks) a target item in the item summary card waterfall display on the terminal, the terminal submits an item information stream retrieval request to the server. This request carries the user's current user information and the target item's information. After receiving the item information stream retrieval request, the server obtains the user's user characteristics and user behavior sequence based on the user information and the target item's information, and obtains the target item's item characteristics. Based on the target item's item characteristics, it determines the items to be recommended. For each item to be recommended, based on the user's user characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the element characteristics of the candidate UI elements in each area of ​​the information page, it determines the correlation between each candidate UI element and the user and the item to be recommended, and based on the correlation between each candidate UI element and the user and the item to be recommended, it determines the target UI elements to be displayed in each area of ​​the information page of each item to be recommended; it obtains the display data of each target UI element; and based on the target UI elements displayed in each area of ​​the information page of each item to be recommended, the display data of the target UI elements, and the layout information of the information page, it renders and generates an item information stream containing the information page of each item to be recommended. The server feeds back the generated item information stream to the terminal so that the item information stream can be displayed on the terminal.

[0043] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0044] Figure 3 This is a flowchart illustrating a dynamic recommendation method for item information flow provided in an exemplary embodiment of this application. The executing entity in this embodiment may be the server mentioned above, such as... Figure 3 As shown, the specific steps of this method are as follows:

[0045] Step S301: Receive the item information stream acquisition request sent when the user performs a trigger operation on the target item, acquire the user characteristics and user behavior sequence, and acquire the item characteristics of the target item.

[0046] One of the triggering actions for a user on a target item is clicking on the target item's summary card in the waterfall layout of the item summary card.

[0047] In practical applications, when a user browses the waterfall of item summary cards displayed on the terminal, when the user clicks on the summary card of any item in the waterfall, the terminal sends an item information stream retrieval request to the server. The item information stream retrieval request carries the user's current user information and the information of the target item corresponding to the summary card clicked by the user.

[0048] The server receives a request to retrieve item information streams, extracts the user information and target item information carried in the request, and obtains the user characteristics and user behavior sequence of the current user based on the user information and target item information, as well as the item characteristics of the target item.

[0049] User characteristics, also known as user tags, are feature information derived from analyzing users' historical behavioral data across various dimensions, such as social attributes, consumption habits, and interests. Examples include a user's identifier, industry, occupation, and preferred categories.

[0050] User behavior sequences are data obtained by recording individual user behaviors on various pages in a time series throughout the lifecycle of a user's visit. They represent a series of events such as clicks, visits, and purchases that occur during a user's daily operation and can be represented as a time series of event sets. They contain the user's fine-grained habits and preferences and are one of the important feature sources for user-level machine learning models.

[0051] Item characteristics refer to the inherent attributes of an item, such as its logo, category, brand, color, model, and capacity.

[0052] In this embodiment, the method for obtaining user features, user behavior sequences, and item features can be implemented using existing technologies for obtaining user attribute features, user behavior sequences, and item attribute features when making item recommendations, and will not be elaborated here.

[0053] Step S302: Determine the items to be recommended based on the characteristics of the target item.

[0054] After obtaining the characteristics of the target item, other items with high relevance to the characteristics of the target item are recalled as items to be recommended, based on the relevance of the characteristics.

[0055] This step can be implemented using any existing method for item recommendation based on the relevance or similarity of item features, which will not be elaborated here.

[0056] Step S303: For each item to be recommended, determine the correlation between each candidate UI element and the user and the item to be recommended based on the user's user characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the component characteristics of the candidate UI elements in each area of ​​the information page; determine the target UI element to be displayed in each area of ​​the information page of the item to be recommended based on the correlation between each candidate UI element and the user and the item to be recommended.

[0057] In this embodiment, the information page for recommended items in the item information stream is broken down into very fine-grained components. The content that can be displayed on the information page is abstracted into UI elements, which are the smallest unit for displaying information and styles. The information page for recommended items is divided into multiple areas, and each area can display one or more UI elements at a time.

[0058] The characteristics of candidate UI elements include static characteristics and statistical characteristics. Static characteristics include element identifier, region identifier, font color, font size, font weight, whether it is sensitive to rights and interests, whether it is sensitive to services, whether it has industry-specific characteristics, number of display rows, and whether it has a time limit. Statistical characteristics include the recent slot conversion rate, the recent UV (Unique Visitor) conversion rate, the recent click-through rate, the recent click UV, the recent inquiry visit rate, and the recent inquiry UV.

[0059] After identifying the items to be recommended, dynamic intelligent UI recommendations are made for each item based on the current user's characteristics, user behavior sequence, and the item's features. The relevance of each candidate UI element to the current user and the recommended item is analyzed, and one or more candidate UI elements with high relevance to both are selected as target UI elements for that area. Target UI elements are those that will be displayed on the item's information page, while other non-target UI elements will not be displayed on the current user's information page for that item.

[0060] This step allows for the dynamic assembly of UI elements displayed on the page information of each item to be recommended to the current user. This results in different UI elements being displayed on the page information of different items recommended to the same user, and different UI elements being displayed on the page information of the same item recommended to different users, thus achieving intelligent UI recommendation of item information flow that is "personalized" for each user.

[0061] For example, Figure 4 Here is an example diagram of an item information flow generated based on the dynamic recommendation method for item information flow in this application, such as... Figure 4 As shown, in the item information stream, the cover area of ​​the first item's information page A displays an item carousel image with a preview, while the cover area of ​​the second item's information page B only displays an item carousel image without a preview. The recommendation reason tags in the tag area of ​​information page A and information page B are different. The action point area of ​​information page A includes two action points: "communication" and "inquiry," while the tag area of ​​information page B includes three action points: "communication," "inquiry," and "place an order."

[0062] Step S304: Obtain the display data for each target UI element.

[0063] After identifying the target UI elements for each area of ​​the information page for each item to be recommended, obtain the display data for each target UI element.

[0064] In this embodiment, the UI element can be the information object to be displayed (such as product selling point tags, merchant qualification tags, etc.) or the style information of the information object to be displayed (such as item carousel with preview, single-line title highlighting, two-line title display, price range highlighting, etc.). If the target UI element defines the style information of the information object to be displayed, then it is necessary to obtain the display data corresponding to the target UI element, that is, the information object to be displayed.

[0065] Step S305: Based on the target UI elements displayed in each area of ​​the information page of each item to be recommended, the display data of the target UI elements, and the layout information of the information page, render and generate an item information stream containing the information page of each item to be recommended.

[0066] After obtaining the target UI elements to be displayed in each area of ​​the information page of each item to be recommended and the display data of the target UI elements, the target UI elements and display data are aggregated to generate the page data of the information page of the item to be recommended based on the layout information of the information page. The page data is then rendered to generate the information page of the item to be recommended. The information pages of the item to be recommended are combined according to the recommendation order of the item to be recommended to generate the item information stream.

[0067] After generating the item information stream, the item information stream is fed back to the user's terminal, where it is displayed for the user to view.

[0068] In this embodiment, the information page for recommended items in the item information stream is broken down into very fine-grained components. The content that can be displayed on the information page is abstracted into UI elements, which are the smallest granularity for displaying information and styles. The information page for recommended items is divided into multiple areas, and each area can display one or more UI elements at a time. When a user triggers an action on a target item on the waterfall page of the item summary card, a request to retrieve the item information stream is sent to the server through the terminal. The server receives the request, obtains the user's user characteristics and user behavior sequence, and obtains the item characteristics of the target item. Based on the item characteristics of the target item, the server determines the item to be recommended. When generating the information page for recommended items, for each item to be recommended, based on the user's characteristics and user behavior sequence, the item's characteristics, and the component characteristics of the candidate UI components in each area of ​​the information page, the relevance of each candidate UI component to the user and the item to be recommended is determined. Based on the relevance of each candidate UI component to the current user and the item to be recommended, one or more candidate UI components with a high relevance to the current user and the item to be recommended are selected from the candidate UI components in each area as the target UI components to be displayed in each area of ​​the information page for that item. The display data for each target UI component is then obtained, and the display data for each target UI component is determined based on the target UI components displayed in each area of ​​the information page for each item to be recommended. The system uses the display data of I-components and target UI components, along with the layout information of the information page, to render and generate an item information stream containing the information page for each item to be recommended. It can dynamically assemble the UI components displayed on the page information for each item to be recommended for the current user. The UI components displayed on the page information for different items recommended to the same user may differ, and the UI components displayed on the page information for the same item recommended to different users may also differ, achieving intelligent recommendation of item information streams tailored to each individual user. This allows for more personalized and accurate recommendations of item information streams for different users (such as B-type users), helping users make decisions while effectively reducing the time cost of repeated clicks, comparisons, and communication.

[0069] In one optional embodiment, the information page includes at least the following area: a tag area. Each candidate UI element in the tag area is used to display a recommendation reason tag, and the recommendation reason tags that can be displayed in the tag area cover at least the following three types of information: product selling points information, merchant qualification information, and whether the product is in stock.

[0070] In this embodiment, multiple recommendation reason tags can be set, covering information such as product selling points, merchant qualifications, and whether the product is in stock (i.e., RTS (Ready to Ship) item). Recommendation reason tags can be enriched and updated at any time according to the needs of actual application. When setting product information, sellers can bind one or more recommendation reason tags to the product.

[0071] For example, the selling points of an item can include at least one of the following: samples available, one-stop service, warranty, traceability, shipping included, overseas after-sales service, 3D virtual showroom, 3D online customization, 2D light customization, etc.

[0072] By setting up rich recommendation reason tags and corresponding candidate UI elements, when dynamically generating information pages for items to be recommended, UI elements with high relevance to the current user can be dynamically recommended from the candidate UI elements of the recommendation reason tags associated with the items to be recommended, so that the information page of the same item can display different recommendation reasons for different users. This can help users make decisions and effectively reduce the time cost caused by users repeatedly clicking to compare and communicate.

[0073] For example, such as Figure 4 As shown, the information page can include a cover area, title area, tag area, price and incentive area, and action point area. Furthermore, the tag area on the information page can be divided into multiple tag sections, each displayed in a different location. Each tag section is an independent area with its own candidate UI elements. For example, the tag area can include a key decision-making information area and a supporting decision-making information area. The key decision-making information area covers two categories of information: whether the product is in stock and its selling points. The supporting decision-making information area covers information such as the merchant's qualifications.

[0074] For example, the candidate UI elements for the cover area include at least one of the following: item carousel with preview (directly displaying the carousel's images and / or videos), item carousel (without preview, supporting left and right switching of images and / or videos), supporting automatic playback under WiFi, and supporting the display of images / videos in smart UI. The candidate UI elements for the title area include at least one of the following: single-line title, single-line title with enhancement, two-line title, two-line title with enhancement, and supporting smart UI titles. The candidate UI elements for the tag area include candidate UI tags corresponding to each recommendation reason tag, and supporting smart UI tags. The candidate UI elements for the price and incentive area include at least one of the following: price range highlight, price range, lowest price highlight, lowest price, tiered price highlight, tiered price, supporting the display of price-related UI forms and discount information, and supporting the display of smart UI price information, etc.

[0075] Candidate UI elements for the action point area include at least one of the following: default action (jump to details page), merchant communication, inquiry (weak style), place order, place order (weak style), inquiry, add to cart.

[0076] In this embodiment, the target UI elements displayed in each area of ​​the information page of the item to be recommended can be dynamically recommended based on the current user's feature information and the feature information of the item to be recommended. This allows for personalized customization of the item information page in the item information stream for different users, making the displayed item information stream more accurately show the information that users care about and the action points that users prefer, facilitating user operation and improving user experience.

[0077] For example, for non-RTS products (i.e., customized items), the information page for non-RTS products can display UI elements for the action point "Inquiry," as well as UI elements for action points such as "Communicate with Merchant" and "Place Order Directly." However, for RTS products, customization is not supported, so the information page for RTS products usually does not display the UI element for the action point "Inquiry." Instead, it can display UI elements for action points such as "Communicate with Merchant," "Add to Cart," and "Place Order Directly."

[0078] In one alternative embodiment, a trained intelligent UI recommendation model can be used to predict the correlation between candidate UI elements and the current user and the items to be recommended.

[0079] In this embodiment, it is assumed that the various areas in the information page are independent and there is no combined reaction between candidate UI elements. The intelligent UI recommendation model predicts the correlation between candidate UI elements and users and recommended items based on the user's feature information, the feature information of the items to be recommended, and the feature information of candidate UI elements.

[0080] Specifically, in step S303 above, for any item to be recommended, the correlation between each candidate UI element and the user and the item to be recommended is determined based on the user's user characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the element characteristics of the candidate UI elements in each area of ​​the information page. This can be achieved in the following way:

[0081] The user's characteristics and behavior sequence, the item characteristics of the item to be recommended, and the component characteristics of each candidate UI element are input into the trained intelligent UI recommendation model. The intelligent UI recommendation model then predicts the correlation between each candidate UI element and the user and the item to be recommended.

[0082] Among them, relevance represents the likelihood of a user selecting a recommended item based on an information page containing candidate UI elements.

[0083] For example, a user's selection behavior regarding a recommended item can be by clicking (entering the details page) or by triggering one or more action points within the action point area, such as adding to the shopping cart or making an inquiry.

[0084] The intelligent UI recommendation model can be implemented using a machine learning-based neural network model, and the model can be trained using a training set determined by a large amount of historical data to obtain a well-trained intelligent UI recommendation model.

[0085] For example, the neural network structure of the intelligent UI recommendation model can adopt a structure of embedding layers superimposed with a multilayer perceptron (MLP). User features, user behavior sequences, product features, and component features are embedded separately, concatenated, and then input into the MLP, which outputs the prediction result.

[0086] In addition, intelligent UI recommendation models can also be implemented using other neural network structures, such as MAB+ feature learning models, deep neural network (DNN) models, user interest models (UIM), etc. It can be any existing neural network that can be trained to predict the influence of information objects or information objects displayed in a specific style on the probability of user behavior towards recommended items (such as click-through rate (CTR), conversion rate (CVR), L-AB, etc.), which will not be elaborated here.

[0087] By using a well-trained intelligent UI recommendation model, the correlation between candidate UI elements and users and recommended items can be accurately predicted based on user characteristics, user behavior sequences, item characteristics of items to be recommended, and component characteristics of candidate UI elements. This allows for precise measurement of the impact of displaying candidate UI elements on the probability of user behavior towards recommended items, thus providing a data foundation for accurately selecting target UI elements to display on information pages.

[0088] Furthermore, after determining the relevance between candidate UI elements and users and recommended items, the candidate UI elements in each region are sorted by relevance based on the relevance of the candidate UI elements, and the best candidate UI elements in the region are selected for display, thus realizing intelligent UI recommendation.

[0089] Specifically, in step S304 above, the target UI element to be displayed in each area of ​​the information page of the item to be recommended is determined based on the relevance of each candidate UI element to the user and the item to be recommended. This can be achieved in the following way:

[0090] For each region, based on the relevance of each candidate UI element in that region to the user and the items to be recommended, and the UI element arrangement rules for that region, n candidate UI elements are selected as the target UI elements to be displayed in that region, where n is an integer and is less than or equal to the threshold for the number of elements to be displayed in that region.

[0091] Optionally, the UI element arrangement rules for this area may include at least one of the following:

[0092] The mutual exclusion relationship of candidate UI elements: the target UI elements displayed in this area do not include candidate UI elements with mutual exclusion relationships; the forced display of candidate UI elements: the target UI elements displayed in this area include candidate UI elements that are forced to be displayed.

[0093] For example, the UI element arrangement rules of any region can set the element mutual exclusion group of that region. Each element mutual exclusion group includes at least two candidate UI elements. Candidate UI elements in the same element mutual exclusion group cannot be displayed in the region at the same time.

[0094] For example, the UI element arrangement rules for any region can set one or more candidate UI elements that must be displayed in that region. The candidate UI elements that must be displayed will definitely be displayed in that region of the information page.

[0095] In addition, the UI component arrangement rules for any area can also set a threshold for the number of components displayed in that area. The component display threshold refers to the maximum number of UI components that can be displayed simultaneously in that area.

[0096] By using UI element arrangement rules for each area, target UI elements that conform to the arrangement rules are selected based on the relevance of candidate UI elements to users and recommended items, thereby improving the flexibility and standardization of UI element display within the area.

[0097] Optionally, at least one region's candidate UI elements have a display level, which indicates how prominent the UI element is when displayed.

[0098] For each region, based on the relevance of each candidate UI element in that region to the user and the item to be recommended, and the UI element arrangement rules of that region, n candidate UI elements are selected as the target UI elements to be displayed in that region. If the item to be recommended meets the preset downgrade conditions, then at least one target UI element is replaced with another candidate UI element with a lower display level according to the preset downgrade conditions.

[0099] The preset degradation conditions can be set and adjusted according to the needs of the actual application scenario, and no specific restrictions are made here.

[0100] For example, taking the candidate UI elements in the title area as including: single-line title, enhanced single-line title, two-line title, and enhanced two-line title, the display levels of "single-line title," "enhanced single-line title," "two-line title," and "enhanced two-line title" can be set to increase sequentially. After determining that the target UI element in the title area of ​​an item to be recommended is "enhanced single-line title," if it is determined that the price of the item to be recommended does not have any preferential information (such as no discounts or coupons), the target UI element in the title area can be replaced with the "single-line title," which has a lower display level.

[0101] Optionally, for each region, based on the relevance of each candidate UI element in that region to the user and the item to be recommended, and the UI element arrangement rules of that region, after selecting n candidate UI elements as the target UI elements to be displayed in that region, if the item to be recommended meets the preset upgrade conditions, then at least one target UI element is replaced with another candidate UI element with a higher display level according to the preset upgrade conditions.

[0102] In this embodiment, by setting preset degradation conditions, when a certain piece of information of the item to be recommended meets the preset degradation conditions, the display method of that piece of information can be downgraded, thereby allowing for flexible adjustment of the display style of the item's information page.

[0103] For example, Figure 5 A framework diagram for dynamically recommending UI elements of an item information flow provided in an exemplary embodiment of this application, such as... Figure 5 As shown, the server provides a material model, including layout materials, area materials, component materials, and data materials for information pages in the item information flow. The server also stores an algorithm recommendation pool, including intelligent UI recommendation algorithms and item recommendation algorithms. When performing intelligent UI recommendations for the item information flow, the server iterates through all candidate UI components (i.e., component materials) in each area to recommend the best combination of UI components within the information page of each item to be recommended. The server can also provide UI component arrangement rules such as mutual exclusion groups, maximum recommendation count, degradation rules, and forced display to improve the flexibility and standardization of UI component combinations within the information page.

[0104] The layout materials include various layout information, each with a corresponding layout template, including information such as the number of rows and columns. The region materials correspond to a region, including a unique identifier for that region (e.g., ...). Figure 5 The area labels shown), the number of candidate UI elements for the area (e.g., Figure 5 The number of elements shown), and the maximum number of recommended UI elements within the area (i.e., the element display threshold). Each element material corresponds to a candidate UI element, including the candidate UI element's display style and bound display data information (such as...). Figure 5The data includes the bound data, component features, and related preset degradation rules. Each data element corresponds to a piece of display data, including the data format, field mapping information, corresponding entity type (such as item, merchant, theme, etc.), and source information. The field mapping information is used to implement the mapping relationship between different fields used to store the same data in different system platforms, and the source information refers to which data source the display data comes from (which can come from multiple different system platforms).

[0105] In addition, such as Figure 5 As shown, the server can also provide a baseline pool, including one or more fixed component combinations, which can be used as a comparative recommendation scheme for A / B experiments, selecting a recommended scheme that is superior to the fixed component combination, or using a recommended scheme with a better fixed component combination.

[0106] In this embodiment, the ability to dynamically display and assemble item information flow is realized. By splitting layout, region, UI element, and data information into atomic information, and combining audience preferences and industry-specific strategies, the display arrangement is adapted to different audiences, effectively reducing the user's decision-making time cost and improving the quality of experience.

[0107] In practical applications, B-type users such as buyers and suppliers often associate their selection of items in a particular category with other categories that share similar usage scenarios. For example, if a buyer clicks on a hiking water bottle in the item summary chart's waterfall layout, they might also be looking to purchase other commonly used hiking items such as trekking poles, backpacks, protective gear, tents, and sleeping bags. Traditional recommendation methods typically only recommend other water bottles based on the user's clicked hiking water bottle, failing to consider the B-type user's desire for a one-stop, interconnected purchasing experience.

[0108] In one alternative embodiment, multiple themed scenarios can be set up based on operational experience, with each themed scenario including items from multiple categories.

[0109] In step S302 above, the items to be recommended are determined based on the characteristics of the target item. This can be achieved in the following way:

[0110] Based on the characteristics of the target item, a first item with the same category as the target item is recalled based on the similarity of the characteristics. Based on the theme scene corresponding to the target item, a second item with a different category than the target item in the same theme scene is recalled. The first item and the second item are then selected as items to be recommended.

[0111] When selecting items to recommend, if the target item triggered by the user belongs to any theme scenario, other categories of items in the same theme scenario can be recommended to the user. This allows for recommendations of related categories of items based on user interests, aggregating items from different categories, saving users time costs of multiple searches, enabling B-type users to accurately source more items and discover additional surprising business opportunities, improving the flexibility of item recommendations, and meeting the one-stop related purchasing needs of B-type users.

[0112] For example, the dynamic recommendation method for item information flow provided in this application can be deployed as a service, such as... Figure 6 As shown, the overall process of the dynamic recommendation method for item information flow is as follows:

[0113] S1. The delivery gateway receives item information stream recommendation requests;

[0114] S2. The delivery gateway requests items to be recommended from the item recommendation service;

[0115] S3, Item Recommendation Service: Recommends items and obtains information about items to be recommended;

[0116] S4. The item recommendation service sends information about the items to be recommended to the delivery gateway.

[0117] S5. The delivery gateway requests smart UI recommendations from the smart UI service SDK, carrying the identifier of the smart UI recommendation service;

[0118] S6. The Smart UI Service SDK requests a Smart UI recommendation from the Smart UI Recommendation Service.

[0119] S7, Intelligent UI Recommendation Service: Recommends intelligent UI and determines intelligent UI control data;

[0120] The intelligent UI control data includes the layout information of the item information flow and the information of the target UI elements displayed in each area of ​​the information page of each item to be recommended.

[0121] S8, the intelligent UI recommendation service feeds back intelligent UI control data to the intelligent UI service SDK;

[0122] S9, the intelligent UI service SDK requests data to be displayed from the search engine;

[0123] S10. Search engine obtains display data;

[0124] S11. The search engine returns display data to the intelligent UI service SDK;

[0125] S12. The Smart UI Service SDK returns smart UI control data and display data to the delivery gateway;

[0126] S13. The delivery gateway recommends the intelligent UI control data and display data to the item information flow page.

[0127] The following example illustrates how a user triggers the "Place Order" action. When a user triggers the "Place Order" action on a specific item's information page within the item's feed, the subsequent processing flow is as follows:

[0128] S14. The delivery gateway receives a request to trigger an order placement action.

[0129] S15. The delivery gateway sends a request to the details service SDK to trigger an order placement action.

[0130] S16. The order processing action is handled by the Detail Service SDK;

[0131] S17. The details service SDK returns the execution result data to the delivery gateway;

[0132] S18. The delivery gateway will display the execution result data on the item information flow page.

[0133] In addition, this application can generate multiple recommended schemes for item information flow. The better recommended scheme can be selected by A / B test splitting method. Specifically, the existing A / B test splitting method can be used, which will not be elaborated here.

[0134] Figure 7 This is a schematic diagram of a dynamic recommendation device for an item information stream provided in an embodiment of this application. The device provided in this embodiment is used to execute a dynamic recommendation method for an item information stream. The item information stream includes information pages for multiple recommended items. The information pages are divided into multiple areas, and each area is used to display at least one UI element.

[0135] like Figure 7 As shown, the dynamic recommendation device 70 for item information flow includes: a feature processing module 71, an item recommendation module 72, an intelligent UI recommendation module 73, a display data acquisition module 74, and a recommendation display module 75.

[0136] The feature processing module 71 is used to receive the item information stream acquisition request sent when the user performs a trigger operation on the target item, acquire the user's user features and user behavior sequence, and acquire the item features of the target item.

[0137] The item recommendation module 72 is used to determine the items to be recommended based on the characteristics of the target item.

[0138] The intelligent UI recommendation module 73 is used to determine the correlation between each candidate UI element and the user and the item to be recommended, based on the user's user characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the element characteristics of the candidate UI elements in each area of ​​the information page, and to determine the target UI element to be displayed in each area of ​​the information page of the item to be recommended, based on the correlation between each candidate UI element and the user and the item to be recommended.

[0139] The display data acquisition module 74 is used to acquire display data for each target UI element.

[0140] The recommendation display module 75 is used to render and generate an item information stream containing the information page of each item to be recommended, based on the target UI elements displayed in each area of ​​the information page of each item to be recommended, the display data of the target UI elements, and the layout information of the information page.

[0141] The apparatus provided in this embodiment can be specifically used to perform the above-described... Figure 3 The specific functions and technical effects of the solutions provided in the corresponding method embodiments will not be elaborated here.

[0142] In an optional embodiment, when determining the correlation between each candidate UI element and the user and the item to be recommended based on the user's user characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the element characteristics of the candidate UI elements in each area of ​​the information page, the intelligent UI recommendation module is further configured to:

[0143] The user's characteristics and behavior sequence, the item characteristics of the item to be recommended, and the component characteristics of each candidate UI element are input into the trained intelligent UI recommendation model. The intelligent UI recommendation model predicts the correlation between each candidate UI element and the user and the item to be recommended. The correlation represents the probability that the user will select the item to be recommended based on the information page containing the candidate UI element.

[0144] In an optional embodiment, when determining the target UI element to be displayed in each area of ​​the information page of the item to be recommended based on the relevance of each candidate UI element to the user and the item to be recommended, the intelligent UI recommendation module is further configured to:

[0145] For each region, based on the relevance of each candidate UI element in that region to the user and the items to be recommended, and the UI element arrangement rules for that region, n candidate UI elements are selected as the target UI elements to be displayed in that region, where n is an integer and is less than or equal to the threshold for the number of elements to be displayed in that region.

[0146] In one alternative embodiment, the UI element arrangement rules for this area include at least one of the following:

[0147] The mutual exclusion relationship of candidate UI elements; the target UI elements displayed in this area do not include candidate UI elements with mutual exclusion relationships.

[0148] Candidate UI elements that are forced to be displayed; the target UI elements displayed in this area include the candidate UI elements that are forced to be displayed.

[0149] In one optional embodiment, at least one region's candidate UI elements have a display level, which indicates the prominence of the UI element when displayed. After selecting n candidate UI elements as target UI elements for display in each region based on the relevance of each candidate UI element in that region to the user and the recommended item, and the UI element arrangement rules for that region, the intelligent UI recommendation module is further configured to:

[0150] If the item to be recommended meets the preset downgrade conditions, then at least one target UI element will be replaced with another candidate UI element with a lower display level according to the preset downgrade conditions.

[0151] In one optional embodiment, when determining the item to be recommended based on the item characteristics of the target item, the item recommendation module is further configured to:

[0152] Based on the characteristics of the target item, a first item with the same category as the target item is recalled based on the similarity of the characteristics. Based on the theme scene corresponding to the target item, a second item with a different category than the target item in the same theme scene is recalled. The first item and the second item are then selected as items to be recommended.

[0153] In one alternative embodiment, the information page includes at least the following area: a tab area.

[0154] Each candidate UI element in the tag area is used to display a recommendation reason tag. The recommendation reason tags that the tag area can display must cover at least the following three types of information: product selling points, merchant qualification information, and whether the product is in stock.

[0155] The device provided in this embodiment can be used to execute the scheme provided in any of the above method embodiments. The specific functions and technical effects that can be achieved will not be described in detail here.

[0156] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an example embodiment of this application. Figure 8 As shown, the electronic device 80 includes a processor 801 and a memory 802 communicatively connected to the processor 801, the memory 802 storing computer execution instructions.

[0157] The processor executes computer execution instructions stored in the memory to implement the solution provided in any of the above method embodiments. The specific functions and technical effects that can be achieved will not be elaborated here.

[0158] This application also provides a computer-readable storage medium storing computer-executable instructions. When executed by a processor, the computer-executable instructions are used to implement the solution provided in any of the above method embodiments. The specific functions and technical effects to be achieved are not described here.

[0159] This application also provides a computer program product, which includes a computer program stored in a readable storage medium. At least one processor of the electronic device can read the computer program from the readable storage medium. The at least one processor executes the computer program to cause the electronic device to perform the solution provided in any of the above method embodiments. The specific functions and technical effects that can be achieved are not described here.

[0160] Furthermore, in some of the processes described in the above embodiments and accompanying drawings, multiple operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The sequence numbers are merely used to distinguish different operations, and the sequence number itself does not represent any execution order. Additionally, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types. "Multiple" means two or more, unless otherwise explicitly specified.

[0161] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0162] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A dynamic recommendation method for item information flow, characterized in that, The item information stream includes multiple information pages for recommended items. Each information page is divided into multiple areas, and each area displays at least one UI element. The method includes: Receive an item information stream acquisition request sent when a user triggers an operation on a target item, acquire the user's user characteristics and user behavior sequence, and acquire the item characteristics of the target item; Based on the characteristics of the target item, determine the item to be recommended; For each item to be recommended, based on the user's characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the component characteristics of the candidate UI components in each area of ​​the information page, the correlation degree between each candidate UI component and the user and the item to be recommended is determined. Based on the correlation degree between each candidate UI component and the user and the item to be recommended, one or more candidate UI components in each area with a high correlation degree with the current user and the item to be recommended are selected as the target UI components for that area. This determines the target UI components displayed in each area of ​​the information page for the item to be recommended. The correlation degree is used to characterize the probability that a user will select an item to be recommended based on the information page containing candidate UI components. Obtain the display data for each of the target UI elements; Based on the target UI elements displayed in each area of ​​the information page of each item to be recommended, the display data of the target UI elements, and the layout information of the information page, an item information stream containing the information page of each item to be recommended is rendered and generated.

2. The method according to claim 1, characterized in that, The step of determining the correlation between each candidate UI element and the user and the item to be recommended based on the user's user characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the element characteristics of the candidate UI elements in each area of ​​the information page includes: The user's user characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the component characteristics of each candidate UI element are input into the trained intelligent UI recommendation model. The intelligent UI recommendation model predicts the correlation degree between each candidate UI element and the user and the item to be recommended. The correlation degree represents the probability that the user will select the item to be recommended based on the information page containing the candidate UI element.

3. The method according to claim 1, characterized in that, The step of determining the target UI element to be displayed in each area of ​​the information page of the item to be recommended based on the correlation between each candidate UI element and the user and the item to be recommended includes: For each region, based on the relevance of each candidate UI element in that region to the user and the item to be recommended, and the UI element arrangement rules for that region, n candidate UI elements are selected as the target UI elements to be displayed in that region, where n is an integer and n is less than or equal to the threshold number of elements to be displayed in that region.

4. The method according to claim 3, characterized in that, The UI element arrangement rules for this area include at least one of the following: The mutual exclusion relationship of candidate UI elements; the target UI elements displayed in this area do not include candidate UI elements with mutual exclusion relationships. Candidate UI elements that are forced to be displayed; the target UI elements displayed in this area include the candidate UI elements that are forced to be displayed.

5. The method according to claim 3, characterized in that, At least one region's candidate UI elements have a display level, which indicates how prominent the UI element is when displayed. After selecting n candidate UI elements as the target UI elements for display in each region based on the relevance of each candidate UI element in that region to the user and the item to be recommended, and the UI element arrangement rules for that region, the process further includes: If the item to be recommended meets the preset downgrade conditions, then at least one target UI element will be replaced with another candidate UI element with a lower display level according to the preset downgrade conditions.

6. The method according to claim 1, characterized in that, The step of determining the item to be recommended based on the characteristics of the target item includes: Based on the characteristics of the target item, a first item with the same category as the target item is recalled based on the similarity of the characteristics, and a second item with a different category in the same theme scene as the target item is recalled based on the theme scene corresponding to the target item. The first item and the second item are selected as items to be recommended.

7. The method according to any one of claims 1-5, characterized in that, The information page includes at least the following areas: a tab area, Each candidate UI element in the tag area is used to display a recommendation reason tag. The recommendation reason tags that the tag area can display must cover at least the following three types of information: product selling points, merchant qualification information, and whether the product is in stock.

8. A dynamic recommendation device for item information flow, characterized in that, The item information stream includes multiple information pages recommending items. Each information page is divided into multiple areas, and each area displays at least one UI element. The device includes: The feature processing module is used to receive an item information stream acquisition request sent when a user triggers an operation on a target item, acquire the user's user features and user behavior sequence, and acquire the item features of the target item. The item recommendation module is used to determine the items to be recommended based on the characteristics of the target item; The intelligent UI recommendation module is used to determine the relevance of each candidate UI element to the user and the item to be recommended for each item to be recommended, based on the user's characteristics and user behavior sequence, the item characteristics of the item to be recommended, and the element characteristics of candidate UI elements in each area of ​​the information page. Based on the relevance of each candidate UI element to the user and the item to be recommended, one or more candidate UI elements in each area with high relevance to the current user and the item to be recommended are selected as target UI elements for that area, thus determining the target UI elements to be displayed in each area of ​​the information page of the item to be recommended. The relevance is used to characterize the probability that the user will select the item to be recommended based on the information page containing candidate UI elements. The display data acquisition module is used to acquire display data for each of the target UI elements. The recommendation display module is used to render and generate an item information stream containing the information page of each item to be recommended, based on the target UI element displayed in each area of ​​the information page of each item to be recommended, the display data of the target UI element, and the layout information of the information page.

9. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

Images