A method for managing recommended content and related devices

Abstract

The application discloses a kind of recommended content management method and related device, specifically related to the machine learning technology of artificial intelligence.The interaction data corresponding to target object is obtained;Then the object portrait is obtained by being statistically carried out in the interaction data based on preset feature dimension;And according to the plurality of features in object portrait, the plurality of features are evaluated to obtain the interaction feature value set;Combining the target feature value is obtained by the income feature value and the interaction feature value set;Further, the target content in recommendation set is determined for pushing according to target feature value and is shown.Thereby, the conversion and income of recommended content in the pushing process are balanced, since the feature dimension is considered by adopting multiple time windows and combining data transmission mode, the integrity of object portrait is guaranteed, and the accuracy of recommended content pushing is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method and apparatus for managing recommended content. Background Technology

[0002] With the rapid development of internet technology, people have increasingly higher demands for recommended content. As the final stage in a recommendation system, haphazard ranking determines the final list of recommendations presented to users and their display order, thus its potential has gradually gained attention in recent years.

[0003] Generally, different forms of recommended content can be mixed and arranged based on strategy rules, such as removing duplicates, scattering results to increase the diversity of recommended results, and forcibly inserting certain types of recommended results, etc. In other words, fixed arrangement rules are used to arrange different forms of recommended content.

[0004] However, due to the differences in various forms of recommended content, users' interest in different recommended content may fluctuate. The mixed ranking by strategy rules may affect the accuracy of recommended content delivery. Summary of the Invention

[0005] In view of this, this application provides a method for managing recommended content, which can effectively improve the accuracy of recommended content delivery.

[0006] The first aspect of this application provides a method for managing recommended content, which can be applied to a system or program in a terminal device that includes a management function for recommended content, specifically including:

[0007] The interaction operations of the target object with different forms of recommended content in the recommendation set are obtained within multiple time windows to determine the interaction data corresponding to the target object.

[0008] The object features and content features in the interaction data are statistically analyzed based on preset feature dimensions to obtain an object profile. The preset feature dimensions are set based on the time window, the interaction operation, and the data transmission method corresponding to the recommended content.

[0009] The multiple features in the object profile are evaluated based on a multi-objective model to obtain a set of interactive feature values;

[0010] The data corresponding to the operation items in the content features are called to perform cumulative revenue statistics in order to determine the revenue feature value corresponding to the recommended content;

[0011] The target feature value is obtained by combining the revenue feature value and the predicted value contained in the interaction feature value set;

[0012] The target content for push is determined in the recommendation set based on the target feature value, and the target content is distributed and displayed in the recommendation interface.

[0013] Optionally, in some possible implementations of this application, obtaining the interaction operations of the target object with different forms of recommendation content in the recommendation set within multiple time windows to determine the interaction data corresponding to the target object includes:

[0014] Obtain the recommendation period corresponding to the target object;

[0015] Gradient configuration is performed based on the recommendation period to determine multiple time windows;

[0016] The interaction operations of the recommended content in the recommendation set within multiple time windows are obtained to determine the interaction data corresponding to the target object.

[0017] Optionally, in some possible implementations of this application, the step of statistically analyzing the object features and content features in the interaction data based on preset feature dimensions to obtain an object profile includes:

[0018] The time period division information indicated by the preset feature dimension is determined based on the historical data corresponding to the target object;

[0019] Determine the network state corresponding to when the target object obtains the recommended content within the statistical time period indicated by the time period segmentation information, so as to determine the data transmission method;

[0020] The interaction operations of the target object with different types of recommended content under the data transmission method are obtained to determine the characteristics of the object;

[0021] The exposure information and association information corresponding to the different types of recommended content under the aforementioned data transmission method are statistically analyzed to determine the content characteristics.

[0022] The object features and the content features are integrated to obtain the object profile.

[0023] Optionally, in some possible implementations of this application, obtaining the interaction operations of the target object with different types of recommended content under the data transmission method to determine the object characteristics includes:

[0024] The data transmission method is compared with the data transmission method in the historical data corresponding to the target object to obtain recommended scenario information;

[0025] Based on the recommended scenario information, the interactive operations of the target object with different types of recommended content under the data transmission method are obtained respectively to determine the object characteristics;

[0026] The step of statistically analyzing the exposure information and association information corresponding to different types of recommended content under the data transmission method to determine the content characteristics includes:

[0027] Based on the recommended scenario information, the exposure information and association information corresponding to the different types of recommended content under the data transmission method are statistically analyzed to determine the content characteristics.

[0028] Optionally, in some possible implementations of this application, determining the time period division information indicated by the preset feature dimension based on the historical data corresponding to the target object includes:

[0029] The interaction frequency information of the target object with the recommended content is determined based on the historical data corresponding to the target object;

[0030] Hotspot time periods are extracted based on the interaction frequency information;

[0031] The hotspot time period is determined as the time period division information indicated by the preset feature dimension.

[0032] Optionally, in some possible implementations of this application, the method further includes:

[0033] Obtain the audience segmentation dimensions, which include age, gender, or activity level;

[0034] Based on the aforementioned audience segmentation dimensions, interactive data statistics are performed across different dimensions to determine the audience profile;

[0035] Determine the confidence parameters corresponding to the target object;

[0036] If the confidence parameter indicates that the target object is not confident, then the object profile is replaced with the crowd profile.

[0037] Optionally, in some possible implementations of this application, the step of calling the data corresponding to the operation item in the content feature to perform cumulative revenue statistics to determine the revenue feature value corresponding to the recommended content includes:

[0038] Determine the recommended content corresponding to the operation item in the content features, and determine the set of associated content for the recommended content corresponding to the operation item;

[0039] Based on the set of associated content, cumulative revenue is statistically analyzed to determine the revenue characteristic value corresponding to the recommended content.

[0040] Optionally, in some possible implementations of this application, the step of calculating the cumulative revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content includes:

[0041] Determine the click information corresponding to the set of associated content;

[0042] If the click information reaches the preset conditions, the cumulative revenue is statistically analyzed based on the set of associated content to determine the revenue feature value corresponding to the recommended content;

[0043] or;

[0044] If the click information does not meet the preset conditions, the predicted value corresponding to the click information will be used as the target feature value.

[0045] Optionally, in some possible implementations of this application, the method further includes:

[0046] Extract the total number of clicks indicated in the object profile;

[0047] If the total number of clicks does not meet the click criteria, the audience profile is invoked to replace the recommended content for all types in the object profile.

[0048] Optionally, in some possible implementations of this application, the method further includes:

[0049] Determine the recommended content for each type indicated in the object profile;

[0050] Recommended content under each type is filtered based on exposure conditions, and recommended content under types that do not meet the exposure conditions is replaced with the user profile.

[0051] Optionally, in some possible implementations of this application, the step of combining the revenue feature value and the predicted value contained in the interaction feature value set to calculate the target feature value includes:

[0052] Determine the content type corresponding to the recommended content;

[0053] Determine the corresponding feature weighting coefficients based on the content type;

[0054] The target feature value is obtained by weighting the predicted values ​​contained in the set of interaction feature values ​​according to the feature weighting coefficients.

[0055] Optionally, in some possible implementations of this application, the recommended content is a heterogeneous card, and the content format of the heterogeneous card includes text and image content, video content, or link content. The recommended interface is a terminal application interface, and the target content is displayed in the focus area of ​​the terminal application interface.

[0056] A second aspect of this application provides a management device for recommended content, comprising:

[0057] The acquisition unit is used to acquire the interaction operations of the target object with different forms of recommendation content in the recommendation set within multiple time windows, so as to determine the interaction data corresponding to the target object;

[0058] The statistical unit is used to perform statistics on the object features and content features in the interaction data based on preset feature dimensions to obtain an object profile. The preset feature dimensions are set based on the time window, the interaction operation, and the data transmission method corresponding to the recommended content.

[0059] An evaluation unit is used to evaluate multiple features in the object profile according to a multi-objective model to obtain a set of interactive feature values.

[0060] The statistical unit is also used to call the data corresponding to the operation items in the content features to perform cumulative revenue statistics, so as to determine the revenue feature value corresponding to the recommended content;

[0061] The management unit is used to calculate the target feature value by combining the revenue feature value and the predicted value contained in the interaction feature value set;

[0062] The management unit is used to determine the target content to be pushed in the recommendation set according to the target feature value, and to distribute the target content in the recommendation interface for display.

[0063] Optionally, in some possible implementations of this application, the acquisition unit is specifically used to acquire the recommendation period corresponding to the target object;

[0064] The acquisition unit is specifically used to perform gradient configuration based on the recommendation period to determine multiple time windows;

[0065] The acquisition unit is specifically used to acquire the interaction operations of the recommended content in the recommendation set within multiple time windows, so as to determine the interaction data corresponding to the target object.

[0066] Optionally, in some possible implementations of this application, the statistical unit is specifically used to determine the time period division information indicated by the preset feature dimension based on the historical data corresponding to the target object;

[0067] The statistical unit is specifically used to determine the network state corresponding to when the target object obtains the recommended content within the statistical period indicated by the time period division information, so as to determine the data transmission method.

[0068] The statistical unit is specifically used to acquire the interaction operations of the target object with different types of recommended content under the data transmission method, so as to determine the characteristics of the object;

[0069] The statistical unit is specifically used to statistically analyze the exposure information and association information corresponding to the different types of recommended content under the data transmission method, so as to determine the content characteristics.

[0070] The statistical unit is specifically used to integrate the object features and the content features to obtain the object profile.

[0071] Optionally, in some possible implementations of this application, the statistical unit is specifically used to compare the data transmission method with the data transmission method in the historical data corresponding to the target object to obtain recommended scenario information;

[0072] The statistical unit is specifically used to obtain the interaction operations of the target object with different types of recommended content under the data transmission method based on the recommendation scenario information, so as to determine the object characteristics;

[0073] The statistical unit is specifically used to statistically analyze the exposure information and association information corresponding to different types of recommended content under the data transmission method based on the recommended scenario information, so as to determine the content features.

[0074] Optionally, in some possible implementations of this application, the statistical unit is specifically used to determine the interaction frequency information of the target object with respect to the recommended content based on the historical data corresponding to the target object;

[0075] The statistical unit is specifically used to extract hotspot time periods based on the interaction frequency information;

[0076] The statistical unit is specifically used to determine the hotspot time period as the time period division information indicated by the preset feature dimension.

[0077] Optionally, in some possible implementations of this application, the statistical unit is specifically used to obtain the audience segmentation dimension, which includes age, gender, or activity level.

[0078] The statistical unit is specifically used to perform interactive data statistics under different dimensions based on the population segmentation dimensions in order to determine the population profile.

[0079] The statistical unit is specifically used to determine the confidence parameters corresponding to the target object;

[0080] The statistical unit is specifically used to replace the object profile with the crowd profile if the confidence parameter indicates that the target object is not confident.

[0081] Optionally, in some possible implementations of this application, the statistical unit is specifically used to determine the recommended content corresponding to the operation item in the content features, and to determine the set of associated content of the recommended content corresponding to the operation item;

[0082] The statistical unit is specifically used to perform cumulative revenue statistics based on the associated content set, so as to determine the revenue feature value corresponding to the recommended content.

[0083] Optionally, in some possible implementations of this application, the management unit is specifically used to determine the click information corresponding to the associated content set;

[0084] The management unit is specifically used to perform cumulative revenue statistics based on the associated content set if the click information reaches a preset condition, so as to determine the revenue feature value corresponding to the recommended content.

[0085] or;

[0086] The management unit is specifically used to take the predicted value corresponding to the click information as the target feature value if the click information does not meet the preset condition.

[0087] Optionally, in some possible implementations of this application, the management unit is specifically used to extract the total number of clicks indicated in the object profile;

[0088] The management unit is specifically used to replace the recommended content for all types in the object profile if the total number of clicks does not meet the click conditions.

[0089] Optionally, in some possible implementations of this application, the management unit is specifically used to determine the recommended content under each type indicated in the object profile;

[0090] The management unit is specifically used to filter recommended content under each type based on exposure conditions, so as to replace recommended content under types that do not meet the exposure conditions with the user profile.

[0091] Optionally, in some possible implementations of this application, the management unit is specifically used to determine the content type corresponding to the recommended content;

[0092] The management unit is specifically used to determine the corresponding feature weighting coefficient based on the content type;

[0093] The management unit is specifically used to perform weighted calculations on the revenue feature value and the predicted value contained in the interaction feature value set according to the feature weighting coefficient, so as to obtain the target feature value.

[0094] A third aspect of this application provides a computer device, comprising: a memory, a processor, and a bus system; the memory is used to store program code; the processor is used to execute the management method of the recommended content described in the first aspect or any one of the first aspects according to instructions in the program code.

[0095] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the management method of the recommended content described in the first aspect or any one of the first aspects.

[0096] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the management method described in the first aspect or various alternative implementations of the first aspect.

[0097] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0098] By acquiring the interaction operations of the target object with different forms of recommended content in the recommendation set within multiple time windows, the interaction data corresponding to the target object is determined. Then, based on preset feature dimensions, the object features and content features in the interaction data are statistically analyzed to obtain an object profile. These preset feature dimensions are set based on time windows, interaction operations, and the data transmission method corresponding to the recommended content. Multiple features in the object profile are evaluated according to a multi-objective model to obtain a set of interaction feature values. Further, the data corresponding to the operation items in the content features are used to calculate the cumulative revenue to determine the revenue feature value corresponding to the recommended content. Then, the revenue feature value and the predicted values ​​contained in the interaction feature value set are combined to calculate the target feature value. Finally, the target content for push notifications in the recommendation set is determined based on the target feature value and displayed on the recommendation interface. This achieves a balance between the conversion and revenue of recommended content during the push process. By using multiple time windows and considering feature dimensions in conjunction with data transmission methods, the completeness of the object profile is ensured, the simulation of fluctuations in user interest content is improved, and the accuracy of recommended content push is enhanced. Attached Figure Description

[0099] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art 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 provided drawings without creative effort.

[0100] Figure 1 Network architecture diagram of the management system for recommended content;

[0101] Figure 2 A flowchart illustrating the management process of recommended content provided in this application embodiment;

[0102] Figure 3 A flowchart illustrating a method for managing recommended content provided in an embodiment of this application;

[0103] Figure 4 A schematic diagram illustrating a scenario for a method of managing recommended content provided in an embodiment of this application;

[0104] Figure 5 A schematic diagram illustrating a scenario for another method of managing recommended content provided in an embodiment of this application;

[0105] Figure 6 A schematic diagram illustrating a scenario for another method of managing recommended content provided in an embodiment of this application;

[0106] Figure 7 A schematic diagram illustrating a scenario for another method of managing recommended content provided in an embodiment of this application;

[0107] Figure 8 A schematic diagram illustrating a scenario for another method of managing recommended content provided in an embodiment of this application;

[0108] Figure 9 A schematic diagram illustrating a scenario for another method of managing recommended content provided in an embodiment of this application;

[0109] Figure 10 A flowchart illustrating another recommended content management method provided for embodiments of this application;

[0110] Figure 11 A flowchart illustrating another recommended content management method provided for embodiments of this application;

[0111] Figure 12 A schematic diagram of the structure of a management device for recommended content provided in an embodiment of this application;

[0112] Figure 13This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application;

[0113] Figure 14 This is a schematic diagram of the structure of a server provided in an embodiment of this application. Detailed Implementation

[0114] This application provides a method and related apparatus for managing recommended content, which can be applied to systems or programs in terminal devices that include recommended content management functions. The method involves acquiring the interaction operations of a target object with different forms of recommended content in a recommendation set within multiple time windows to determine the interaction data corresponding to the target object. Then, based on preset feature dimensions, the object features and content features in the interaction data are statistically analyzed to obtain an object profile. These preset feature dimensions are set based on time windows, interaction operations, and the data transmission method corresponding to the recommended content. Multiple features in the object profile are evaluated according to a multi-objective model to obtain a set of interaction feature values. Further, the data corresponding to the operation items in the content features are called to perform cumulative revenue statistics to determine the revenue feature value corresponding to the recommended content. Then, the revenue feature value and the predicted value contained in the interaction feature value set are combined to calculate the target feature value. Finally, the target content for push notifications in the recommendation set is determined based on the target feature value, and the target content is displayed on the recommendation interface. This achieves a balance between the conversion and revenue of recommended content during the push process. By using multiple time windows and considering feature dimensions in conjunction with data transmission methods, the completeness of the object profile is ensured, the simulation of fluctuations in user interest content is improved, and the accuracy of recommended content push is enhanced.

[0115] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0116] First, some terms that may appear in the embodiments of this application will be explained.

[0117] Rerank: refers to the process of re-ranking or rearranging the candidate list, which is the stage of determining the final recommendation list and its display order based on the candidate set generated during the recall and ranking stages of the recommendation system.

[0118] Multi-objective optimization refers to optimization with two or more objective functions, aiming to find a sorting method that makes the overall optimization of multiple objectives.

[0119] Heterogeneous cards: refer to different types of topic (item) cards in the recommendation system that have significant differences in content, display format, and revenue value.

[0120] User card preferences: refers to a profile obtained by statistically analyzing user behavior data, which can depict the degree of interest a user has in different cards.

[0121] It should be understood that the recommended content management method provided in this application can be applied to systems or programs in terminal devices that include recommended content management functions, such as content recommendation applications. Specifically, the recommended content management system can run on systems such as... Figure 1 In the network architecture shown, such as Figure 1 The diagram shows the network architecture of the recommended content management system. As can be seen, the recommended content management system can manage recommended content from multiple information sources. Specifically, the system generates corresponding interaction data on the server through the interaction of the terminal with the recommended content. This interaction data is then parsed to obtain the corresponding object profile, which in turn indicates the process of further content push. Figure 1 The document shows various terminal devices, which can be computer devices. In real-world scenarios, more or fewer types of terminal devices may participate in the management of recommended content. The specific number and types depend on the actual scenario and are not limited here. Additionally, Figure 1 The diagram shows one server, but in real-world scenarios, multiple servers can be involved, especially in scenarios involving multi-objective model training and interaction. The specific number of servers depends on the actual scenario.

[0122] In this embodiment, the server can be a standalone 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, CDN, and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, and the terminal and server can be connected to form a blockchain network; this application does not impose any restrictions.

[0123] It is understood that the aforementioned content recommendation management system can run on personal mobile terminals, such as content recommendation applications, or it can run on servers, or it can run on third-party devices to provide content recommendation management and obtain the management and processing results of the information source's recommended content. Specifically, the content recommendation management system can run as a program on the aforementioned devices, or it can run as a system component of the aforementioned devices, or it can run as a cloud service program. The specific operating mode depends on the actual scenario and is not limited here.

[0124] With the rapid development of internet technology, people have increasingly higher demands for recommended content. As the final stage in a recommendation system, haphazard ranking determines the final list of recommendations presented to users and their display order, thus its potential has gradually gained attention in recent years.

[0125] Generally, different forms of recommended content can be mixed and arranged based on strategy rules, such as removing duplicates, scattering results to increase the diversity of recommended results, and forcibly inserting certain types of recommended results, etc. In other words, fixed arrangement rules are used to arrange different forms of recommended content.

[0126] However, due to the differences in various forms of recommended content, users' interest in different recommended content may fluctuate. The mixed ranking by strategy rules may affect the accuracy of recommended content delivery.

[0127] To address the aforementioned problems, this application proposes a method for managing recommended content, employing machine learning (ML) technology. Machine learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory, among others. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0128] Specifically, this method is applied to Figure 2 The workflow framework for managing recommended content is shown below, such as... Figure 2The diagram shown is a flowchart of a recommended content management process provided in an embodiment of this application. Users interact with recommended content (heterogeneous cards) on the terminal side (clicking, downloading, staying, etc.), thereby generating corresponding interaction data on the server. The server analyzes the feature dimensions based on the interaction data to generate an object profile, and statistically analyzes the object profile to obtain the feature values ​​of user content push, thereby sorting the recommended content and distributing it on the interface.

[0129] Understandably, different optimization goals of a recommendation system may be mutually restrictive; for example, if one metric increases, another may drop significantly. This phenomenon is particularly severe in this business scenario because it contains various heterogeneous cards (text / image, short video, short content, social, text / image following, video following, etc.), and some cards have floating consumption. This means that the time spent and advertising revenue generated by a single user click are not only from the content viewed that time, but also from the revenue generated from continuously viewing different content after clicking into the floating layer. Different types of cards have significant differences in time spent and advertising revenue.

[0130] Therefore, this application proposes a multi-objective mixed ranking method that can balance the mutual influence of clicks, duration, and advertising revenue in recommendation scenarios with multiple heterogeneous cards, and can achieve simultaneous increase of all indicators or increase of the main objective without decrease of other objectives.

[0131] It is understood that the method provided in this application can be a program written as processing logic in a hardware system, or it can be a recommended content management device, implemented in an integrated or external manner. As one implementation, the recommended content management device obtains the interaction operations of a target object with different forms of recommended content in the recommendation set within multiple time windows to determine the interaction data corresponding to the target object; then, it statistically analyzes the object features and content features in the interaction data based on preset feature dimensions to obtain an object profile. These preset feature dimensions are set based on time windows, interaction operations, and the data transmission method corresponding to the recommended content; and it evaluates multiple features in the object profile according to a multi-objective model to obtain a set of interaction feature values; further, it calls the data corresponding to the operation items in the content features to perform cumulative revenue statistics to determine the revenue feature value corresponding to the recommended content; then, it combines the revenue feature value and the predicted value contained in the set of interaction feature values ​​to calculate the target feature value; finally, it determines the target content to be pushed in the recommendation set based on the target feature value and distributes the target content for display on the recommendation interface. This achieves a balance between conversion and revenue of recommended content during the push process. By employing multiple time windows and combining data transmission methods to consider feature dimensions, it ensures the completeness of the target audience profile, improves the simulation of fluctuations in user interest content, and enhances the accuracy of recommended content push.

[0132] The solutions provided in this application relate to machine learning technology in artificial intelligence, and are specifically illustrated through the following embodiments:

[0133] Based on the above process architecture, the management method for the recommended content in this application will be introduced below. Please refer to [link / reference]. Figure 3 , Figure 3 The flowchart illustrates a method for managing recommended content provided in this application embodiment. This method can be executed by a terminal, a server, or both. This application embodiment includes at least the following steps:

[0134] 301. Obtain the interaction operations of the target object with different forms of recommendation content in the recommendation set within multiple time windows to determine the interaction data corresponding to the target object.

[0135] In this embodiment, the recommended content is a heterogeneous card. The content format of the heterogeneous card includes text and image content, video content, or link content. Specifically, it can be text and image, short video, short content, social media, follow text and image, follow video, jump link, and other media formats. Figure 4 The scene shown, Figure 4This is a schematic diagram of a scenario for a method of managing recommended content provided in an embodiment of this application. The diagram shows different forms of recommended content contained in the recommendation interface A1. That is, the media formats of content 1-6 can be the same or different, and the specific media formats depend on the actual scenario. The recommended content is used to display in the recommendation interface, which can be a terminal application interface, thereby facilitating the further recommendation conversion process.

[0136] It is understood that the target of the recommended content can be a user, an account, or other representative object identifier. In the following embodiments, the target object is the target user as an example for illustration, but this is not a limitation.

[0137] Specifically, since users' interests fluctuate, data can be collected by dividing the data into multiple time windows and time periods. First, the recommendation period corresponding to the target object is obtained; then, gradient configuration is performed based on the recommendation period to determine multiple time windows; and finally, the interaction operations of the recommended content in the recommendation set within multiple time windows are obtained to determine the interaction data corresponding to the target object.

[0138] It is understandable that the time window serves as the basis for statistical analysis of user interactions, thereby reflecting user preferences across different time dimensions. These preferences can then be used to select recommended content, thus enabling the method in this embodiment to be applied to user preferences across different time dimensions and improving the accuracy of recommended content selection.

[0139] In one possible scenario, in order to depict users’ card preferences in the long term, short term and real time (different time dimensions), the card profile collects user behavior data from three time windows: (1) Annual profile: collects data for one year in advance from the current day; (2) Monthly profile: collects data for one month in advance from the current day; (3) Recent (session) profile: collects data from the user’s last 50 refreshes, thereby ensuring the comprehensiveness of the recommendation cycle corresponding to the target object, thereby ensuring the accuracy of the content in different recommendation cycles.

[0140] 302. Based on preset feature dimensions, perform statistical analysis on object features and content features in the interaction data to obtain object profiles.

[0141] In this embodiment, the preset feature dimensions are set based on the time window, interactive operation, and data transmission method corresponding to the recommended content. Specifically, the statistics of the object profile are to count the time period, network status, and card consumed for each consumption behavior of each object in different time windows, and build the basic data of the profile. In addition, the object features are the features corresponding to the target object, and the set of these features is the object profile. For example, when the target object is the target account, the object features are the relevant features of the recommended content corresponding to the target account, thereby obtaining the account features, and the corresponding object profile can also be called the account profile. Or, when the target object is the target user, the object features are the user's features, and the corresponding object profile can also be called the user profile. The following embodiment is an example of the target object being the target user, and no limitation is made here.

[0142] The time period refers to a further granular division of the time window, thereby creating a user profile for different time periods within that window. For example, if the time window is the most recent month, the time period could include the morning (9-11 am), afternoon (2-5 pm), and evening (7-10 pm) of each day within that month. Based on these time periods, user consumption habits can be statistically analyzed to obtain user preferences in the short term (month), improving the comprehensiveness of the user profile and ensuring the accuracy of subsequent content recommendations.

[0143] Therefore, we can first determine the time period division information indicated by the preset feature dimensions based on the historical data corresponding to the target object; then determine the network status corresponding to the target object when it obtains recommended content within the statistical time period indicated by the time period division information, so as to determine the data transmission method, such as dividing the network status into WIFI, 4G, 3G, 2G, and unknown network, thereby creating profile data under different data transmission methods; and obtain the target object's interaction operations with different types of recommended content under the data transmission methods to determine the object characteristics; then, respectively, statistically analyze the exposure information and association information corresponding to different types of recommended content under the data transmission methods to determine the content characteristics; and finally, integrate the object characteristics and content characteristics to obtain the object profile.

[0144] The description of this object profile takes into account the significant differences in consumption habits of different groups of people in different time periods and network conditions in the business scenario. Therefore, this embodiment has made more detailed statistics based on time periods and network conditions.

[0145] In one possible scenario, time periods are divided as follows: 1. Each day is divided into four segments by hour: early morning (2-5 AM), morning (6-11 AM), afternoon (12 PM-6 PM), and evening (7 PM-1 AM the next day); 2. Each week is divided into weekdays and weekends: weekdays (Monday to Friday, excluding Friday evening) and weekends (Friday evening, Saturday, and Sunday). Since user consumption habits are more specific on Friday evenings, they are also included in the weekends, thereby increasing the salience of features within each time period to facilitate the recommendation and segmentation of content.

[0146] Optionally, the above time period division is based on data sparsity, resulting in four segments per day. However, this division can be more refined. Specifically, one can first determine the interaction frequency of the target audience with the recommended content based on historical data; then extract peak time periods based on this interaction frequency information; and finally, define these peak time periods as time period divisions indicated by preset feature dimensions. This yields time period data with rich interaction data, avoiding data statistics for periods with no data, such as late-night periods, thus ensuring the validity of the data within each time period.

[0147] Optionally, regarding the statistical process for the aforementioned data transmission methods, considering that different WiFi networks correspond to different usage scenarios and therefore different recommended content (e.g., recommended content differs between company WiFi and home WiFi), the data transmission method can be compared with the historical data of the target object to obtain recommendation scenario information (determining whether it is a common scenario, such as a company network). Then, based on the recommendation scenario information, the interaction operations of the target object with different types of recommended content under each data transmission method are obtained to determine object characteristics. Furthermore, based on the recommendation scenario information, the exposure information and association information corresponding to different types of recommended content under each data transmission method are statistically analyzed to determine content characteristics, thereby ensuring the accuracy of features under different network scenarios. For example, whether the WiFi in the network status is a commonly used WiFi network, because some users have different consumption habits on company WiFi and home WiFi, thus ensuring the accuracy of features for that scenario.

[0148] For example, an object profile is the basic data used to build a profile for each user by statistically analyzing the time period, network status, and cards used for each consumption behavior within a time window. For instance, a partial monthly profile of user A might include:

[0149] Network status: WIFI.

[0150] Time of day: Weekend evening.

[0151] Exposure (number): 30 images and text, 10 short videos, 20 short videos, 5 social media posts, and 15 short content posts.

[0152] Clicks (number): Text and images 5, short videos 3, short videos 6, social media 0, short content 8.

[0153] Duration (seconds): Text and images 200, short videos 321, short videos 680, social media 0, short content 220.

[0154] Driven by (number): Text and images 5, short videos 26, short videos 12, social media 0, short content 31.

[0155] Among them, object features include user preferences, while content profiles include the number of exposures, clicks, click-through rates, number of interactions, and consumption duration corresponding to the recommended content. The extraction process of each of the above feature dimensions is explained below.

[0156] Specifically, for impressions, which refers to the number of times a card is shown, to ensure the validity of the impression count, it can be set that the card must show 50% of its area to be considered an exposure. Additionally, for clicks, which refers to the number of times a card is clicked, it can be set that the card's duration of use must be greater than or equal to 6 seconds to be considered a valid click, thus filtering out invalid clicks such as clickbait titles.

[0157] Additionally, while click-through rate (CTR) can be directly calculated by dividing clicks by impressions, the CTR obtained by directly dividing clicks by impressions is unreliable for some users whose cards receive fewer impressions. Therefore, Wilson Score can be used for calculation, as shown in the following formula:

[0158]

[0159]

[0160] Where pos represents the number of clicks; n represents the number of impressions; and z represents the confidence parameter, which can take the value 2, i.e., 95% confidence level.

[0161] Click preferences in the user profile, which characterize a user's preference for a certain type of card, are calculated using the Wilson Score formula mentioned above, since some users may have a high or low click rate for all cards. In this formula, pos represents the number of clicks for a certain type of card, and n represents the total number of clicks.

[0162] The driving number is the click conversion corresponding to the card. However, considering that some cards have a floating layer, which jumps to related content, the driving number is the number of clicks for text and image cards without a floating layer; while for video cards with a floating layer, the driving number is the total number of videos consumed in the floating layer.

[0163] For consumption duration, it refers to the total consumption time of the user on each type of card; for video cards with a floating layer, it includes the cumulative consumption time within the floating layer.

[0164] The single-click impact factor, calculated by dividing the number of impacts by the number of clicks, is used to characterize the advertising revenue that a single user click can bring, because advertising revenue is positively correlated with the number of times the user scrolls up in the overlay.

[0165] The single-click consumption time is calculated by dividing the consumption time by the number of clicks, and is used to depict the time-based revenue that a user can generate from a single click.

[0166] Optionally, the above description outlines the process of creating an object profile. However, for users with limited data, the description may be inaccurate. To avoid the impact of this data on the content recommendation process, a user profile can be used instead. This involves first obtaining the user segmentation dimensions, such as age, gender, or activity level; then, performing statistical analysis of interaction data under different dimensions based on the user segmentation dimensions to determine the user profile; and finally, determining the confidence parameter corresponding to the target object. If the confidence parameter indicates that the target object is not confident, the object profile is replaced with a user profile.

[0167] In one possible scenario, when object profiling fails or the strategy determines that the profiling data is unreliable, audience profiling can be used instead of object profiling. Audience profiling uses the same statistical dimensions as object profiling, except that it doesn't track the consumption behavior of individual users, but rather segments all users into different audiences based on various dimensions, and then performs statistics based on these audience segments. For example, the segmentation dimensions might be as follows:

[0168] Age: 0-6 years, 7-12 years, 13-15 years, 16-18 years, 19-22 years, 23-25 ​​years, 26-30 years, 31-40 years, 41-50 years, and over 50 years.

[0169] Gender: Male, Female, Unknown.

[0170] Activity level: Divided into four categories based on the number of clicks a user has made in the last 30 days: no exposure, low activity, medium activity, and high activity.

[0171] By using alternative settings for audience profiles, the validity of the profile data for each individual is ensured.

[0172] 303. Evaluate multiple features in the object profile based on the multi-objective model to obtain a set of interactive feature values.

[0173] In this embodiment, a multi-objective model is one with two or more objective functions, for example... Figure 5 The scene shown, Figure 5This is a schematic diagram of another method for managing recommended content provided in this application embodiment. The diagram shows that the feature layer extracts object features, content features, and context features (floating layer relationships) and inputs them into a multi-objective model to make click prediction values ​​and duration prediction values. That is, the multi-objective model includes click tasks and duration tasks, which are scored separately for recommendation reference.

[0174] Furthermore, after calculating the predicted click and duration values, it's necessary to balance revenue, i.e., to obtain more time and advertising revenue. Because of the layered nature of the cards, the rules for setting revenue differ, specifically:

[0175] (1) Regarding single recommended content.

[0176] In this embodiment, a single recommended content is a recommended content (card) that does not have a floating layer or jump relationship. That is, the time revenue corresponding to the recommended content and the additional advertising collection can be calculated directly.

[0177] (2) Recommended content regarding the existence of floating layers.

[0178] In this embodiment, the main application scenario is the information flow service of the content push application. For heterogeneous cards (text, short video, and short video cards), after clicking on the short video card, users can enter an immersive video overlay scene, where they can scroll up infinitely to consume more videos. Moreover, an advertisement will be inserted after consuming a few videos. However, when clicking on the text card, it is just a text and image details page with an advertisement, which cannot drive more consumption.

[0179] Therefore, in this business scenario, text / image and video cards have inherently significant differences in duration and advertising revenue due to their product formats, specifically as follows: Figure 6 As shown, Figure 6 This is a schematic diagram illustrating another scenario of a recommended content management method provided in this application embodiment. The diagram shows the distribution of the duration (in seconds) of a user's single click on a card and the number of videos consumed, with the horizontal axis representing the quantiles. The diagram shows that a heavy video user can consume nearly one hour with a single click, driving consumption of over 100 videos, almost 10 times the consumption duration of a single click by a heavy text / image user. Furthermore, Figure 7 This is a schematic diagram illustrating another method for managing recommended content provided in this application. The diagram shows that advertising revenue, in terms of ad impressions, is dozens of times higher than that of a single click on a text / image card. Therefore, when considering multi-objective optimization of clicks, duration, and advertising revenue, a fusion strategy must be designed to address the consumption differences within the floating layer, as detailed in step 304.

[0180] In one possible scenario, such as Figure 8 As shown, Figure 8This is a schematic diagram of another method for managing recommended content provided in this application embodiment. The diagram shows that after clicking on recommended content B1, one can enter floating content B2. By sliding or clicking on the interface corresponding to floating content B2, one can adjust to more different floating content, which are related content. Thus, the content features of these related content are all included in recommended content B1 to improve the relevance between content.

[0181] 304. Call the data corresponding to the operation items in the content features to perform cumulative revenue statistics in order to determine the revenue feature value corresponding to the recommended content.

[0182] In this embodiment, the statistical process for cumulative revenue means that for a single recommended content, only its own time conversion and advertising revenue are considered, while for overlay content, related content needs to be considered.

[0183] Specifically, to determine the revenue feature value of floating layer content, we can identify the recommended content corresponding to the operation item in the content feature, and determine the set of related content of the recommended content corresponding to the operation item; then, based on the set of related content, we can perform cumulative revenue statistics to determine the revenue feature value corresponding to the recommended content, thereby ensuring the accuracy of the revenue description of floating layer content.

[0184] Optionally, to ensure the effectiveness of the revenue description, when a user has no clicks on a given day, priority is given to ensuring that clicks are only considered based on the click task score; after a user clicks on a given day, further consideration is given to obtaining more time and advertising revenue to determine the click information corresponding to the associated content set; specifically, if the click information meets the preset conditions (e.g., the number of clicks is greater than 5), then the cumulative revenue is statistically analyzed based on the associated content set to determine the revenue feature value corresponding to the recommended content; and when the click information does not meet the preset conditions, then the predicted value corresponding to the click information is used as the target feature value.

[0185] This embodiment targets heterogeneous cards with significant differences in duration and advertising revenue. It combines a deep learning model and revenue prediction strategy to perform multi-objective ranking, thereby improving click, duration, and advertising revenue metrics and balancing the expectations among multiple objectives.

[0186] Optionally, after the multi-objective model outputs the predicted values ​​of user clicks and duration for each candidate item, if the user has no clicks on the day, in order to prioritize clicks, only the click task score is considered. After the user has clicks on the day, more duration and advertising revenue are then considered.

[0187] Understandably, regarding the difference in revenue from overlay cards, the fusion strategy considers not only the predicted click and duration tasks for each candidate item output by the multi-objective model, but also the duration and ad revenue generated by a single click calculated from the card profile. Because ad revenue is positively correlated with the number of swipes in the overlay, and the eCPM of the ads to be consumed in the overlay is unknown, the driving force of a single click is used to approximate the estimated ad revenue. The final fusion score is only used for ranking, so this approximation is comparable, and there's no need to consider differences in absolute values.

[0188] Optionally, a separate model can be trained to predict the expected revenue of a user's single click to enter the floating layer, thereby improving the efficiency of obtaining revenue feature values.

[0189] 305. Calculate the target feature value by combining the predicted values ​​contained in the set of return feature values ​​and interaction feature values.

[0190] In this embodiment, the target feature value can be calculated by weighting the predicted values ​​contained in the set of benefit feature values ​​and interaction feature values.

[0191] Optionally, since different card contents have different focuses, targeted designs can be made. That is, firstly, the content type corresponding to the recommended content is determined; then, the corresponding feature weighting coefficient is determined based on the content type; then, the predicted values ​​contained in the revenue feature value and interaction feature value set are weighted according to the feature weighting coefficient to obtain the target feature value, thereby improving the adaptability of the target feature value for different forms of content recommendation.

[0192] 306. Determine the target content to be pushed in the recommendation set based on the target feature value, and distribute the target content in the recommendation interface for display.

[0193] In this embodiment, the recommended content in the recommendation set can be sorted based on the target feature value to obtain the target content at the top, and the target content can be distributed and displayed in the recommendation interface. Specifically, it can be displayed in the center so that it is easy for users to notice.

[0194] As described in the above embodiments, by acquiring the interaction operations of the target object with different forms of recommended content in the recommendation set within multiple time windows, the interaction data corresponding to the target object is determined. Then, based on preset feature dimensions, the object features and content features in the interaction data are statistically analyzed to obtain an object profile. These preset feature dimensions are set based on time windows, interaction operations, and the data transmission method corresponding to the recommended content. Multiple features in the object profile are evaluated according to a multi-objective model to obtain a set of interaction feature values. Further, the data corresponding to the operation items in the content features are called to perform cumulative revenue statistics to determine the revenue feature value corresponding to the recommended content. Then, the revenue feature value and the predicted value contained in the interaction feature value set are combined to calculate the target feature value. Finally, the target content for push in the recommendation set is determined based on the target feature value, and the target content is distributed and displayed on the recommendation interface. This achieves a balance between the conversion and revenue of recommended content during the push process. By using multiple time windows and considering feature dimensions in conjunction with data transmission methods, the completeness of the object profile is ensured, the simulation degree of fluctuations in user interest content is improved, and the accuracy of recommended content push is enhanced.

[0195] The following is combined Figure 9 This section explains how to recommend content based on a trained multi-objective model. The recommended content can be media objects, which may include videos, images, and heterogeneous cards, etc. See also... Figure 9 , Figure 9 This is a schematic diagram illustrating another method for managing recommended content provided in this application. The multi-objective model includes a feature mapping layer, a feature extraction layer, a feature concatenation layer, and a prediction layer. In practical applications, when recommending media objects based on the multi-objective model, the server obtains the user data and content data of the media object to be recommended, and then inputs the user data and content data into the multi-objective model.

[0196] Multi-objective models use feature mapping layers to process user data and content data separately, obtaining feature vectors for the corresponding user data and content data. Specifically, this can be achieved through one-hot encoding or by using a pre-trained feature mapping model.

[0197] After obtaining the feature vectors of the corresponding user data and content data, a feature extraction layer is used to extract features from these vectors, resulting in the feature vector of the media object to be recommended. Specifically, the feature extraction layer consists of a wide layer, a DNN layer, and a shared NFM layer. Therefore, the DNN layer can perform implicit feature crossing on the feature vectors of the user data and content data to extract high-order feature vectors; the NFM layer can perform explicit feature crossing on the feature vectors of the user data and content data and sum them to obtain a multi-dimensional feature vector; and the wide layer can linearly sum the feature vectors of the user data and content data based on weights to output a feature vector with reduced dimensionality.

[0198] After obtaining the feature vectors of the media objects to be recommended, a feature concatenation layer concatenates the vectors to obtain a concatenated vector. Based on this concatenated vector, a prediction layer predicts the interaction features, resulting in the feature prediction results for the corresponding media objects to be recommended. Specifically, the prediction layer can be an artificial neural network model that uses activation functions to predict interaction features and obtain the feature prediction results. This prediction layer can be either regression prediction or classification prediction. When the prediction layer is a regression prediction, it uses a first activation function (e.g., a regression function) to perform regression processing and predict the feature prediction results for each interaction feature. When the prediction layer is a classification prediction, it uses a second activation function (e.g., a softmax classification function) to perform classification processing and predict the feature prediction results for each interaction feature.

[0199] Therefore, based on the feature prediction results output by the multi-objective model, recommendations are made for the media objects to be recommended. Specifically, the feature prediction results can be the estimated click-through rate, so the media objects to be recommended can be recommended based on the magnitude of the estimated click-through rate.

[0200] The following describes an exemplary application of the embodiments of the present invention in a practical application scenario.

[0201] In the application of the multi-objective model, the object features and content features in the object profile depict the user's preference for various heterogeneous cards (text and images, short videos, short content, social, text and images followed, video followed) in this business at different times and under different network conditions. These features serve as features of the model layer and are used to predict card revenue in the multi-objective layer.

[0202] Most of them are based on user preferences for the content itself (categories, tags, items), without detailed descriptions of user preferences for different card types at different times and in different contexts. In this business scenario, the consumption habits of different groups of people vary greatly. For example, although they are all heavy video users, some users only consume text and image cards during the day on weekdays and only consume video cards in the evening or on weekends.

[0203] Multi-objective optimization is based on the revenue of a single item. However, in the scenario presented in this paper, some cards have floating layer consumption. This means that the revenue from a single user click is not only the revenue from watching the content clicked, but also the revenue from watching different content after clicking into the floating layer. Therefore, the multi-objective optimization effect is poor.

[0204] This embodiment constructs an object profile that depicts the degree of user preference for various heterogeneous cards (text, short video, short content, social, followed text, followed video) in this service at different times and under different network conditions. This profile serves as a feature of the mixed-ranking model and is used for multi-objective revenue prediction, thereby improving the accuracy of the multi-objective model and user experience, and solving the problem of inaccurate characterization of card preferences in different contexts.

[0205] This embodiment addresses the challenge of optimizing heterogeneous cards with significant differences in duration and advertising revenue by designing a deep learning model and revenue prediction strategy. This multi-objective optimization improves click-through rate, duration, and advertising revenue metrics, thus resolving the difficulty of optimizing multi-objective metrics for heterogeneous cards with vastly different revenues.

[0206] The following explanation uses a revenue assessment scenario as an example. Please refer to [link / reference]. Figure 10 , Figure 10 A flowchart illustrating another recommended content management method provided in this application embodiment, which includes at least the following steps:

[0207] 1001. Did the user click on it that day?

[0208] In this embodiment, after the multi-objective model outputs the predicted values ​​of user clicks and duration for each candidate item, if the user does not click on any item that day, in order to prioritize ensuring that clicks are only considered based on the click task score, and only after the user clicks on any item that day, will the model consider obtaining more duration and advertising revenue, thereby improving the credibility of the revenue feature value.

[0209] 1002. Calculate the feature value corresponding to the click.

[0210] In this embodiment, if there are no clicks, it means that the content has not yet generated revenue. The number of clicks can be retained, and the target feature value can be characterized based on the predicted value corresponding to the number of clicks, thereby improving the credibility of the target feature value.

[0211] 1003. Estimate revenue based on the target profile.

[0212] In this embodiment, in the process of estimating revenue, to address the issue of unreliability in object profile data, when a user is an extremely inactive user or has very few exposures on a certain type of card, a crowd profile is used instead of an individual profile.

[0213] 1004. Total number of clicks on the object profile ≤ 5.

[0214] 1005. The card data used in the object profiling adopts the crowd profiling data.

[0215] In this embodiment, the total number of clicks indicated in the object profile is first extracted; if the total number of clicks does not meet the click condition (total number of clicks ≤ 5), the audience profile is called to replace the recommended content for all types in the object profile, and the specific value depends on the actual scenario.

[0216] 1006. Determine the confidence level of each card for the user.

[0217] In this embodiment, since the push logic for different types of cards (recommended content) is different, it is necessary to determine the confidence level of each type of card separately.

[0218] 1007. The number of exposures of the object portrait card is ≤3.

[0219] 1008. Object Profile: This category of card data uses audience profile data.

[0220] In this embodiment, the recommended content under each type indicated in the object profile is first determined; then the recommended content under each type is filtered based on the exposure conditions, so that the recommended content under the type that does not meet the exposure conditions (exposure number ≤ 3) (such as text and images) is replaced with the audience profile, and the specific value depends on the actual scenario.

[0221] 1009. Calculate the target feature value.

[0222] In this embodiment, the target feature values ​​are weighted and calculated based on the data after confidence filtering, thereby constructing an object profile that describes the user's preference for various heterogeneous cards (text, images, short videos, short content, social, followed text, followed videos) in this business at different times and under different network conditions. This profile serves as a feature of the mixed-ranking model and is used for multi-objective revenue prediction, improving the accuracy of the multi-objective model and user experience, and solving the problem of inaccurate characterization of card preferences in different contexts.

[0223] In addition, for heterogeneous cards with huge differences in duration and advertising revenue, a deep learning model and revenue prediction strategy were designed to perform multi-objective optimization, improve click, duration and advertising revenue indicators, and solve the problem of multi-objective optimization of heterogeneous cards with huge revenue differences.

[0224] In one possible scenario, using this embodiment for content push significantly improves key metrics such as user click DAU, consumption time, and advertising revenue.

[0225] In another possible scenario, the interface distribution for push content can be targeted, such as... Figure 11 As shown, Figure 11 A flowchart illustrating another recommended content management method provided in this application embodiment, which includes at least the following steps:

[0226] 1101. Determine the target feature values ​​corresponding to the recommended content in the recommendation set.

[0227] In this embodiment, the calculation of the target feature value corresponding to the recommended content is described in [reference needed]. Figure 3 The illustrated embodiment will not be described in detail here.

[0228] 1102. Update the target feature values ​​based on the target application type indicated by the recommendation interface to obtain the recommended feature values.

[0229] In this embodiment, since different target application types have preferences for different recommended content, such as short video applications focusing on pushing short videos and news applications focusing on pushing text and images, the target feature values ​​can be weighted and updated to obtain the recommended feature values, so as to ensure the adaptability of the recommended feature values ​​to the application scenario.

[0230] 1103. Based on the focus area corresponding to the target application type, fill in the content according to the recommended feature value to push the recommended content.

[0231] In this embodiment, the target content is displayed in the focus area of ​​the terminal application interface. This focus area can be the center, edge, or other location of the interface. The specific location can be set according to the hotspot area corresponding to the target application type. For example, if the center of a short video application is the focus area, the content with the highest recommendation feature value will be filled into this area to push the recommended content, thereby ensuring the adaptability of the content and the application and further improving the accuracy of the recommended content.

[0232] To better implement the above-described solutions of the embodiments of this application, related apparatus for implementing the above solutions is also provided below. Please refer to... Figure 12 , Figure 12 This application provides a schematic diagram of the structure of a management device for recommended content, the management device 1200 including:

[0233] The acquisition unit 1201 is used to acquire the interaction operations of the target object with different forms of recommendation content in the recommendation set within multiple time windows, so as to determine the interaction data corresponding to the target object;

[0234] The statistical unit 1202 is used to perform statistics on the object features and content features in the interaction data based on preset feature dimensions to obtain an object profile. The preset feature dimensions are set based on the time window, the interaction operation, and the data transmission method corresponding to the recommended content.

[0235] Evaluation unit 1203 is used to evaluate multiple features in the object profile according to the multi-objective model to obtain a set of interactive feature values;

[0236] The statistical unit 1202 is also used to call the data corresponding to the operation items in the content features to perform cumulative revenue statistics, so as to determine the revenue feature value corresponding to the recommended content;

[0237] The management unit 1204 is used to calculate the target feature value by combining the revenue feature value and the predicted value contained in the set of interaction feature values.

[0238] The management unit 1204 is used to determine the target content to be pushed in the recommendation set according to the target feature value, and to distribute the target content in the recommendation interface for display.

[0239] Optionally, in some possible implementations of this application, the acquisition unit 1201 is specifically used to acquire the recommendation period corresponding to the target object;

[0240] The acquisition unit 1201 is specifically used to perform gradient configuration based on the recommendation period to determine multiple time windows;

[0241] The acquisition unit 1201 is specifically used to acquire the interaction operations of the recommended content in the recommendation set within multiple time windows, so as to determine the interaction data corresponding to the target object.

[0242] Optionally, in some possible implementations of this application, the statistical unit 1202 is specifically used to determine the time period division information indicated by the preset feature dimension based on the historical data corresponding to the target object;

[0243] The statistical unit 1202 is specifically used to determine the network status when the target object obtains the recommended content within the statistical period indicated by the time period division information, so as to determine the data transmission method.

[0244] The statistical unit 1202 is specifically used to acquire the interaction operations of the target object with different types of recommended content under the data transmission mode, so as to determine the characteristics of the object;

[0245] The statistical unit 1202 is specifically used to statistically analyze the exposure information and association information corresponding to the different types of recommended content under the data transmission method, so as to determine the content characteristics.

[0246] The statistical unit 1202 is specifically used to integrate the object features and the content features to obtain the object profile.

[0247] Optionally, in some possible implementations of this application, the statistical unit 1202 is specifically used to compare the data transmission method with the data transmission method in the historical data corresponding to the target object to obtain recommended scenario information;

[0248] The statistical unit 1202 is specifically used to obtain the interaction operations of the target object with different types of recommended content under the data transmission method based on the recommendation scenario information, so as to determine the object characteristics;

[0249] The statistical unit 1202 is specifically used to statistically analyze the exposure information and association information corresponding to different types of recommended content under the data transmission method based on the recommended scenario information, so as to determine the content features.

[0250] Optionally, in some possible implementations of this application, the statistical unit 1202 is specifically used to determine the interaction frequency information of the target object with respect to the recommended content based on the historical data corresponding to the target object;

[0251] The statistical unit 1202 is specifically used to extract hotspot time periods based on the interaction frequency information;

[0252] The statistical unit 1202 is specifically used to determine the hotspot period as the time period division information indicated by the preset feature dimension.

[0253] Optionally, in some possible implementations of this application, the statistical unit 1202 is specifically used to obtain the population segmentation dimension, which includes age, gender, or activity level;

[0254] The statistical unit 1202 is specifically used to perform interactive data statistics under different dimensions based on the population segmentation dimension in order to determine the population profile.

[0255] The statistical unit 1202 is specifically used to determine the confidence parameters corresponding to the target object;

[0256] The statistical unit 1202 is specifically used to replace the object profile with the crowd profile if the confidence parameter indicates that the target object is not confident.

[0257] Optionally, in some possible implementations of this application, the statistical unit 1202 is specifically used to determine the recommended content corresponding to the operation item in the content features, and to determine the set of associated content of the recommended content corresponding to the operation item;

[0258] The statistical unit 1202 is specifically used to perform cumulative revenue statistics based on the associated content set, so as to determine the revenue feature value corresponding to the recommended content.

[0259] Optionally, in some possible implementations of this application, the management unit 1204 is specifically used to determine the click information corresponding to the associated content set;

[0260] The management unit 1204 is specifically used to perform cumulative revenue statistics based on the associated content set if the click information reaches a preset condition, so as to determine the revenue feature value corresponding to the recommended content.

[0261] or;

[0262] The management unit 1204 is specifically used to take the predicted value corresponding to the click information as the target feature value if the click information does not meet the preset conditions.

[0263] Optionally, in some possible implementations of this application, the management unit 1204 is specifically used to extract the total number of clicks indicated in the object profile;

[0264] The management unit 1204 is specifically used to replace the recommended content for all types in the object profile if the total number of clicks does not meet the click conditions.

[0265] Optionally, in some possible implementations of this application, the management unit 1204 is specifically used to determine the recommended content under each type indicated in the object profile;

[0266] The management unit 1204 is specifically used to filter the recommended content under each type based on exposure conditions, so as to replace the recommended content under the type that does not meet the exposure conditions with the audience profile.

[0267] Optionally, in some possible implementations of this application, the management unit 1204 is specifically used to determine the content type corresponding to the recommended content;

[0268] The management unit 1204 is specifically used to determine the corresponding feature weighting coefficient based on the content type;

[0269] The management unit 1204 is specifically used to perform weighted calculations on the revenue feature value and the predicted value contained in the interaction feature value set according to the feature weighting coefficient, so as to obtain the target feature value.

[0270] By acquiring the interaction operations of the target object with different forms of recommended content in the recommendation set within multiple time windows, the interaction data corresponding to the target object is determined. Then, based on preset feature dimensions, the object features and content features in the interaction data are statistically analyzed to obtain an object profile. These preset feature dimensions are set based on time windows, interaction operations, and the data transmission method corresponding to the recommended content. Multiple features in the object profile are evaluated according to a multi-objective model to obtain a set of interaction feature values. Further, the data corresponding to the operation items in the content features are used to calculate the cumulative revenue to determine the revenue feature value corresponding to the recommended content. Then, the revenue feature value and the predicted values ​​contained in the interaction feature value set are combined to calculate the target feature value. Finally, the target content for push notifications in the recommendation set is determined based on the target feature value and displayed on the recommendation interface. This achieves a balance between the conversion and revenue of recommended content during the push process. By using multiple time windows and considering feature dimensions in conjunction with data transmission methods, the completeness of the object profile is ensured, the simulation of fluctuations in user interest content is improved, and the accuracy of recommended content push is enhanced.

[0271] This application also provides a terminal device, such as... Figure 13 The diagram shown is a structural schematic of another terminal device provided in an embodiment of this application. For ease of explanation, only the parts related to the embodiment of this application are shown. For specific technical details not disclosed, please refer to the method section of the embodiment of this application. The terminal can be any terminal device including mobile phones, tablets, personal digital assistants (PDAs), point-of-sale (POS) terminals, in-vehicle computers, etc. Taking a mobile phone as an example:

[0272] Figure 13 This is a block diagram illustrating a portion of the structure of a mobile phone related to the terminal provided in the embodiments of this application. (Reference) Figure 13 The mobile phone includes components such as a radio frequency (RF) circuit 1310, a memory 1320, an input unit 1330, a display unit 1340, a sensor 1350, an audio circuit 1360, a wireless fidelity (WiFi) module 1370, a processor 1380, and a power supply 1390. Those skilled in the art will understand that... Figure 13 The mobile phone structure shown does not constitute a limitation on the mobile phone and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0273] The following is combined Figure 13 A detailed introduction to each component of a mobile phone:

[0274] RF circuit 1310 can be used for receiving and transmitting signals during information transmission or calls. Specifically, it receives downlink information from the base station and processes it with processor 1380; additionally, it transmits uplink data to the base station. Typically, RF circuit 1310 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low-noise amplifier (LNA), a duplexer, etc. Furthermore, RF circuit 1310 can also communicate wirelessly with networks and other devices. The aforementioned wireless communication can use any communication standard or protocol, including but not limited to Global System for Mobile Communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Message Service (SMS), etc.

[0275] The memory 1320 can be used to store software programs and modules. The processor 1380 executes various mobile phone functions and data processing by running the software programs and modules stored in the memory 1320. The memory 1320 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 1320 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.

[0276] The input unit 1330 can be used to receive input numerical or character information, and generate key signal inputs related to user settings and function control of the mobile phone. Specifically, the input unit 1330 may include a touch panel 1331 and other input devices 1332. The touch panel 1331, also known as a touch screen, can collect touch operations performed by the user on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or near the touch panel 1331, and air touch operations within a certain range on the touch panel 1331), and drive the corresponding connection devices according to a pre-set program. Optionally, the touch panel 1331 may include two parts: a touch detection device and a touch controller. The touch detection device detects the user's touch position and the signal generated by the touch operation, and transmits the signal to the touch controller; the touch controller receives touch information from the touch detection device, converts it into touch point coordinates, sends it to the processor 1380, and can receive and execute commands sent by the processor 1380. Furthermore, the touch panel 1331 can be implemented using various types of sensors, including resistive, capacitive, infrared, and surface acoustic wave sensors. In addition to the touch panel 1331, the input unit 1330 may also include other input devices 1332. Specifically, these other input devices 1332 may include, but are not limited to, one or more of the following: a physical keyboard, function keys (such as volume control buttons, power buttons, etc.), a trackball, a mouse, and a joystick.

[0277] The display unit 1340 can be used to display information input by the user or information provided to the user, as well as various menus of the mobile phone. The display unit 1340 may include a display panel 1341, which may optionally be configured as a liquid crystal display (LCD), organic light-emitting diode (OLED), or similar form. Further, a touch panel 1331 may cover the display panel 1341. When the touch panel 1331 detects a touch operation on or near it, it transmits the information to the processor 1380 to determine the type of touch event. Subsequently, the processor 1380 provides corresponding visual output on the display panel 1341 according to the type of touch event. Although in Figure 13 In this embodiment, the touch panel 1331 and the display panel 1341 are two separate components to realize the input and output functions of the mobile phone. However, in some embodiments, the touch panel 1331 and the display panel 1341 can be integrated to realize the input and output functions of the mobile phone.

[0278] The mobile phone may also include at least one sensor 1350, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor. The ambient light sensor can adjust the brightness of the display panel 1341 according to the ambient light level, and the proximity sensor can turn off the display panel 1341 and / or the backlight when the phone is moved to the ear. As a type of motion sensor, an accelerometer sensor can detect the magnitude of acceleration in various directions (generally three axes). When stationary, it can detect the magnitude and direction of gravity and can be used for applications that recognize the phone's posture (such as landscape / portrait switching, related games, magnetometer posture calibration), vibration recognition-related functions (such as pedometer, taps), etc. Other sensors that may be configured in the mobile phone, such as gyroscopes, barometers, hygrometers, thermometers, and infrared sensors, will not be described in detail here.

[0279] Audio circuit 1360, speaker 1361, and microphone 1362 provide an audio interface between the user and the mobile phone. Audio circuit 1360 converts received audio data into electrical signals and transmits them to speaker 1361, where speaker 1361 converts them into sound signals for output. On the other hand, microphone 1362 converts collected sound signals into electrical signals, which are received by audio circuit 1360, converted into audio data, and then processed by processor 1380 before being transmitted via RF circuit 1310 to, for example, another mobile phone, or the audio data can be output to memory 1320 for further processing.

[0280] WiFi is a short-range wireless transmission technology. Mobile phones, through the WiFi module 1370, can help users send and receive emails, browse web pages, and access streaming media, providing users with wireless broadband internet access. Although Figure 13 WiFi module 1370 is shown, but it is understood that it is not an essential component of a mobile phone and can be omitted as needed without changing the essence of the invention.

[0281] The processor 1380 is the control center of the mobile phone, connecting various parts of the phone through various interfaces and lines. It executes various functions and processes data by running or executing software programs and / or modules stored in the memory 1320, and by calling data stored in the memory 1320. Optionally, the processor 1380 may include one or more processing units; optionally, the processor 1380 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the aforementioned modem processor may also not be integrated into the processor 1380.

[0282] The mobile phone also includes a power supply 1390 (such as a battery) that supplies power to various components. Optionally, the power supply can be logically connected to the processor 1380 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system.

[0283] Although not shown, mobile phones may also include a camera, Bluetooth module, etc., which will not be described in detail here.

[0284] In this embodiment of the application, the processor 1380 included in the terminal also has the function of performing the various steps of the page processing method described above.

[0285] This application also provides a server; please refer to [link / reference]. Figure 14 , Figure 14 This is a schematic diagram of a server structure provided in an embodiment of this application. The server 1400 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 1422 (e.g., one or more processors) and memory 1432, and one or more storage media 1430 (e.g., one or more mass storage devices) for storing application programs 1442 or data 1444. The memory 1432 and storage media 1430 can be temporary or persistent storage. The program stored in the storage media 1430 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the server. Furthermore, the CPU 1422 may be configured to communicate with the storage media 1430 and execute the series of instruction operations in the storage media 1430 on the server 1400.

[0286] Server 1400 may also include one or more power supplies 1426, one or more wired or wireless network interfaces 1450, one or more input / output interfaces 1458, and / or one or more operating systems 1441, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0287] The steps performed by the management device in the above embodiments can be based on this Figure 14 The server structure shown.

[0288] This application also provides a computer-readable storage medium storing management instructions for recommended content, which, when executed on a computer, cause the computer to perform the aforementioned actions. Figures 2 to 11 The steps performed by the management device for recommended content in the method described in the illustrated embodiment.

[0289] This application also provides a computer program product that includes management instructions for recommended content. When run on a computer, it causes the computer to perform the aforementioned actions. Figures 2 to 11 The steps performed by the management device for recommended content in the method described in the illustrated embodiment.

[0290] This application embodiment also provides a recommended content management system, which may include... Figure 12 The management device for recommended content in the described embodiments, or Figure 13 The terminal device in the described embodiments, or Figure 14 The server described.

[0291] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0292] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0293] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0294] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0295] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a content management device, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0296] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for managing recommended content, characterized in that, include: The interaction operations of the target object with different forms of recommended content in the recommendation set are obtained within multiple time windows to determine the interaction data corresponding to the target object; the recommended content includes heterogeneous cards. Based on preset feature dimensions, the object features, content features, and context features in the interaction data are statistically analyzed to obtain an object profile. The preset feature dimensions are set based on the time window, the interaction operation, and the data transmission method corresponding to the recommended content. The context features are used to characterize the floating layer relationship of the recommended content. The multiple features in the object profile are evaluated based on a multi-objective model to obtain a set of interactive feature values; Determine the recommended content corresponding to the operation item in the content features, and determine the set of associated content of the recommended content corresponding to the operation item; Based on the set of related content, the cumulative revenue is statistically analyzed to determine the revenue characteristic value corresponding to the recommended content; The associated content in the associated content set includes the overlay content of the recommended content corresponding to the operation item; the revenue feature value corresponding to the recommended content includes the revenue feature value of the overlay content. Determine the content type corresponding to the recommended content; Determine the corresponding feature weighting coefficients based on the content type; The target feature value is obtained by weighting the predicted values ​​contained in the set of interaction feature values ​​according to the feature weighting coefficients. The target content for push is determined in the recommendation set based on the target feature value, and the target content is distributed and displayed in the recommendation interface.

2. The method according to claim 1, characterized in that, The step of obtaining the target object's interaction operations with different forms of recommendation content in the recommendation set within multiple time windows to determine the interaction data corresponding to the target object includes: Obtain the recommendation period corresponding to the target object; Gradient configuration is performed based on the recommendation period to determine multiple time windows; The interaction operations of the recommended content in the recommendation set within multiple time windows are obtained to determine the interaction data corresponding to the target object.

3. The method according to claim 1, characterized in that, The step of statistically analyzing the object features and content features in the interaction data based on preset feature dimensions to obtain an object profile includes: The time period division information indicated by the preset feature dimension is determined based on the historical data corresponding to the target object; Determine the network state corresponding to when the target object obtains the recommended content within the statistical time period indicated by the time period segmentation information, so as to determine the data transmission method; The interaction operations of the target object with different types of recommended content under the data transmission method are obtained to determine the characteristics of the object; The exposure information and association information corresponding to the different types of recommended content under the aforementioned data transmission method are statistically analyzed to determine the content characteristics. The object features and the content features are integrated to obtain the object profile.

4. The method according to claim 3, characterized in that, The step of obtaining the target object's interaction operations with different types of recommended content under the data transmission method to determine the object's characteristics includes: The data transmission method is compared with the data transmission method in the historical data corresponding to the target object to obtain recommended scenario information; Based on the recommended scenario information, the interactive operations of the target object with different types of recommended content under the data transmission method are obtained respectively to determine the object characteristics; The step of statistically analyzing the exposure information and association information corresponding to different types of recommended content under the data transmission method to determine the content characteristics includes: Based on the recommended scenario information, the exposure information and association information corresponding to the different types of recommended content under the data transmission method are statistically analyzed to determine the content characteristics.

5. The method according to claim 3, characterized in that, The step of determining the time period division information indicated by the preset feature dimension based on the historical data corresponding to the target object includes: The interaction frequency information of the target object with the recommended content is determined based on the historical data corresponding to the target object; Hotspot time periods are extracted based on the interaction frequency information; The hotspot time period is determined as the time period division information indicated by the preset feature dimension.

6. The method according to claim 3, characterized in that, The method further includes: Obtain the audience segmentation dimensions, which include age, gender, or activity level; Based on the aforementioned audience segmentation dimensions, interactive data statistics are performed across different dimensions to determine the audience profile; Determine the confidence parameters corresponding to the target object; If the confidence parameter indicates that the target object is not confident, then the object profile is replaced with the crowd profile.

7. The method according to claim 1, characterized in that, The step of calculating cumulative revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content includes: Determine the click information corresponding to the set of associated content; If the click information reaches the preset conditions, the cumulative revenue is statistically analyzed based on the set of associated content to determine the revenue feature value corresponding to the recommended content; or; If the click information does not meet the preset conditions, the predicted value corresponding to the click information will be used as the target feature value.

8. The method according to claim 7, characterized in that, The method further includes: Extract the total number of clicks indicated in the object profile; If the total number of clicks does not meet the click criteria, the audience profile is invoked to replace the recommended content for all types in the object profile.

9. The method according to claim 8, characterized in that, The method further includes: Determine the recommended content for each type indicated in the object profile; Recommended content under each type is filtered based on exposure conditions, and recommended content under types that do not meet the exposure conditions is replaced with the user profile.

10. The method according to claim 1, characterized in that, The content of the heterogeneous cards includes text and images, video content, or links. The recommendation interface is a terminal application interface, and the target content is displayed in the focus area of ​​the terminal application interface.

11. A device for managing recommended content, characterized in that, include: The acquisition unit is used to acquire the interaction operations of the target object with different forms of recommended content in the recommendation set within multiple time windows, so as to determine the interaction data corresponding to the target object; the recommended content includes heterogeneous cards. The statistical unit is used to perform statistics on object features, content features, and context features in the interaction data based on preset feature dimensions to obtain an object profile. The preset feature dimensions are set based on the time window, the interaction operation, and the data transmission method corresponding to the recommended content. The context features are used to characterize the floating layer relationship of the recommended content. An evaluation unit is used to evaluate multiple features in the object profile according to a multi-objective model to obtain a set of interactive feature values. The statistical unit is further configured to determine the recommended content corresponding to the operation item in the content features, and to determine the set of associated content of the recommended content corresponding to the operation item; and to perform cumulative revenue statistics based on the set of associated content to determine the revenue feature value corresponding to the recommended content; The associated content in the associated content set includes the overlay content of the recommended content corresponding to the operation item; the revenue feature value corresponding to the recommended content includes the revenue feature value of the overlay content. The management unit is used to determine the content type corresponding to the recommended content; determine the corresponding feature weighting coefficient based on the content type; and perform weighted calculation on the revenue feature value and the predicted value contained in the interaction feature value set according to the feature weighting coefficient to obtain the target feature value. The management unit is used to determine the target content to be pushed in the recommendation set according to the target feature value, and to distribute the target content in the recommendation interface for display.

12. The apparatus according to claim 11, characterized in that, The acquisition unit is specifically used for: Obtain the recommendation period corresponding to the target object; Gradient configuration is performed based on the recommendation period to determine multiple time windows; The interaction operations of the recommended content in the recommendation set within multiple time windows are obtained to determine the interaction data corresponding to the target object.

13. The apparatus according to claim 11, characterized in that, The statistical unit is specifically used for: The time period division information indicated by the preset feature dimension is determined based on the historical data corresponding to the target object; Determine the network state corresponding to when the target object obtains the recommended content within the statistical time period indicated by the time period segmentation information, so as to determine the data transmission method; The interaction operations of the target object with different types of recommended content under the data transmission method are obtained to determine the characteristics of the object; The exposure information and association information corresponding to the different types of recommended content under the aforementioned data transmission method are statistically analyzed to determine the content characteristics. The object features and the content features are integrated to obtain the object profile.

14. The apparatus according to claim 13, characterized in that, The statistical unit is specifically used for: The data transmission method is compared with the data transmission method in the historical data corresponding to the target object to obtain recommended scenario information; Based on the recommended scenario information, the interactive operations of the target object with different types of recommended content under the data transmission method are obtained respectively to determine the object characteristics; Based on the recommended scenario information, the exposure information and association information corresponding to the different types of recommended content under the data transmission method are statistically analyzed to determine the content characteristics.

15. The apparatus according to claim 13, characterized in that, The statistical unit is specifically used for: The interaction frequency information of the target object with the recommended content is determined based on the historical data corresponding to the target object; Hotspot time periods are extracted based on the interaction frequency information; The hotspot time period is determined as the time period division information indicated by the preset feature dimension.

16. The apparatus according to claim 13, characterized in that, The statistical unit is specifically used for: Obtain the audience segmentation dimensions, which include age, gender, or activity level; Based on the aforementioned audience segmentation dimensions, interactive data statistics are performed across different dimensions to determine the audience profile; Determine the confidence parameters corresponding to the target object; If the confidence parameter indicates that the target object is not confident, then the object profile is replaced with the crowd profile.

17. The apparatus according to claim 11, characterized in that, The management unit is specifically used for: Determine the click information corresponding to the set of associated content; If the click information reaches the preset conditions, the cumulative revenue is statistically analyzed based on the set of associated content to determine the revenue feature value corresponding to the recommended content; or; If the click information does not meet the preset conditions, the predicted value corresponding to the click information will be used as the target feature value.

18. The apparatus according to claim 17, characterized in that, The management unit is specifically used for: Extract the total number of clicks indicated in the object profile; If the total number of clicks does not meet the click criteria, the audience profile is invoked to replace the recommended content for all types in the object profile.

19. The apparatus according to claim 18, characterized in that, The management unit is specifically used for: Determine the recommended content for each type indicated in the object profile; Recommended content under each type is filtered based on exposure conditions, and recommended content under types that do not meet the exposure conditions is replaced with the user profile.

20. A computer device, characterized in that, The computer device includes a processor and memory: The memory is used to store program code; the processor is used to execute the method for managing recommended content as described in any one of claims 1 to 10 according to the instructions in the program code.

21. A computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method for managing recommended content as described in any one of claims 1 to 10.

22. A computer program product, characterized in that, The method includes computer instructions stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method for managing recommended content as described in any one of claims 1 to 10.

Images