Content display method and device, electronic equipment, storage medium and program product

Abstract

The invention discloses a content display method and device, electronic equipment, a storage medium and a program product. The content display method and device can be applied to artificial intelligence fields such as computer vision, natural language processing and machine learning and can also be applied to large model fields such as pre-training models. The method comprises the following steps: displaying a subscription page, wherein the subscription page displays a plurality of subscription contents; determining read subscription content and to-be-read subscription content in the subscription page, wherein the to-be-read subscription content is displayed at an associated position of the read subscription content; in the subscription page, recommended content is displayed at the associated position of the to-be-read subscription content, and the recommended content is content determined by the read subscription content and the to-be-read subscription content. The display effect of the recommended content can be improved, so that the content touch rate of the recommended content is improved, and the user experience is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, specifically to a content display method, apparatus, electronic device, storage medium, and program product. Background Technology

[0002] In recent years, with the rapid development and continuous innovation of internet technology, content recommendation services have become an important means for content platforms to attract users and improve user experience. For example, content platforms can use machine learning and artificial intelligence algorithms to recommend content to users in a personalized way.

[0003] However, in current content recommendation services, recommended content and subscribed content are usually displayed separately on different pages or in different areas, resulting in poor display effects and difficulty in arousing users' interest in viewing them. This leads to a low probability of recommended content being exposed or viewed, i.e., low content reach. Summary of the Invention

[0004] This application provides a content display method, apparatus, electronic device, storage medium, and program product, which can improve the display effect of recommended content, thereby increasing the content reach rate of recommended content and enhancing user experience.

[0005] This application provides a content display method, including: displaying a subscription page, the subscription page displaying multiple subscription contents; determining read subscription contents and unread subscription contents in the subscription page, the unread subscription contents being displayed at an associated position of the read subscription contents; and displaying recommended content in the subscription page and at an associated position of the unread subscription contents, the recommended content being content determined by the read subscription contents and the unread subscription contents.

[0006] This application also provides a content display method, including: obtaining read subscription content and unread subscription content from a plurality of subscription content displayed on a subscription page; obtaining at least one candidate content, wherein the unread subscription content is displayed at an associated position of the read subscription content; determining recommended content from the candidate content based on the read subscription content and the unread subscription content; and determining the display position of the recommended content on the subscription page according to the associated position of the unread subscription content on the subscription page, so as to display the recommended content on the subscription page.

[0007] This application embodiment also provides a content display device, including: a display unit for displaying a subscription page, the subscription page displaying multiple subscription contents; a subscription content determination unit for determining read subscription contents and unread subscription contents in the subscription page, the unread subscription contents being displayed at an associated position of the read subscription contents; the display unit is further configured to display recommended content in the subscription page at an associated position of the unread subscription contents, the recommended content being content determined by the read subscription contents and the unread subscription contents.

[0008] In some implementations, the subscription content determination unit is specifically configured to: in response to a viewing operation of any of the subscription content in the subscription page, determine the subscription content corresponding to the viewing operation as the read subscription content, and determine the subscription content associated with the read subscription content as the unread subscription content.

[0009] This application embodiment also provides a content display device, including: an acquisition unit, configured to acquire read subscription content and unread subscription content from a plurality of subscription content displayed on a subscription page, and acquire at least one candidate content, wherein the unread subscription content is displayed at an associated position of the read subscription content; a recommended content determination unit, configured to determine recommended content from the candidate content based on the read subscription content and the unread subscription content; and a display position determination unit, configured to determine the display position of the recommended content on the subscription page according to the associated position of the unread subscription content on the subscription page, so as to display the recommended content on the subscription page.

[0010] In some implementations, the recommended content determination unit includes a first recommended content determination subunit and a second recommended content determination subunit, comprising: the first recommended content determination subunit, configured to predict the content reach rate of the candidate content to the object to be recommended based on the object information of the object to be recommended, the read subscription content, and the subscription content to be read, wherein the object to be recommended is the object corresponding to the subscription page; and the second recommended content determination subunit, configured to determine recommended content from the candidate content based on the content reach rate.

[0011] In some implementations, predicting the content reach rate of the candidate content to the object to be recommended based on the object information of the object to be recommended, the read subscription content, and the subscription content to be read includes: predicting a first viewing probability of the object to be recommended viewing the candidate content based on the association between the object information of the object to be recommended and the candidate content; predicting the exposure rate of the candidate content on the subscription page based on the association between the candidate content, the read subscription content, and the subscription content to be read; and determining the content reach rate of the candidate content to the object to be recommended by combining the first viewing probability of the candidate content and the exposure rate of the candidate content on the subscription page.

[0012] In some implementations, predicting the exposure rate of the candidate content on the subscription page based on the correlation between the candidate content, the read subscription content, and the unread subscription content includes: obtaining a second viewing probability of the target object viewing the unread subscription content, and obtaining first viewing operation data of the unread subscription content; using pre-trained exposure rate prediction parameters, based on the second viewing probability of the candidate content, the read subscription content, the unread subscription content, and the first viewing operation data of the unread subscription content, to predict the exposure rate of the candidate content on the subscription page.

[0013] In some implementations, determining recommended content from the candidate content based on the content reach rate includes: obtaining a second viewing probability of the target audience viewing the subscribed content to be read; comparing the content reach rate with the second viewing probability of the subscribed content to be read, so as to determine recommended content from the candidate content based on the comparison result.

[0014] In some implementations, comparing the content reach rate with the second viewing probability of the subscribed content to determine recommended content from the candidate content based on the comparison result includes: determining candidate content from the candidate content, wherein the content reach rate of the candidate content exceeds the second viewing probability of the subscribed content to be read; and determining recommended content from the candidate content based on the content reach rate of the candidate content.

[0015] In some embodiments, the content display device further includes a subscription unit, which is configured to: obtain object information of the object to be recommended based on the subscription operation of the object to be recommended to the subscribed content; and predict the second viewing probability of the object to be recommended viewing the subscribed content based on the association between the object information of the object to be recommended and the subscribed content.

[0016] In some implementations, obtaining the second viewing probability of the object to be recommended viewing the subscribed content to be read includes: obtaining the second viewing probability of the object to be recommended viewing the subscribed content to be read from the second viewing probability of the subscribed content.

[0017] In some implementations, the object information includes second viewing operation data of the object to be recommended on the subscribed content, and the content display device further includes an update unit, which is used to: update the second viewing operation data of the object to be recommended on the subscribed content on the subscription page according to the viewing operation of the object to be recommended on the subscription page, so as to obtain the updated object information of the object to be recommended, and the updated object information is used to predict the content reach rate.

[0018] This application also provides an electronic device, including a processor and a memory, wherein the memory stores a plurality of instructions; the processor loads instructions from the memory to execute steps in any of the content display methods provided in this application.

[0019] This application also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute steps in any of the content display methods provided in this application.

[0020] This application also provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the steps in any of the content display methods provided in this application.

[0021] This application embodiment can display a subscription page, which displays multiple subscription contents; determine the read subscription contents and unread subscription contents on the subscription page, and display the unread subscription contents in an associated position on the read subscription contents; display recommended content on the subscription page in an associated position on the unread subscription contents, and the recommended content is content determined by the read subscription contents and the unread subscription contents.

[0022] In this application, recommended content and its display location are determined based on the unread subscription content. This ensures that the recommended content aligns with the unread subscription content on the subscription page in terms of content information and display location, allowing the recommended content to integrate more naturally and seamlessly into the user's reading experience. This improves the display effect of the recommended content, enhances the user's immersion in the subscription page content, and increases their willingness to continue reading, thereby increasing the content reach of the recommended content and improving the user experience. Specifically, after the unread subscriber views the subscription content, the recommended content is determined based on the read subscription content and the unread subscription content displayed in association with it. This ensures that the recommended content better matches the needs and interests of the subscriber on the subscription page, further improving the display effect of the recommended content, increasing its reach, and enhancing the user experience. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0024] Figure 1a This is a schematic diagram of a scenario illustrating the content display method provided in an embodiment of this application;

[0025] Figure 1b This is a flowchart illustrating the content display method provided in an embodiment of this application;

[0026] Figure 1c This is a schematic diagram of the subscription page provided in an embodiment of this application;

[0027] Figure 1d This is a schematic diagram of the detailed page for displaying the subscription page as provided in the embodiments of this application;

[0028] Figure 1e This is a schematic diagram of the update subscription page provided in an embodiment of this application;

[0029] Figure 2 This is a flowchart illustrating a content display method provided in another embodiment of this application;

[0030] Figure 3a This is a schematic diagram of the content display system provided in the embodiments of this application;

[0031] Figure 3b This is a flowchart illustrating a content display method provided in yet another embodiment of this application;

[0032] Figure 4 This is a schematic diagram of the structure of the content display device provided in the embodiments of this application;

[0033] Figure 5 This is a schematic diagram of the structure of a content display device provided in another embodiment of this application;

[0034] Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0035] 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. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] This application provides a content display method, apparatus, electronic device, storage medium, and program product.

[0037] Specifically, the content display device can be integrated into an electronic device, such as a terminal or server. The terminal can be a mobile phone, tablet, smart Bluetooth device, laptop, or personal computer (PC); the server can be a single server or a server cluster consisting of multiple servers.

[0038] In some embodiments, the content display device may also be integrated into multiple electronic devices, such as multiple servers, with multiple servers implementing the content display method of this application.

[0039] In some embodiments, the server may also be implemented as a terminal.

[0040] For example, refer to Figure 1aThis content display method is applied to a content display system, which includes a terminal and a server. The terminal displays a subscription page containing multiple subscribed content items. It identifies read and unread subscription content on the subscription page, with the unread content displayed in an associated position within the read content. Recommended content, determined by the read and unread content, is displayed in the associated position within the subscription page and the unread content. The server retrieves the read and unread subscription content from the multiple subscription content items displayed on the terminal's subscription page, and retrieves at least one candidate content item. Based on the read and unread content items, it determines recommended content from the candidate content. Finally, based on the associated position within the subscription page and the unread content item, it determines the display position of the recommended content on the subscription page, thus displaying the recommended content on the terminal's subscription page.

[0041] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0042] The following sections provide detailed descriptions. It should be noted that the order of the following embodiments is not intended to limit the preferred order of the embodiments. It is understood that in the specific embodiments of this application, user-related data such as object information of the object to be recommended, subscription content, candidate content, recommended content, viewing operations, scrolling operations, and subscription operations are involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0043] Artificial intelligence (AI) is a technology that uses digital computers to simulate human perception of the environment, acquisition of knowledge, and use of that knowledge. This technology can enable machines to possess functions similar to human perception, reasoning, and decision-making. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly include computer vision, speech processing, natural language processing, as well as machine learning / deep learning, autonomous driving, and intelligent transportation.

[0044] Computer vision (CV) is a technology that uses computers to perform operations such as recognition, measurement, and further processing of target images, replacing the human eye. Computer vision technology typically includes image processing, image recognition, image semantic understanding, image retrieval, virtual reality, augmented reality, simultaneous localization and mapping (SLAM), autonomous driving, intelligent transportation, and other technologies. It also includes common biometric recognition technologies such as facial recognition and fingerprint recognition. For example, image processing techniques include image colorization and image outline extraction.

[0045] Natural Language Processing (NLP) is an important field within computer science and artificial intelligence. It studies the theories and methods for enabling effective communication between humans and computers using natural language. NLP is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field involves natural language—the language people use in daily life—and thus it has a close relationship with linguistic research. NLP techniques typically include text processing, semantic understanding, machine translation, question answering, and knowledge graphs.

[0046] Machine Learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instruction-based learning.

[0047] Autonomous driving technology typically includes high-precision maps, environmental perception, behavior decision-making, path planning, motion control, and other technologies, and autonomous driving technology has broad application prospects.

[0048] With the research and advancement of artificial intelligence (AI) technology, AI is being studied and applied in various fields, such as smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, autonomous driving, drones, robots, smart healthcare, smart customer service, vehicle networking, and smart transportation. It is believed that with the development of technology, AI will be applied in more fields and play an increasingly important role.

[0049] Pre-trained models, also known as foundational models or large models, refer to deep neural networks (DNNs) with a large number of parameters. These DNNs are trained on massive amounts of unlabeled data. Leveraging the function approximation capabilities of large-parameter DNNs, Proximity-Based Transformers (PTMs) extract common features from the data. Through fine-tuning, efficient parameter fine-tuning (PEFT), and prompt-tuning techniques, they are suitable for downstream tasks. Therefore, pre-trained models can achieve ideal results in small-shot or zero-shot scenarios. PTMs can be categorized according to the data modality they process, such as language models (ELMO, BERT, GPT), visual models (Swin-transformer, ViT, V-MOE), speech models (VALL-E), and multimodal models (ViBERT, CLIP, Flamingo, Gato). Multimodal models refer to models that establish feature representations for two or more data modalities. Pre-trained models are important tools for outputting AI-generated content (AIGC) and can also serve as a general interface connecting multiple task-specific models.

[0050] In this embodiment, a content display method is provided, such as... Figure 1b As shown, this content display method is applied to the terminal, and the specific process is as follows:

[0051] 110. Display the subscription page, which shows multiple subscription items.

[0052] In this context, a subscription page refers to the page or interface on a content platform used to browse subscribed content. Subscribed content refers to the content published by the subscribed entity, such as articles, images, videos, or other types of multimedia content. The subscribed entity can be any target on the content platform that can be subscribed to, such as the entity or topic that publishes content on the platform, and so on.

[0053] This application's embodiments can be applied to content recommendation scenarios on content platforms. The content of the content platform can differ in different application scenarios. For example, the content of a news publishing platform can be news articles published by news media or other entities, or news articles within news topics. The content of a video sharing platform can be videos published by video publishers or other entities, or videos within interactive topics. The content of a social media application platform can be multimedia content published by content publishers or other entities such as public accounts, video accounts, or other specific accounts, or multimedia content within interactive topics.

[0054] Users of a content platform can subscribe to accounts of subscribed entities or specific topics using their user accounts for subsequent actions. For example, after a user opens the subscription page through an application running on their terminal, the terminal can display something like... Figure 1c The subscription page shown displays preview cards of the latest subscription content A through C from the multiple objects the user has subscribed to.

[0055] 120. Determine the read subscription content and unread subscription content on the subscription page. Unread subscription content will be displayed in the same location as the read subscription content.

[0056] In this context, "read subscription content" refers to subscription content that has already been viewed. "Unread subscription content" refers to subscription content displayed in the location associated with the read subscription content; this can be unviewed subscription content or previously viewed subscription content. The associated location of the read subscription content refers to the position on the subscription page that is related to its display location. For example, the associated location of the read subscription content can be within a preset range, such as above, below, to the left, or to the right of the read subscription content, etc.

[0057] In some implementations, the associated location of read subscription content is a location determined by a specified page orientation and associated with the display position of the read subscription content on the subscription page.

[0058] For example, in some implementations, the specified page orientation is related to the arrangement of subscribed content on the subscription page. For instance, when the subscribed content on the subscription page is arranged sequentially from top to bottom, the default page orientation is downward, and the subscription content to be read can be the next subscription content located below the already read subscription content. Alternatively, when the subscribed content on the subscription page is arranged sequentially from left to right, the default page orientation is right, and the subscription content to be read can be the next subscription content located to the right of the already read subscription content.

[0059] For example, in some implementations, the specified page direction is determined based on the subscription content based on the historical scrolling operations on the subscription page. For instance, the scrolling operations performed by the user on the subscription page after opening it can be recorded, including the scrolling direction, the number of operations in each scrolling direction, or the operation duration. The scrolling direction with the most cumulative operations or the scrolling direction with the longest cumulative operation duration can be used as the specified page direction.

[0060] In some implementations, currently viewed subscription content can be designated as read subscription content. This allows recommended content determined based on read subscription content to better align with current needs and interests, increasing click-through rates and exposure, thereby improving content reach. Specifically, determining read subscription content and unread subscription content on the subscription page includes:

[0061] In response to a viewing action on any subscribed content on the subscription page, the subscribed content corresponding to the viewing action is identified as read subscribed content, and the subscribed content associated with the read subscribed content is identified as unread subscribed content.

[0062] The "view" action refers to the action of viewing subscribed content on the subscription page. Viewing actions can include, but are not limited to, one or more combinations of methods such as clicking, keyboard shortcuts, gestures, voice commands, and shaking.

[0063] For example, a user can click to view a preview card of any subscription content on the subscription page, such as the preview card of subscription content A. The subscription content A that has been clicked to view is the subscription content that has been read. On the subscription page, the unviewed subscription content B displayed in the associated position of subscription content A is the subscription content to be read.

[0064] In some implementations, a detailed page of the subscription content can be displayed on the terminal in response to a viewing operation of the subscription content on the subscription page. For example, such as Figure 1d The diagram shown illustrates the process of redirecting to the detailed page of the subscription page. When a user clicks on the preview card of subscription content A in the subscription page, the detailed page of subscription content A shown in the diagram will be displayed on the terminal.

[0065] 130. On the subscription page, in the relevant position of the subscription content to be read, the recommended content is determined by the subscription content already read and the subscription content to be read.

[0066] For example, the associated position of the subscription content to be read can be a position within a preset range of the subscription content to be read, such as the location of the subscription content to be read, above, below, to the left or to the right of the subscription content to be read, etc., so that the subscription page can display the subscription content and recommended content in a mixed manner by displaying the recommended content at the associated position of the subscription content to be read in the subscription page.

[0067] It should be noted that directly displaying subscription content and recommended content together on the subscription page would lead to a cluttered and poorly presented display, causing user resistance and resulting in a low probability of recommended content being exposed or viewed, i.e., low content reach. However, in this embodiment, recommended content and its display position are determined based on the subscription content to be read, ensuring that the recommended content is consistent with the subscription content to be read on the subscription page in terms of content information and display position. This allows recommended content to dynamically and generatively appear on the subscription page as the user views the subscription content, making the recommended content more naturally and smoothly integrated into the user's reading experience. This improves the display effect of recommended content, enhances the user's immersion in reading the subscription page content and their willingness to continue reading, thereby increasing the click-through rate and exposure rate of recommended content, i.e., increasing the content reach rate of recommended content, avoiding the waste of computing power in the content recommendation service, and improving the user experience.

[0068] In some implementations, the associated location of the subscription content to be read is a location determined by a specified page orientation and associated with the display position of the subscription content on the subscription page.

[0069] In some implementations, the original display position of the subscription content to be read on the subscription page can be used as the display position of the candidate content on the subscription page. Specifically, displaying recommended content in the associated position of the subscription content to be read includes: displaying recommended content on the subscription page at the original display position of the subscription content to be read, and displaying the subscription content to be read at an updated display position, where the updated display position is the position where the original display position of the subscription content to be read on the subscription page has been moved according to a specified page direction. For example, it can be as follows: Figure 1e The diagram shown illustrates the updated subscription page. Recommended content is displayed in the original subscription page (i.e., the subscription page before the operation of viewing read subscription content) at the position of the unread subscription content (i.e., the original display position). At the same time, the unread subscription content in the original subscription page and the subscription content arranged after it are arranged sequentially to update the subscription page, resulting in the updated subscription page shown in the diagram.

[0070] In some implementations, the updated subscription page displaying recommended content may be displayed in response to an exit operation on the details page of read subscription content, or in response to a view operation or refresh operation on the subscription page.

[0071] The content display scheme provided in this application can be applied to various content recommendation scenarios. For example, taking a content platform as an example, a subscription page is displayed, showing multiple subscribed contents; the read and unread subscription contents on the subscription page are determined, and the unread subscription contents are displayed in the associated position of the read subscription contents; recommended content is displayed in the associated position of the unread subscription contents on the subscription page, and the recommended content is determined by the read and unread subscription contents.

[0072] As can be seen from the above, in this embodiment, the recommended content and its display position are determined based on the subscription content to be read, so that the recommended content is consistent with the subscription content to be read on the subscription page in terms of content information and display position. This makes the recommended content more naturally and smoothly integrated into the user's reading experience of the subscription content, thereby improving the display effect of the recommended content, enhancing the user's immersion in reading the subscription page content and their willingness to continue reading, thus increasing the content reach rate of the recommended content and improving the user experience. Specifically, in this embodiment, after the target audience views the subscription content, the recommended content is determined based on the read subscription content and the subscription content to be read displayed in association with the read subscription content. This makes the recommended content more aligned with the needs and interests of the target audience on the subscription page, further improving the display effect of the recommended content, increasing the content reach rate of the recommended content, and improving the user experience.

[0073] In this embodiment, a content display method is provided, such as... Figure 2 As shown, this content display method is applied to the server, and the specific process is as follows:

[0074] 210. Retrieve read subscription content and unread subscription content from multiple subscription content displayed on the subscription page, and retrieve at least one candidate content. The unread subscription content is displayed in the associated position of the read subscription content.

[0075] Candidate content refers to a set of content that is intended to be recommended to the target audience. For example, candidate content can be a set of content that a content platform selects and matches through recommendation algorithms to provide to users who may be interested in it, or it can be a set of content manually set according to application scenarios or actual needs.

[0076] For example, after a user clicks on a preview card of any subscription content on the subscription page, such as the preview card of subscription content A, the user can obtain subscription content A and the subscription content B to be read located in the associated position of subscription content A based on the viewing operation.

[0077] In some implementations, events corresponding to viewing operations can be detected in real time to provide immediate content recommendations, thereby improving the timeliness of content recommendations and enhancing user experience. Specifically, this involves obtaining read and unread subscription content from multiple subscriptions displayed on the subscription page, as well as obtaining at least one candidate content, including:

[0078] Create event detection code, which is used to detect viewing events triggered by viewing operations;

[0079] Place the event detection code into the application corresponding to the subscription page to detect and view events on the subscription page;

[0080] When a view event is detected, retrieve the read subscription content and the unread subscription content from the multiple subscription content displayed on the subscription page, and retrieve at least one candidate content.

[0081] In this context, a viewing event refers to an event triggered by a viewing action. For example, event detection code can be pre-designed to detect click events triggered by user clicks. This event detection code is then placed into the application of the content platform running on the terminal. When a user clicks on any subscribed content on the subscription page, a corresponding click event is triggered. The event detection code can capture this click event in real time and send the data to a backend server for processing via a real-time data stream processing system (such as Apache Kafka or Amazon Kinesis). On the backend server, a real-time computing framework (such as Apache Flink or Spark Streaming) is used to process the real-time data stream to perform content recommendation tasks. A real-time data stream is a continuously generated data stream, typically containing real-time events, logs, sensor data, etc. Real-time data streams require real-time processing and analysis to enable immediate action after data generation. Real-time data stream processing systems and real-time computing frameworks can help efficiently process and analyze real-time data streams, enabling real-time data-driven applications and services.

[0082] In some implementations, content with a causal or event chain relationship to read and subscribed content can be used as candidate content to increase the relevance between recommended content and the user's currently viewed and subscribed content. This makes the recommended content more aligned with current needs and interests, increasing click-through rates and exposure, thereby improving content reach and avoiding wasted computing power in the content recommendation service. Specifically, at least one candidate content is obtained, including:

[0083] Content that is related to the read and subscribed content will be selected as candidate content. The relationship can include at least one of the following: causal relationship or event chain relationship.

[0084] In this context, causality refers to a cause-and-effect relationship between multiple pieces of content; that is, the occurrence or change of one piece of content leads to the occurrence or change of another piece of content. For example, an article on a content platform introducing healthy eating practices and another article introducing the benefits of healthy eating have a causal relationship.

[0085] In this context, an event chain relationship refers to the correlation between a series of events. In this embodiment, an event chain relationship represents the ability of multiple pieces of content to be connected sequentially in time to form an event chain. For example, multiple video slices obtained from segmenting a long video have an event chain relationship. Alternatively, if a user first subscribes to article A, then views article B, and subsequently views article C, these events form an event chain between articles A, B, and C; that is, articles A, B, and C have an event chain relationship.

[0086] For example, in this embodiment, various contents of the content platform (including subscribed content and other content) can be stored in a content database for easy retrieval and recommendation. The content tags of each content stored in the content database can be compared with the content tags of read and subscribed content to determine whether there is a causal relationship or event chain relationship based on the content tags. This yields candidate content that has complementary information or a sequential event relationship with the recommended content. The less interference these candidate contents cause to the user, the more easily they are accepted by the user. This determination of causal relationship or event chain relationship can be made through a content tag relationship graph, or it can be done manually by comparing content tags.

[0087] In some implementations, content tags can be added to the content to facilitate candidate content searching and content recommendation based on these tags. Content tags are metadata used to describe the subject, characteristics, or attributes of the content. Content tags can differ in different application scenarios. For example, tags such as article type, region, subject, author, keywords, and publication time can be added to article-type content; tags such as video type, theme, author, shot, and publication time can be added to video-type content; and tags such as image content, shooting location, author, color, and publication time can be added to image-type content. In the embodiments of this application, the content tags for any content may include tag information added by the content creator, administrator, or other entities, or may include keywords or feature information extracted from the content.

[0088] 220. Based on read and unread subscription content, determine recommended content from the candidate content.

[0089] For example, the relevance of candidate content to read and subscribed content, such as cosine similarity or correlation coefficient, can be calculated, and recommended content can be selected from the candidate content based on the calculated relevance.

[0090] In some implementations, recommended content can be determined based on viewed read subscription content, unread subscription content displayed in association with the read subscription content, and object information of the target audience. This makes the recommended content more aligned with the needs and interests of the target audience, increasing the click-through rate and exposure rate of the recommended content, thereby improving the content reach and avoiding wasted computing power in the content recommendation service. Specifically, determining recommended content from candidate content based on read subscription content and unread subscription content includes:

[0091] Based on the object information of the object to be recommended, the read subscription content, and the subscription content to be read, predict the content reach rate of the candidate content to the object to be recommended. The object to be recommended is the object corresponding to the subscription page.

[0092] Recommended content is determined from candidate content based on content reach.

[0093] The "object to be recommended" refers to the entity to which content will be recommended. This object can be an individual, organization, brand, or other entity. For example, the object to be recommended could be a user associated with the subscription page displayed on the terminal of the content platform, such as a user account logged into that subscription page.

[0094] Among them, the object information of the object to be recommended refers to the information related to the object to be recommended, such as the object's historical behavior, object tags (such as interests), attribute information, etc.

[0095] Content reach rate refers to the probability that content is accessed. In some implementations, content reach rate can characterize the joint probability of content being shown to the target audience (i.e., exposure rate) and the probability of content being viewed by the target audience (i.e., viewing probability). In other words, content reach rate can be a combination of exposure rate and viewing probability.

[0096] For example, with the user's individual permission or consent, information about the target audience, the subscription content to be read, and candidate content can be obtained. This information, along with the content of the candidate content, can be input into a pre-trained neural network model to predict the reach rate of the candidate content. The pre-trained neural network model can be one or a combination of convolutional neural networks (CNNs), recurrent neural networks (RNNs), attention-based neural networks, and fully connected networks. Alternatively, the correlation between the candidate content and the target audience's information and subscription content, such as cosine similarity or correlation coefficient, can be calculated, and this correlation can be used as the predicted reach rate of the candidate content. Candidate content can be ranked based on its reach rate, and the candidate content with the highest reach rate or multiple candidate content with high reach rates can be recommended. Alternatively, candidate content with a reach rate exceeding a preset threshold can be recommended.

[0097] In some implementations, after the target user views the subscribed content, the relevant viewing operation data can be updated based on the user's viewing action to obtain updated user information, thereby achieving more accurate and real-time content recommendation. Specifically, the user information includes second viewing operation data of the target user on the subscribed content. The content display method also includes:

[0098] Based on the viewing actions of the target audience on the subscription page, update the second viewing action data of the target audience on the subscription content to obtain the updated target audience information. The updated target audience information is used to predict the content reach rate.

[0099] View operation data refers to data related to viewing content. For example, view operation data may include one or more of the following: object identifier, operation time, number of views, and viewing duration. Secondary view operation data refers to data on the viewing actions of the target audience on the subscribed content.

[0100] For example, taking the view operation as a click operation, when a user clicks on the preview card of any subscribed content on the subscription page, the operation time of the click operation, the number of click operations, and the viewing duration of the subscribed content can be recorded in the view operation data of the object information to update the object information of the object to be recommended. The updated object information can be used to predict the content reach rate of candidate content of the object to be recommended.

[0101] In some implementations, the object information of the objects to be recommended can be updated periodically. This ensures that the object information in the content platform remains up-to-date, and by setting a reasonable update cycle, system load can be reduced, system response speed improved, and user experience enhanced. Specifically, the object information includes the second viewing operation data of the objects to be recommended on the subscribed content. The content display method further includes: updating the second viewing operation data of the objects to be recommended on the subscribed content according to a preset update cycle to obtain the updated object information of the objects to be recommended. The updated object information is used to predict the content reach rate. The preset update cycle can be a pre-set update cycle that can be set according to actual needs or application scenarios. For example, the object information of all objects in the content platform can be updated every 30 seconds.

[0102] In some implementations, the content reach rate of candidate content can be comprehensively measured by combining the probability of the target audience viewing the candidate content and the exposure rate of the candidate content. This ensures that the content reach rate reflects the likelihood of the candidate content being viewed and exposed by the target audience, thereby more accurately and comprehensively evaluating the candidate content and improving the accuracy of content recommendations. Specifically, based on the target audience's information, read subscriptions, and unread subscriptions, the content reach rate of candidate content is predicted, including:

[0103] Based on the relationship between the object information of the object to be recommended and the candidate content, predict the probability of the object to be recommended viewing the candidate content first.

[0104] Based on the relationship between candidate content, read subscription content, and unread subscription content, predict the exposure rate of candidate content on the subscription page;

[0105] By combining the first-view probability of candidate content with the exposure rate of candidate content on the subscription page, the content reach rate of candidate content to be recommended is determined.

[0106] Here, the viewing probability refers to the probability that an object will view the content. The viewing probability reflects an object's tendency to view the content; the higher the viewing probability, the greater the likelihood that the content will be viewed by the object. In this embodiment, the first viewing probability of the object to be recommended viewing the candidate content represents the probability that the object to be recommended will view the candidate content. The exposure rate of the candidate content on the subscription page refers to the probability that it will be exposed to the object to be recommended within the subscription content. The higher the exposure rate, the greater the probability that the candidate content will be exposed to the object to be recommended within the subscription content. This exposure rate is related to the relevance between the candidate content and the subscription content of the object to be recommended. It can be understood that the higher the relevance between the candidate content and the subscription content of the object to be recommended, the higher the probability that the candidate content will be exposed to the object to be recommended within the subscription content, i.e., the higher the exposure rate on the subscription page.

[0107] For example, the object information of the recommended object and candidate content can be input into a pre-trained viewing probability prediction model, so that the pre-trained viewing probability prediction model can predict the first viewing probability of the recommended object viewing the candidate content. Similarly, candidate content, read subscription content, and unread subscription content can be input into a pre-trained exposure rate prediction model, so that the exposure rate prediction model can predict the exposure rate of the candidate content on the subscription page. The pre-trained viewing probability prediction model and the pre-trained exposure rate prediction model can be one or more combinations of Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), attention-based neural networks, and fully connected networks.

[0108] Alternatively, the relevance between the object information of the recommended object and the candidate content can be calculated, such as cosine similarity or correlation coefficient, and the calculated relevance can be used as the first probability that the recommended object will view the candidate content. Furthermore, the relevance between the candidate content and read / subscribed content as well as unread / subscribed content can be calculated, such as cosine similarity or correlation coefficient, and the calculated relevance can be used as the exposure rate of the candidate content on the subscription page.

[0109] In this embodiment, for any candidate content, the first viewing probability of the recommended object to view the candidate content and the exposure rate of the candidate content on the subscription page can be combined using one or more operations such as addition and multiplication to obtain the content reach rate of the candidate content to the recommended object. In this way, the content reach rate of all candidate content can be calculated.

[0110] In some implementations, the content reach rate of a candidate content can be obtained by multiplying its first view probability and exposure rate. Specifically, determining the content reach rate of a candidate content for a recommended object by combining its first view probability and exposure rate on the subscription page includes: using the product of the candidate content's first view probability and its exposure rate on the subscription page as the content reach rate of the candidate content for the recommended object. For example, for any candidate recommended object, the content reach rate of the candidate recommended object = the candidate content's first view probability × the candidate content's exposure rate on the subscription page.

[0111] In some implementations, the correlation between the object to be predicted and the candidate recommended objects can be captured by pre-trained viewing probability prediction parameters using the merged features of object information and candidate content, thereby improving the accuracy of the predicted viewing probability. Specifically, based on the correlation between the object information of the object to be recommended and the candidate content, the first viewing probability of the object to be recommended viewing the candidate content is predicted, including:

[0112] Obtain object characteristics and candidate content characteristics from object information;

[0113] The first merged feature is obtained by concatenating the object features and the content features of the candidate content.

[0114] By using pre-trained view probability prediction parameters, the view probability of the first merged feature is predicted to obtain the first view probability of the candidate content to be recommended.

[0115] Here, object features refer to the feature representation of object information. Content features of candidate content refer to the feature representation of candidate content. In this embodiment, the object features of the object to be recommended can be feature representations extracted from object information. The content features of candidate content can be feature representations extracted from candidate content, and / or feature representations extracted from external information carried by the candidate content, such as tags or other information. The first merged feature refers to the feature obtained by concatenating object features and content features of candidate content.

[0116] Here, the view prediction probability parameter refers to the parameter used to predict the view probability. Pre-trained view probability prediction parameters can refer to view probability prediction parameters learned through the training process. For example, view prediction probability parameters may include a set of weight parameters that can be used for linear classification or fully connected processing, and so on.

[0117] For example, a pre-trained viewing probability prediction model may include a feature extraction network and a viewing probability prediction network, with the pre-trained viewing probability prediction parameters being the network parameters of the viewing probability prediction network. The feature extraction network can extract object features and candidate content features from the object information. The extracted object features and content features are then concatenated and input into the viewing probability prediction network to predict the viewing probability, outputting the first viewing probability of the object to be recommended for each candidate content. This application does not limit the specific form of the feature extraction network; for example, the feature extraction network can be one or more combinations of neural networks that can be used for feature extraction, such as convolutional neural networks (CNN), recurrent neural networks (RNN), and neural networks based on attention mechanisms.

[0118] In some implementations, the view probability prediction network can be a fully connected network. The view probability prediction parameters can be those of a fully connected network, such as weight parameters, bias parameters, etc. For example, a pre-trained fully connected network, such as an MLP (Multi-Layer Perceptron), can be used to input merged features. This allows the network to learn the complex relationship between input features and view probabilities through fully connected processing, mapping the merged features of object information and recommended content to the output value. Then, an activation function, such as the sigmoid function, is used in the output layer of the fully connected network to map the output value to the 0-1 range, yielding the view probability. The MLP is a basic feedforward neural network consisting of at least three layers of neurons: an input layer, hidden layers, and an output layer. For each layer, linear and non-linear transformations are combined for fully connected processing.

[0119] In some implementations, the exposure rate of candidate content can be measured by combining the second viewing probability of the content to be read, the viewing operation data of the already read content, and the candidate content. The second viewing probability of the content to be read and the viewing operation data of the already read content can accurately and quantitatively reflect the interest and engagement of the target audience with the already read content and the candidate content, thereby improving the accuracy of the predicted exposure rate of the candidate content. Specifically, based on the correlation between the candidate content, the already read content, and the content to be read, the exposure rate of the candidate content on the subscription page is predicted, including:

[0120] Get the second viewing probability of the recommended object to view the subscribed content to be read, and get the first viewing operation data of the subscribed content to be read;

[0121] By using pre-trained exposure rate prediction parameters, and based on the second viewing probability of candidate content, read subscription content, unread subscription content, and the first viewing operation data of unread subscription content, the exposure rate of candidate content on the subscription page is predicted.

[0122] Here, the second viewing probability of the target object viewing the subscribed content represents the probability that the target object will view the subscribed content. In some implementations, the second viewing probability of the target object viewing the subscribed content can be predicted based on the association between the target object's object information and the subscribed content. For example, in some implementations, a pre-trained viewing probability prediction model can be used to predict both the first and second viewing probabilities. The target object's object information and candidate content can be input into the pre-trained viewing probability prediction model to predict the second viewing probability.

[0123] The first viewing operation data refers to the data on viewing operations of the content to be subscribed to. In this embodiment, the first viewing operation data may include only the data on the viewing operations of the recommended object on the content to be subscribed to, or it may include the data on the viewing operations of all objects on the content to be subscribed to, depending on the actual application scenario or needs.

[0124] Exposure prediction parameters refer to the parameters used to predict exposure rates. Pre-trained exposure prediction parameters can refer to exposure prediction parameters learned through the training process. For example, exposure prediction parameters may include a set of weight parameters that can be used for linear classification or fully connected processing, and so on.

[0125] For example, a pre-trained exposure prediction model can include an exposure prediction network. This network can predict the first view probability of the recommended object viewing the candidate content based on the second view probability of candidate content, read subscription content, unread subscription content, and the second view operation of unread subscription content.

[0126] In some implementations, view operation tags can be added to content to retrieve first view operation data. View operation tags are tags used to record relevant data about viewing operations performed on content. For example, the view operation tags for any content may include one or more of the following: the object identifier of the subscribed object viewing the content, the operation time, the number of views, and the viewing duration, where the subscribed object refers to the object subscribed to the content. It should be noted that when a subscribed object views any content, the relevant data in the view operation tags for that content can be updated.

[0127] In some implementations, when a target user subscribes to any content, the second viewing probability of that user viewing the newly added subscription can be predicted in real time, thus pre-determining the second viewing probabilities for all subscriptions. This allows the target user to directly and quickly obtain the second viewing probability of the subscribed content from the pre-determined second viewing probabilities when performing a viewing operation, thereby improving the response speed of content recommendation and achieving real-time content recommendation. Specifically, before obtaining the second viewing probability of the target user viewing the subscribed content, the process further includes:

[0128] Based on the subscription operations of the target audience to the subscribed content, obtain the object information of the target audience to be recommended;

[0129] Based on the relationship between the object information of the object to be recommended and the subscribed content, predict the probability of the object to be recommended viewing the subscribed content a second time;

[0130] Obtain the second probability of the recommended user viewing the subscribed content, including:

[0131] From the second viewing probability of the subscribed content, obtain the second viewing probability of the recommended object viewing the subscribed content to be read.

[0132] In this context, a subscription operation refers to the act of subscribing to content. Subscription operations can include, but are not limited to, one or more combinations of methods such as clicking, keyboard shortcuts, gesture operations, voice commands, and shaking. For example, in this embodiment, one can subscribe to the content published by an entity or topic that publishes content on a content platform by following or subscribing to that entity or topic; this content is the subscribed content.

[0133] For example, when a user subscribes to any content publisher on a content platform, they can access the content published by that publisher (i.e., the subscribed content) and the user's object information. A view probability prediction network is used to predict the view probability of the acquired object information and subscribed content to obtain a second view probability for each subscribed content published by that publisher. In practical applications, a data table can be created to store the second view probabilities of subscribed content. After a user performs a subscription operation, the latest predicted second view probability of the subscribed content is stored in this data table. This data table can also store the content identifier of the subscribed content and its corresponding second view probability. To obtain the second view probability of any subscribed content (such as content to be read), the second view probability of that subscribed content can be retrieved from the data table based on its content identifier.

[0134] In some implementations, the accuracy of predicted viewing probabilities can be improved by capturing the association between the object to be predicted and the subscribed content using pre-trained viewing probability prediction parameters and merged features of object information and subscribed content. Specifically, based on the association between the object information of the object to be recommended and the subscribed content, a second viewing probability of the object to be recommended viewing the subscribed content is predicted, including:

[0135] Obtain object characteristics and content characteristics of subscribed content to acquire object information;

[0136] The second merged feature is obtained by concatenating the object features and the content features of the subscribed content.

[0137] By using pre-trained view probability prediction parameters, the view probability of the second merged feature is predicted to obtain the second view probability of the object to be recommended viewing the subscribed content.

[0138] Here, the content features of the subscribed content refer to the feature representation of the subscribed content. In this embodiment, the content features of the subscribed content can be feature representations extracted from the subscribed content, and / or feature representations extracted from external information carried by the subscribed content, such as tags or other information. The second merged feature refers to the feature obtained by concatenating the object features and the content features of the subscribed content.

[0139] For example, a pre-trained viewing probability prediction model may include a feature extraction network and a viewing probability prediction network, with the pre-trained viewing probability prediction parameters being the network parameters of the viewing probability prediction network. The feature extraction network can extract object features from the object information and content features from the subscribed content. The extracted object features and content features are then concatenated and input into the viewing probability prediction network to predict the viewing probability, outputting a second viewing probability for each candidate content to be recommended. This application does not limit the specific form of the feature extraction network; for example, the feature extraction network can be one or more combinations of neural networks that can be used for feature extraction, such as convolutional neural networks (CNN), recurrent neural networks (RNN), and neural networks based on attention mechanisms.

[0140] In some implementations, pre-trained exposure rate prediction parameters can be used to capture the correlation between candidate content and both read and unread subscriptions based on the combined features of candidate content features, read subscriptions, and unread subscriptions, thereby improving the accuracy of exposure rate prediction. Specifically, the content display method further includes:

[0141] Obtain the content features of candidate content and the content features of read and subscribed content;

[0142] Using pre-trained exposure rate prediction parameters, and based on the second view probability of candidate content, read subscription content, and unread subscription content, as well as the first view operation data of unread subscription content, the exposure rate of candidate content on the subscription page is predicted, resulting in the exposure rate of candidate content on the subscription page, including:

[0143] The third merged feature is obtained by concatenating the content features of candidate content, the content features of read and subscribed content, the second viewing probability of the recommended object to view the subscribed content to be read, and the first viewing operation data of the subscribed content to be read.

[0144] Exposure rate prediction is performed on the third merged feature using pre-trained exposure rate prediction parameters, which is the exposure rate of candidate content on the subscription page.

[0145] Among them, the content features of read subscription content refer to the feature representation of read subscription content. In this embodiment of the application, the content features of read subscription content can be feature representations extracted from read subscription content, and / or feature representations extracted from external information such as tags or other information carried by read subscription content. The third merged feature refers to the feature obtained by concatenating the content features of candidate content, the content features of read subscription content, the second viewing probability of the recommended object viewing the subscribed content to be read, and the first viewing operation data of the subscribed content to be read.

[0146] For example, a pre-trained exposure rate prediction model can include a feature extraction network and an exposure rate prediction network. The pre-trained exposure rate prediction parameters are the network parameters of the exposure rate prediction network. The feature extraction network can extract content features of candidate content and read / subscribed content. Then, the extracted object features, content features, the second view probability of the subscribed content, and the first view operation data of the subscribed content are concatenated and input into the exposure rate prediction network to predict the exposure rate, outputting the exposure rate of the candidate content on the subscription page.

[0147] In some implementations, the exposure rate prediction network can be a fully connected network. The exposure rate prediction parameters can be those of the fully connected network, such as weight parameters, bias parameters, etc. For example, a third merged feature can be input into a pre-trained fully connected network, such as an MLP. The fully connected network learns the complex relationship between the input features and the exposure rate through fully connected processing, mapping the content features of the candidate content, the read subscription content, and the merged features of the unread subscription content to the output value. Then, an activation function such as the sigmoid function is used in the output layer of the fully connected network to map the output value to the range of 0-1 to obtain the exposure rate of the candidate content on the subscription page.

[0148] In some implementations, subscribed content can be sorted according to its viewing probability, and then displayed sequentially on the subscription page. This prioritizes displaying content more likely to be viewed, increasing the probability of interaction between the user and the subscription page, thereby improving content reach and user experience. Specifically, this content display method further includes:

[0149] Based on the second viewing probability of the subscribed content, the subscribed content of the recommended object is sorted to obtain the sorting result, which is used to indicate the display order of the subscribed content on the subscription page.

[0150] The sorting result can be ordered from highest to lowest according to the second viewing probability. For example, the sorting result can be a data table that associates the content identifiers of the subscribed content with the corresponding second viewing probabilities, with the data in the table arranged from highest to lowest according to the second viewing probability.

[0151] For example, for any given user, the content identifiers of all the content they have subscribed to can be stored in a data table, ordered from highest to lowest according to a second viewing probability. When the user opens the subscription page through an application running on their terminal, the content identifiers of the subscribed content can be read sequentially from the data table. Preview cards of the subscribed content corresponding to the read content identifiers are then displayed on the subscription page in the order they are read, thus showcasing the user's subscribed content on the subscription page in descending order of the second viewing probability.

[0152] In some implementations, a viewing probability prediction model can be obtained by jointly training samples corresponding to the first viewing probability and the second viewing probability. This allows the pre-trained viewing probability prediction model to be used to predict both the first and second viewing probabilities, thereby increasing the generalization ability of the pre-trained model and reducing the complexity of viewing probability prediction. Specifically, this content display method also includes:

[0153] Obtain the training sample set and the viewing probability prediction parameters to be trained. The training sample set includes multiple first samples and second samples. The first samples are the samples corresponding to the candidate content, and the second samples are the samples corresponding to the subscribed content.

[0154] The viewing probability prediction parameters are trained using the first and second samples to obtain the pre-trained viewing probability prediction parameters.

[0155] Here, the first sample corresponds to the first viewing probability, and the second sample corresponds to the second viewing probability. For example, any first sample may include object information and candidate content for any object, and any second sample may include object information and subscribed content for any object. Pre-trained viewing probability prediction parameters are used to predict the first and second viewing probabilities.

[0156] For example, samples in the training sample set can be prepared randomly or according to simple rules. The training sample set can include multiple batches of training samples, which can be used to train the viewing probability prediction model (i.e., the viewing probability prediction parameters) in multiple rounds. Each batch of training samples can include multiple first samples and multiple second samples, so that in each round of training, samples corresponding to candidate content and samples corresponding to subscribed content are used simultaneously for training. For the specific training process and principle, please refer to the inference process of the pre-trained viewing probability prediction model mentioned above, which will not be repeated here. During training, the difference between the model output and the true label (i.e., loss) can be calculated using a loss function. Based on this loss, optimization algorithms such as gradient descent can be used to update the model parameters of the viewing probability prediction model to minimize the loss function until a preset termination condition is met, thus obtaining the pre-trained viewing probability prediction model (i.e., the pre-trained viewing probability prediction parameters).

[0157] In some implementations, multiple training samples can be used to train the exposure rate prediction model (i.e., the exposure rate prediction parameters to be trained) to obtain a pre-trained exposure rate prediction model (i.e., the pre-trained exposure rate prediction parameters). The specific training process and principles can be found in the inference process of the pre-trained view probability prediction model described above, and will not be repeated here. The training samples can include candidate content, read subscription content, and unread subscription content. During training, the difference between the model output and the true label (i.e., the loss) can be calculated using a loss function. Based on this loss, optimization algorithms such as gradient descent can be used to update the model parameters of the exposure rate prediction model to minimize the loss function until a preset termination condition is met, thus obtaining the pre-trained exposure rate prediction model (i.e., the pre-trained exposure rate prediction parameters).

[0158] It should be noted that, in order to optimize the performance of the pre-trained view probability prediction model and the pre-trained exposure rate prediction model, when updating the model parameters of the view probability prediction model and the exposure rate prediction model during training, the network parameters of the feature extraction network and the view probability prediction network in the view probability prediction model can be updated simultaneously, as can the network parameters of the feature extraction network and the exposure rate prediction network in the exposure rate prediction model.

[0159] In some implementations, the recommended content rate can be improved by filtering candidate content based on the content reach rate of the content to be recommended and the predicted viewing probability of the subscribed content. Specifically, determining recommended content from candidate content based on content reach rate includes:

[0160] Obtain the second probability of the recommended user viewing the subscribed content to be read;

[0161] The content reach rate is compared with the second view probability of the subscribed content to determine recommended content from the candidate content based on the comparison results.

[0162] For example, the content reach rate of candidate content can be compared with the second view probability of the subscribed content to be read. Candidate content with a content reach rate greater than the second view probability of the subscribed content to be read can be selected as recommended content. In this way, by comparing the reach rate of candidate content with the predicted view probability of the subscribed content to be read, more likely candidate content and recommended content can be filtered from the candidate content. This ensures that when recommended content is displayed on the subscription page, it has a higher chance of being viewed than unread subscribed content located in the same position as read subscription content, thus improving the content reach rate of recommended content.

[0163] In some implementations, candidate content that is more likely to be viewed is filtered from the candidate content by comparing the reach rate of candidate content with the predicted viewing probability of the subscribed content to be read. And, based on the content reach rate, recommended content with a higher reach rate is filtered from the candidate content to improve the content reach rate of the recommended content. Specifically, comparing the content reach rate with a second viewing probability of the subscribed content to determine recommended content from the candidate content based on the comparison result includes:

[0164] From the candidate content, select the content to be selected, and the content reach rate of the selected content exceeds the second view probability of the content to be subscribed to;

[0165] Recommended content is determined from the candidate content based on its reach.

[0166] For example, the content reach rate of candidate content can be compared with the second view probability of the content to be subscribed to. Candidate content with a content reach rate greater than the second view probability of the content to be subscribed to is selected as candidate content. Then, based on the content reach rates of multiple candidate content, the candidate content is sorted, and the candidate content with the highest content reach rate or multiple candidate content with high content reach rates is recommended. Alternatively, candidate content with a content reach rate higher than a preset threshold can be selected as recommended content.

[0167] In some implementations, to more accurately determine candidate content, a control parameter can be introduced to select candidate content from the pool of candidate content. Specifically, the content reach rate of the candidate content is greater than the product of the second viewing probability of the subscribed content and the control parameter. For example, candidate content with a content reach rate greater than (the second viewing probability of the subscribed content × α) can be selected as candidate content. Here, α is the control parameter, i.e., the hyperparameter, and its value ranges from 0 to 1. The specific value of α can be determined by prior preferences or by finding the optimal value through A / B testing.

[0168] 230. Determine the display position of the recommended content on the subscription page based on the associated position of the subscribed content to be read on the subscription page, so as to display the recommended content on the subscription page.

[0169] For example, the associated position of the subscription content to be read on the subscription page can be used as the display position of the recommended content on the subscription page. For example, the associated position of the subscription content to be read can be a position within a preset range of the subscription content to be read, such as the current position of the subscription content to be read, above, below, to the left or to the right of the subscription content to be read, etc.

[0170] The content recommendation scheme provided in this application can be applied to various content recommendation scenarios. For example, taking a content platform as an example, it obtains read subscription content and unread subscription content from multiple subscription content displayed on the subscription page, and obtains at least one candidate content, with the unread subscription content displayed in the associated position of the read subscription content; based on the read subscription content and the unread subscription content, it determines recommended content from the candidate content; and according to the associated position of the unread subscription content on the subscription page, it determines the display position of the recommended content on the subscription page, so as to display the recommended content on the subscription page.

[0171] As can be seen from the above, in this embodiment, the recommended content and its display position are determined based on the subscription content to be read, so that the recommended content is consistent with the subscription content to be read on the subscription page in terms of content information and display position. This makes the recommended content more naturally and smoothly integrated into the user's reading experience of the subscription content, thereby improving the display effect of the recommended content, enhancing the user's immersion in reading the subscription page content and their willingness to continue reading, thus increasing the content reach rate of the recommended content and improving the user experience. Specifically, in this embodiment, after the target audience views the subscription content, the recommended content is determined based on the read subscription content and the subscription content to be read displayed in association with the read subscription content. This makes the recommended content more aligned with the needs and interests of the target audience on the subscription page, further improving the display effect of the recommended content, increasing the content reach rate of the recommended content, and improving the user experience.

[0172] The method described in the above embodiments will be further described in detail below.

[0173] In this embodiment, the method of this application embodiment will be described in detail using a content recommendation scenario applied to a social media application platform as an example.

[0174] The content display method of this application embodiment can be applied to, for example, Figure 3a The content display system shown includes a client and a server, with the server including a recommendation server module and an operation module.

[0175] like Figure 3b As shown, the specific process of one content display method is as follows:

[0176] 300. Display the subscription page on the terminal corresponding to the target audience. The subscription page displays multiple subscription items.

[0177] In this embodiment, the target audience for recommendation can be users of a social media application platform. For example, when any user, such as user A, opens a subscription page through an application (i.e., a client) running on a terminal, in response to this operation, the terminal can obtain the subscription page information of the content subscribed to by user A from the server, and display the subscription page on the terminal. This subscription page displays preview cards of the latest subscription content from multiple content publishers subscribed to by user A, such as public accounts and video accounts. For example, in this embodiment, the subscription page can be the message box interface of a subscription account, and the preview cards can be message cards in the message box interface of the subscription account.

[0178] 310. In response to a viewing operation on any subscription content on the subscription page, the terminal determines the subscription content corresponding to the viewing operation as read subscription content, and determines the subscription content associated with the read subscription content as unread subscription content.

[0179] 320. The server retrieves read subscription content and unread subscription content from the multiple subscription content displayed on the subscription page, and retrieves at least one candidate content.

[0180] For example, when user A clicks on the preview card of any subscribed content (i.e., read subscribed content) on the subscription page via a terminal, in response to this viewing action, the terminal can retrieve detailed page information of the read subscribed content from the server to display the detailed page of the read subscribed content on the terminal. Simultaneously, based on this viewing action, the server can trigger a content recommendation task, using the operations module to retrieve content from the content database of the social media application platform that has a causal relationship or event chain relationship with the read subscribed content as candidate content, thus obtaining a set of candidate content.

[0181] 330. The server predicts the probability that the recommended object will view the candidate content first, based on the association between the object information of the object to be recommended and the candidate content.

[0182] For example, the recommendation server module can call a pre-trained viewing probability prediction model, input user A's user information and the relevant information of each candidate content into the pre-trained viewing probability prediction model, and after processing by the feature extraction network and the viewing probability prediction network, obtain the first viewing probability of user A viewing each candidate content.

[0183] 340. The server obtains the second probability of viewing the subscribed content to be recommended by the target audience.

[0184] For example, when a user subscribes to content, the recommendation server module can instantly invoke a pre-trained viewing probability prediction model to determine the second viewing probability of the user viewing the subscribed content. Thus, when the user views already read subscribed content, the second viewing probability of viewing the unread subscribed content can be instantly and directly obtained from the determined second viewing probability of the subscribed content. For instance, when user A subscribes to any content (such as content A), the recommendation server module can invoke the pre-trained viewing probability prediction model, inputting user A's user information and relevant information about subscribed content A into the model. After processing by the feature extraction network and the viewing probability prediction network, the second viewing probability of user A viewing subscribed content A is obtained and stored in a data table corresponding to user A. This data table is used for the second viewing probabilities of all subscribed content. When the unread subscribed content is content A, the second viewing probability of subscribed content A can be directly obtained from this data table.

[0185] 350. The server uses pre-trained exposure rate prediction parameters to predict the exposure rate of candidate content based on the second viewing probability of candidate content, read subscription content, unread subscription content, and the first viewing operation data of unread subscription content, and obtains the exposure rate of candidate content on the subscription page.

[0186] For example, the recommendation server module can call a pre-trained exposure rate prediction model, inputting the relevant information of each candidate content in the candidate content set, the relevant information of read subscription content, the second viewing probability of the subscription content to be read, and the first viewing operation data of the subscription content to be read into the pre-trained exposure rate prediction model. After processing by the feature extraction network and the exposure rate prediction network, the exposure rate of each candidate content in the candidate content set on the subscription page is obtained.

[0187] 360. The server combines the first viewing probability of candidate content with the exposure rate of candidate content on the subscription page to determine the content reach rate of candidate content to be recommended.

[0188] For example, for any candidate content, the recommendation server module can multiply the probability of user A viewing that candidate content by the candidate content's exposure rate on the subscription page to obtain the content reach rate of that candidate content for user A. In this way, the content reach rate of all candidate content can be calculated.

[0189] 370. The server compares the content reach rate with the second view probability of the subscribed content to determine recommended content from the candidate content based on the comparison results.

[0190] For example, the recommendation server module can compare the content reach rate of all candidate content with the second view probability corresponding to the content to be subscribed to, and select candidate content whose content reach rate is greater than the second view probability of the content to be subscribed to as the candidate content. In this embodiment, the candidate content is determined by introducing the control formula "(first view probability of candidate content × exposure rate of candidate content on the subscription page) > (second view probability of the content to be subscribed to × α)", where α is a control parameter, i.e., a hyperparameter, and the value of α ranges from 0 to 1. Then, based on the content reach rate of multiple candidate content, the candidate content is sorted, and the candidate content with the highest content reach rate is selected as the recommended content.

[0191] 380. The server determines the display position of the recommended content on the subscription page based on the association position of the subscription content to be read on the subscription page.

[0192] For example, the recommendation server module can use the display position of the content to be read on the subscription page as the display position of candidate content on the subscription page, so that recommended content is displayed in the same position as the content to be read. Simultaneously, the content to be read on the original subscription page, as well as any subscription content following it, are arranged sequentially to update the subscription page, resulting in an updated subscription page. The recommendation server module can then send the updated subscription page information to user A's corresponding terminal.

[0193] 390. The terminal displays recommended content in the relevant position of the subscription page and the subscription content to be read.

[0194] For example, after user A clicks to exit the detailed page of read subscription content displayed on the terminal, in response to the exit operation, an updated subscription page with recommended content can be displayed on user A's terminal. In this embodiment, subscription content and recommended content are displayed together on the subscription page to reduce the number of times users switch between different tabs or areas, allowing users to browse and discover content of interest in a unified information flow, making it easier for users to find information of interest and improving user experience. Furthermore, while browsing subscription content, users can also access recommended content based on user information and subscription content, increasing the chances of users discovering new content, thereby increasing content exposure and consumption. It is understood that in practical application scenarios, users' scrolling operations on the subscription page are usually random and rapid; therefore, the user's current viewing operation of subscription content has a higher confidence level, and displaying recommended content based on the user's viewing operation is more effective.

[0195] As can be seen from the above, in this embodiment, after the target user views the subscribed content, the recommended content and its display position are determined based on the unread subscribed content. This ensures that the recommended content is consistent with the unread subscribed content on the subscription page in terms of content information and display position, allowing the recommended content to integrate more naturally and smoothly into the user's reading experience. This improves the display effect of the recommended content, enhances the user's immersion in reading the subscription page content, and increases their willingness to continue reading, thereby increasing the content reach rate of the recommended content and improving the user experience. Specifically, in this embodiment, after the target user views the subscribed content, the recommended content is determined based on the read subscribed content and the unread subscribed content displayed in association with it. This makes the recommended content more aligned with the needs and interests of the target user on the subscription page, further improving the display effect of the recommended content, increasing the content reach rate of the recommended content, and enhancing the user experience.

[0196] To better implement the above methods, this application also provides a content display device, which can be integrated into an electronic device, such as a terminal or server. The terminal can be a mobile phone, tablet computer, smart Bluetooth device, laptop computer, or personal computer; the server can be a single server or a server cluster composed of multiple servers.

[0197] For example, in this embodiment, the method of this application embodiment will be described in detail by taking the content display device specifically integrated into the terminal as an example.

[0198] For example, such as Figure 4 As shown, the content display device may include a display unit 410 and a subscription content determination unit 420, as follows:

[0199] (a) Display unit 410

[0200] Used to display the subscription page, which shows multiple subscription items.

[0201] The display unit is also used to display recommended content in the subscription page and in the associated position of the subscription content to be read. The recommended content is determined by the subscription content already read and the subscription content to be read.

[0202] (II) Subscription Content Determination Unit 420

[0203] This is used to identify read subscription content and unread subscription content on the subscription page. Unread subscription content is displayed in the same location as read subscription content.

[0204] In some implementations, the subscription content determination unit is specifically used to: in response to a viewing operation on any subscription content in the subscription page, determine the subscription content corresponding to the viewing operation as read subscription content, and determine the subscription content associated with the read subscription content as unread subscription content.

[0205] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0206] As can be seen from the above, the content display device of this embodiment includes a display unit and a subscription content determination unit. The display unit is used to display a subscription page, which displays multiple subscription contents. The subscription content determination unit is used to determine the read subscription contents and the unread subscription contents on the subscription page, with the unread subscription contents displayed in an associated position with the read subscription contents. The display unit is also used to display recommended content in the subscription page and in an associated position with the unread subscription contents, the recommended content being content determined by the read subscription contents and the unread subscription contents.

[0207] Therefore, in this embodiment, recommended content and its display position are determined based on the subscription content to be read, ensuring that the recommended content is consistent with the subscription content to be read on the subscription page in terms of content information and display position. This allows the recommended content to integrate more naturally and smoothly into the user's reading experience, improving the display effect of the recommended content, enhancing the user's immersion in the subscription page content and their willingness to continue reading, thereby increasing the content reach rate of the recommended content and improving the user experience. Specifically, in this embodiment, after the target audience views the subscription content, the recommended content is determined based on the read subscription content and the related subscription content to be read, making the recommended content more aligned with the needs and interests of the target audience on the subscription page, further improving the display effect of the recommended content, increasing the content reach rate of the recommended content, and improving the user experience.

[0208] To better implement the above methods, this application also provides a content display device, which can be integrated into an electronic device, such as a terminal or server. The terminal can be a mobile phone, tablet computer, smart Bluetooth device, laptop computer, or personal computer; the server can be a single server or a server cluster composed of multiple servers.

[0209] For example, in this embodiment, the method of this application embodiment will be described in detail by taking the content display device as specifically integrated into the server.

[0210] For example, such as Figure 5As shown, the content display device may include an acquisition unit 510, a recommended content determination unit 520, and a display position determination unit 530, as follows:

[0211] (I) Acquisition Unit 510

[0212] This is used to retrieve read subscription content and unread subscription content from multiple subscription content displayed on the subscription page, as well as to retrieve at least one candidate content, with the unread subscription content displayed in the associated position of the read subscription content.

[0213] (II) Recommendation Content Determination Unit 520

[0214] It is used to determine recommended content from candidate content based on read subscription content and unread subscription content.

[0215] In some implementations, the recommended content determination unit includes a first recommended content determination subunit and a second recommended content determination subunit, including: the first recommended content determination subunit, which is used to predict the content reach rate of candidate content for the object to be recommended based on the object information of the object to be recommended, the read subscription content, and the subscription content to be read, wherein the object to be recommended is the object corresponding to the subscription page; and the second recommended content determination subunit, which is used to determine recommended content from the candidate content based on the content reach rate.

[0216] In some implementations, the content reach rate of candidate content for the recommended object is predicted based on the object information of the object to be recommended, the read subscription content, and the subscription content to be read. This includes: predicting the first viewing probability of the recommended object viewing the candidate content based on the association between the object information of the object to be recommended and the candidate content; predicting the exposure rate of the candidate content on the subscription page based on the association between the candidate content, the read subscription content, and the subscription content to be read; and determining the content reach rate of the candidate content for the recommended object by combining the first viewing probability of the candidate content and the exposure rate of the candidate content on the subscription page.

[0217] In some implementations, the exposure rate of candidate content on the subscription page is predicted based on the relationship between candidate content, read subscription content, and unread subscription content. This includes: obtaining the second viewing probability of the recommended object viewing the unread subscription content, and obtaining the first viewing operation data of the unread subscription content; using pre-trained exposure rate prediction parameters, based on the second viewing probability of candidate content, read subscription content, and unread subscription content, and the first viewing operation data of the unread subscription content, the exposure rate of candidate content on the subscription page is predicted.

[0218] In some implementations, determining recommended content from candidate content based on content reach includes: obtaining a second viewing probability of the target audience viewing the subscribed content; comparing the content reach with the second viewing probability of the subscribed content to determine recommended content from candidate content based on the comparison result.

[0219] In some implementations, comparing content reach rate with a second viewing probability of the subscribed content to determine recommended content from candidate content based on the comparison result includes: determining candidate content from candidate content, wherein the content reach rate of the candidate content exceeds the second viewing probability of the subscribed content; and determining recommended content from the candidate content based on the content reach rate of the candidate content.

[0220] In some implementations, the content display device further includes a subscription unit, which is used to: obtain object information of the object to be recommended based on the object's subscription operation to the subscribed content; and predict a second viewing probability of the object to be recommended viewing the subscribed content based on the association between the object information of the object to be recommended and the subscribed content.

[0221] In some implementations, obtaining the second viewing probability of the target object viewing the subscribed content includes: obtaining the second viewing probability of the target object viewing the subscribed content from the second viewing probability of the subscribed content.

[0222] In some embodiments, the content display device further includes a sorting unit, which is used to: sort the subscription content of the recommended object according to the second viewing probability of the subscription content, and obtain a sorting result, which is used to indicate the display order of the subscription content on the subscription page.

[0223] In some implementations, the object information includes second viewing operation data of the object to be recommended on the subscribed content. The content display device further includes an update unit, which is used to: update the second viewing operation data of the object to be recommended on the subscribed content on the subscription page according to the viewing operation of the object to be recommended on the subscription page, so as to obtain the updated object information of the object to be recommended. The updated object information is used to predict the content reach rate.

[0224] (III) Unit 530 for determining the display location

[0225] This is used to determine the display position of recommended content on the subscription page based on the associated position of the subscribed content to be read, so as to display the recommended content on the subscription page.

[0226] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0227] As can be seen from the above, the content display device of this embodiment includes an acquisition unit, a recommended content determination unit, and a display position determination unit. The acquisition unit is used to acquire read subscription content and unread subscription content from multiple subscription contents displayed on the subscription page, and to acquire at least one candidate content, with the unread subscription content displayed at an associated position of the read subscription content. The recommended content determination unit is used to determine recommended content from the candidate content based on the read subscription content and the unread subscription content. The display position determination unit is used to determine the display position of the recommended content on the subscription page according to the associated position of the unread subscription content on the subscription page, so as to display the recommended content on the subscription page.

[0228] Therefore, in this embodiment, recommended content and its display position are determined based on the subscription content to be read, ensuring that the recommended content is consistent with the subscription content to be read on the subscription page in terms of content information and display position. This allows the recommended content to integrate more naturally and smoothly into the user's reading experience, improving the display effect of the recommended content, enhancing the user's immersion in the subscription page content and their willingness to continue reading, thereby increasing the content reach rate of the recommended content and improving the user experience. Specifically, in this embodiment, after the target audience views the subscription content, the recommended content is determined based on the read subscription content and the related subscription content to be read, making the recommended content more aligned with the needs and interests of the target audience on the subscription page, further improving the display effect of the recommended content, increasing the content reach rate of the recommended content, and improving the user experience.

[0229] This application also provides an electronic device, which can be a terminal, a server, or other similar device. The terminal can be a mobile phone, tablet computer, smart Bluetooth device, laptop computer, personal computer, etc.; the server can be a single server or a server cluster composed of multiple servers, etc.

[0230] In some embodiments, the content display device may also be integrated into multiple electronic devices, such as multiple servers, with multiple servers implementing the content display method of this application.

[0231] In this embodiment, a server will be used as an example for detailed description. For example, ... Figure 6 As shown, it illustrates a schematic diagram of the server structure involved in an embodiment of this application. Specifically:

[0232] The server may include components such as a processor 610 with one or more processing cores, a memory 620 with one or more computer-readable storage media, a power supply 630, an input module 640, and a communication module 650. Those skilled in the art will understand that... Figure 6 The server architecture shown does not constitute a limitation on the server and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. Wherein:

[0233] The processor 610 is the control center of the server, connecting various parts of the server via various interfaces and lines. It performs various server functions and processes data by running or executing software programs and / or modules stored in the memory 620, and by accessing data stored in the memory 620. In some embodiments, the processor 610 may include one or more processing cores; in some embodiments, the processor 610 may integrate an application processor and a modem processor, wherein the application processor primarily handles the operating system, user interface, and applications, while the modem processor primarily handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 610.

[0234] The memory 620 can be used to store software programs and modules. The processor 610 executes various functional applications and data processing by running the software programs and modules stored in the memory 620. The memory 620 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the server, etc. In addition, the memory 620 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 620 may also include a memory controller to provide the processor 610 with access to the memory 620.

[0235] The server also includes a power supply 630 that supplies power to the various components. In some embodiments, the power supply 630 can be logically connected to the processor 610 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 630 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0236] The server may also include an input module 640, which can be used to receive input numeric or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0237] The server may also include a communication module 650. In some embodiments, the communication module 650 may include a wireless module, through which the server can perform short-range wireless transmission, thereby providing users with wireless broadband internet access. For example, the communication module 650 can be used to help users send and receive emails, browse web pages, and access streaming media.

[0238] Although not shown, the server may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 610 in the server loads the executable files corresponding to the processes of one or more applications into the memory 620 according to the following instructions, and the processor 610 runs the applications stored in the memory 620 to realize various functions, as follows:

[0239] The subscription page is displayed, showing multiple subscribed content items. Read and unread subscription items are identified on the subscription page, with unread subscription items displayed in the corresponding position to the read subscription items. Recommended content is displayed in the corresponding position to the unread subscription items on the subscription page; the recommended content is determined by the read and unread subscription items.

[0240] Alternatively, retrieve read subscription content and unread subscription content from multiple subscription content displayed on the subscription page, and retrieve at least one candidate content, with the unread subscription content displayed in the associated position of the read subscription content; based on the read subscription content and the unread subscription content, determine recommended content from the candidate content; and determine the display position of the recommended content on the subscription page according to the associated position of the unread subscription content on the subscription page, so as to display the recommended content on the subscription page.

[0241] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0242] As can be seen from the above, in this embodiment, the recommended content and its display position are determined based on the subscription content to be read, so that the recommended content is consistent with the subscription content to be read on the subscription page in terms of content information and display position. This makes the recommended content more naturally and smoothly integrated into the user's reading experience of the subscription content, thereby improving the display effect of the recommended content, enhancing the user's immersion in reading the subscription page content and their willingness to continue reading, thus increasing the content reach rate of the recommended content and improving the user experience. Specifically, in this embodiment, after the target audience views the subscription content, the recommended content is determined based on the read subscription content and the subscription content to be read displayed in association with the read subscription content. This makes the recommended content more aligned with the needs and interests of the target audience on the subscription page, further improving the display effect of the recommended content, increasing the content reach rate of the recommended content, and improving the user experience.

[0243] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0244] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute steps in any of the content display methods provided in embodiments of this application. For example, the instructions can execute the following steps:

[0245] The subscription page is displayed, showing multiple subscribed content items. Read and unread subscription items are identified on the subscription page, with unread subscription items displayed in the corresponding position to the read subscription items. Recommended content is displayed in the corresponding position to the unread subscription items on the subscription page; the recommended content is determined by the read and unread subscription items.

[0246] Alternatively, retrieve read subscription content and unread subscription content from multiple subscription content displayed on the subscription page, and retrieve at least one candidate content, with the unread subscription content displayed in the associated position of the read subscription content; based on the read subscription content and the unread subscription content, determine recommended content from the candidate content; and determine the display position of the recommended content on the subscription page according to the associated position of the unread subscription content on the subscription page, so as to display the recommended content on the subscription page.

[0247] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0248] According to one aspect of this application, a computer program product or computer program is provided, comprising a computer program or instructions that, when executed by a processor, implement the steps of the methods provided in the various optional implementations of the above embodiments. The computer program / instructions are stored in a computer-readable storage medium. A processor of an electronic device reads the computer program / instructions from the computer-readable storage medium and executes the computer program / instructions, causing the electronic device to perform the methods provided in the various optional implementations of the above embodiments.

[0249] Since the instructions stored in the storage medium can execute the steps of any of the content display methods provided in the embodiments of this application, the beneficial effects that any of the content display methods provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0250] The foregoing has provided a detailed description of a content display method, apparatus, electronic device, storage medium, and program product provided in the embodiments of this application. 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 methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A content display method, characterized in that, include: The subscription page is displayed, showing multiple subscription items; The read subscription content and unread subscription content in the subscription page are determined, and the unread subscription content is displayed in the associated position of the read subscription content; Recommended content is displayed in the subscription page at the location associated with the subscription content to be read. The recommended content is determined by the subscription content already read and the subscription content to be read.

2. The content display method as described in claim 1, characterized in that, Determining the read subscription content and unread subscription content on the subscription page includes: In response to a viewing operation of any of the subscription content on the subscription page, the subscription content corresponding to the viewing operation is determined as the read subscription content, and the subscription content associated with the read subscription content is determined as the unread subscription content.

3. A content display method, characterized in that, include: The system retrieves read subscription content and unread subscription content from multiple subscription content displayed on the subscription page, and retrieves at least one candidate content, wherein the unread subscription content is displayed at the associated position of the read subscription content; Based on the read subscription content and the unread subscription content, recommended content is determined from the candidate content; Based on the associated position of the subscription content to be read on the subscription page, the display position of the recommended content on the subscription page is determined, so as to display the recommended content on the subscription page.

4. The content display method as described in claim 3, characterized in that, The step of determining recommended content from the candidate content based on the read subscription content and the unread subscription content includes: Based on the object information of the object to be recommended, the read subscription content, and the subscription content to be read, the content reach rate of the candidate content to the object to be recommended is predicted, where the object to be recommended is the object corresponding to the subscription page; Based on the content reach rate, recommended content is determined from the candidate content.

5. The content display method as described in claim 4, characterized in that, The method of predicting the content reach rate of the candidate content to the object to be recommended based on the object information of the object to be recommended, the read subscription content, and the subscription content to be read includes: Based on the association between the object information of the object to be recommended and the candidate content, predict the first probability that the object to be recommended will view the candidate content; Based on the relationship between the candidate content, the read subscription content, and the unread subscription content, predict the exposure rate of the candidate content on the subscription page; The content reach rate of the candidate content to the object to be recommended is determined by combining the first viewing probability of the candidate content and the exposure rate of the candidate content on the subscription page.

6. The content display method as described in claim 5, characterized in that, The step of predicting the exposure rate of the candidate content on the subscription page based on the relationship between the candidate content, the read subscription content, and the unread subscription content includes: Obtain the second viewing probability of the recommended object viewing the subscribed content to be read, and obtain the first viewing operation data of the subscribed content to be read; Using pre-trained exposure rate prediction parameters, based on the candidate content, the read subscription content, the second viewing probability of the unread subscription content, and the first viewing operation data of the unread subscription content, the exposure rate of the candidate content on the subscription page is predicted.

7. The content display method as described in claim 4, characterized in that, The step of determining recommended content from the candidate content based on the content reach rate includes: Obtain the second probability that the recommended object will view the subscribed content to be read; The content reach rate is compared with the second view probability of the subscribed content to be read, so as to determine recommended content from the candidate content based on the comparison result.

8. The content display method as described in claim 7, characterized in that, The step of comparing the content reach rate with the second view probability of the subscribed content to be read, in order to determine recommended content from the candidate content based on the comparison result, includes: From the candidate content, selectable content is determined, wherein the content reach rate of the selectable content exceeds the second viewing probability of the subscribed content; Recommended content is determined from the candidate content based on the content reach rate of the candidate content.

9. The content display method as described in claim 6 or 7, characterized in that, Before obtaining the second viewing probability of the object to be recommended viewing the subscribed content, the method further includes: Based on the subscription operation of the object to be recommended to the subscribed content, obtain the object information of the object to be recommended; Based on the association between the object information of the object to be recommended and the subscribed content, predict the second viewing probability of the object to be recommended viewing the subscribed content; The step of obtaining the second viewing probability of the target object viewing the subscribed content includes: The second viewing probability of the target object viewing the subscribed content is obtained from the second viewing probability of the subscribed content.

10. The content display method as described in claim 4, characterized in that, The object information includes the second viewing operation data of the object to be recommended on the subscribed content, and the method further includes: Based on the viewing operation of the target object on the subscription page, the second viewing operation data of the target object on the subscription content is updated to obtain the updated object information of the target object. The updated object information is used to predict the content reach rate.

11. A content display device, characterized in that, include: A display unit is used to display a subscription page, which shows multiple subscription items; The subscription content determination unit is used to determine the read subscription content and the unread subscription content in the subscription page, wherein the unread subscription content is displayed in the associated position of the read subscription content; The display unit is also used to display recommended content in the subscription page at the associated position of the subscription content to be read, wherein the recommended content is determined by the subscribed content and the subscription content to be read.

12. A content display device, characterized in that, include: The acquisition unit is used to acquire read subscription content and unread subscription content from multiple subscription content displayed on the subscription page, and to acquire at least one candidate content, wherein the unread subscription content is displayed at the associated position of the read subscription content; The recommended content determination unit is used to determine recommended content from the candidate content based on the read subscription content and the unread subscription content; The display position determination unit is used to determine the display position of the recommended content on the subscription page based on the associated position of the subscribed content to be read on the subscription page, so as to display the recommended content on the subscription page.

13. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing multiple instructions; the processor loads instructions from the memory to perform the steps of the content display method as described in any one of claims 1 to 10.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to perform the steps of the content display method according to any one of claims 1 to 10.

15. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the processor, they implement the steps of the content display method according to any one of claims 1 to 10.

Images