Content display method and apparatus, and electronic device and storage medium

Abstract

A content display method, which is executed by means of an electronic device. The method comprises: displaying a content browsing interface, which is used for displaying a pushed information stream to a recipient user (S31); and in response to a content switching operation for the content browsing interface, displaying in the content browsing interface dedicated promotion content, which is generated on the basis of pushed media content carried in the information stream, jumping to a content fragment pointed to by dedicated key information of a user degree-of-interest matching level in the dedicated promotion content, and playing the content fragment, wherein the dedicated promotion content is used for describing dedicated key information of different levels matching the recipient user, each piece of dedicated key information is comprehensively generated on the basis of original key information in the pushed media content and content preference information of the recipient user, and the user degree-of-interest matching level is determined on the basis of the degree of interest of the recipient user for the information stream (S32).

Description

Content display methods, devices, electronic devices and storage media

[0001] Related applications

[0002] This application claims priority to Chinese patent application filed on October 22, 2024, with application number 202411475996.1, entitled "A method, apparatus, electronic device and storage medium for displaying multimedia content", the entire contents of which are incorporated herein by reference. Technical Field

[0003] This application relates to the field of Internet technology, and in particular to a content display method, apparatus, electronic device, and storage medium. Background Technology

[0004] With the widespread use of the internet and electronic devices, promoting products, services, brands, or concepts through media content (such as video ads) has become a major promotional method. However, in the current content delivery industry, many media content providers still face challenges in attracting audiences with compelling content and attractive benefits.

[0005] On the one hand, due to limited production budgets, insufficient creativity, or a lack of understanding of the needs and interests of the target audience, the production quality of much media content is uneven, resulting in low-quality content that fails to effectively stimulate viewer interest. Furthermore, some media content lacks originality and relies heavily on similar presentation techniques and visual elements, making it easy for viewers to become bored and unwilling to click for more information. On the other hand, in this information-saturated era, viewers are faced with a vast amount of information and content, making their attention increasingly scarce and their criteria for selecting media content increasingly stringent. Therefore, even content with some appeal may be overlooked among a sea of ​​other content.

[0006] Therefore, how to effectively increase the click-through rate of media content is an urgent issue to be addressed. Summary of the Invention

[0007] This application provides a content display method, apparatus, electronic device, and storage medium to improve the click-through rate of media content.

[0008] This application provides a content display method, including:

[0009] The content browsing interface is used to display the pushed information stream to the user receiving the push notification.

[0010] In response to the content switching operation of the content browsing interface, exclusive promotional content generated based on the push media content carried by the information stream is displayed in the content browsing interface, and the user is redirected to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback;

[0011] The exclusive promotional content is used to describe various levels of exclusive key information that match the target user; each exclusive key information is generated based on the original key information in the push media content and the target user's content preference information; the user interest matching level is determined based on the target user's interest in the information stream.

[0012] Another content display method provided in this application embodiment includes:

[0013] When a user browses the information stream in the content browsing interface, the push media content to be displayed carried in the information stream is preloaded.

[0014] Extract the original key information from the pushed media content and obtain the content preference information of the pushed user;

[0015] Based on the original key information and the content preference information, exclusive promotional content is generated for the pushed user, wherein the exclusive promotional content is used to describe different levels of exclusive key information that match the pushed user;

[0016] The exclusive promotional content is sent to the target user, and in response to the target user's content switching operation on the content browsing interface, the exclusive promotional content is displayed on the content browsing interface, and the user is redirected to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback; wherein, the user interest matching level is determined based on the target user's interest in the information stream.

[0017] This application provides a content display device, comprising:

[0018] The display unit is used to display a content browsing interface; the content browsing interface is used to display the pushed information stream to the pushed user.

[0019] The response unit is used to respond to the content switching operation of the content browsing interface, display exclusive promotional content generated based on the push media content carried by the information stream in the content browsing interface, and jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback;

[0020] The exclusive promotional content is used to describe various levels of exclusive key information that match the target user; each exclusive key information is generated based on the original key information in the push media content and the target user's content preference information; the user interest matching level is determined based on the target user's interest in the information stream.

[0021] Optionally, the level of each of the exclusive key information is associated with the content preference information of the pushed user, and the level represents the user's attention index to the corresponding exclusive key information.

[0022] An electronic device provided in this application includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of any of the above-described content display methods.

[0023] This application provides a computer-readable storage medium including a computer program. When the computer program is run on an electronic device, the computer program is used to cause the electronic device to perform the steps of any of the above-described content display methods.

[0024] This application provides a computer program product, which includes a computer program stored in a computer-readable storage medium. When the processor of an electronic device reads the computer program from the computer-readable storage medium, the processor executes the computer program, causing the electronic device to perform the steps of any of the above-described content display methods.

[0025] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the published drawings without creative effort.

[0027] Figure 1 is a schematic diagram of an application scenario in an embodiment of this application;

[0028] Figure 2A is a schematic diagram of an advertising display container displaying an advertisement in a related technology;

[0029] Figure 2B is a schematic diagram of another advertising display container in the related technology for displaying advertisements;

[0030] Figure 3 is a flowchart illustrating the implementation of a content display method according to an embodiment of this application;

[0031] Figure 4 is a schematic diagram of the matching relationship between screen swiping state and interest level in an embodiment of this application;

[0032] Figure 5 is a schematic diagram of matching selling points according to the screen swiping speed of the pushed user in an embodiment of this application;

[0033] Figure 6 is a schematic diagram of another method of matching selling points based on the screen swiping speed of the pushed user in an embodiment of this application;

[0034] Figure 7 is a schematic diagram of a selling point carousel in an embodiment of this application;

[0035] Figure 8 is a schematic diagram of the advertisement display in the first type of browser in the embodiments of this application;

[0036] Figure 9 is a schematic diagram of the advertisement display in the second type of browser in the embodiments of this application;

[0037] Figure 10 is a schematic diagram of the advertisement display in a third browser according to an embodiment of this application;

[0038] Figure 11 is a schematic diagram of the advertisement display in the fourth browser in the embodiments of this application;

[0039] Figure 12 is a schematic diagram of the advertisement display in the fifth type of browser in the embodiments of this application;

[0040] Figure 13 is a schematic diagram of the advertisement display in the sixth type of browser in the embodiments of this application;

[0041] Figure 14 is a schematic diagram of a specific key information time point highlighting method in an embodiment of this application;

[0042] Figure 15 is a schematic diagram of mutual selection of advertising selling points in an embodiment of this application;

[0043] Figure 16 is a flowchart illustrating another content display method provided in this application embodiment;

[0044] Figure 17 is a schematic diagram of a method for extracting the preference tags and their exclusive key information of different subgroups in an embodiment of this application;

[0045] Figure 18 is a schematic diagram of an optional multi-terminal interaction implementation in an embodiment of this application;

[0046] Figure 19 is a schematic diagram of the composition structure of a content display device in an embodiment of this application;

[0047] Figure 20 is a schematic diagram of the composition structure of another content display device in an embodiment of this application;

[0048] Figure 21 is a schematic diagram of the hardware structure of an electronic device using an embodiment of this application;

[0049] Figure 22 is a schematic diagram of the hardware structure of another electronic device using an embodiment of this application. Detailed Implementation

[0050] 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.

[0051] The following describes some of the concepts involved in the embodiments of this application.

[0052] Content browsing interface: This is a user-facing interface used to display the pushed information stream to the recipient. Specifically, it refers to the visual area on an application or website where users can view and interact with various content, such as the homepage of a news app, the product list page of an e-commerce platform, or the recommendation page for short videos. The content browsing interface can contain various elements, such as text, images, videos, and links, arranged according to a specific layout and structure for easy browsing and navigation.

[0053] Information feed: This refers to a series of content items dynamically displayed in a content browsing interface according to a certain logic and order. These content items can take various forms, such as articles, images, videos, user posts, and advertisements; in this article, they are referred to as media content. They are typically personalized recommendations based on user interests, behaviors, and preferences. The characteristic of an information feed is that the content is constantly updated, and users can switch between content using scrolling or swiping to browse new content.

[0054] Key information refers to the core elements or important features extracted from media content, such as the selling points of an advertisement, the unique functions of a product, and the main highlights of an event. These aim to attract users' attention and stimulate their interest. Specifically, selling points refer to the unprecedented, original, or unique characteristics of the product in the advertisement. When implemented in marketing strategies and tactics, these characteristics are transformed into benefits and utilities that consumers can accept and recognize, thus achieving the goals of product sales and brand building.

[0055] Specifically, this application mainly involves two types of key information: original key information and personalized key information. Original key information refers to the core elements or important characteristics of the original pushed media content, and is a general type of information applicable to different users. Personalized key information, on the other hand, is generated based on the original key information and the content preference information of the pushed user; it is specific to the pushed user and is a type of personalized information.

[0056] Personalized promotional content: This is content generated specifically for the target user, combining the content of the push media with the target user's content preferences. It is personalized promotional content designed to increase user interest and engagement. The personalized promotional content in this application typically includes multiple levels of personalized key information, each level specifically reflecting the target user's preferences and needs.

[0057] The design concept of the embodiments of this application is briefly introduced below:

[0058] In the information age, using media content to promote products, services, brands, or ideas has become a primary promotional method. For example, to increase awareness and purchase of their products, manufacturers typically create promotional media content, such as video advertisements, and display them on well-known public platforms like web pages.

[0059] As described in the background section, in the current advertising industry, many video ads still face challenges in attracting audiences to click through compelling content and attractive benefits.

[0060] To overcome the aforementioned problems, a novel advertising display method has been proposed in related technologies. This method utilizes diverse advertising display containers to simultaneously showcase video ads, selling points, and conversion buttons within a single screen. Figures 2A and 2B illustrate this approach, demonstrating how ads are displayed in a browser using advertising display containers. Figures 2A and 2B show the video ad and selling points on the right side of the interface, while Figure 2B shows the conversion button on the left. This advertising display method helps advertisers quickly convey key advertising information (where selling points need to be manually uploaded by the advertiser) and also helps users quickly and easily access this key information, thereby achieving higher commercial conversion rates. Advertising display containers for other applications (apps) are similar and will not be elaborated upon further.

[0061] While this type of advertising display container solution can centrally present key information and improve user perception efficiency, it also has some potential drawbacks, such as information overload, heavy interaction load, compromised user experience consistency, high loading and running requirements, and the need for manual intervention. A detailed analysis follows:

[0062] This type of ad display container, which centrally displays video ads, selling points, and conversion buttons, can easily lead to an overload of information that users struggle to digest in a short time, resulting in information confusion. Furthermore, users may need to expend extra effort to understand and operate multiple interactive elements, increasing complexity and learning costs. Additionally, if this display method clashes with the app's overall design style, it may disrupt user habits and affect the consistency of the user experience. Moreover, video ads typically require high bandwidth and processing power for smooth playback, which can pose a challenge to low-end devices, potentially causing slow loading or stuttering. Finally, advertisers need to fill in selling point information, display copy, and images on the server when uploading, which is cumbersome and makes it impossible to create personalized displays based on user preferences, failing to attract different user groups with different ad benefits.

[0063] In view of this, this application proposes a content display method, apparatus, electronic device, and storage medium. Specifically, when browsing an information stream product, a user can trigger a content switching operation in the content browsing interface to switch the information stream currently displayed. In the embodiments of this application, the information stream can carry push media content to be displayed. When the user triggers the content switching operation, and the switched content browsing interface needs to display the push media content, the client in this application does not directly display the original push media content, but rather displays exclusive promotional content generated based on this push media content, tailored to the user. For different users, different exclusive promotional content can be generated based on the same push media content, achieving a highly personalized content presentation and a personalized content display effect for different users.

[0064] Furthermore, this exclusive promotional content specifically includes content fragments pointed to by various levels of exclusive key information matched with the target user. Unlike the universality of the original key information in the push media content, this exclusive key information is generated by combining the original key information in the push media content with the target user's content preference information. For different users, it can match the preferences and interests of different users. For a single user, it can better match the target user's interests and consumption habits, thus optimizing the matching degree between the push media content and the target user.

[0065] Based on this, when displaying personalized promotional content to the target user, this application directly locates and plays the content segment pointed to by the personalized key information at the user's interest level (i.e., the level matching that interest) within the personalized promotional content, according to the target user's interest level in the pushed information stream. In other words, for the same target user, if their interest in the current information stream content differs, content segments pointed to by different levels of personalized key information will be prioritized for that user to attract clicks and generate consumption behavior for the pushed media content, thereby increasing click-through rates. Clearly, this avoids the aforementioned problems of information overload, heavy interaction load, compromised user experience consistency, high loading and running requirements, and the need for manual intervention in delivery.

[0066] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0067] Figure 1 illustrates an application scenario according to an embodiment of this application. The application scenario includes a terminal device 110 and a server 120.

[0068] In this application embodiment, the terminal device 110 includes, but is not limited to, mobile phones, tablets, laptops, desktop computers, e-book readers, smart voice interaction devices, smart home appliances, and in-vehicle terminals. The terminal device may have a client related to information flow products installed. This client can be software (such as a browser, short video software, etc.), or a webpage, mini-program, etc. The server 120 is the server corresponding to the software, webpage, mini-program, etc., or a server specifically used for media content display control; this application does not impose specific limitations. The server 120 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0069] It should be noted that the content display methods in the various embodiments of this application can be executed by an electronic device, which can be a terminal device 110 or a server 120. That is, the method can be executed by the terminal device 110 or the server 120 alone, or by both the terminal device 110 and the server 120. For example, when executed by both the terminal device 110 and the server 120, the terminal device 110 can be equipped with the aforementioned client, such as a browser. The user receiving the push can browse the pushed information stream through the content browsing interface in the browser. This information stream can carry the pushed media content to be displayed. When the user receiving the push triggers a content switching operation on the content browsing interface, switching the information stream displayed on the current content browsing interface, assuming that the original content browsing interface after the switch needs to display the pushed media content, the client in this application does not directly display the original pushed media content, but displays exclusive promotional content generated based on this pushed media content for the user receiving the push. This exclusive promotional content is generated by the server 120 and returned to the client. The specific implementation method is as follows:

[0070] The client displays a content browsing interface to the pushed user, which is used to show the pushed information stream to the pushed user;

[0071] During the browsing process of the pushed user, the server 120 will preload the pushed media content to be displayed carried by the information stream; extract each original key information in the pushed media content and obtain the content preference information of the pushed user; generate exclusive promotional content for the pushed user based on each original key information and content preference information, wherein the exclusive promotional content is used to describe the exclusive key information of different levels that match the pushed user.

[0072] Then, server 120 sends the exclusive promotional content to the client of the target user. After the client responds to the target user's content switching operation on the content browsing interface through terminal device 110, the exclusive promotional content generated based on the push media content carried by the information stream can be displayed on the content browsing interface, and the user can jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback.

[0073] In one alternative implementation, the terminal device 110 and the server 120 can communicate via a communication network.

[0074] In one alternative implementation, the communication network is a wired network or a wireless network.

[0075] It should be noted that Figure 1 is only an example, and in reality, the number of terminal devices and servers is not limited, and no specific limitation is made in this embodiment.

[0076] In this embodiment of the application, when there are multiple servers, the multiple servers can form a blockchain, and the servers are nodes on the blockchain; as disclosed in the content display method of this embodiment, the content-related information and user-related information involved can be stored on the blockchain, such as information flow data, data related to exclusive promotional content, user content preference information, etc.

[0077] Furthermore, the embodiments of this application can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence (AI), smart transportation, and assisted driving. Specifically, they involve browsing-related businesses of information flow products in these scenarios. A few examples are briefly listed below:

[0078] (a) Information flow advertising on cloud technology platforms:

[0079] On cloud technology platforms, when users browse technical articles, solutions, and the latest news, the system can push relevant advertisements based on users' interests and behavioral data, such as the latest cloud computing services, security tools, or data analysis solutions.

[0080] (II) Information flow advertising in intelligent transportation applications:

[0081] In intelligent transportation applications, when users view traffic information, navigation routes, or traffic news, the system can push relevant advertisements based on the user's travel habits and preferences, such as nearby gas stations, car repair services, or traffic violation inquiry services.

[0082] (III) In-feed advertising for driver assistance systems:

[0083] In driver assistance systems, when users are using navigation and driver assistance functions, the system can push relevant advertisements based on the user's driving habits and destination information, such as restaurants, hotels, attractions, or emergency rescue services along the way. For example, when a user sets a destination for a long trip, the system may push advertisements for rest stops, gas stations, or specialty restaurants along the route.

[0084] In the scenarios listed above, the content display method proposed in this application can be used to display these advertisements to users. Specifically, the content display method proposed in this application intelligently processes the original advertisements that are originally pushed to the target user, generating personalized advertisements that meet the target user's own needs. For example, when the system pushes an advertisement for a specialty restaurant to a user, it can highlight the dishes, discount information, or dining experience that the user is most interested in based on the user's eating habits, thereby increasing the user's click-through rate.

[0085] It should be noted that the above-listed practical application scenarios are just simple examples. Other related application scenarios are also applicable to the embodiments of this application, and will not be described in detail here.

[0086] Furthermore, it should be emphasized that the specific embodiments of this application involve user-related data, such as the content preference information and user behavior data listed above. When the above 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.

[0087] The following describes the content display method provided by the exemplary embodiments of this application in conjunction with the application scenarios described above and with reference to the accompanying drawings. It should be noted that the above application scenarios are only shown to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited in any way in this respect.

[0088] Referring to Figure 3, this is a flowchart of an implementation of a content display method provided in an embodiment of this application, applied to a client. The specific implementation process of this method is as follows: S31-S32:

[0089] S31: Display content browsing interface; the content browsing interface is used to display the pushed information stream to the pushed user.

[0090] An information feed is a collection of content that a user browses on a client. Media content is the basic unit that constitutes an information feed, including text, images, videos, etc. It can be divided into two main categories: general content (such as news, articles, and user sharing) and advertising content (such as product promotion and brand promotion).

[0091] Regular content refers to non-advertising content generated by the platform or users within the information feed. This content includes news, articles, videos, images, and user-shared content, aiming to provide information, entertainment, or social interaction. Regular content primarily aims to meet users' reading, viewing, or interaction needs, enhancing user experience and platform activity. This type of content typically does not have commercial promotional purposes, focusing more on content quality and user interests.

[0092] Advertising content refers to media content inserted into news feeds that aims to promote products, services, brands, or ideas. This content is typically provided by advertisers, reviewed by the platform, and then displayed to users. It may include images, videos, text, and interactive elements, and usually carries a clear "advertisement" label. In practice, advertising content is designed to attract users' attention, deliver specific marketing messages, and encourage users to take action, such as clicking links, downloading apps, or purchasing products; it is a type of promotional media content.

[0093] Of course, the media content used for promotion in the information feed, in addition to advertising content, also includes, but is not limited to, the following types:

[0094] (1) Sponsored Content: Content sponsored by brands or companies is usually of high quality and professionalism, aiming to provide valuable information while implicitly promoting the brand. Examples include industry reports and expert interviews published by brands. (2) Collaborative Content: Content generated by the platform in collaboration with creators or brands, aiming to enhance brand exposure through content marketing. Examples include review videos and feature articles created by well-known bloggers in collaboration with brands. (3) Content Recommended to Users: Content pushed based on users' interests and behaviors, which may include recommendations for certain products or services, but is not necessarily advertising in the traditional sense. Examples include articles, videos, or products that users may be interested in. (4) Event Promotion: Promotional content for events held by the platform or brands. Examples include limited-time offers, lucky draws, and announcements of online and offline events.

[0095] It should be noted that the push media content in this application embodiment can be the promotional media content listed above, or it can be ordinary content; this document does not specifically limit it. The following mainly uses advertising content as an example, but other content is also applicable and will not be elaborated on here.

[0096] The client in this embodiment can be a client that supports information stream browsing, allowing users to browse the information stream through the content browsing interface within the client. Specifically, the client can be a browser, short video platform, social media platform, instant messaging application, content aggregation platform, knowledge sharing community, lifestyle sharing platform, etc., and this document does not impose any specific limitations on it.

[0097] When users consume information feed products, advertising content is generally displayed in a mixed manner within the feed. For example, in browsers, based on users' browsing history and interests, relevant video ads can be inserted into web page content, search results, or recommendation pages to improve ad effectiveness and user experience. On short video platforms, video ads related to users' interests can typically be inserted into the content users browse to increase ad effectiveness. On content aggregation platforms, personalized news and articles can be pushed based on users' reading habits, and related video ads will also be interspersed in the information feed. For instance, in knowledge-sharing communities, in addition to text answers, video content is also supported, and video ads targeting specific questions or topics will appear in their information feed. On lifestyle sharing platforms, users can discover and share shopping experiences, travel experiences, and other content, and can also find brand-partnered video ads in their information feed; and so on.

[0098] In this article, the process by which users view content, information, and advertisements is collectively referred to as the consumption process. The related behaviors during this process are collectively referred to as consumption behaviors, and the habits of users during this process are collectively referred to as consumption habits, and so on.

[0099] It should be noted that the client and content browsing scenarios listed above are just simple examples, and this article does not make any specific limitations on them.

[0100] S32: In response to content switching operations on the content browsing interface, display exclusive promotional content generated based on the push media content carried by the information stream in the content browsing interface, and jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback.

[0101] When users consume content through news feed products, promotional push media content (such as advertisements) is generally displayed in a mixed layout within the news feed. Specifically, mixed layout refers to the method of arranging promotional push media content alongside the media content that users normally consume. For example, advertisements are inserted into the news feed, alternating with non-advertising content to achieve a more natural and seamless display effect. During browsing, users will trigger certain interactive behaviors to switch content based on their interests and preferences, matching their current consumption habits.

[0102] In practical applications, different information flow products differ in their interface design, user groups, and usage scenarios. Therefore, the specific triggering methods for content switching operations also vary. In other words, in this embodiment, the content switching operation of an information flow product can include various user interaction behaviors, which are typically used for navigation between different content items. Several content switching operations are briefly listed below:

[0103] Content switching operation 1: swipe operation.

[0104] Specifically, this includes, but is not limited to, swiping up and down, and swiping left and right. Swiping up and down is the most common way to switch content, allowing users to browse different items. In some layouts, users can also swipe left and right to switch between different content sections or cards.

[0105] When a user switches to pushed media content via a swipe, this application, in terms of display logic, first uses AI to intelligently understand the original information of the pushed media content, extract the original key information, and combine it with the user's content preferences to generate personalized promotional content containing different levels of exclusive key information. Anchor points for exclusive key information are added to the personalized promotional content. Finally, based on the user's swipe state, different levels of exclusive key information are matched and displayed. This content is derived from AI understanding and AI generation, which will be more in line with the user's interests and consumption habits, and promote user clicks, conversions and other behaviors.

[0106] The display logic for other content switching operations described below is similar and will not be repeated here.

[0107] Content switching operation 2: Click operation.

[0108] For example, users can switch content by clicking the "Next Page" or "Previous Page" button on the screen.

[0109] Content switching operation three: scrolling operation.

[0110] For example, on a desktop, users can use the mouse wheel to scroll through pages and browse different content items; on a laptop, users can use touchpad scrolling gestures to switch content, and so on.

[0111] Content switching operation four: Keyboard operation.

[0112] For example, users can use the arrow keys (up, down, left, right) on the keyboard to navigate and switch content. In addition, some applications support using specific keyboard shortcuts to quickly switch content.

[0113] Content switching operation five: voice commands.

[0114] In applications that support voice interaction, users can switch content using voice commands, such as saying "next" or "previous".

[0115] Content switching operation six: Gesture operation.

[0116] In some applications, users can use pinch gestures to zoom in or out of content, thereby indirectly switching between content. Alternatively, users can use two-finger swipes to quickly browse multiple content items, and so on.

[0117] The operation methods listed above can be selected and combined according to the specific information flow product and user interface design to provide a richer and more flexible user experience. Furthermore, the operation methods listed above are merely simple examples; other content switching operations not mentioned herein are also applicable to the embodiments of this application and will not be elaborated upon here.

[0118] In this embodiment, when the user triggers a content switching operation on the content browsing interface and switches the information stream displayed on the current content browsing interface, assuming that the original content browsing interface needs to display the pushed media content, the client in this application does not directly display the original pushed media content, but displays exclusive promotional content generated based on this pushed media content for the user. This exclusive promotional content is generated by the server and returned to the client.

[0119] Specifically, the exclusive promotional content is used to describe different levels of exclusive key information that are matched with the target user; each exclusive key information is generated based on the original key information in the push media content and the target user's content preference information.

[0120] Content preference information describes a user's level of interest and preference for specific types of content. It can also describe some basic information about the user (such as age, gender, and region). Specifically, it can be derived from the analysis of user behavior data (such as browsing history, viewing preferences, and interaction history), or it can be obtained by combining geographical location, social network relationships, user feedback, and contextual information. This content preference information can be in the form of personal tags, which may include age, gender, region, consumption habits, hobbies, etc. Of course, it can also be in other forms, which are not specifically limited in this article.

[0121] For example, a movie enthusiast prefers science fiction and action movies; a fitness guru enjoys fitness tutorials and healthy eating content; a travel enthusiast is passionate about exploring new places and pays attention to travel guides and experience sharing; a tech geek is interested in the latest technology products and digital product reviews; a food blogger enjoys cooking tutorials and food; and so on.

[0122] It is important to emphasize that this application adheres to the principle of compliance and strictly complies with relevant data protection and privacy regulations in the process of collecting and using user data. This helps to protect the legitimate rights and interests of users and ensures that content owners and developers are free from legal disputes.

[0123] Furthermore, this application protects user privacy and data security by establishing a trust mechanism. Specifically, by providing transparent and secure content display services, this helps build a trust mechanism between users and content creators / developers, improves the reputation and image of the entire content ecosystem, and lays a solid foundation for the long-term development of the industry.

[0124] Key information in this application refers to core elements or key features extracted from media content, such as the selling points of an advertisement, the unique features of a product, or the main highlights of an event, which are intended to attract users' attention and stimulate their interest.

[0125] In the embodiments of this application, key information includes, but is not limited to, the following forms: cover images and text, posters, videos, charts, audio, interactive content, text summaries, and tags. This document does not specifically limit these forms.

[0126] For example, consider key information in plain text format. A smartwatch advertisement might highlight its selling points as "24-hour heart rate monitoring, sleep quality analysis, and 14-day ultra-long battery life." Another example is an electric toothbrush whose unique features include "31,000 vibrations per minute, intelligent pressure sensing, and 5 cleaning modes." Yet another example is a music festival promotion whose main highlights are "live performances by renowned domestic and international bands, three consecutive days, and a free camping area."

[0127] Specifically, this application mainly involves two types of key information: original key information and personalized key information. Original key information refers to the core elements or important characteristics of the original push media content, such as the selling points uploaded by advertisers. It is not targeted at any specific user; for different users, the original key information of this push media content is the same, belonging to a type of universal information. Personalized key information, on the other hand, is generated based on the original key information and the content preference information of the target user. It is personalized key information specific to the target user; for different users, the personalized key information can be different, belonging to a type of personalized information.

[0128] In practical applications, there can be a certain correspondence between the exclusive key information in the personalized promotional content and the original key information in the pushed media content. For example, one exclusive key information corresponds to one original key information, one exclusive key information corresponds to multiple original key information, and multiple exclusive key information corresponds to one original key information, etc. In the latter case, each exclusive key information directly corresponds to a specific original key information, which is easy for users to understand; while in the latter case, one exclusive key information encompasses multiple related original key information, providing users with more comprehensive information at once; and in the latter case, when one exclusive key information corresponds to multiple original key information, the same original key information can be split into multiple different exclusive key information to meet the needs of different users. This article does not impose specific limitations on this.

[0129] Similarly, in the embodiments of this application, the content preference information of different users can be the same or different (specifically, it can be partially different or completely different). Therefore, different exclusive promotional content can be generated based on the same push media content to achieve highly personalized content display.

[0130] Furthermore, in step S32, when displaying the newly generated exclusive promotional content to the push recipient, the push recipient's interest in the information flow will be used to determine an exclusive key information that matches the user's interest level from the exclusive key information at different levels, and the user will be redirected to that exclusive key information section for playback.

[0131] Optionally, the level of each exclusive key information is associated with the content preference information of the pushed user, and the level represents the user's attention index to the corresponding exclusive key information.

[0132] In this embodiment, for a given user, each piece of exclusive key information and its level can be dynamically adjusted based on the user's content preferences. The level represents the user's attention index to the corresponding exclusive key information. Generally, the closer a piece of exclusive key information matches the user's preferences, the higher the user's attention index and the higher the level. The attention index represents the degree of attention, reflecting the attractiveness of the exclusive key information to the user. The greater the attractiveness, the higher the user's level of attention to the exclusive key information.

[0133] Each piece of exclusive key information and its level can be generated by the terminal device or by the server. For details on the generation method, please refer to the relevant instructions on the server side, which will not be repeated here.

[0134] Specifically, multiple exclusive key information items for the same pushed media content and the pushed user can be divided into two levels, such as the highest level: S-level exclusive key information, and ordinary levels: A / B / C, etc., different exclusive key information items.

[0135] Of course, multiple exclusive key information for the same push media content and the push recipient can also be divided into multiple levels, that is, each exclusive key information corresponds to a level, such as 5: highest level: S-level exclusive key information, second highest level: A-level exclusive key information, middle level: B-level exclusive key information, second lowest level: C-level exclusive key information, and lowest level: D-level exclusive key information.

[0136] It should be noted that the above level division method is just a simple example, that is, each level corresponds to at least one exclusive key information, and it is allowed for one level to correspond to multiple exclusive key information. This article does not make specific restrictions on this.

[0137] For example, a user named Xiao Li frequently browses health food information online, showing a strong interest in organic and low-sugar foods. For Xiao Li, a certain health food brand's advertisement offers the following three different levels of exclusive selling points (i.e., exclusive key information):

[0138] A: Organic certification (highest level): Since Xiao Li is very concerned about food safety and quality and has a high demand for organic food, "organic certification" has become the highest level of selling point.

[0139] B: Low-sugar formula (medium level): Xiao Li occasionally searches for low-sugar foods to control her daily sugar intake, so "low-sugar formula" is a selling point for the medium level.

[0140] C: Special Offers (Low Level): Although promotional information can attract some consumers, according to Xiao Li's browsing history, he is relatively low in price sensitivity. Therefore, "Special Offers" is set as a low-level selling point.

[0141] Looking at it further, when displaying media content to different users (different groups with different preferences), the focus can be on the current user's preferences to accurately extract and design the content layout, highlighting the content that the user is interested in, and achieving a personalized display effect.

[0142] Therefore, in the above implementation, by associating the level of each exclusive key information with the preferences of the user receiving the push notification, personalized optimization of key information is achieved. Based on this, and combined with the user's interest in the current information stream, a specific level of exclusive key information is determined for playback. This approach helps users find content of interest more quickly. Simultaneously, accurate personalized recommendations help improve content click-through rates and conversion rates, thereby bringing higher returns on investment to content producers.

[0143] In this embodiment of the application, when displaying exclusive promotional content to the pushed user, in order to improve the viewing experience and interaction rate of the pushed user, the exclusive key information to be displayed first can be determined according to the user's interest in the pushed information stream, and the user can directly jump to the content segment pointed to by the exclusive key information for playback.

[0144] It is important to emphasize that the interest level in this article specifically refers to the user's interest in the information feed pushed to the content browsing interface at the moment before the user is about to browse the pushed media content, or in a short period of time before that moment. Hereafter, it will be referred to as the user's interest level in the current content browsing interface / current content, or simply as the current interest level.

[0145] The user's level of interest in the current content browsing interface can be determined through at least one of the following methods:

[0146] Method 1: Determined based on the content switching operations of the user receiving the push notification on the content browsing interface.

[0147] In this approach, interest level is negatively correlated with the trigger state of content switching operations.

[0148] The trigger state refers to the specific characteristics of the content switching operation performed by the user on the content browsing interface, including but not limited to the switching speed, switching range, and switching frequency.

[0149] Interest level I can be calculated using the following formula: I = ab × Vc × Ad × F, where a is the preset base interest level value, V represents the switching speed, b is the weighting coefficient of the switching speed, A represents the switching amplitude, c is the weighting coefficient of the switching amplitude, F represents the switching frequency, and d is the weighting coefficient of the switching frequency. The weighting coefficients b, c, and d can be adjusted and optimized according to the actual situation, and b + c + d ≤ 1.

[0150] Taking the swipe operation as an example of content switching, the switching speed refers to the speed at which the user performs the swipe operation on the content browsing interface, the switching amplitude refers to the distance the user's finger or touch device moves on the screen when performing the swipe operation on the content browsing interface, and the switching frequency refers to the number of times the user performs the content swipe operation per unit of time, reflecting the rhythm and frequency of the user's content browsing.

[0151] Assuming a user can trigger a swipe gesture to switch screen content, the terminal device captures and recognizes the user's swipe gesture in real time, including information such as swipe direction, speed, and distance. Specifically, when determining a user's interest in the current information stream content based on their swipe gesture, the following rules can be set:

[0152] For example, the faster the swipe up, the more likely it is that the user finds the content uninteresting or irrelevant and wants to skip it quickly, indicating a lower level of interest in the current content. Conversely, the slower the swipe up, the higher the user's interest in the current content.

[0153] For example, a larger upward swipe usually means that the user wants to quickly browse multiple pieces of content rather than stay on one piece, which may also indicate that the user has a lower level of interest in the current content. Conversely, a smaller upward swipe indicates that the user has a higher level of interest in the current content.

[0154] For example, if a user swipes the screen quickly multiple times in a short period of time, it indicates that the user has low interest in the current content. Conversely, if a user swipes the screen less frequently over a longer period of time, it indicates that the user has high interest in the current content.

[0155] Of course, the above rules can also be combined. For example, small and slow swipes may indicate that the user is carefully reading or watching the current content, and therefore has a high level of interest. Conversely, large and rapid swipes indicate that the user has a low level of interest and wants to quickly skip the current content; and so on. This will not be elaborated further here.

[0156] Figure 4 illustrates a matching relationship between screen swiping state and interest level in an embodiment of this application. For example, when a user is browsing news information, the news information page currently being viewed is a content browsing interface as described in this application. During browsing, the terminal device can determine the user's interest level in the current content by recognizing the screen swiping speed (hereinafter referred to as swiping speed). For example, if the swiping speed is slow, the user is considered to be interested in the current content, with a high level of interest; conversely, if the swiping speed is fast, the user is considered to be uninterested in the current content, with a low level of interest.

[0157] Figure 4 shows the screen swiping state in terms of swiping speed. Of course, the same applies to swiping range, etc., which will not be elaborated on here.

[0158] In addition, trigger states can also include the complexity of the path a user jumps between different content, the number of times they revisit the content, etc. For example, a complex jump path may indicate that the user is looking for a specific type of content, while a simple path may indicate that the user is satisfied with the current content. For example, frequent revisit behavior of users indicates that the user has a high interest in specific content, etc. This article does not make specific limitations on this.

[0159] Of course, by taking all these trigger states into account, we can more accurately assess a user's interest in the current content.

[0160] Method 2: Determined based on the duration of the user's gaze on the content browsing interface.

[0161] In this method, interest level is positively correlated with the duration of gaze.

[0162] Interest level I can be calculated using the following formula: I = e × T, where T represents the duration of gaze dwell and e is a preset weighting coefficient, the value of which can be adjusted and optimized according to the actual situation.

[0163] In this embodiment, when a user browses content, the terminal device can detect the user's gaze lingering on the screen. For example, eye-tracking devices or cameras can be used to capture the user's gaze movement data, recording the user's gaze point and gaze duration in real time. Through data processing and analysis, the duration of the user's gaze lingering on the current screen is calculated, thereby assessing the user's interest in the current content. After recording the user's gaze point and gaze duration in real time, the data processing and analysis process is as follows: First, the recorded data is preprocessed to remove noise and outliers; then, based on the position and duration of the gaze point, the user's dwell time in each area of ​​the screen is determined; finally, the dwell time in each area is summed to obtain the duration of the user's gaze lingering on the current screen, thereby assessing the user's interest in the current content.

[0164] Alternatively, one could calculate the average viewing time of users receiving push notifications over a period of time to assess their interest in the current content. A longer viewing time generally indicates a higher level of interest in the current content.

[0165] For example, if a user spends a relatively long time in front of a short video when browsing pushed content, it is considered that the user has a high level of interest in that short video; conversely, if the time spent in front of the video is short, the user has a low level of interest.

[0166] Method 3: Determined based on the user's interaction with the information flow in the content browsing interface.

[0167] In this approach, interest level is positively correlated with the number of interactive behaviors; the more interactive behaviors, the higher the interest level.

[0168] Interactive behaviors include, but are not limited to, liking, commenting, sharing, and saving.

[0169] Interest level can be calculated using the following formula: I = μn, where I represents the calculated interest level and n represents the number of interaction behaviors. When counting interaction behaviors, the system counts each user action such as liking, commenting, sharing, and saving, and then sums these counts to obtain the total number of interaction behaviors. μ is a preset weighting coefficient, ranging from 0 to 1. Its value can be adjusted and optimized according to actual conditions. For example, a suitable μ value can be determined through experimental statistics based on the impact of different types of interaction behaviors on user interest level.

[0170] For example, if a user who is receiving a push notification likes and comments on the currently pushed short video content multiple times, it indicates that they have a high level of interest in the current content.

[0171] For example, if a user who received a push notification likes, comments, or shares the article while browsing content on a social media platform, it indicates a high level of interest and appreciation for the content. Conversely, if the user doesn't interact at all, the level of interest is considered low.

[0172] In addition, interactive behaviors may include zooming in or out, indicating whether the user zoomed in or out on certain content, which usually means that the user is particularly interested in certain details.

[0173] It should be noted that the methods of determining the user's current level of interest by recognizing screen swiping speed or eye lingering detection, as listed above, are just simple examples. Other methods of determining interest are also applicable to the embodiments of this application, and will not be described in detail here.

[0174] In this embodiment of the application, when displaying exclusive promotional content to users, the user's interest in the pushed information stream is determined based on different user experience states (such as touch state, gaze duration, etc. listed above). This allows for the priority playback of exclusive key information at different levels to attract user clicks and generate consumption behavior for the pushed media content, thereby increasing the click-through rate.

[0175] In other words, when playing content segments pointed to by exclusive key information matching the user's interest level, the determined user interest matching level will be different depending on the current interest level. Assuming that multiple exclusive key information segments in this exclusive promotional content are divided into two levels, such as the highest level listed above: S-level exclusive key information, and ordinary levels: A / B / C... etc., then interest ranges can be preset (or interest thresholds can be set directly). Different interest ranges will prioritize displaying exclusive key information of different levels. Therefore, step S32 can be specifically divided into the following two cases:

[0176] Playback scenario 1: If the interest level falls within the preset interest level range, the player will be redirected to the exclusive promotional content, where the highest-level exclusive key information will be displayed.

[0177] Playback scenario 2: If the interest level does not fall within the preset interest level range, the player will randomly jump to any content segment pointed to by any exclusive key information (excluding the highest level) for playback, or randomly jump to any content segment pointed to by any exclusive key information among the various exclusive key information for playback.

[0178] Specifically, when making random jumps, a uniform random method can be used, that is, selecting evenly from multiple exclusive key information; a weighted random method can also be used, selecting according to the weight of each exclusive key information, with content segments with higher weights having a higher probability of being selected. This weight can be related to level, user's historical behavior, etc., which are not specifically limited in this article; another example is the use of time window random method, selecting content segments within a relatively recent time period; and so on.

[0179] Playback scenario two is further divided into two sub-scenarios:

[0180] Sub-scenario 1: Randomly play any content segment pointed to by any exclusive key information (including the highest level exclusive key information).

[0181] In sub-case one, any of the exclusive key information here can be one of the S / A / B / C... listed above. Even if the user receiving the push notification has a high level of interest in the current content, they may randomly jump to the content segment pointed to by the highest level exclusive key information for playback.

[0182] Sub-scenario 2: Randomly play the content segment pointed to by any exclusive key information (excluding the highest level exclusive key information).

[0183] In sub-case two, any of the exclusive key information here can be one of A / B / C... listed above. That is, when the user receiving the push has a high level of interest in the current content, the content segment pointed to by the S-level exclusive key information will not be played first, but will be randomly jumped to the content segment pointed to by any of the exclusive key information in A / B / C... for playback.

[0184] Suppose the personalized promotional content contains five key pieces of information. These five pieces of information belong to two levels. As mentioned above, the interest level of the target user can be divided into two interest ranges, with each range corresponding to a level. For example, an interest level in the range of 0-50 indicates that the target user's current interest level is low, corresponding to playback scenario one above. An interest level in the range of 50-100 indicates that the target user's current interest level is high, corresponding to playback scenario two above. Specifically, it could be one of the sub-scenarios within playback scenario two.

[0185] It should be noted that the above are just simple examples, and other situations are also applicable to the embodiments of this application, which will not be elaborated on in detail here.

[0186] Furthermore, the above example uses the division of multiple exclusive key information in this exclusive promotional content into two levels. If the multiple exclusive key information in this exclusive promotional content is divided into more levels, assuming each exclusive key information has one level, then before implementing step S32, more interest intervals can be divided, each interest interval can be matched with a level, and according to the interest interval to which the current interest belongs, the user is redirected to the exclusive promotional content, and the content segment pointed to by the exclusive key information of the level that matches that interval is played.

[0187] For example, suppose the personalized promotional content contains five personalized keywords, each with a different level. The user's screen scrolling speed can be divided into five speed ranges, each corresponding to an interest level. For instance, users with an interest level of 80-100 will be redirected to the lowest-level personalized keyword content; those with 60-80 will be redirected to the next lowest level; those with 40-60 will be redirected to the middle level; those with 20-40 will be redirected to the next highest level; and those with 0-20 will be redirected to the highest level, and so on.

[0188] It should be noted that the above are just simple examples, and other situations are also applicable to the embodiments of this application, which will not be elaborated on in detail here.

[0189] In practical applications, considering that the user experience states listed in this article (such as screen scrolling speed, gaze duration, etc.) can reflect the user's interest in the current content to a certain extent, in order to better match the user's information flow consumption habits, different state intervals can be set directly according to different user experience states to match different levels of exclusive key information display, which is more conducive to pushing more interesting key information to users at the right time.

[0190] As shown in Figure 4, when a user swipes the screen at a slower speed, lower-level, exclusive key information can be displayed, giving the user more time to focus on it. Conversely, when the user swipes faster, their gaze is less focused on the screen, allowing higher-level key information to attract their attention and encourage clicks. The specific implementation of this process will be illustrated below and will not be repeated here.

[0191] Specifically, in this approach, one or more speed ranges can be preset, which can be understood as interest ranges. Similarly, when using other types of user experience states (such as gaze duration) to determine interest, the corresponding state ranges (such as duration ranges) can also be directly understood as interest ranges, etc. This article does not make specific limitations on this.

[0192] The following example demonstrates how to determine the interest level of a user based on their screen swiping speed. Interest level is negatively correlated with the screen swiping speed of the user. In other words, a speed range related to screen swiping speed can be set, and by analyzing whether the user's swiping speed falls within the preset speed range, it can be determined which level of exclusive key information to prioritize displaying.

[0193] Taking the push media content as the advertising content and the key information as the selling points of the advertising content as an example, the electronic device in this application has the ability to map scrolling speed to video positioning. This ability can calculate the video time segment of the exclusive selling point corresponding to the user's interest level within the exclusive advertising content based on the user's scrolling speed, so as to position the device at the corresponding location for playback. For example, fast scrolling will jump to a pre-set (or AI-detected) highlight segment, while slow scrolling will position the device within a more recent time segment. A highlight segment refers to the video time segment of the exclusive advertising content within the exclusive advertising content, calculated by the electronic device based on the user's scrolling speed and matching the user's interest level. This is a pre-set (or AI-detected) video segment that highlights the key selling points of the advertisement and attracts the user's attention, and will be played when the user scrolls the screen quickly.

[0194] The following example illustrates this:

[0195] Figure 5 illustrates a method of matching selling points based on the screen swiping speed of the user in this application. The diagram in Figure 5 divides selling points into two levels: S-level selling points and A / B / C… selling points. S-level selling points represent the strongest selling point, followed by A / B / C… in priority order. Furthermore, three speed ranges are defined: screen swiping speed ≥ 5s / screen (slow), 5s / screen ≤ screen swiping speed ≤ 2s / screen (relatively slow), and screen swiping speed ≤ 2s / screen (relatively fast).

[0196] On the user side, when a user browses content after receiving the push notification, the merchant side will locate and display different advertising selling points based on different user experience states (such as different scrolling speeds of the user). These selling points (specifically, the exclusive key information and corresponding content fragments at the user interest matching level mentioned in this article, which can be extracted and generated by AI) and the merchant side can also identify the user's screen scrolling speed to display corresponding selling points to attract users to click and convert. Specifically, the exclusive advertisement generated for the user will play from the located selling points, and then continue to rotate to the next selling point. After all selling points have been displayed, all selling point fragments will be rotated in sequence (such as in chronological order, in order of level, etc.). The specific process of rotating the display is described in the following embodiment and will not be repeated here.

[0197] The following is a brief explanation of the relationship between the different speed ranges mentioned in Figure 5 and the user interest matching level:

[0198] On the product side, during the process of a user consuming pushed advertising content, based on the detection of scrolling speed in the preceding process, when the user scrolls to the advertising slot, the corresponding selling point information will be displayed in that advertising slot, as shown in the several cases listed in Figure 5. The specific determination and display methods are as follows:

[0199] The first method: When the screen swiping speed of the user being pushed is less than or equal to 2 seconds per screen, or when the screen swiping speed of the user being pushed is greater than or equal to 5 seconds per screen, any one of the selling points A / B / C.... can be randomly displayed. That is, when it is recognized that the user being pushed has a high level of interest in the current screen content, ordinary selling points are randomly displayed to the user being pushed.

[0200] The second approach is to display the top-tier selling points when the screen swiping speed of the target user is ≤2s / screen. This means that when it is recognized that the target user has low interest in the current screen content, the strongest selling point of the advertisement is directly displayed to the target user.

[0201] In the solution shown in Figure 5, by intelligently recognizing the user's swiping speed, the system can accurately capture the user's interests and display highly relevant exclusive selling points. This personalized display method allows users to experience more personalized and thoughtful service, thereby increasing their acceptance and favorability of the advertisement.

[0202] It should be noted that the situations listed in Figure 5 above are just simple examples. If there are more different levels of advertising selling points, more different speed ranges can be divided. For example, if the screen scrolling speed is ≤2s / screen, it is positioned at the highest level of advertising selling point; if the screen scrolling speed is ≤5s / screen, it is positioned at the second highest level of advertising selling point; if the screen scrolling speed is ≤10s / screen, it is positioned at the second lowest level of advertising selling point; if the screen scrolling speed is ≥10s / screen, it is positioned at the lowest level of advertising selling point, and so on. The specific settings can be flexibly set according to actual needs, and will not be elaborated on in this article.

[0203] In summary, the above briefly describes how, in this embodiment of the application, the user's interest in the current content can influence which specific key information is first displayed to the user when they switch to the pushed media content. Furthermore, to ensure the user receives more comprehensive key information, in step S32, after the content segment indicated by the key information matching the user's interest level has finished playing, the content segment indicated by the alternative key information can continue to play.

[0204] Specifically, the first step is to determine the priority display of exclusive key information content segments based on the recipient's interest in the current content. Once this segment finishes playing, the next segment can be played, which is linked to by alternative exclusive key information within the same promotional content. During this step, one or more content segments can be played; this is not specifically limited in this document.

[0205] Taking the continuation of a content segment as an example, the client can randomly play an alternative exclusive key information content segment, or play them according to priority or the time sequence of these exclusive key information content segments, etc. This article does not make specific restrictions on this.

[0206] Taking the continued playback of multiple content segments as an example, in this application, the client can continue playing other content segments according to at least one of the following strategies:

[0207] Playback Strategy 1: Random Playback.

[0208] Each time, a content segment is randomly selected from the remaining exclusive key information and played until all the content segments of each exclusive key information have been played.

[0209] For example, a personalized promotional content has a total of 4 personalized key information content segments, denoted as S, A, B, and C, with the priority order as: S>A>B>C.

[0210] In step S32, the A-level exclusive key information content segment is displayed first. After the playback is completed, the C-level exclusive key information content segment is played randomly. After the playback is completed, the S-level exclusive key information content segment is played randomly. After the playback is completed, the B-level exclusive key information content segment is played randomly.

[0211] In this embodiment, a content segment is randomly selected from the remaining exclusive key information for playback each time, until all content segments for each exclusive key information have been played. This method increases the diversity and novelty of the content, preventing users from becoming bored due to monotonous content. Random playback also ensures that users receive different information at different times, improving the overall content coverage and user engagement.

[0212] Playback Strategy Two: Sort by Interest.

[0213] Based on the remaining exclusive key information, prioritize each content segment and play the content segments that the pushed user may be more interested in first, until all the content segments for each exclusive key information have been played.

[0214] Taking the above example, in step S32, the A-level exclusive key information content segment is displayed first. After the playback is completed, the S-level exclusive key information content segment is played. After the playback is completed, the B-level exclusive key information content segment is played. After the playback is completed, the C-level exclusive key information content segment is played.

[0215] In this embodiment, the remaining exclusive key information is prioritized, and content segments that the pushed content user may be more interested in are played first, until all content segments for each exclusive key information have been played. This approach can more accurately meet the user's personalized needs and improve user satisfaction and engagement with the content. By prioritizing the display of highly interesting content, user attention can be quickly attracted, enhancing the user experience.

[0216] Playback Strategy 3: Rotate all segments.

[0217] For example, a customized promotional content has a total of 4 exclusive key information content segments, denoted as S, A, B, and C, in chronological order: S->C->B->A.

[0218] In step S32, the A-level exclusive key information content segment is displayed first. After the A-level exclusive key information content segment is played, the S-level exclusive key information content segment is played. After the S-level exclusive key information content segment is played, the C-level exclusive key information content segment is played. After the C-level exclusive key information content segment is played, the B-level exclusive key information content segment is played.

[0219] In this embodiment of the application, this approach ensures that users can access all key information multiple times, deepening their impression and improving the information absorption rate. The carousel strategy can also provide a consistent content experience when users visit multiple times, enhancing their memory and cognition, and improving the overall influence of the content.

[0220] In summary, by continuing to play content segments linked to optional key information, not only does this increase content diversity and novelty, preventing users from losing interest due to monotony, but it also precisely meets users' personalized needs, improving user engagement and satisfaction. Simultaneously, it ensures users have multiple exposures to all key information, reinforcing their understanding and enhancing information absorption and the overall user experience.

[0221] Of course, the playback strategies listed above can also be combined. For example, when combining playback strategy one and playback strategy three, one possible playback method is:

[0222] First, play the content segment pointed to by the selected exclusive key information randomly according to playback strategy one; after all the content segments pointed to by each exclusive key information have been played, play the content segments pointed to by each exclusive key information in turn according to playback strategy three.

[0223] The specific process is as follows:

[0224] The initial playback involves prioritizing the playback of exclusive key information snippets that match the user's interests, based on the user's level of interest. This quickly attracts the user's attention, provides the most relevant information, and improves the user's initial engagement and satisfaction.

[0225] Then, random playback is performed. After the content segments matching the user's interest level have finished playing, a previously unplayed exclusive key information content segment is randomly selected and played. This step is repeated until all exclusive key information content segments have been played. This method increases the diversity and freshness of the content, prevents users from losing interest due to monotonous content, and ensures that users can access a variety of key information.

[0226] Finally, a carousel is played. After all content segments have finished playing, all exclusive key information segments are played in sequence to ensure that the pushed content can be accessed by the user multiple times. This method ensures that the user can be exposed to all key information multiple times, deepens the impression, improves the information absorption rate and the overall user satisfaction.

[0227] For example, a customized promotional content has a total of 4 exclusive key information content segments, denoted as S, A, B, and C, in chronological order: S->C->B->A.

[0228] In step S32, the A-level exclusive key information content segment is displayed first. After it finishes playing, the B-level exclusive key information content segment is played randomly. After that, the S-level exclusive key information content segment is played randomly. After that, the C-level exclusive key information content segment is played randomly. After all the content segments have finished playing, the S-level exclusive key information content segment, C-level exclusive key information content segment, B-level exclusive key information content segment, A-level exclusive key information content segment, and so on, in chronological order.

[0229] Taking the cases listed in Figure 5 as examples, Figure 6 shows another schematic diagram of matching selling points based on the screen swiping speed of the pushed user in this application embodiment. The judgment and display methods in Figure 6 are the same as those listed in Figure 5. Figure 6 is just an exemplary visualization from the perspective of the client interface.

[0230] Suppose that the personalized ad generated for the target user has three unique selling points, denoted as Selling Point 1, Selling Point 2, and Selling Point 3. Selling Point 3 is the strongest selling point, while Selling Points 1 and 2 are of the same level, both considered ordinary selling points. As shown in Figure 6, depending on the target user's scrolling speed, the video ad content with different selling points can be previewed, as detailed below:

[0231] When the screen swiping speed of the user receiving the push notification is ≤5s / screen, or when 5s / screen ≤ the screen swiping speed of the user receiving the push notification is ≤2s / screen, one of the selling points will be randomly played in the corresponding ad slot, as shown in Figure 6, which shows the random display of selling point 1. After the selling point at the user's interest matching level has finished playing, the next selling point will be randomly displayed, as shown in Figure 6, which shows the random display of selling point 2.

[0232] When the screen swiping speed of the user receiving the push notification is ≤2s / screen, the strongest selling point will be displayed in the corresponding ad slot, as shown in Figure 6, which shows selling point 3.

[0233] In either case, after all the selling points have been displayed, all selling point segments will be rotated in sequence, as shown in Figure 7, which is a schematic diagram of a selling point rotation in an embodiment of this application. For the above three selling points, selling point 3 -> selling point 2 -> selling point 1 can be rotated in sequence. The rotation order is not specifically limited in this document.

[0234] Next, two real-world advertising examples will be used to illustrate the display process of the exclusive promotional content in the embodiments of this application:

[0235] Advertising Case 1:

[0236] Assume the user receiving the push notification (denoted as User X1) is a news-conscious user who is also a heavy follower of new phone information. User X1 is currently scrolling through news at a relatively slow swipe speed. When they reach the ad slot, they are pushed an ad showcasing a new phone from Brand C. Based on the content display method described in this application, the original ad is not directly shown to User X1. Instead, a personalized ad is generated based on the original ad and User X1's content preferences. This personalized ad contains three selling points, which are also tailored to User X1's preferences, including extracted content related to the new phone's performance and selling points, such as:

[0237] Such features include fast response time, satellite communication, and powerful performance.

[0238] Referring to the example of the relationship between screen swiping speed and interest level shown in Figure 4 above, if user X1's current swiping speed is relatively slow, it means that user X1 has a high level of interest in the current content. Using sub-case one in playback case two above, the three selling points of the exclusive advertisement can be randomly played and displayed at this time. First, one selling point is randomly played. After the user finishes watching the first selling point, the next selling point is randomly displayed, and so on. After all the selling points are displayed, all selling point segments are played in turn.

[0239] For example, the content segment pointed to by selling point 1 is first played randomly, as shown in Figure 8, which is a schematic diagram of the advertisement display in the first type of browser in this application embodiment. Figure 8 shows that user X1 views information at a relatively slow swipe speed. When swiping to the advertisement slot, as shown in area S81 of Figure 8, selling point 1 is randomly displayed to user X1. Selling point 1 mainly refers to the launch time of the phone, reminding the user that the C brand phone is available for purchase today. Specifically, in the area shown in S81, in addition to playing the content segment pointed to by selling point 1, user X1 can also be prompted that selling point 1 is generated by AI intelligent extraction, as shown in S82. Furthermore, playback time points can be marked on the playback progress bar, as shown in S83.

[0240] After selling point 1 finishes playing, the content segment pointed to by selling point 2 is played randomly, as shown in Figure 9. This is a schematic diagram of the advertisement display in the second type of browser in this application embodiment. Figure 9 shows the process of automatically jumping to the next selling point - selling point 2 after selling point 1 finishes playing. Selling point 2 mainly emphasizes the satellite calling function of the mobile phone. Similarly, in the area shown in S91, in addition to playing the content segment pointed to by selling point 2, the user X1 can also be prompted that selling point 2 is generated by AI intelligent extraction, as shown in S92. In addition, the playback time point can also be marked on the playback progress bar, as shown in S93.

[0241] After selling point 2 finishes playing, the content segment pointed to by selling point 3 is played randomly, as shown in Figure 10. This is a schematic diagram of the advertisement display in the third type of browser in this application embodiment. Figure 10 shows the process of automatically jumping to the next selling point - selling point 3 after selling point 2 finishes playing. Selling point 3 mainly emphasizes the phone's powerful performance, ultra-reliable architecture, ultra-high-definition imaging, etc. Similarly, in the area shown in S101, in addition to playing the content segment pointed to by selling point 3, the user can also be prompted X1 that selling point 3 is AI-generated intelligent selling point 3, as shown in S102. In addition, the playback time point can also be marked on the playback progress bar, as shown in S103.

[0242] It should be noted that the advertising selling points, their display methods, and display order listed above are just simple examples, and this article does not make any specific limitations.

[0243] Advertising Case 2:

[0244] Suppose another user (user X2) is a food enthusiast who enjoys deals and benefits. User X2 is currently scrolling through information at a relatively fast pace. When they reach the ad slot, they are presented with a promotional ad for a certain package deal. Based on the content display method described in this application, the original ad is not directly shown to User X2. Instead, a personalized ad is generated based on the original ad and User X2's content preferences. This personalized ad may also have three selling points, tailored to User X2's preferences. For example, when the server uses AI to understand the original ad content, it focuses on extracting the discount information. When generating key selling points, the AI ​​also emphasizes the discount information and related text, visually amplifying and highlighting this information. The visual layout design also uses high-contrast information elements so that User X2 immediately understands the discount information.

[0245] Referring to the example of the relationship between swiping speed and interest in Figure 4 above, if user X2's current swiping speed is relatively fast, it means that user X2 has a low level of interest in the current content. In the case of playback scenario one described above, the highest level of selling point information in the exclusive advertisement should be played first. After the user has finished viewing the selling point, the next selling point should be displayed randomly, and so on. After all the selling points have been displayed, all selling point segments should be played in rotation.

[0246] Figure 11 illustrates the ad display in the fourth browser embodiment of this application. Figure 11 shows user X2 quickly scrolling up to view information. When scrolling to the ad slot, as shown in area S111 of Figure 11, the strongest selling point—such as selling point 1—is prioritized for user X2. Selling point 1 primarily refers to the limited-time price discount information for the package, reminding the user that the Maixx package is only 10 yuan today. Specifically, in the area shown in S111, in addition to playing the content segment indicated by selling point 1, user X2 can also be prompted that selling point 1 is generated by AI intelligent extraction, as shown in S112. Furthermore, playback time points can be marked on the playback progress bar, as shown in S113.

[0247] After the strongest selling point - Selling Point 1 - finishes playing, the content segment pointed to by Selling Point 2 is played randomly, as shown in Figure 12, which is a schematic diagram of the advertisement display in the fifth type of browser in this application embodiment. Figure 12 illustrates the process of automatically jumping to the next selling point - Selling Point 2 after Selling Point 1 finishes playing. Selling Point 2 mainly includes the featured products and promotional information of the package, reminding users of the watermelon popping boba drink's summer surprise, buy one get one free. Specifically, in the area shown in S121, in addition to playing the content segment pointed to by Selling Point 2, the user can also be prompted that Selling Point 2 is generated by AI intelligent extraction, as shown in S122. Furthermore, playback time points can be marked on the playback progress bar, as shown in S123.

[0248] After selling point 2 finishes playing, the content segment pointed to by selling point 3 is played randomly, as shown in Figure 13, which is a schematic diagram of the advertisement display in the sixth type of browser in this application embodiment. Figure 12 shows the process of automatically jumping to the next selling point—selling point 3—after selling point 2 finishes playing. Selling point 3 mainly provides information on the takeaway discount for this package, reminding users of free takeaway delivery. Specifically, in the area shown in S131, in addition to playing the content segment pointed to by selling point 3, the user can also be prompted that selling point X2 is AI-generated selling point 3, as shown in S132. Furthermore, playback time points can be marked on the playback progress bar, as shown in S133.

[0249] It should be noted that the advertising selling points, their display methods, and display order listed above are just simple examples, and this article does not make any specific limitations.

[0250] Through this multi-stage playback strategy, the system can not only quickly attract users' attention, but also ensure that users receive the most relevant and valuable information at different time periods, ensuring that users can fully understand all important information and not miss any important content, thereby improving user satisfaction and engagement.

[0251] It should be noted that the playback strategies or combinations thereof listed above are merely simple examples. Other combinations are also applicable to the embodiments of this application. For example, playback strategy two and playback strategy three can be combined, i.e., initial playback is performed first, i.e., based on the interest level of the pushed user, exclusive key information content segments matching the user's interest level are played first; then, after the content segments matching the user's interest level have finished playing, they are played in order of priority until all exclusive key information content segments have been played; finally, a loop is played, i.e., after all content segments have been played, all exclusive key information content segments are played in a loop to ensure that the pushed user can access all key information multiple times; and so on.

[0252] Of course, other playback strategies and combinations thereof are also applicable to the embodiments of this application, and will not be described in detail here.

[0253] In step S32, in addition to displaying exclusive promotional content in the current content browsing interface and jumping to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback, the time points associated with each exclusive key information can also be highlighted on the playback progress bar of the exclusive promotional content.

[0254] The time point associated with each unique key piece of information can be the start time of that unique key piece of information within the unique promotional content, that is, the time when the content segment pointed to by that unique key piece of information begins to play. For example, unique key pieces of information A start at the 3rd second, B starts at the 6th second, and C starts at the 8th second.

[0255] Of course, this point in time can also be an important node or a climax in the content segment pointed to by the specific key information, etc., but this article does not make specific limitations on it.

[0256] For ease of distinction, the time point for associating exclusive key information at the user interest matching level can be denoted as the first time point, and the time point for associating alternative exclusive key information can be denoted as the second time point (also known as the alternative time point). In the embodiments of this application, the first time point and the second time point can be highlighted in the same way or in different ways, and this document does not impose any specific limitations.

[0257] There are many ways to highlight time points, such as: color marking; graphic identification, which refers to adding specific icons or symbols, such as asterisks, dots, or arrows, to the time point; highlighting, which can be achieved by color changes, enlargement, or adding borders; text prompts, which display short text prompts near the time point to explain the content corresponding to that time point; and adding animation effects, such as blinking, pulse, or gradient effects at the time point, which can be automatically highlighted when the user first loads the app, or triggered when the user interacts (such as when the mouse hovers over the screen).

[0258] In this paper, different colors, graphics, text prompts, and animation effects can be used to distinguish between the first and second time points, but no specific restrictions are imposed on this.

[0259] These methods effectively highlight the time points corresponding to key information, helping users quickly locate and jump to content segments of interest.

[0260] Figure 14 illustrates a method for highlighting specific key information time points in an embodiment of this application. In Figure 14, the time points of each specific key information point are highlighted with a magnified effect. Assume the specific promotional video is a 10-second specific advertisement generated from a 20-second original advertisement (i.e., the pushed media content), containing three specific key information points: A, B, and C, with time points distributed as follows: second 3 (00:03), second 6 (00:06), and second 8 (00:08). These time points are magnified on the playback progress bar.

[0261] It should be noted that the above-listed display methods for time points are just simple examples. Other display methods are also applicable to the embodiments of this application, and will not be described in detail here.

[0262] The above implementation provides an intuitive visual cue to help users better understand and manage content.

[0263] Based on this, users can independently select, quickly locate, and jump to content segments of interest, improving information absorption and retention. In other words, this application supports users selecting between multiple specific key information segments. One possible implementation method is as follows:

[0264] In response to a first-type trigger operation for any alternative time point, the user is redirected to the corresponding content segment in the exclusive promotional content for playback. The alternative time points are time points associated with alternative exclusive key information other than the exclusive key information of the user interest matching level.

[0265] Here, "corresponding content fragment" refers to the content fragment pointed to by the exclusive key information associated with "any alternative time point".

[0266] Taking Figure 14 as an example, if the time point associated with the exclusive key information of the user interest matching level is 00:03, then the alternative time points include 00:06 and 00:08. Among them, the content segment pointed to by 00:06 is the content segment pointed to by exclusive key information B, and the content segment pointed to by 00:08 is the content segment pointed to by exclusive key information C.

[0267] The first type of triggering operation is used to distinguish it from the second type of triggering operation. In essence, both are time-based triggering operations, such as clicking, touching, long pressing, and double-tapping. It is only necessary to note that the specific actions of the first type of triggering operation and the second type of triggering operation are different. For example, the first type of triggering operation is a click and the second type of triggering operation is a long press, or the first type of triggering operation is a long press and the second type of triggering operation is a double-tapping, etc. It can be flexibly set according to the scenario and actual needs. This article does not make specific restrictions on it.

[0268] This application supports push recipients in selecting between multiple exclusive key information items. Specifically, when a push recipient browses the pushed media content, the push recipient is directly shown exclusive promotional content generated based on this pushed media content, and is preferentially redirected to the content segment pointed to by the exclusive key information at the user's interest matching level within that exclusive promotional content.

[0269] Meanwhile, this application supports highlighting the time points associated with each exclusive key information in the playback progress bar of the exclusive promotional content. In this way, the pushed users can jump to the corresponding content segment in the exclusive promotional content for playback by performing the first type of trigger operation for any of the alternative time points.

[0270] Taking video ads as an example, the exclusive key information can be understood as the advertising selling points, assuming the first type of triggering operation is a click. As shown in Figure 15, it is a schematic diagram of mutual selection of advertising selling points in an embodiment of this application. For example, selling point 1 is played first, and three time points are highlighted in the playback progress bar, namely selling point 1, selling point 2 and selling point 3. During the playback of the content segment pointed to by selling point 1, if the user clicks the third time point (i.e., the time point of selling point 3), they can jump to the content segment pointed to by selling point 3 to start playback.

[0271] Similarly, if a user clicks on the second time point (i.e., the time point of selling point 2) during the playback of the content segment pointed to by selling point 1, the user will be redirected to the content segment pointed to by selling point 2 to start playback.

[0272] Of course, after the content segment indicated by selling point 1 finishes playing, other selling points can continue to play, such as the content segment indicated by selling point 2. During this playback, if the user clicks on the third time point (i.e., the time point of selling point 3), they will be redirected to the content segment indicated by selling point 3. Similarly, if the user clicks on the first time point (i.e., the time point of selling point 1), they will be redirected to the content segment indicated by selling point 1, and so on. This article does not impose any specific limitations on this.

[0273] In the above implementation, by highlighting each time point on the playback progress bar, intuitive visual prompts are provided to users, which can more effectively convey key information, improve overall user satisfaction and participation, and provide users with a flexible navigation method, ensuring that users can independently select and quickly locate content segments of interest, increasing interactivity and user experience, and improving the absorption and retention of information.

[0274] Furthermore, considering that this application directly displays personalized promotional content newly generated based on the pushed media content to users during the browsing process, some users may have a need to view the original pushed media content. Therefore, this application also supports users to view the original key information based on personalized key information. One optional implementation method is as follows:

[0275] In response to a second type of triggering operation at any point in time, the pushed media content is displayed, and the user is redirected to the pushed media content to play a content segment associated with the corresponding exclusive key information.

[0276] In this embodiment, there is a certain correspondence between the exclusive key information and the original key information, as detailed in the above embodiments, which will not be repeated here. Therefore, "content fragment associated with the corresponding exclusive key information" refers to the original content fragment in the pushed media content where the original key information corresponds to any given time point. Specifically, when one exclusive key information corresponds to multiple original key information pieces, playback will preferentially jump to the original key information piece with the earliest playback time.

[0277] The second type of triggering operation can be a click, touch, long press, double-click, etc., as detailed in the above embodiments.

[0278] Assuming the second type of triggering operation is a long press, in this embodiment of the application, if the user who received the push wants to view the original key information corresponding to a certain exclusive key information, they can long press the time point corresponding to the exclusive key information. The client responds to the second type of triggering operation of the user who received the push for that time point, displays the push media content, and jumps to the push media content to play the content segment associated with the exclusive key information (that is, the original key information content segment corresponding to the exclusive key information).

[0279] Suppose the pushed media content is a 10-minute tech product review video, and the generated personalized promotional content is a 3-minute advertisement containing three personalized key information: A - 3 seconds, B - 6 seconds, C - 8 seconds, and the corresponding original key information is: A' - 3 minutes, B' - 6 minutes, C' - 8 minutes.

[0280] When a user is watching exclusive promotional content, the content segment of exclusive key information A, which is matched with the user's interest level, is played at the initial 3-second mark. The user can then click at the 6-second mark to jump to the content segment of exclusive key information B. Similarly, the user can click at the 8-second mark to jump to the content segment of exclusive key information C.

[0281] Of course, if users want to view the content related to these exclusive key information in the original tech product review video, they can long-press the 3-second mark to display the original tech product review video and jump to the 3-minute mark to play the content segment related to exclusive key information A. Similarly, users can long-press the 6-second mark to display the original tech product review video and jump to the 6-minute mark to play the content segment related to exclusive key information B. Likewise, users can long-press the 8-second mark to display the original tech product review video and jump to the 8-minute mark to play the content segment related to exclusive key information C.

[0282] Based on the above implementation method, users can be provided with personalized promotional content while also viewing the original media content. Users can freely choose to view either promotional content or original content according to their interests and needs, improving user experience and satisfaction. Simultaneously, this method enhances information transparency, allowing users to trace back to the original content, verify the authenticity of promotional information, and increase trust. Furthermore, flexible triggering actions can enhance user interactivity and engagement, improving content appeal and dissemination effectiveness.

[0283] In summary, this solution can display corresponding promotional information based on different users' preferences and behavioral characteristics. This means content creators can reach their target audience more accurately, increasing the exposure and click-through rate of media content, thereby achieving higher conversion rates. Furthermore, by intelligently optimizing ad display and reducing invalid exposure and waste, the solution helps lower the overall cost of media content. At the same time, higher promotional effectiveness also means content creators can obtain greater returns with lower investment. The implementation of this solution will bring numerous benefits, including improved user experience, enhanced content promotion effectiveness, innovative development in the media industry, and protection of user privacy and data security.

[0284] It should be noted that the above is an introduction to the content display method on the client side of this application. The following is a further introduction to the content display method on the server side of this application:

[0285] Referring to Figure 16, this is a flowchart of another content display method provided in this application embodiment, applied to a server. The specific implementation process of this method is as follows: S161 to S163:

[0286] S161: For pushed media content, extract the original key information from the pushed media content and obtain the content preference information of the pushed users.

[0287] Similar to the above description on the client side, the push media content in this application embodiment can be advertising content, sponsored content, event promotion content, or other media content used for promotion, which will not be repeated here.

[0288] Optionally, the push media content can come from a designated material library, which stores media content uploaded by several content owners (such as advertisers, high-quality content creators, etc.) for promotion. These content owners can cooperate with the platforms supported by this application (such as advertising platforms) through a contract, and upload some media content that needs to be promoted to the platform regularly or irregularly.

[0289] For example, advertisers (enterprises or individuals) can enter into cooperation agreements with advertising platforms or media outlets, becoming their official advertising clients. This partnership typically involves a series of service terms and agreements, including but not limited to the scope, timing, format, budget allocation, payment methods, data sharing, and performance evaluation standards for advertising campaigns. For advertising platforms or media outlets, signed advertisers are a significant source of revenue; for advertisers, it's a crucial way to leverage professional platform or media resources to more effectively achieve marketing goals such as brand promotion, product marketing, and user growth. Advertising data sharing refers to the arrangements in the cooperation agreement between the advertiser and the advertising platform or media outlet regarding the sharing of relevant data (such as user behavior data and advertising performance data) during the advertising campaign. Reasonable data sharing helps advertisers and platforms optimize their advertising strategies.

[0290] Based on the above, the platform can store this media content in the material library and deliver the media content according to the requirements of the content owner. For example, different advertisements can be pushed to different users at different times or time periods. The pushed media content can be displayed in the information flow browsed by the user in a mixed manner.

[0291] Specifically, when content creators sign contracts and upload media content, they generally need to provide basic information such as the title, description, creative materials (images, videos, etc.), budget, and bidding strategy. Furthermore, to improve the accuracy and effectiveness of the media content, it's necessary to tag it. These tags indicate the category of the media content. Specifically, content creators can tag ads based on the media content and target market, such as product type, industry, and interests. These tags help the platform more accurately match media content with potential audiences.

[0292] Taking advertising content as an example, advertising tags can be divided into a multi-level structure. For example: First-level tags indicate categories, such as e-commerce, entertainment, education, and travel; second-level tags indicate product categories, such as fast-moving consumer goods, cosmetics, food, meals, and electronics. Keywords can also be added, such as new product launches, skin whitening, weight loss, discounts, Double Eleven discounts, and spending-based reduction activities. Through this multi-level tagging system, the platform can more accurately push advertising content to users who are interested, thereby improving click-through rates and conversion rates.

[0293] Of course, when uploading media content, content creators can use the tools provided by the platform to select their target audience, including multiple dimensions such as age, gender, geographical location (e.g., first-tier cities, second-tier cities), interests, and consumption behavior. The platform can leverage big data and algorithm models to help content creators accurately target their audience. Furthermore, they can use the platform's tools to set the timing (e.g., specific dates or time periods) and frequency (e.g., daily display limit) of their media content to ensure it reaches the target audience at the optimal time and frequency, and so on.

[0294] This application takes into account that user preferences are constantly changing. In order to increase the click-through rate of media content, when the target user browses the pushed media content through the information flow product, what is displayed to the target user is not the original pushed media content, but exclusive promotional content generated based on the pushed media content and targeted at the target user.

[0295] In this application embodiment, the generation timing of the exclusive promotional content can be varied and determined based on different dimensions and conditions to ensure accurate content delivery and optimal results. The exclusive promotional content can be generated before the user browses the original push media content or in real-time during the user's browsing process. Therefore, the generation timing of the exclusive promotional content includes, but is not limited to, the following, which are briefly explained below using advertising as an example:

[0296] Timing 1: User dimension, generated in real time when users browse information flow products.

[0297] For example, when a user is browsing information feed products such as social media and news apps, the system can preload the ad to be pushed to the user just before the user is about to view it, based on the user's real-time behavior and interests. The system can then generate and display newly generated personalized ad content based on this ad.

[0298] Timing 2: Content owner dimension, generated at the first moment of the delivery time period specified by the content owner, or before the delivery time period.

[0299] Advertisers can pre-set the time period for ad placement. For example, if they want to promote a product during the Double Eleven event (such as November 1st to November 11th), they can obtain the interests and preferences of the target audience in advance on November 1st or October 31st. Combined with the original ad, they can generate exclusive ad content for the target audience in advance. In this way, users can directly browse the newly generated exclusive ad content in real time.

[0300] Timing 3: User dimension, generated in advance based on user behavior habits.

[0301] For example, if we know from the behavior of the target user that they browse social media, news apps, and other information feed products between 7 and 8 pm every day, then we can generate personalized advertising content for that user a certain amount of time in advance (such as half a day in advance) before 7 pm that day. After that, the user can directly view the newly generated personalized advertising content in real time.

[0302] Timing 4: User dimension, generated after a user places an order for related products.

[0303] For example, after a user places an order for a certain food item, new personalized advertising content can be generated for that user from food-related ads in a designated material library using the method proposed in this article. After that, the user can directly view the newly generated personalized advertising content in real time, and so on.

[0304] It is important to emphasize that this application adheres to the principle of compliance and strictly complies with relevant data protection and privacy regulations in the process of collecting and using user data. This helps to protect the legitimate rights and interests of users and ensures that content owners and developers are free from legal disputes.

[0305] Furthermore, it should be noted that the above-mentioned timing for generating exclusive promotional content is merely a simple example, and this article does not impose any specific limitations on it. Any timing for generation is applicable to the embodiments of this application, and will not be elaborated upon here.

[0306] The following explanation will primarily focus on the first opportunity listed above:

[0307] In practical applications, when users browse information streams through content browsing interfaces, the client and server sides preload the media content to be displayed carried by the information stream.

[0308] For example, in this embodiment of the application, the push media content to be displayed for promotion will be preloaded. In this way, when the user needs to browse the push media content, the client side will read the basic information of the push media content (including: title, video tags, and related promotional information, etc.) as the basic input for recognition and understanding, and transmit the content to the server side. The server will analyze the information and combine it with the user's existing preference tags to regenerate personalized AI-enhanced promotional content to improve the overall content conversion efficiency.

[0309] Specifically, during the preloading process of the original push media content, the server can use AI technology (such as natural language processing, image recognition, video analysis, etc.) to understand the push media content and extract the basic information of the push media content, such as the title, video tags, and related products, services or brands, concepts, etc. This basic information will serve as the original reference for generating new exclusive promotional content later.

[0310] In addition, the server will further extract the core original key information of the pushed media content, including but not limited to selling points, discounts, benefits, and perks, such as the new product launch time, coupon information, and performance demonstration of the mobile phone products mentioned above. These core selling point information will play an important guiding role in the subsequent generation of exclusive key information (such as copywriting and slogan description).

[0311] The content preference information of the pushed users and how they are obtained can be found in the relevant descriptions on the client side mentioned above, and will not be repeated here.

[0312] S162: Generate exclusive promotional content for the target user based on the original key information and content preference information. The exclusive promotional content describes the different levels of exclusive key information that match the target user.

[0313] To ensure accurate delivery of media content, content creators typically define a specific audience in the early stages, ensuring that this audience aligns with the products, services, brands, or philosophies promoted in the content. However, in reality, there are situations where a content creator's services cover different user groups. In such cases, if the content can be tailored to the interests, consumption habits, and language preferences of different groups, it is easier to stimulate their purchasing intentions and help the content creator achieve the desired promotional results.

[0314] Therefore, in an optional implementation, in order to generate personalized promotional content for the target users and further optimize the promotional effect, S162 can be implemented according to the following process, including the following steps S1621 to S1624 (not shown in Figure 16):

[0315] S1621: Combine the content preference information of the pushed users to determine the subgroup to which the pushed users belong.

[0316] In this step, we can specifically analyze the content preference information of the pushed users to understand their interests, needs and preferences. Based on this information, we can construct detailed preference description features of the pushed users to gain a more intuitive understanding of their characteristics and needs.

[0317] In this application embodiment, the preference description features include, but are not limited to, the following aspects:

[0318] Basic attributes: such as age, gender, and geographical location; interests and hobbies: such as music, movies, technology, and sports; behavioral characteristics: such as browsing history, purchase records, and search keywords; social relationships: such as friends circle and following list.

[0319] It is important to emphasize that this application adheres to the principle of compliance and strictly complies with relevant data protection and privacy regulations in the process of collecting and using user data. This helps to protect the legitimate rights and interests of users and ensures that content owners and developers are free from legal disputes.

[0320] Furthermore, this application protects user privacy and data security by establishing a trust mechanism. Specifically, by providing transparent and secure content display services, this helps build a trust mechanism between users and content creators / developers, improves the reputation and image of the entire content ecosystem, and lays a solid foundation for the long-term development of the industry.

[0321] Furthermore, this application also considers that in practical applications, media content delivery is generally targeted at an audience, which refers to the specific group of people that the content creator hopes to attract and influence through the content. Based on this, the audience can be subdivided into different subgroups according to the individual preferences of each user within the target audience for the pushed media content. This allows for the identification of the subgroup to which the pushed user belongs. Using this method, personalized key information can be customized for each subgroup, achieving a more precise and customized effect.

[0322] In order to determine the subgroup to which the push user belongs, the audience is divided into different subgroups. Clustering algorithms (such as K-means, DBSCAN) or rule-based methods can be used to classify the push user into one or more subgroups.

[0323] When using the K-means algorithm, the clustering parameter K (i.e., the number of subgroups) can be determined using the elbow method. The specific process for calculating the sum of squared errors (SSE) within clusters for different K values ​​is as follows: For each K value, the data points of the pushed users are assigned to K clusters. The squared distance from each data point to the centroid of its cluster is calculated. Then, these squared distances for all data points are summed to obtain the SSE for that K value. For example, suppose there are n data points, and the distance from the i-th data point to the centroid of its cluster is d. i ,but By continuously changing the K value and calculating the corresponding SSE, a curve showing the relationship between the K value and SSE is plotted, and the K value with a slowing downward trend in SSE is selected as the optimal value.

[0324] For the DBSCAN algorithm, it is necessary to determine the neighborhood radius ε and the minimum number of points MinPts. The method for selecting a suitable ε value by calculating the distance distribution between data points is as follows: First, calculate the distance between any two data points, for example, using the Euclidean distance formula. Here, x and y are two data points, and m is the dimension of the data points. These distance values ​​are then sorted, and a distance distribution curve is plotted. The shape of the curve is observed, and a suitable ε value is chosen such that the slope of the distance distribution curve changes significantly around that ε value. MinPts can be adjusted based on the data density and the expected subpopulation size.

[0325] Then, the following steps can be performed:

[0326] S1622: Match each original key information with the content requirements corresponding to the subgroup.

[0327] Before this step, you can classify and organize the extracted original key information (such as advertising selling points) in advance to ensure that each original key information is clear, unambiguous, and can highlight the unique value of the product, service, brand, or concept.

[0328] Building upon this foundation, this step involves analyzing the content needs of the subgroups to which the pushed users belong (obtainable through user preferences, etc.). The original key information of the pushed media content (such as advertising selling points) is matched with the content needs of this subgroup to obtain matching results. These results represent the degree of matching between each piece of original key information and the content need. This degree of matching can be obtained by calculating the similarity between each piece of original key information and the content need features of the subgroup. Specifically, cosine similarity can be used, where the original key information and the subgroup content need features are represented as vectors, and the cosine value between the two vectors is calculated. The closer the cosine value is to 1, the higher the similarity. Alternatively, Euclidean distance can be used, calculating the Euclidean distance between the two vectors; the smaller the distance, the higher the similarity.

[0329] For example, if this subgroup is particularly interested in the battery life and appearance of technology products, then the matching degree of the original key information related to battery life and appearance will also be relatively high. Similarly, if this subgroup is price-sensitive and has high requirements for quality, then the matching degree of the original key information related to these two aspects will also be relatively high; and so on.

[0330] S1623: Based on the matching results and the content fragments in the pushed media content pointed to by each original key information, generate exclusive promotional content composed of each exclusive key information.

[0331] In this step, based on the matching results, the most relevant segments to the needs of the subgroup are extracted from the pushed media content. This is then used to generate personalized promotional content, which can highlight the key information that the subgroup cares about most, thereby improving the attractiveness and relevance of the content. For example, if a subgroup is price-sensitive, the product's cost-effectiveness can be highlighted; if a subgroup has high requirements for quality, the product's quality assurance can be emphasized.

[0332] The specific generation process can be as follows: First, extract the segments most relevant to the needs of the subgroup from the pushed media content. Then, sort these segments according to their matching degree, with segments having higher matching degrees having higher priority. The specific method for sorting these segments based on matching degree is as follows: Use the matching degree corresponding to each segment as the basis for sorting, and employ a sorting algorithm (such as quicksort, bubble sort, etc.) to sort the segments. Taking quicksort as an example, select a benchmark segment, place segments with a matching degree greater than the benchmark segment to its right, and segments with a matching degree less than the benchmark segment to its left. Then, recursively sort the segments on both sides until all segments are arranged in an ordered manner. In this way, segments with higher matching degrees have higher priority.

[0333] Then, for each content fragment corresponding to the original key information, creative copywriting is created based on the preferences and needs of the subgroup. Creative copywriting can employ template matching and keyword replacement methods. Some copywriting templates are predefined, and keywords in the templates are replaced based on the interests of the target users and the original key information to generate personalized copywriting. Simultaneously, natural language generation techniques, such as a Transformer-based model, are used to generate attractive copywriting based on the target users' interests and the original key information. The specific process is as follows: First, the target users' interests and the original key information are encoded into vector representations that the model can process. These are then input into a Transformer-based model, which consists of multiple encoder and decoder layers. The encoder layer performs feature extraction and representation learning on the input information, while the decoder layer gradually generates new copywriting content based on the encoder's output and previously generated parts of the copywriting. During the generation process, the model weights different information based on pre-trained parameters and an attention mechanism to ensure that the generated copywriting is relevant to the target users' interests and the original key information and is attractive. Finally, the generated copywriting is decoded and converted into natural language text.

[0334] In terms of visual elements, image enhancement algorithms, such as histogram equalization and sharpening, can be used for images and videos to improve image clarity and color saturation. For color selection, appropriate color schemes can be chosen based on the preferences of the target users and the content theme, such as using complementary or analogous colors to enhance visual effects. Layout design can employ grid layouts or the golden ratio principle to make the content more organized and aesthetically pleasing. Finally, the processed content fragments are combined into customized promotional content.

[0335] In this embodiment, generating exclusive promotional content specifically involves generating content fragments corresponding to each exclusive key piece of information. Specifically, creative copywriting can be developed based on the original key information and its corresponding content fragments, highlighting the selling points most relevant to the target users. Creative copywriting can employ template matching and keyword replacement methods. Predefined copywriting templates are used, and keywords in the templates are replaced based on the target users' interests and the original key information to generate personalized copy. Simultaneously, natural language generation technologies, such as Transformer-based models, can be used, taking the target users' interests and the original key information as input, to generate attractive copy. Visual elements, such as images, videos, colors, and layouts, are optimized to enhance the visual appeal of the content. For images and videos, image enhancement algorithms, such as histogram equalization and sharpening, can be used to improve image clarity and color saturation. In color selection, appropriate color schemes can be chosen based on the target users' preferences and the content theme, such as using complementary or analogous colors to enhance visual effects. Layout design can employ grid layouts or the golden ratio principle to make the content more organized and aesthetically pleasing. When refining and simplifying content, text summarization algorithms, such as the extractive TextRank algorithm, can be used to extract key sentences, ensuring concise and clear information delivery. Combining the target user's historical behavior and content preferences with relevant supplementary information or interactive elements can increase content appeal and interactivity. For example, recommending related product combinations based on a user's historical purchase history; adding interactive buttons, such as polls and raffles, can improve user engagement.

[0336] By using one or more of the processing methods listed above, it can be ensured that the generated personalized key information is more in line with the preferences of the target users and better meets their needs, thereby improving the click-through rate and conversion rate.

[0337] Taking advertising as an example, when writing creative copy, you should write attractive and targeted advertising copy for each subgroup, ensuring that the copy is concise and clear, can quickly grab the user's attention, and guide them to generate a purchase intention.

[0338] When designing visual elements, it's essential to incorporate user preference tags to create advertising visuals that align with those preferences, such as colors, patterns, and images. These visual elements should enhance the advertisement's appeal and recognizability. Analyzing user preferences during this process ensures the design better integrates with users' daily habits, using more relevant and familiar descriptions, or highlighting key selling points or coupon information to attract user clicks.

[0339] Specifically, before creating creative copy or designing personalized visual elements for different subgroups, a pre-configured correspondence between each subgroup and its unique key information extraction method can be established based on the different content needs of each subgroup. Establishing this correspondence helps ensure that the advertising content accurately meets the needs of different subgroups, thereby improving the attractiveness and conversion rate of the advertisement.

[0340] Specifically, this extraction method can be in the form of a set of keywords (statements), as shown in Figure 17. Based on this, for different subgroups, the original advertising information can be analyzed and extracted using a pre-configured extraction method based on the understanding of the large model and the comprehensive analysis of user preference tags, generating exclusive advertising content containing exclusive selling points.

[0341] Figure 17 illustrates a method for extracting preference tags and their specific key information for different subgroups in an embodiment of this application. Figure 17 lists six subgroups, which will be described below:

[0342] The first type: teenage users.

[0343] The preference tags of the teenage user subgroup are generally characterized by a high level of interest in entertainment news, games, and celebrity activities.

[0344] Considering their preference for novel and exciting content, and their high demands for gameplay and visual effects, the corresponding keywords for extraction include, but are not limited to: unique gameplay / cool visuals, emphasizing the product's unique gameplay and cool visuals to attract the attention of teenagers.

[0345] Let's abbreviate the preference tags for this subgroup as Q, the keyword extraction method as H, and the AI ​​word grouping capability as A. Taking creative copywriting as an example, by organically combining the above two elements and adding subjects, verbs, objects, and adjectives, we can generate exclusive selling point copy that makes the advertising content more marketable and attractive. This can be simply expressed by the following formula:

[0346] Exclusive selling point copy = Q1 + Q2 + Q3... + H1 + H2 + H3... + A1 + A2 + A3...

[0347] Among them, Q i (i = 1, 2, 3, ...) represents the i-th preference tag of the target audience. For example, for teenage users, Q1 might represent "entertainment news," Q2 might represent "games," etc.; H j (j = 1, 2, 3, ...) represents the j-th keyword in a specific keyword extraction method. For example, for teenage users, H1 might be "unique gameplay," and H2 might be "cool visuals," etc.; A k (k = 1, 2, 3, ...) represents the k-th word-grouping ability of AI. AI can use these preference tags and keywords to generate fluent and attractive copywriting, such as combining elements like "latest gaming phone", "ultimate gaming fun", "cool graphics", and "smooth combat" into complete copywriting.

[0348] The formula indicates that by integrating multiple preference tags of the target audience (Q1, Q2, Q3...), specific keyword extraction methods (H1, H2, H3...), and AI's various word grouping capabilities (A1, A2, A3...), more attractive and targeted advertising copy can be generated.

[0349] Example: The latest gaming phone brings you the ultimate gaming experience, with stunning graphics and exhilarating combat!

[0350] Similarly, the generation of posters and video clips with unique selling points can be achieved by combining multiple preference tags of the target audience (Q1, Q2, Q3...), specific keyword extraction methods (H1, H2, H3...), and other AI capabilities, such as graphic design and video production, to generate more attractive and targeted posters and video clips.

[0351] The second type: college student users.

[0352] The preference tags of this subgroup of college student users are generally characterized by their interest in the latest relevant news, such as exam information and study abroad guides.

[0353] Given their high demand for products and services that can improve learning efficiency and grades, the corresponding keywords for extraction include, but are not limited to: improving learning efficiency / grade improvement / actual experience, emphasizing the practicality of the product and its positive impact on learning.

[0354] Taking creative copywriting as an example, here's an example: Intelligent learning assistant, helping you prepare for exams efficiently and easily meet exam challenges!

[0355] The third type: working professionals.

[0356] The preferences of this subgroup of working professionals are generally reflected in their interest in tutorials and experience sharing on workplace skills, as well as content related to healthy eating and exercise.

[0357] Given their high demand for products and services that can improve work efficiency and teamwork, the corresponding keywords for extraction include, but are not limited to: business efficiency improvement / teamwork enhancement / efficiency improvement, emphasizing the product's professionalism and its positive impact on workplace life.

[0358] Taking creative copywriting as an example, here's an example: Efficient office software improves team collaboration and helps you easily cope with workplace challenges!

[0359] The fourth group: middle-aged and elderly people.

[0360] The preferences of the middle-aged and elderly subgroup are generally characterized by a fondness for light and enjoyable cultural and entertainment content, such as opera, dance, and health preservation methods.

[0361] Considering their high demand for products that can improve their quality of life, the corresponding keywords for extraction include, but are not limited to: intelligent massage / muscle relaxation / sleep improvement / actual experience, emphasizing the comfort of the product and its positive impact on health.

[0362] Taking creative copywriting as an example, here's an example: Smart massage chairs relieve muscle fatigue, improve sleep quality, and allow you to enjoy a comfortable life!

[0363] The fifth type: male users.

[0364] The preference tags of male users generally show that they have a high interest in the latest technology products and information, sports events, financial news, and electronic technology.

[0365] Given their high demand for high-performance and advanced technology products, the corresponding keywords for extraction include, but are not limited to: latest configuration / innovative experience / price advantage / advanced technology / superior performance, emphasizing the advanced technology and superior performance of the products.

[0366] Taking creative copywriting as an example, here's an example: Brand C, Model D mobile phone, leading technology, the future is in your hands, let's explore cutting-edge technology together!

[0367] The sixth type: female users.

[0368] The preference tags of female users generally show that they have a high interest in celebrity gossip, fashion trends, beauty tutorials, home organization, etc.

[0369] Considering their high demands for aesthetics and user experience, the corresponding keywords for extraction include, but are not limited to: makeup effect / makeup application experience / color and elegant visuals / exquisite packaging / excellent usage effect, emphasizing the product's aesthetics and usage effectiveness.

[0370] Taking creative copywriting as an example, here's an example: New lipstick, rich in color, smooth application, making you radiant!

[0371] It should be noted that the subgroups listed above are just simple examples; other user groups are organized in a similar way and will not be elaborated on here.

[0372] In the embodiments of this application, the user preference-based advertising push model listed above can achieve precise targeting and personalized delivery of advertising content, effectively improving the accuracy and attractiveness of advertisements. By conducting in-depth analysis of the preferences of different subgroups, pre-configuring the correspondence between different subgroups and key information extraction methods, and combining this with the original information provided by the advertiser, highly targeted advertising content can be generated. This model not only enhances the attractiveness and conversion rate of advertisements but also improves user satisfaction, helping advertisers achieve the goal of selling advertisements. In the future, with the continuous development of AI technology, this model will play an increasingly important role in advertising and marketing.

[0373] It should be noted that the processing methods listed above are just simple examples. In addition, other methods for generating exclusive promotional content based on various original key information and the content preference information of the pushed users are also applicable to the embodiments of this application, and will not be described in detail here.

[0374] In the above implementation, users are divided into different subgroups and matched according to the content needs of the subgroups. This segmentation and matching mechanism can better capture the specific needs of different user groups, generate more targeted and exclusive promotional content, and improve the accuracy of the content.

[0375] In addition, this application also determines the level of exclusive key information based on the matching results. The level of each exclusive key information reflects the user's attention index to the information, ensuring that exclusive key information of different levels is displayed first in different situations, thereby further improving the user experience.

[0376] This approach not only improves content relevance and user satisfaction, but also enhances user engagement and interaction, thereby increasing platform activity and user loyalty.

[0377] Specifically, through the aforementioned sub-steps S1621 to S1623, personalized promotional content tailored to the targeted users can be generated. Furthermore, the following sub-step S1624 can also be executed during this process:

[0378] S1624: Determine the level of each specific key information based on the matching results.

[0379] The level of each exclusive key information represents the attention index of the pushed user to the corresponding exclusive key information. The specific definition is as described in the above embodiment, and will not be repeated here.

[0380] In this embodiment of the application, the matching result can represent the degree of matching between each original key information and the content requirement. Secondly, there is a certain correspondence between the original key information and the exclusive key information. Based on this, the degree of matching between each exclusive key information and the content requirement can actually be analyzed. The higher the degree of matching, the higher the attention index of the pushed user to the exclusive key information, and correspondingly, the higher the level of the exclusive key information.

[0381] In the above implementation, by segmenting users into different subgroups and matching them according to the content needs of each subgroup, the resulting personalized key information is more accurate. This segmentation and matching mechanism can better capture the specific needs of different user groups, providing more targeted content and improving the accuracy of the generated personalized promotional content. Furthermore, by determining the level of personalized key information based on the matching results, the level of each piece of personalized key information can highly accurately reflect the user's level of attention to that information, ensuring that different levels of personalized key information are prioritized for display in different situations, further enhancing the user experience.

[0382] By following the steps above, you can generate personalized promotional content for the targeted users, thereby increasing click-through rates and conversion rates and optimizing promotional effectiveness.

[0383] In this embodiment of the application, in order to improve the display efficiency of pushed media content, the server can use AI recognition to understand the main segments of the pushed media content during preloading, thereby extracting the original key information. Then, through Artificial Intelligence Generated Content (AIGC), combined with the content preference information of the pushed user, dynamic exclusive promotional content (including but not limited to products, titles, benefits, etc.) is generated. The exclusive key information in the exclusive promotional content can specifically include videos, copywriting, poster images, etc.

[0384] Furthermore, the server will also break down the advertising selling points into different levels based on the attractiveness of the specific key information. For example, advertising selling points can be divided into S-level selling points, and A / B / C, etc. In this way, this application can display different advertising selling points according to different user experience states.

[0385] Specifically, when generating personalized promotional content based on an AI model, another optional implementation of step S162 is as follows:

[0386] The content preference information and various original key information are directed to content segments in the pushed media content and input into the trained generative model; based on the generative model, personalized promotional content composed of various exclusive key information is generated.

[0387] The specific working process of the generative model can be as follows: After receiving content preference information and original content fragments corresponding to original key information as input, the generative model first encodes the input data, converting it into feature vectors that the model can understand. Then, the neural network layer inside the model performs non-linear transformations and feature extraction on these feature vectors, learning the mapping relationship between the input data and the specific key information. In this process, the model performs calculations and inferences based on pre-trained parameters, continuously adjusting the output results to better match expectations. Finally, the model decodes the processed feature vectors into various specific key information and the content fragments they point to; these content fragments constitute the specific promotional content.

[0388] The trained generative model takes as input content preference information and original content fragments corresponding to the original key information. Based on this model, it can generate various unique key information and the content fragments they point to, which together form the exclusive promotional content. Correspondingly, this generative model can be trained on a large number of labeled sample key information and sample content preference information related to a large number of sample users.

[0389] Furthermore, the model's input can also be content preference information and pushed media content. Correspondingly, this generative model can be trained based on a large amount of labeled sample media content related to sample users and sample content preference information. Sample media content, one of the input data used to train the generative model, is labeled media content selected from a large amount of sample user-related data. Based on this content and sample content preference information, the generative model is trained, enabling it to learn the ability to generate specific key information based on user preferences. Sample key information is the raw key information extracted from the sample media content. During model training, it is used as input along with sample content preference information, generating specific key information and the content fragments they point to as output, allowing the model to learn how to generate specific key information based on user preferences.

[0390] In this embodiment of the application, the key information of the sample refers to the original key information extracted from the sample media content, while the sample content preference information is for the sample users and represents the sample users (such as historical users of a certain platform / platforms).

[0391] The extracted key information from the samples can be cleaned, deduplicated, and formatted. Then, key features can be extracted from this processed information. For example, Natural Language Processing (NLP) techniques can be used to extract key features from advertising selling points, such as keywords, phrases, and sentiment. These features can then be used to annotate the key information of the samples. The content preference information of the sample users (such as preference tags) is categorized and standardized. For instance, the preference tags are converted into numerical or vector representations to facilitate processing by machine learning models. Similarly, the inputs used in model applications will also undergo corresponding processing.

[0392] Based on the above, the original content fragments corresponding to the sample users' content preference information and the labeled sample key information can be used as inputs, and the outputs can be each exclusive key information and the content fragments it points to. The model can then be trained to learn how to generate corresponding exclusive key information based on the user's preferences.

[0393] During model training, reinforcement learning can be used to optimize the generation strategy of the generative model, making it more attentive to user preferences and thus improving the quality of the generated personalized key information.

[0394] In simple terms, first, define the environment (the interaction environment between the user and the platform) and the agent (the AI ​​model that generates unique key information). Next, define the state space, which is the contextual information when the agent generates unique key information, including user preferences, historical behavioral data, etc. Then, define the action space, which is the specific operations the agent performs to generate unique key information, such as selecting key information or adjusting copy. Finally, define the reward function, which provides positive or negative rewards based on user feedback and behavioral data (such as click-through rate and conversion rate). The state space is the set of contextual information in which the agent generates unique key information during reinforcement learning's optimization of the generative model. It includes user preferences, historical behavioral data, etc. The agent makes decisions based on the information in the state space, choosing appropriate actions to generate unique key information. The action space is the set of specific operations the agent can take to generate unique key information during the reinforcement learning optimization of the generative model, such as selecting key information or adjusting copy. The agent selects actions from the action space based on the state space to generate unique key information and obtain rewards.

[0395] The reward function can be defined as R = w1 × C + w2 × T + w3 × I, where R represents the reward value, used to measure the overall effect of the personalized key information generated by the model; C represents the click-through rate (CTR), which is the ratio of the number of times users click on the personalized key information to the number of times it is displayed; T represents the conversion rate, which is the proportion of users who actually make a purchase after clicking on the personalized key information; and I represents the number of user interactions, such as the total number of likes, comments, and shares. w1, w2, and w3 are the weights of CTR, conversion rate, and interaction count, respectively, and w1 + w2 + w3 = 1. The weights can be determined using a multi-round A / B testing method. In each round of testing, a certain proportion of traffic is allocated to different weight combinations, and the reward value R under each weight combination is recorded. After multiple rounds of testing, the weight combination that maximizes the reward value R is selected as the final weight value. If the business objective focuses more on CTR, the range of w1 should be appropriately increased when initially setting the weight combination; if the focus is more on conversion rate, the range of w2 should be increased. Business objective is a factor to be considered when determining the weights of various factors in the reward function (such as the weights of click-through rate, conversion rate, and number of interactions). It reflects the expected results to be achieved. Different business objectives will affect the initial setting of the weights in order to achieve different business needs.

[0396] During model training, the agent progressively optimizes its behavior through exploration and exploitation strategies. Based on the collected rewards, it updates the strategy using reinforcement learning algorithms (such as Q-learning and Deep Q-Network (DQN)) to better cater to user preferences. The exploration and exploitation strategies are as follows: exploration involves the agent randomly selecting an action with a certain probability to discover new behaviors that may bring high rewards; exploitation involves the agent selecting the action it currently believes will bring the greatest reward based on its learned experience. In the early stages of training, the probability of exploration is higher; as training progresses, the probability of exploration gradually decreases, while the probability of exploitation gradually increases.

[0397] The Q-learning algorithm maintains a Q-table to record the Q-value (i.e., the expected cumulative reward) for each state and action. In each interaction, the agent selects an action based on the current state, executes the action to obtain a reward and the next state, and then updates the algorithm according to the Q-learning formula: Q(s,a) = Q(s,a) + α[r + γmax]. a′ The Q-table is updated by [Q(s′,a′)-Q(s,a)], where s represents the current state, a represents the action performed, r represents the reward received, s′ represents the next state, α is the learning rate, and γ is the discount factor.

[0398] Deep Q-Networks (DQNs) are an extension of Q-learning that uses neural networks to approximate the Q-value function. During training, DQN takes the state as input and outputs the Q-value for each action through the neural network. The agent selects actions based on these Q-values, receives a reward and the next state after executing the action, and then stores these experiences in an experience replay pool. A batch of experiences is randomly drawn from the experience replay pool for training. A mean squared error loss function is used. To update the parameters θ of the neural network, where N is the batch size and θ - These are the parameters of the target network, which periodically copies and updates its parameters from the main network. By continuously updating the neural network's parameters, DQN can learn better strategies to focus on user preferences. The Mean Squared Error Loss Function (MSE) is a loss function used during the training of Deep Q-Networks (DQNs) to update the neural network parameters. It measures the model's prediction error by calculating the average of the squared errors between the predicted and target values, thereby enabling the model to continuously optimize and learn better strategies to focus on user preferences.

[0399] Through this reinforcement learning method, the generative model can continuously optimize its generation strategy to generate more personalized key information that better matches user preferences, thereby improving the click-through rate, conversion rate, and return on investment of media content and achieving better promotional results.

[0400] Subsequently, the trained generative model can be used to generate new, personalized key information based on content preference information and the content segments pointed to by the original key information in the pushed media content. Furthermore, the trained model can be used to evaluate the quality of the generated personalized key information, including semantic rationality, innovativeness, and attractiveness. Based on the evaluation results, the model can be further optimized, or the generated results can be improved, and so on.

[0401] In this embodiment, the generative model is a deep learning model, mainly based on NLP. NLP technology is the foundation for understanding and generating text data, and is crucial for extracting key information from advertising selling points and generating new selling point descriptions.

[0402] Specifically, the generative model can be: a sequence generation model based on a transformer, a generative model, a variational autoencoder (VAE), etc.

[0403] Among them, sequence generation models based on Transformer, such as the third-generation generative pre-trained Transformer (GPT-3), have powerful text generation capabilities. They can generate coherent and semantically correct text based on context. They perform well in generating advertising selling points and can generate high-quality copy. Based on the input content preferences and original key information, they can generate coherent and semantically correct exclusive promotional content, thereby increasing user interest and engagement.

[0404] Generative models, such as Generative Adversarial Networks (GANs), can generate realistic new data through adversarial training. This is suitable for generating innovative advertising selling points, creating more creative and attractive personalized promotional content, and improving user click-through rates and conversion rates.

[0405] VAE is another generative model that generates new data by learning the latent distribution of the data. In ad selling point generation, VAE can be used to generate diverse content that matches user preferences.

[0406] It should be noted that the generative models listed above are just simple examples. Other generative models are also applicable to the embodiments of this application, such as sequence-to-sequence (Seq2Seq) models, recurrent neural networks (RNN) and their variants, long short-term memory networks (LSTM), gated recurrent units (GRU), etc., which will not be described in detail here.

[0407] In practical applications, the choice of generative model depends on the specific task requirements and scenario. For example, if high-quality, coherent text needs to be generated, a transformer-based model (such as GPT-3 or T5) can be chosen. If creative and diverse content needs to be generated, a generative model (such as GAN or VAE) can be selected. These models can be used to more effectively extract and generate advertising selling points, improve promotional effectiveness, and so on. This article does not make specific limitations on these aspects.

[0408] In short, the entire process from identifying the original push media content to generating personalized promotional content can be completed by AI models. By leveraging AI-extracted and AIGC-generated key information, combined with popular elements and creativity, media content that is more in line with current aesthetics and context can be generated. This helps to enhance the attractiveness and reach of the media content used for promotion, making such media content no longer a burden for users, but a valuable source of information.

[0409] Taking push media content as an example of advertising, the above implementation method combines AI recognition and AIGC technologies, demonstrating the advertising industry's strength in technological innovation. This will set a new benchmark for the industry and drive the entire advertising sector towards a more intelligent and personalized direction. Furthermore, this solution will inspire more advertisers and developers to explore similar technological applications, thereby expanding the application scenarios and boundaries of advertising. For example, similar personalized advertising display methods will have great potential in fields such as education, e-commerce, and finance.

[0410] To further optimize the performance of the generative model and improve the accuracy and appeal of the generated personalized key information, this application proposes a data-driven and feedback mechanism to continuously optimize the model. The data-driven and feedback mechanism is a method for optimizing the performance of the generative model. By collecting feedback data from various users on the personalized key information generated based on the generative model, such as click-through rate, conversion rate, viewing time, number of interactions, and specific user opinions and suggestions, this feedback data is analyzed to evaluate the performance of the generative model and identify its strengths and weaknesses. Then, based on the evaluation results, the parameters of the generative model are adjusted, new algorithms or technologies are introduced, the model output is dynamically adjusted, and the optimization effect can be verified through A / B testing to select the optimal solution, thereby improving the accuracy and appeal of the generated personalized key information.

[0411] An optional implementation method is as follows:

[0412] Collect feedback data from each user on the unique key information generated based on the generative model; iterate and optimize the generative model based on the collected feedback data.

[0413] Feedback data refers to users' reactions and evaluations of the generated exclusive key information (specifically, exclusive promotional content), including but not limited to: click-through rate, conversion rate, viewing time, number of interactions (such as likes, comments, and shares), as well as users' specific opinions and suggestions.

[0414] Specifically, user feedback data on the exclusive key information generated based on the generative model can be collected through various channels (such as online surveys, user comments, social media interactions, etc.).

[0415] Once the feedback data has been collected, it can be analyzed. This includes statistically analyzing positive and negative user feedback, analyzing trends in key metrics such as click-through rate and conversion rate, interpreting specific user feedback, and understanding users' views and suggestions regarding personalized key information.

[0416] Based on the above analysis results, the performance of the generative model can be evaluated, its advantages and disadvantages can be identified, and then the parameters of the generative model can be adjusted and the model optimized based on the evaluation results.

[0417] For example, based on an advertisement for a smartwatch, the model performs well in generating descriptions of the smartwatch's health monitoring functions, but falls short in generating descriptions of battery life. In this case, the model's weight for health monitoring functions can be increased, and the logic for generating battery life descriptions can be improved.

[0418] Furthermore, new algorithms or technologies can be introduced to enhance the model's generative capabilities and adaptability. For example, more advanced deep learning models (such as BERT and T5) can be used to generate higher-quality, proprietary key information.

[0419] Building upon the above, the output of the generative model can be dynamically adjusted based on real-time feedback data to ensure that the generated key information consistently meets user needs. Furthermore, A / B testing of different generative models and strategies can be conducted to verify optimization effects and select the optimal solution.

[0420] In the above implementation, based on past ad click data, the extraction methods for highlighting key selling points and the image display effects of generative AI are continuously optimized. Leveraging the computational advantages of large-scale models, high-click-rate materials can be continuously iterated and optimized to consistently ensure high click-through rates and conversion rates, thereby increasing content creators' trust in the platform and the likelihood of long-term cooperation. For users, efficient, clear, and accurate information display helps them make decisions more quickly, enhancing their content experience.

[0421] In summary, by continuously collecting user feedback and monitoring model performance, problems with the model can be identified and resolved in a timely manner, thereby continuously improving the accuracy and appeal of the generated personalized key information.

[0422] S163: Send exclusive promotional content to the target user and respond to the target user's content switching operation on the content browsing interface. Display the exclusive promotional content on the content browsing interface and jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback; wherein, the content browsing interface is used to display the pushed information stream to the target user, and the information stream carries the pushed media content; the user interest matching level is determined according to the target user's interest in the information stream.

[0423] In this embodiment of the application, after generating the exclusive promotional content corresponding to the pushed media content based on the above method, the server will send it back to the client, which will then display it to the pushed user.

[0424] In order to achieve more precise content delivery and further expand the promotion effect of pushed media content, in one optional implementation, the pushed media content can also be pushed to relevant users on other platforms based on the content preference information of the pushed user and various exclusive key information matched with the pushed user.

[0425] In this embodiment, based on the content preference information of the pushed user, a preference description feature can be constructed. For the feature dimensions such as basic attributes (e.g., age, gender, geographic location), interests (e.g., music, movies, technology, sports), behavioral characteristics (e.g., browsing history, purchase records, search keywords), and social relationships (e.g., friend circle, following list), the following quantification methods are adopted: For basic attributes, age can be assigned values ​​according to age groups; gender can be encoded using binary codes; and geographic location can be assigned values ​​based on regional importance or activity. Interests can be assigned values ​​after normalization based on the user's browsing time and interaction frequency for different types of content. Behavioral characteristics can be assigned values ​​based on weighted calculations of the importance and frequency of the behavior. Social relationships can be quantified based on the number of friends and interaction intensity. These quantified features are integrated into a multi-dimensional vector to form the preference description feature of the pushed user.

[0426] Based on this, by analyzing the preferences of the pushed users, relevant users on other platforms can be identified. These relevant users generally share similar interests, behavioral characteristics, and basic attributes with the pushed users.

[0427] In other words, "relevant users" here can be users from other platforms whose preferences and descriptive characteristics match their preferences.

[0428] Specifically, there are many ways to identify users related to other platforms in this application embodiment. A few are listed below:

[0429] Identification Method 1: Similarity Calculation.

[0430] This approach uses machine learning algorithms (such as K-nearest neighbors and collaborative filtering) to calculate the similarity between the preferences of users on other platforms and the preferences of the user receiving the push notifications. Based on this, a similarity threshold can be set to filter out users with similarity scores higher than the threshold as relevant users. This identification method can comprehensively consider multiple features and has high accuracy.

[0431] Taking the K-nearest neighbors algorithm as an example, the specific calculation process is as follows: The preference description features of the user receiving the push notification and other platform users are represented as vectors U and V, respectively. The similarity S between the two is calculated using the Euclidean distance formula: Where n is the dimension of the feature vector, U i and V iLet S and S represent the values ​​of the user receiving the push notification and the user on the other platform in the i-th feature dimension, respectively. The smaller the similarity value S, the more similar the preferences of the two users are. A similarity threshold t is set. When S ≤ t, the user on the other platform is considered to have a high similarity to the user receiving the push notification and can be regarded as a relevant user.

[0432] Recognition Method 2: Tag Matching.

[0433] First, tags are extracted from the content preference information of the target users, such as "tech product enthusiast," "sports enthusiast," and "music enthusiast." Then, the tags of the target users are matched with tags of users on other platforms to identify users with the same or similar tags. This identification method is simple to implement and has low computational cost.

[0434] For example, if the user receiving the push notification is interested in technology products, then users on other platforms who are also interested in technology products can be considered relevant users.

[0435] Both of the above methods are based on the content preference information of the recipient. In addition to the above two methods, it is also possible to identify the recipient based on their behavior, as shown in the following identification method three:

[0436] Identification Method 3: Behavioral Analysis.

[0437] Specifically, this involves analyzing the behavioral patterns of the targeted users on the current platform, such as browsing time and interaction frequency, to filter out users on other platforms who exhibit similar behaviors. This method can capture dynamic behavioral characteristics of users and is suitable for scenarios with abundant behavioral data.

[0438] It should be noted that the above identification methods can be used in combination, and the above-listed implementation methods are just simple examples. In addition, other identification methods are also applicable to the embodiments of this application, and will not be described in detail here.

[0439] Based on the above, the media content can then be pushed to the identified relevant users. Here, "other platforms" refers to other platforms related to content display besides the client the recipient is logged into (i.e., the current platform), such as social media, search engines, and e-commerce platforms.

[0440] Based on the above implementation methods, by leveraging user preferences and customized key information (such as advertising selling points), push media content (such as advertising content) can be precisely delivered to the target audience, achieving accurate content delivery and improving click-through rates and conversion rates. Furthermore, under this delivery method, the personalized promotional content seen by users is highly matched to their interests, reducing interference from irrelevant content and enhancing the user experience.

[0441] In addition, media content can be pushed to relevant users through various channels such as social media, search engines, and e-commerce platforms, which expands the coverage and influence of the content, increases the exposure of the content, and improves users' awareness.

[0442] In addition to the methods described above, to further optimize the promotional effect of pushed media content, in one optional implementation, the actual delivery effect of the pushed media content can also be detected; based on the actual delivery effect, at least one of the following can be adjusted:

[0443] The original key information corresponding to the pushed media content, and the delivery strategy for the pushed media content.

[0444] Taking push media content as advertising content as an example, this application supports continuous monitoring of the actual performance of advertising, specifically monitoring key performance indicators, including click-through rate, conversion rate, user engagement, return on investment (ROI), and user feedback. Based on the actual performance, the advertising selling points (i.e., original key information) and advertising strategies can be adjusted in a timely manner to optimize advertising effectiveness.

[0445] When adjusting advertising selling points based on actual campaign performance, the following methods are included, but are not limited to:

[0446] Method 1 for adjusting selling points: Optimize advertising creative.

[0447] The optimization of advertising creative includes, but is not limited to, copywriting adjustments, visual design, and content updates.

[0448] Specifically, you can optimize ad copy based on user feedback and click-through rate data to make it more attractive and persuasive; in addition, improve the visual elements of the ads, such as image and video quality and color scheme, to enhance the user's visual experience; furthermore, you can regularly update ad content to keep it fresh and attract users' continued attention.

[0449] Method 2 for adjusting selling points: Highlight the core selling points.

[0450] Specifically, you can highlight the product's unique features and advantages, such as technological innovation, user experience, and cost-effectiveness; increase user reviews and use cases to enhance the credibility and appeal of the advertisement.

[0451] In addition to adjusting the advertising selling points, the ad format and style can also be modified. By testing different types of ad formats, such as video ads, image and text ads, and banner ads, the most attractive format for users can be found. The length of the ads can also be adjusted, such as shortening or lengthening the video ad, to increase the percentage of users who watch the entire ad. Furthermore, based on user feedback, interactive elements can be added to the ads, such as surveys and prize draws, to increase user engagement and brand loyalty.

[0452] It should be noted that the aforementioned adjustments to the advertising format, structure, and interactive elements can be applied to the original push media content or to exclusive promotional content; this article does not impose any specific limitations on this.

[0453] When adjusting the campaign strategy based on actual campaign performance, the following methods may be used, including but not limited to:

[0454] Strategy adjustment method 1: Optimize distribution channels.

[0455] In addition to the multi-channel advertising methods listed above, you can also adjust the advertising ratio based on the performance of each channel, prioritizing the advertising on the best-performing channels.

[0456] Strategy Adjustment Method Two: Precisely Target the Audience.

[0457] Specifically, based on user preferences, the audience can be segmented into multiple subgroups, and personalized targeting strategies can be developed for each subgroup. In addition, the audience can be dynamically adjusted based on actual targeting results, eliminating inefficient audiences and increasing efficient audiences.

[0458] Strategy adjustment method three: Adjust the delivery time.

[0459] Specifically, you can analyze user activity times and choose to run ads during the peak user activity periods to improve click-through rates and conversion rates. In addition, you can adjust the timing and frequency of ad placements based on factors such as seasons and holidays.

[0460] Strategy Adjustment Method Four: Optimize Budget.

[0461] Specifically, allocate the budget reasonably based on the actual campaign performance to avoid ineffective campaigns; in addition, optimize bidding strategies to improve the competitiveness of advertisements and reduce the cost per click.

[0462] It should be noted that the above-listed adjustment methods are just simple examples. Other adjustment methods are also applicable to the embodiments of this application, and will not be repeated here.

[0463] In the above implementation, by continuously monitoring the effectiveness of content delivery, it is possible to understand the content's performance in real time and quickly identify its strengths and weaknesses. Based on this feedback, the original key information of the content can be adjusted in a timely manner to improve user interest and engagement. Simultaneously, adjusting the delivery strategy can enhance the content's reach and effectiveness. Through these adjustments, the promotional effect of the content can be significantly improved, increasing user click-through rates and conversion rates, ultimately achieving better marketing goals.

[0464] It should be noted that the above is an introduction to the content display method on the server side in this application. In short, the technical solution of this application covers product interaction, product form, and server technology implementation. The overall technical logic can be summarized into the following three sides: user side, product side (i.e., client side), and backend side (i.e., server side). Taking the generation of exclusive advertising content when preloading advertisements as an example, one optional interaction logic for these three sides is as follows:

[0465] Referring to Figure 18, which is a schematic diagram of an optional multi-terminal interaction implementation in an embodiment of this application, the specific implementation method is as follows:

[0466] First, on the user side, users browse information streams. For example, users can view various information on a browser, and the information streams they browse may contain advertisements to be displayed.

[0467] Building upon the user's perspective, the product side is responsible for preloading advertisements. Specifically, the product side preloads ad segments suitable for the user's consumption and sends them back to the backend. In addition, the product side can also send user behavior data (referring to data permitted by the user) back to the backend (not shown in Figure 18).

[0468] In addition, the product side will also identify and detect the user's screen swiping speed in real time and make several judgments as shown in Figure 18. Based on different judgment results, when the user views and previews the advertisement, the ad player will display the selling points user interface (UI), progress bar positioning point style and other different representations based on the materials issued by the backend side. For details, please refer to the above embodiments. Repeated parts will not be repeated.

[0469] Building upon the product side, the backend receives the aforementioned data and extracts the core selling points (i.e., original key information) of the advertisement using AI preprocessing technology. Specifically, this extraction process involves parsing the advertisement's original information, including intelligent analysis and understanding of the title, copy, and visuals, and identifying the original information, including but not limited to core selling points, image recognition, and promotional information. The backend also analyzes user preferences and personal tags to customize advertising content more suitable for users. Subsequently, combining the advertisement's selling points and user preferences, the backend extracts content that aligns with user preferences. Based on this, different levels of selling point slices (i.e., content fragments pointed to by exclusive key information) are generated according to user preferences, such as S, A, B, and C levels. The S level represents the selling point that is most attractive to the user and requires comprehensive identification based on their preferences, as detailed in the above embodiment, which will not be repeated here. Finally, the generated advertising selling point slices (referring to exclusive advertising content containing different levels of exclusive selling points and corresponding content fragments) and other materials are sent back to the product-side client for display. In the context of advertising as the medium for pushing content, a selling point segment refers to a content fragment that categorizes the selling points of an advertisement into different levels, such as S, A, B, and C, based on the preferences of the target user. The S level represents the selling point that is most attractive to that user. These selling point segments contain exclusive selling points at different levels and corresponding content fragments, forming customized advertising content. This content is generated on the backend and then sent back to the product client for display.

[0470] In addition, it should be noted that during the process of generating exclusive advertising content on the backend side, interaction with the product side is also possible, so that the product side can display the progress to the user, such as prompting the user "Exclusive selling points are being generated..." If the user wants to view and preview the advertisement during the generation process, the original advertisement is displayed to the user. If the user wants to view and preview the advertisement after it has been generated, the newly generated exclusive advertisement is displayed to the user.

[0471] Returning to the user side, users can view and preview advertisements, and view the advertisement details page. For specific implementation methods, please refer to the above embodiments, and repeated details will not be described again.

[0472] In addition, the product side of this application also supports users to freely select or view multiple advertising selling points in rotation until the end of the advertising browsing process. For specific implementation methods, please refer to the above embodiments, and repeated parts will not be described again.

[0473] Throughout the entire process, AI technology was fully utilized, not only to help understand and analyze advertising content, but also to take into full account users' personal preferences and behavioral data, aiming to provide a more personalized advertising experience.

[0474] In summary, by combining these processes and technologies, this application proposes an interactive advertising display solution that matches user interaction behavior. This solution constructs an efficient and intelligent technical process for generating advertising selling points, enabling personalized information recommendations for different users. Specifically, by intelligently analyzing users' screen swiping speed, it dynamically displays diverse advertising selling points refined by artificial intelligence technology and generated by AIGC. This not only optimizes the matching degree between advertisements and users, improving advertising conversion rates and user satisfaction, but also provides advertisers with more accurate market insights by collecting and analyzing user behavioral data, thereby supporting more effective marketing strategies. Furthermore, this solution demonstrates the innovative application of artificial intelligence and AIGC technology in the advertising field, foreshadowing new trends in industry development and driving the advancement of related technologies.

[0475] Based on the same inventive concept, this application also provides a content display device. As shown in FIG19, which is a structural schematic diagram of the content display device 1900, it may include:

[0476] Display unit 1901 is used to display a content browsing interface; the content browsing interface is used to display the pushed information stream to the pushed user.

[0477] The response unit 1902 is used to respond to the content switching operation of the content browsing interface, display exclusive promotional content generated based on the push media content carried by the information stream in the content browsing interface, and jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback.

[0478] The exclusive promotional content is used to describe various levels of exclusive key information that match the target user; each exclusive key information is generated based on the original key information in the push media content and the target user's content preference information; the user interest matching level is determined based on the target user's interest in the information stream.

[0479] Optionally, the level of each of the exclusive key information is associated with the content preference information of the pushed user, and the level represents the user's attention index to the corresponding exclusive key information.

[0480] Optionally, the response unit 1902 is specifically used for:

[0481] If the interest level falls within a preset interest level range, the user will be redirected to the exclusive promotional content, where the highest-level exclusive key information will be used to play the content segment.

[0482] If the interest level does not fall within the preset interest level range, the player will randomly jump to any content segment pointed to by any exclusive key information other than the highest level, or randomly jump to any content segment pointed to by any exclusive key information among the various exclusive key information segments.

[0483] Optionally, the response unit 1902 is further configured to determine the degree of interest by at least one of the following methods:

[0484] The interest level is determined based on the content switching operation of the pushed user on the content browsing interface, wherein the interest level is negatively correlated with the triggering state of the content switching operation;

[0485] The interest level is determined based on the duration of the user's gaze on the content browsing interface, wherein the interest level is positively correlated with the duration of gaze.

[0486] Optionally, the response unit 1902 is further configured to:

[0487] After the content segment pointed to by the exclusive key information of the user interest matching level has finished playing, the content segment pointed to by the alternative exclusive key information will continue to play.

[0488] Optionally, the response unit 1902 is specifically used for:

[0489] The program will randomly redirect to the content segment indicated by the selected exclusive key information for playback.

[0490] After all the content segments pointed to by each specific key information have been played, the content segments pointed to by each specific key information will be played in turn.

[0491] Optionally, the response unit 1902 is further configured to:

[0492] The playback progress bar of the exclusive promotional content highlights the time points associated with each exclusive key piece of information;

[0493] In response to a first-type trigger operation for any alternative time point, the user is redirected to the corresponding content segment in the exclusive promotional content for playback. The alternative time points are time points associated with alternative exclusive key information other than the exclusive key information of the user interest matching level.

[0494] Optionally, the response unit 1902 is further configured to:

[0495] The playback progress bar of the exclusive promotional content highlights the time points associated with each exclusive key piece of information;

[0496] In response to a second type of triggering operation at any point in time, the pushed media content is displayed, and the user is redirected to the pushed media content to play a content segment associated with the corresponding exclusive key information.

[0497] Based on the same inventive concept, this application also provides another content display device. As shown in Figure 20, which is a structural schematic diagram of the content display device 2000, it may include:

[0498] Analysis unit 2001 is used to extract various original key information from the pushed media content and obtain the content preference information of the pushed user.

[0499] The generation unit 2002 is used to generate exclusive promotional content for the pushed user based on the original key information and the content preference information, wherein the exclusive promotional content is used to describe the exclusive key information at different levels that match the pushed user;

[0500] Feedback unit 2003 is used to send the exclusive promotional content to the target user, and in response to the target user's content switching operation on the content browsing interface, display the exclusive promotional content in the content browsing interface, and jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback; wherein, the content browsing interface is used to display the pushed information stream to the target user, the information stream carrying the pushed media content; the user interest matching level is determined according to the target user's interest in the information stream.

[0501] Optionally, the generation unit 2002 is specifically used for:

[0502] Based on the content preference information of the pushed users, determine the subgroup to which the pushed users belong;

[0503] Match the original key information with the content requirements corresponding to the subgroup;

[0504] Based on the matching results and the content segments in the pushed media content pointed to by each of the original key information, the exclusive promotional content composed of each exclusive key information is generated.

[0505] Optionally, the generation unit 2002 is further configured to:

[0506] The level of each exclusive key information is determined based on the matching result, wherein the level of each exclusive key information represents the attention index of the pushed user to the corresponding exclusive key information.

[0507] Optionally, the generation unit 2002 is specifically used for:

[0508] The content preference information and the original key information are directed to the content segments in the pushed media content and input into the trained generation model;

[0509] Based on the generation model, the exclusive promotional content is generated, which consists of the various exclusive key information.

[0510] The generative model is trained based on labeled key information of the samples and sample content preference information.

[0511] Optionally, the device further includes:

[0512] The first optimization unit 2004 is used to collect feedback data from each user on the exclusive key information generated based on the generation model;

[0513] The generative model is iteratively optimized based on the collected feedback data.

[0514] Optionally, the device further includes:

[0515] The second optimization unit 2005 is used to detect the actual delivery effect of the pushed media content;

[0516] Based on the actual delivery results, adjust at least one of the following:

[0517] The original key information corresponding to the pushed media content and the delivery strategy of the pushed media content.

[0518] Optionally, the device further includes:

[0519] Promotion unit 2006 is used to push the push media content to relevant users on other platforms based on the content preference information of the push recipient and various exclusive key information matched with the push recipient.

[0520] Specifically, when browsing information feed products, users can trigger a content switching operation in the content browsing interface to change the information feed currently displayed. In this embodiment, the information feed can carry push media content to be displayed. When the user triggers the content switching operation, and the switched content browsing interface needs to display the push media content, the client in this application does not directly display the original push media content, but displays exclusive promotional content generated based on this push media content for the user. For different users, different exclusive promotional content can be generated based on the same push media content, achieving a highly personalized content presentation and a personalized content display effect for different users.

[0521] Furthermore, this exclusive promotional content specifically includes content fragments pointed to by various levels of exclusive key information matched with the target user. Unlike the universality of the original key information in the push media content, this exclusive key information is generated by combining the original key information in the push media content with the target user's content preference information. For different users, it can match the preferences and interests of different users. For a single user, it can better match the target user's interests and consumption habits, thus optimizing the matching degree between the push media content and the target user.

[0522] Based on this, when displaying personalized promotional content to the target user, this application will directly locate and play the content segment pointed to by the personalized key information at the user's interest level (i.e., the level matching that interest) within the personalized promotional content, according to the target user's interest level in the pushed information stream. In other words, for the same target user, if their interest in the current information stream content differs, content segments pointed to by different levels of personalized key information will be prioritized for that user to attract clicks, generate consumption behavior for the pushed media content, and thus increase click-through rates.

[0523] For ease of description, the above sections are divided into modules (or units) according to their functions and described separately. Of course, in implementing this application, the functions of each module (or unit) can be implemented in one or more software or hardware components.

[0524] 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.

[0525] Having introduced the content demonstration method and apparatus of exemplary embodiments of this application, we will now introduce an electronic device according to another exemplary embodiment of this application.

[0526] Those skilled in the art will understand that various aspects of this application can be implemented as a system, method, or program product. Therefore, various aspects of this application can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, collectively referred to herein as a "circuit," "module," or "system."

[0527] Based on the same inventive concept as the above-described method embodiments, this application also provides an electronic device. In one embodiment, the electronic device may be a server, such as the server 120 shown in FIG1. ​​In this embodiment, the structure of the electronic device may be as shown in FIG21, including a memory 2101, a communication module 2103, and one or more processors 2102.

[0528] The memory 2101 is used to store computer programs executed by the processor 2102. The memory 2101 may mainly include a program storage area and a data storage area. The program storage area may store the operating system and programs required to run instant messaging functions, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0529] Memory 2101 may be volatile memory, such as random-access memory (RAM); memory 2101 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 2101 may be any other medium capable of carrying or storing a desired computer program having the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 2101 may be a combination of the above-described memories.

[0530] Processor 2102 may include one or more central processing units (CPUs) or digital processing units, etc. Processor 2102 is used to implement the above-described display method when calling computer programs stored in memory 2101.

[0531] The communication module 2103 is used to communicate with terminal devices and other servers.

[0532] This embodiment does not limit the specific connection medium between the memory 2101, communication module 2103, and processor 2102. In Figure 21, the memory 2101 and processor 2102 are connected via a bus 2104, which is depicted as a thick line. The connection methods between other components are merely illustrative and not intended to be limiting. The bus 2104 can be an address bus, data bus, control bus, etc. For ease of description, only one thick line is used in Figure 21, but this does not imply that there is only one bus or one type of bus.

[0533] The memory 2101 stores a computer storage medium, which in turn stores computer-executable instructions for implementing the content display method of this application embodiment. The processor 2102 is used to execute the above-described content display method, as shown in FIG16.

[0534] In another embodiment, the electronic device may also be other electronic devices, such as the terminal device 110 shown in FIG1. ​​In this embodiment, the structure of the electronic device may be as shown in FIG22, including: a communication component 2210, a memory 2220, a display unit 2230, a camera 2240, a sensor 2250, an audio circuit 2260, a Bluetooth module 2270, a processor 2280, and other components.

[0535] The communication component 2210 is used to communicate with the server. In some embodiments, it may include a Circuit-Wireless Fidelity (WiFi) module, which is a short-range wireless transmission technology. Electronic devices can use the WiFi module to help users send and receive information.

[0536] The memory 2220 can be used to store software programs and data. The processor 2280 executes various functions of the terminal device 110 and performs data processing by running the software programs or data stored in the memory 2220. The memory 2220 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. The memory 2220 stores an operating system that enables the terminal device 110 to run. In this application, the memory 2220 can store the operating system and various applications, and may also store computer programs that execute the methods shown in the embodiments of this application.

[0537] The display unit 2230 can also be used to display information input by the user or information provided to the user, as well as various menus of the terminal device 110, in a graphical user interface (GUI). Specifically, the display unit 2230 may include a display screen 2232 disposed on the front of the terminal device 110. The display screen 2232 may be configured as a liquid crystal display, a light-emitting diode, or the like. The display unit 2230 can be used to display the content browsing interface, etc., as described in the embodiments of this application.

[0538] The display unit 2230 can also be used to receive input digital or character information and generate signal inputs related to user settings and function control of the terminal device 110. Specifically, the display unit 2230 may include a touch screen 2231 disposed on the front of the terminal device 110, which can collect touch operations of the user on or near it, such as clicking buttons, dragging scroll boxes, etc.

[0539] The touchscreen 2231 can be placed on top of the display screen 2232, or the touchscreen 2231 and the display screen 2232 can be integrated to realize the input and output functions of the terminal device 110. After integration, it can be referred to as a touch display screen. In this application, the display unit 2230 can display the application and the corresponding operation steps.

[0540] Camera 2240 can be used to capture still images, which users can then share via an application. There can be one or multiple cameras 2240. The object being photographed is projected onto a photosensitive element through a lens, generating an optical image. The photosensitive element can be a charge-coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS) phototransistor. The photosensitive element converts the light signal into an electrical signal, which is then transmitted to a processor 2280 to be converted into a digital image signal.

[0541] The terminal device may also include at least one sensor 2250, such as an accelerometer 2251, a proximity sensor 2252, a fingerprint sensor 2253, and a temperature sensor 2254. The terminal device may also be equipped with other sensors such as a gyroscope, barometer, hygrometer, thermometer, infrared sensor, light sensor, and motion sensor.

[0542] Audio circuitry 2260, speaker 2261, and microphone 2262 provide an audio interface between the user and terminal device 110. Audio circuitry 2260 converts received audio data into electrical signals, which are then transmitted to speaker 2261, where they are converted into sound signals for output. Terminal device 110 may also be equipped with volume buttons for adjusting the volume of the sound signal. On the other hand, microphone 2262 converts collected sound signals into electrical signals, which are received by audio circuitry 2260, converted into audio data, and then output to communication component 2210 for transmission to, for example, another terminal device 110, or to memory 2220 for further processing.

[0543] Bluetooth module 2270 is used to interact with other Bluetooth devices that also have Bluetooth modules via the Bluetooth protocol. For example, a terminal device can establish a Bluetooth connection with a wearable electronic device (such as a smartwatch) that also has a Bluetooth module through Bluetooth module 2270, thereby exchanging data.

[0544] The processor 2280 is the control center of the terminal device, connecting various parts of the terminal through various interfaces and lines. It executes software programs stored in the memory 2220 and calls data stored in the memory 2220 to perform various functions and process data of the terminal device. In some embodiments, the processor 2280 may include one or more processing units; the processor 2280 may also integrate an application processor and a baseband processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the baseband processor mainly handles wireless communication. It is understood that the baseband processor may not be integrated into the processor 2280. In this application, the processor 2280 can run the operating system, applications, user interface display and touch response, and the content display method of the embodiments of this application. Furthermore, the processor 2280 is coupled to the display unit 2230.

[0545] In some possible implementations, various aspects of the content display method provided in this application may also be implemented in the form of a program product, which includes a computer program. When the program product is run on an electronic device, the computer program is used to cause the electronic device to perform the steps in the content display method according to the various exemplary embodiments of this application described above. For example, the electronic device may perform the steps shown in FIG3 or FIG16.

[0546] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0547] The program product of the embodiments of this application may employ a portable compact disc read-only memory (CD-ROM) and include a computer program, and may run on an electronic device. However, the program product of this application is not limited thereto. In this document, the readable storage medium may be any tangible medium that contains or stores a program that may be used by or in conjunction with a command execution system, apparatus, or device.

[0548] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a readable computer program. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with a command execution system, apparatus, or device.

[0549] Computer programs contained on readable media may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0550] Computer programs for performing the operations of this application can be written in any combination of one or more programming languages, including user-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The computer program can execute entirely on the user's electronic device, partially on the user's electronic device, as a standalone software package, partially on the user's electronic device and partially on a remote electronic device, or entirely on a remote electronic device or server. In cases involving remote electronic devices, the remote electronic device can be connected to the user's electronic device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external electronic device (e.g., via the Internet using an Internet service provider).

[0551] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0552] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0553] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing a computer-usable computer program.

[0554] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0555] These computer program commands may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the commands stored in the computer-readable storage medium produce an article of manufacture including command means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0556] These computer program commands may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the commands executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0557] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0558] In summary, this application provides a content display method, apparatus, electronic device, computer-readable storage medium, and computer program product. By displaying a content browsing interface for showing a pushed information stream to the recipient user, and upon receiving a content switching operation on this interface, it displays personalized promotional content generated based on the pushed media content carried in the information stream, and jumps to a content segment pointed to by personalized key information at the user's interest matching level within the personalized promotional content for playback. The personalized key information is generated by combining the original key information of the pushed media content with the content preference information of the recipient user. This method achieves personalized content customization, using user preference information to filter and integrate key information, enabling the system to provide customized content according to each user's unique needs. Technically, it reduces the workload of the system in processing and pushing irrelevant information, improving information processing efficiency. Simultaneously, because the displayed content accurately matches user interests, it reduces the time cost for users to search and filter information, improving the speed and convenience of information acquisition.

[0559] Each level of exclusive key information is associated with the content preferences of the recipient user, and this level represents the user's level of interest in the corresponding exclusive key information. This design allows the system to hierarchically sort exclusive key information based on user interests. Technically, the system can optimize the storage and retrieval of key information based on this hierarchical mechanism. For example, higher-level exclusive key information can be stored in more easily accessible storage areas, allowing for faster retrieval when displaying content, reducing data reading and display latency, and improving content display response speed. Furthermore, this hierarchical sorting enables the system to utilize computing resources more efficiently when processing large amounts of key information, avoiding indiscriminate processing of all key information and improving resource utilization efficiency.

[0560] When the user's interest level falls within a preset interest range, the system redirects to the content segment linked to the highest-level exclusive keyword information within the personalized promotional content. This preset interest range is determined based on big data analysis and machine learning algorithms. Within this range, the system can accurately determine the user's high level of interest in the content. Displaying the highest-level exclusive keyword information at this point technically allows for more precise fulfillment of the user's interest needs. The system can optimize the redirection algorithm based on the user's interest level and the level of the keyword information, reducing time consumption and data processing during the redirection process, improving accuracy and speed, and thus enhancing the user experience.

[0561] If the user's interest level does not fall within the preset interest level range, the system will randomly jump to any content segment pointed to by any of the exclusive key information (excluding the highest level) for playback, or randomly jump to any content segment pointed to by any of the exclusive key information for playback. Technically, this random jump mechanism increases the system's flexibility and versatility. The system can implement random jumps using a simple and efficient random number generation algorithm that does not consume excessive system resources. Random jumps can provide users with more content choices when their interest level is low, increasing the likelihood of them discovering new points of interest. Simultaneously, random jumps can balance the display opportunities of various key information segments, preventing some key information from being overlooked due to prolonged lack of display, thus improving information utilization.

[0562] Interest level is determined by either the user's content switching actions on the content browsing interface (with a negative correlation between interest level and the trigger state of these actions) or the duration of the user's gaze on the content browsing interface (with a positive correlation between interest level and gaze duration). Employing multiple methods to comprehensively determine interest level improves the accuracy of interest assessment. The system can utilize sensor technology (such as cameras and eye trackers) to acquire the user's gaze duration and an event monitoring mechanism to record the user's content switching actions. Through the fusion and analysis of this multi-source data, the system can gain a more comprehensive understanding of the user's interests, providing a more accurate basis for subsequent content recommendations and navigation, reducing misjudgments and inaccurate recommendations, and improving the system's intelligence and accuracy.

[0563] After the content segment pointed to by the user's interest-matching key information has finished playing, the system continues playing content segments pointed to by alternative key information. This operation technically achieves continuous playback and expanded display of content. The system can automatically call up the next content segment corresponding to alternative key information after the content segment pointed to by one key information has finished playing, reducing manual steps for the user and improving playback continuity and smoothness. At the same time, this continuous playback method also improves the system's efficiency in utilizing content resources, avoiding resource idleness and waste.

[0564] When playing the content segment pointed to by the selected exclusive key information, the system randomly jumps to that segment for playback. After all the content segments pointed to by each exclusive key information have finished playing, the system sequentially rotates through the content segments pointed to by each exclusive key information. This random jump and rotation mechanism technically increases the diversity and cyclicality of content display. The system can implement these two playback methods using a loop algorithm and a random number generator. This algorithm is simple and efficient to implement and will not significantly impact system performance. Random jumps provide users with different content experiences, while rotation ensures that all key information has a chance to be displayed, improving content exposure and utilization.

[0565] The playback progress bar of the exclusive promotional content highlights the time points associated with each key piece of information. Responding to a first-type trigger action at any selected time point, the user is redirected to the corresponding content segment within the promotional content. Technically, this time-point highlighting and redirection feature enhances user interactivity and convenience. The system can implement this functionality by adding markers and event monitoring mechanisms to the progress bar. Users can quickly jump to content segments of interest by clicking on time points, reducing the time and effort required to find content and improving information acquisition efficiency. Simultaneously, this interaction method also improves system responsiveness and user experience.

[0566] The playback progress bar of the exclusive promotional content highlights each exclusive key and the associated time point of each piece of information. In response to a second type of trigger action at any given time point, the system displays the pushed media content and jumps to the playback of the content segment associated with the corresponding exclusive key information. This feature technically enables flexible switching between exclusive promotional content and the original pushed media content. The system can quickly locate and display the relevant content segment when the user triggers an action through data association and jump algorithms. This switching function increases the flexibility and completeness of content display, allowing users to choose to view different forms of content according to their needs, improving the comprehensiveness and accuracy of information acquisition.

[0567] Another content display method involves first extracting key information from the pushed media content and obtaining the content preference information of the target users. Then, based on this information, personalized promotional content is generated for the target users. Finally, this personalized promotional content is sent to the target users and displayed when they switch content, redirecting them to the content segment indicated by the personalized key information matching their interests. Personalizing content from its source requires the use of natural language processing and machine learning algorithms. By extracting key information, the system can structure the content, facilitating subsequent analysis and matching. The process of generating personalized promotional content is a complex algorithm optimization process that generates the most suitable content based on user preferences and content characteristics. This personalized content generation method improves the targeting and relevance of content, reduces the generation and push of irrelevant content, and improves the system's processing efficiency and resource utilization.

[0568] When generating personalized promotional content for targeted users, the system combines the content preferences of the targeted users to determine their subgroups. It then matches each original key piece of information with the content needs of those subgroups. Based on the matching results and the content segments within the push media content linked to each original key piece of information, personalized promotional content is generated, composed of specific key pieces of information. Technically, determining subgroups requires clustering algorithms to group users with similar interests into the same subgroup. Matching the original key pieces of information with the content needs of the subgroups requires similarity calculation algorithms, such as cosine similarity algorithms. Through this matching and generation process, the system can more accurately generate personalized promotional content for different subgroups, improving the accuracy and effectiveness of content recommendations. Simultaneously, this subgroup-based content generation method can optimize the system's storage and processing architecture, improving its scalability and performance.

[0569] The system determines the level of each exclusive key message based on the matching results. The level of each exclusive key message represents the level of attention the recipient user has for that specific key message. Technically, determining the level requires using classification algorithms from machine learning to categorize and rank the exclusive key messages based on the matching results. This ranking mechanism allows the system to manage and display key information more effectively. Higher-level exclusive key messages receive priority processing and display within the system, improving the efficiency of key message delivery and user attention. Simultaneously, the ranking mechanism helps the system optimize resource allocation, avoiding the average processing of all key messages and improving overall system performance.

[0570] Content preference information and various original key information are fed into a pre-trained generative model, which then generates personalized promotional content composed of these unique key information segments. This generative model is trained using labeled sample key information and sample content preference information. Technically, training the generative model requires deep learning algorithms, such as neural network algorithms. Through training with a large amount of sample data, the model learns the intrinsic relationship between user preference patterns and content features. When generating personalized key information, the model can quickly and accurately generate content that meets user needs based on the input content preference information and original key information. This model-based content generation method improves the efficiency and quality of content generation, reduces manual intervention, and enhances the system's automation and intelligence.

[0571] The system collects feedback data from various users regarding the personalized key information generated based on the generative model. Based on this feedback, the model is iteratively optimized. Technically, collecting feedback data requires a robust data collection and storage system to ensure data accuracy and completeness. Iterative optimization of the generative model necessitates the use of machine learning model update algorithms, such as gradient descent. By continuously collecting feedback data and optimizing the model, the system can adapt to changing user needs and content updates. The optimized model generates more attractive and relevant personalized key information, improving the accuracy and effectiveness of the content, while also enhancing the system's stability and reliability.

[0572] The system monitors the actual effectiveness of pushed media content delivery and adjusts the original key information or delivery strategy based on this performance. Technically, monitoring effectiveness requires a data analysis and monitoring system to collect and analyze metrics such as impressions and click-through rates in real time. Adjusting the original key information and delivery strategy based on these metrics necessitates the use of intelligent decision-making algorithms, such as reinforcement learning. Through real-time monitoring and dynamic adjustment of delivery effectiveness, the system can continuously optimize content and delivery strategies, improving content dissemination and resource utilization efficiency. Furthermore, this dynamic adjustment mechanism allows the system to better adapt to different user groups and market environments.

[0573] Based on the content preferences of the recipient user and various unique key information matching that user, the system pushes media content to relevant users on other platforms. Technically, this requires establishing a cross-platform data transmission and sharing mechanism to ensure the secure and efficient delivery of content across different platforms. The system needs to interface with other platforms through data interfaces and protocols to achieve data interaction and synchronization. Simultaneously, the system also needs to accurately identify and filter relevant users, pushing content to those most likely to be interested. Cross-platform push can expand the reach of content, increase its exposure and influence, and also improve the system's compatibility and openness.

[0574] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0575] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A content display method, executed by an electronic device, the method comprising: Displays the content browsing interface; The content browsing interface is used to display the pushed information stream to the user; In response to the content switching operation of the content browsing interface, exclusive promotional content generated based on the push media content carried by the information stream is displayed in the content browsing interface, and the user is redirected to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback; The exclusive promotional content is used to describe various levels of exclusive key information that match the target user; each exclusive key information is generated based on the original key information in the push media content and the target user's content preference information; the user interest matching level is determined based on the target user's interest in the information stream.

2. The method as described in claim 1, wherein the level of each of the exclusive key information is associated with the content preference information of the pushed user, and the level represents: the attention index of the pushed user to the corresponding exclusive key information.

3. The method as described in claim 1 or 2, wherein the step of jumping to and playing the content segment pointed to by the exclusive key information at the user interest matching level in the exclusive promotional content includes: If the interest level falls within a preset interest level range, the user will be redirected to the exclusive promotional content, where the highest-level exclusive key information will be used to play the content segment.

4. The method according to any one of claims 1 to 3, wherein the method further comprises: If the interest level does not fall within the preset interest level range, the player will randomly jump to any content segment pointed to by any exclusive key information other than the highest level, or randomly jump to any content segment pointed to by any exclusive key information among the various exclusive key information segments.

5. The method of any one of claims 1 to 4, wherein the degree of interest is determined by at least one of the following methods: determining, according to a content switching operation of the pushed user on the content browsing interface, wherein, The level of interest is negatively correlated with the triggering state of the content switching operation; or, The interest level is determined based on the duration of the user's gaze on the content browsing interface, wherein the interest level is positively correlated with the duration of gaze.

6. The method according to any one of claims 1 to 5, wherein the method further comprises: After the content segment pointed to by the exclusive key information of the user interest matching level has finished playing, the content segment pointed to by the alternative exclusive key information will continue to play.

7. The method as described in claim 6, wherein continuing to play the content segment pointed to by the alternative exclusive key information includes: The program will randomly redirect to the content segment indicated by the selected exclusive key information for playback. After all the content segments pointed to by each specific key information have been played, the content segments pointed to by each specific key information will be played in turn.

8. The method according to any one of claims 1 to 7, wherein the step of jumping to and playing the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content further includes: The playback progress bar of the exclusive promotional content highlights the time points associated with each exclusive key piece of information; Furthermore, the method further includes: In response to a first-type trigger operation for any alternative time point, the user is redirected to the corresponding content segment in the exclusive promotional content for playback. The alternative time points are time points associated with alternative exclusive key information other than the exclusive key information of the user interest matching level.

9. The method according to any one of claims 1 to 8, wherein the step of jumping to and playing the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content further includes: The playback progress bar of the exclusive promotional content highlights the time points associated with each exclusive key piece of information; Furthermore, the method further includes: In response to a second type of triggering operation at any point in time, the pushed media content is displayed, and the user is redirected to the pushed media content to play a content segment associated with the corresponding exclusive key information.

10. A content display method, performed by an electronic device, the method comprising: For pushed media content, extract each original key piece of information from the pushed media content and obtain the content preference information of the user receiving the push; Based on the original key information and the content preference information, exclusive promotional content is generated for the pushed user, wherein the exclusive promotional content is used to describe different levels of exclusive key information that match the pushed user; The exclusive promotional content is sent to the target user, and in response to the target user's content switching operation on the content browsing interface, the exclusive promotional content is displayed on the content browsing interface, and the user is redirected to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback; wherein, the content browsing interface is used to display the pushed information stream to the target user, and the information stream carries the pushed media content; the user interest matching level is determined based on the target user's interest in the information stream.

11. The method of claim 10, wherein generating personalized promotional content for the target user based on the original key information and the content preference information comprises: Based on the content preference information of the pushed users, determine the subgroup to which the pushed users belong; Match the original key information with the content requirements corresponding to the subgroup; Based on the matching results and the content segments in the pushed media content pointed to by each of the original key information, the exclusive promotional content composed of each exclusive key information is generated.

12. The method of claim 11, wherein generating the exclusive promotional content composed of the exclusive key information based on the matching result and the content fragments in the pushed media content pointed to by the original key information further includes: The level of each exclusive key information is determined based on the matching result, wherein the level of each exclusive key information represents the attention index of the pushed user to the corresponding exclusive key information.

13. The method according to any one of claims 10 to 12, wherein generating personalized promotional content for the target user based on the original key information and the content preference information comprises: The content preference information and the original key information are directed to the content segments in the pushed media content and input into the trained generation model; Based on the generation model, the exclusive promotional content is generated, which consists of the various exclusive key information. The generative model is trained based on labeled key information of the samples and sample content preference information.

14. The method of claim 13, further comprising: Collect feedback data from each user on the exclusive key information generated based on the generative model; The generative model is iteratively optimized based on the collected feedback data.

15. The method of any one of claims 10 to 14, wherein the method further comprises: Detect the actual delivery effect of the pushed media content; Based on the actual delivery results, adjust at least one of the following: The original key information corresponding to the pushed media content or the delivery strategy of the pushed media content.

16. The method of any one of claims 10 to 15, wherein the method further comprises: Based on the content preference information of the target user and various unique key information matched with the target user, the push media content is pushed to relevant users on other platforms.

17. A content display device, comprising: The display unit is used to display the content browsing interface; The content browsing interface is used to display the pushed information stream to the user; The response unit is used to respond to the content switching operation of the content browsing interface, display exclusive promotional content generated based on the push media content carried by the information stream in the content browsing interface, and jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback; The exclusive promotional content is used to describe various levels of exclusive key information that match the target user; each exclusive key information is generated based on the original key information in the push media content and the target user's content preference information; the user interest matching level is determined based on the target user's interest in the information stream.

18. A content display device, comprising: The analysis unit is used to extract various original key information from the pushed media content and obtain the content preference information of the pushed user. The generation unit is used to generate exclusive promotional content for the pushed user based on the original key information and the content preference information, wherein the exclusive promotional content is used to describe the exclusive key information at different levels that match the pushed user; The feedback unit is used to send the exclusive promotional content to the target user and, in response to the target user's content switching operation on the content browsing interface, display the exclusive promotional content on the content browsing interface and jump to the content segment pointed to by the exclusive key information of the user interest matching level in the exclusive promotional content for playback; wherein, the content browsing interface is used to display the pushed information stream to the target user, the information stream carrying the pushed media content; the user interest matching level is determined according to the target user's interest in the information stream.

19. An electronic device comprising a processor and a memory, wherein, The memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of any of the methods described in claims 1 to 16.

20. A computer-readable storage medium comprising a computer program that, when executed on an electronic device, causes the electronic device to perform the steps of any one of the methods of claims 1 to 16.

21. A computer program product comprising a computer program stored in a computer-readable storage medium; wherein when a processor of an electronic device reads the computer program from the computer-readable storage medium, the processor executes the computer program, causing the electronic device to perform the steps of any one of claims 1 to 16.

Images