CN115239356A

CN115239356A - Recommended content management method and related device

Info

Publication number: CN115239356A
Application number: CN202110441837.XA
Authority: CN
Inventors: 王昕睿; 刘晓光
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2021-04-23
Filing date: 2021-04-23
Publication date: 2022-10-25

Abstract

The application discloses a recommended content management method and a related device, and particularly relates to an artificial intelligence machine learning technology. Obtaining interactive data corresponding to a target object; then, counting the interactive data based on preset feature dimensions to obtain an object portrait; evaluating a plurality of characteristics in the object portrait according to the multi-target model to obtain an interactive characteristic value set; combining the income characteristic value and the interaction characteristic value set to obtain a target characteristic value; and then determining the target content for pushing in the recommendation set according to the target characteristic value and displaying the target content. Therefore, conversion and income of the recommended content in the pushing process are balanced, and due to the fact that multiple time windows are adopted and the characteristic dimension is considered in combination with a data transmission mode, the integrity of the portrayal of the object is guaranteed, and the accuracy of the recommended content pushing is improved.

Description

Recommended content management method and related device

Technical Field

The present application relates to the field of computer technologies, and in particular, to a method and a related apparatus for managing recommended content.

Background

With the rapid development of internet technology, people have higher and higher requirements for recommended content. Shuffling is a final stage in a recommendation system for recommending content, and can determine a recommendation list and a presentation order thereof, which are finally presented to a user, so that the potential thereof is gradually emphasized in recent years.

Generally, different forms of recommended content may be arranged in a mixed manner based on policy rules, for example, different types of policies such as removing duplication, scattering results, increasing diversity of recommended results, and enforcing a certain type of recommended results are eliminated, that is, different forms of recommended content are arranged by using a fixed arrangement rule.

However, due to differences of different types of recommended content, the interest levels of users for different recommended content may fluctuate, and the accuracy of recommended content push may be affected by mixedly arranging policy rules.

Disclosure of Invention

In view of this, the present application provides a method for managing recommended content, which can effectively improve the accuracy of pushing the recommended content.

A first aspect of the present application provides a method for managing recommended content, which may be applied to a system or a program including a function of managing recommended content in a terminal device, and specifically includes:

acquiring interactive operation of a target object on different forms of recommended content in a recommendation set in a plurality of time windows to determine interactive data corresponding to the target object;

counting object features and content features in the interactive data based on preset feature dimensions to obtain an object portrait, wherein the preset feature dimensions are set based on the time window, the interactive operation and a data transmission mode corresponding to the recommended content;

evaluating a plurality of characteristics in the object image according to the multi-target model to obtain an interaction characteristic value set;

calling data corresponding to the operation items in the content characteristics to carry out statistics on accumulated profits so as to determine a benefit characteristic value corresponding to the recommended content;

calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set to obtain a target characteristic value;

and determining target content used for pushing in the recommendation set according to the target characteristic value, and distributing the target content in a recommendation interface for displaying.

Optionally, in some possible implementation manners of the present application, the obtaining of interactive operations of the target object on different types of recommended content in the recommendation set within a plurality of time windows to determine interactive data corresponding to the target object includes:

acquiring a recommendation cycle corresponding to the target object;

performing gradient configuration based on the recommended period to determine a plurality of time windows;

and acquiring interactive operation of the recommended content in the recommendation set in a plurality of time windows to determine the interactive data corresponding to the target object.

Optionally, in some possible implementation manners of the present application, the counting, based on a preset feature dimension, object features and content features in the interaction data to obtain an object representation includes:

determining time interval division information indicated by the preset feature dimension according to historical data corresponding to the target object;

determining a corresponding network state when the target object acquires the recommended content within a statistical time period indicated by the time period division information to determine the data transmission mode;

acquiring interactive operation of the target object on different types of recommended contents in the data transmission mode to determine the object characteristics;

respectively counting exposure information and associated information corresponding to the different types of recommended content in the data transmission mode to determine the content characteristics;

and integrating the object features and the content features to obtain the object portrait.

Optionally, in some possible implementation manners of the present application, the obtaining of the interactive operation of the target object on the different types of recommended content in the data transmission manner to determine the object characteristics includes:

comparing the data transmission mode with a data transmission mode in historical data corresponding to the target object to obtain recommended scene information;

respectively acquiring interactive operation of the target object on different types of recommended contents in the data transmission mode based on the recommendation scene information to determine the object characteristics;

the respectively counting exposure information and associated information corresponding to the different types of recommended content in the data transmission mode to determine the content characteristics includes:

and respectively counting exposure information and associated information corresponding to the different types of recommended content in the data transmission mode based on the recommended scene information so as to determine the content characteristics.

Optionally, in some possible implementation manners of the present application, the determining, according to the historical data corresponding to the target object, time period division information indicated by the preset feature dimension includes:

determining interaction frequency information of the target object for the recommended content according to the historical data corresponding to the target object;

extracting hotspot time periods based on the interaction frequency information;

and determining the hotspot time interval as the time interval division information indicated by the preset characteristic dimension.

Optionally, in some possible implementations of the present application, the method further includes:

acquiring a crowd segmentation dimension, wherein the crowd segmentation dimension comprises age, gender or liveness;

performing interactive data statistics under different dimensions based on the crowd segmentation dimension to determine a crowd image;

determining a confidence parameter corresponding to the target object;

replacing the object representation with the crowd representation if the confidence parameter indicates that the target object is not trusted.

Optionally, in some possible implementation manners of the present application, the invoking data corresponding to the operation item in the content feature to perform statistics of accumulated revenue so as to determine a revenue feature value corresponding to the recommended content includes:

determining recommended content corresponding to the operation item in the content characteristics, and determining a related content set of the recommended content corresponding to the operation item;

and carrying out statistics of accumulated revenue based on the associated content set to determine the revenue characteristic value corresponding to the recommended content.

Optionally, in some possible implementation manners of the present application, the performing statistics of accumulated revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content includes:

determining click information corresponding to the associated content set;

if the click information reaches a preset condition, carrying out accumulated profit statistics based on the associated content set so as to determine the profit characteristic value corresponding to the recommended content;

or;

and if the click information does not reach the preset condition, taking a predicted value corresponding to the click information as a target characteristic value.

extracting the total number of clicks indicated in the object picture;

and if the total number of clicks does not meet the click condition, calling the crowd portrait to replace recommended contents of all types in the object portrait.

determining recommended content in the object representation under each type indicated;

and screening the recommended contents under each type based on exposure conditions so as to replace the recommended contents under the types which do not meet the exposure conditions by the crowd portraits.

Optionally, in some possible implementation manners of the present application, the calculating, by combining the profit feature value and a predicted value included in the set of interaction feature values, to obtain a target feature value includes:

determining a content type corresponding to the recommended content;

determining a corresponding feature weighting factor based on the content type;

and carrying out weighted calculation on the income characteristic value and a predicted value contained in the interactive characteristic value set according to the characteristic weighting coefficient so as to obtain the target characteristic value.

Optionally, in some possible implementation manners of the present application, the recommended content is a heterogeneous card, a content format of the heterogeneous card includes image-text content, video content, or link content, the recommended interface is a terminal application interface, and the target content is displayed in a focus area of the terminal application interface.

A second aspect of the present application provides a recommended content management apparatus, including:

the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring interactive operation of a target object on different forms of recommended contents in a recommendation set in a plurality of time windows so as to determine interactive data corresponding to the target object;

the statistical unit is used for carrying out statistics on object features and content features in the interactive data based on preset feature dimensions to obtain an object portrait, and the preset feature dimensions are set based on the time window, the interactive operation and a data transmission mode corresponding to the recommended content;

the evaluation unit is used for evaluating a plurality of characteristics in the object image according to the multi-target model to obtain an interactive characteristic value set;

the statistical unit is further configured to call data corresponding to an operation item in the content feature to perform statistics on accumulated revenue so as to determine a revenue feature value corresponding to the recommended content;

the management unit is used for calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set so as to obtain a target characteristic value;

and the management unit is used for determining the target content used for pushing in the recommendation set according to the target characteristic value and distributing the target content in a recommendation interface for displaying.

Optionally, in some possible implementation manners of the present application, the obtaining unit is specifically configured to obtain a recommendation cycle corresponding to the target object;

the obtaining unit is specifically configured to perform gradient configuration based on the recommended period to determine multiple time windows;

the obtaining unit is specifically configured to obtain interactive operations of recommended content in the recommendation set in a plurality of time windows, so as to determine the interactive data corresponding to the target object.

Optionally, in some possible implementation manners of the present application, the statistical unit is specifically configured to determine, according to historical data corresponding to the target object, time period division information indicated by the preset feature dimension;

the statistical unit is specifically configured to determine a corresponding network state when the target object acquires the recommended content within a statistical time period indicated by the time period division information, so as to determine the data transmission manner;

the statistical unit is specifically configured to acquire interactive operations of the target object on different types of recommended content in the data transmission manner to determine the object characteristics;

the statistical unit is specifically configured to separately count exposure information and associated information corresponding to the different types of recommended content in the data transmission manner to determine the content characteristics;

the statistical unit is specifically configured to integrate the object features and the content features to obtain the object portrait.

Optionally, in some possible implementation manners of the present application, the statistical unit is specifically configured to compare the data transmission manner with a data transmission manner in the historical data corresponding to the target object to obtain recommended scene information;

the statistical unit is specifically configured to respectively obtain interactive operations of the target object on different types of recommended content in the data transmission mode based on the recommendation scene information, so as to determine the object characteristics;

the statistical unit is specifically configured to separately count exposure information and associated information corresponding to the different types of recommended content in the data transmission mode based on the recommended scene information, so as to determine the content characteristics.

Optionally, in some possible implementation manners of the present application, the statistical unit is specifically configured to determine, according to the historical data corresponding to the target object, interaction frequency information of the target object for the recommended content;

the statistical unit is specifically configured to extract a hotspot time period based on the interaction frequency information;

the statistical unit is specifically configured to determine the hotspot time interval as the time interval division information indicated by the preset feature dimension.

Optionally, in some possible implementation manners of the present application, the statistical unit is specifically configured to obtain a crowd segmentation dimension, where the crowd segmentation dimension includes age, gender, or liveness;

the statistical unit is specifically used for carrying out interactive data statistics under different dimensions based on the crowd segmentation dimensions to determine crowd images;

the statistical unit is specifically configured to determine a confidence parameter corresponding to the target object;

the statistical unit is specifically configured to replace the object portrait with the crowd portrait if the confidence parameter indicates that the target object is not trusted.

Optionally, in some possible implementation manners of the present application, the statistical unit is specifically configured to determine recommended content corresponding to the operation item in the content feature, and determine an associated content set of the recommended content corresponding to the operation item;

the statistical unit is specifically configured to perform statistics on accumulated revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content.

Optionally, in some possible implementation manners of the present application, the management unit is specifically configured to determine click information corresponding to the associated content set;

the management unit is specifically configured to, if the click information reaches a preset condition, perform statistics on accumulated revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content;

or;

the management unit is specifically configured to, if the click information does not meet the preset condition, take a predicted value corresponding to the click information as a target feature value.

Optionally, in some possible implementations of the present application, the management unit is specifically configured to extract a total number of clicks indicated in the object representation;

the management unit is specifically configured to, if the total number of clicks does not satisfy a click condition, invoke the crowd portrait to replace recommended contents of all types in the object portrait.

Optionally, in some possible implementations of the present application, the management unit is specifically configured to determine recommended content in each type indicated in the object representation;

the management unit is specifically configured to filter the recommended content under each type based on an exposure condition, so as to replace the recommended content under the type that does not meet the exposure condition with the crowd portrait.

Optionally, in some possible implementation manners of the present application, the management unit is specifically configured to determine a content type corresponding to the recommended content;

the management unit is specifically configured to determine a corresponding feature weighting coefficient based on the content type;

the management unit is specifically configured to perform weighted calculation on the income eigenvalue and the predicted value included in the interaction eigenvalue set according to the characteristic weighting coefficient, so as to obtain the target eigenvalue.

A third aspect of the present application provides a computer device comprising: a memory, a processor, and a bus system; the memory is used for storing program codes; the processor is configured to execute the method for managing recommended content according to any one of the first aspect and the first aspect, according to instructions in the program code.

A fourth aspect of the present application provides a computer-readable storage medium having stored therein instructions, which, when executed on a computer, cause the computer to execute the method for managing recommended content according to the first aspect or any one of the first aspects.

According to an aspect of the application, a computer program product or computer program is provided, comprising computer instructions, the computer instructions being stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the method for managing recommended content provided in the first aspect or the various alternative implementations of the first aspect.

According to the technical scheme, the embodiment of the application has the following advantages:

interactive data corresponding to the target object are determined by obtaining interactive operations of the target object on different forms of recommended contents in the recommendation set in a plurality of time windows; then, counting object features and content features in the interactive data based on preset feature dimensions to obtain an object portrait, wherein the preset feature dimensions are set based on a time window, interactive operation and a data transmission mode corresponding to recommended content; evaluating a plurality of characteristics in the object portrait according to the multi-target model to obtain an interactive characteristic value set; further calling data corresponding to the operation items in the content characteristics to carry out statistics on accumulated profits so as to determine a benefit characteristic value corresponding to the recommended content; then, calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set to obtain a target characteristic value; and further determining target content for pushing in the recommendation set according to the target characteristic value, and distributing the target content in a recommendation interface for displaying. Therefore, conversion and income of the recommended content in the pushing process are balanced, and due to the fact that multiple time windows are adopted and the characteristic dimensions are considered in combination with a data transmission mode, the completeness of portrayal of the object is guaranteed, the simulation degree of fluctuation of the user interest content is improved, and the accuracy of the recommended content pushing is improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the embodiments or the prior art descriptions will be briefly described below, it is obvious that the drawings in the following description are only the embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.

FIG. 1 is a diagram of a network architecture in which a management system for recommending content operates;

fig. 2 is a flowchart illustrating management of recommended content according to an embodiment of the present disclosure;

fig. 3 is a flowchart of a recommended content management method according to an embodiment of the present application;

fig. 4 is a scene schematic diagram of a management method for recommended content according to an embodiment of the present application;

fig. 5 is a scene schematic diagram of another recommended content management method according to an embodiment of the present application;

fig. 6 is a scene schematic diagram of another recommended content management method according to an embodiment of the present application;

fig. 7 is a scene schematic diagram of another recommended content management method according to an embodiment of the present application;

fig. 8 is a scene schematic diagram of another recommended content management method according to an embodiment of the present application;

fig. 9 is a scene schematic diagram of another recommended content management method according to an embodiment of the present application;

fig. 10 is a flowchart of another recommended content management method according to an embodiment of the present application;

fig. 11 is a flowchart of another recommended content management method according to an embodiment of the present application;

fig. 12 is a schematic structural diagram of a recommended content management apparatus according to an embodiment of the present application;

fig. 13 is a schematic structural diagram of a terminal device according to an embodiment of the present application;

fig. 14 is a schematic structural diagram of a server according to an embodiment of the present application.

Detailed Description

The embodiment of the application provides a recommended content management method and a related device, which can be applied to a system or a program containing a recommended content management function in terminal equipment, and interactive data corresponding to a target object is determined by acquiring interactive operation of the target object on different forms of recommended content in a recommendation set in a plurality of time windows; then, counting object features and content features in the interactive data based on preset feature dimensions to obtain an object portrait, wherein the preset feature dimensions are set based on a time window, interactive operation and a data transmission mode corresponding to recommended content; evaluating a plurality of characteristics in the object portrait according to the multi-target model to obtain an interactive characteristic value set; further calling data corresponding to the operation items in the content characteristics to carry out statistics on accumulated profits so as to determine a benefit characteristic value corresponding to the recommended content; then, calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set to obtain a target characteristic value; and further determining target content for pushing in the recommendation set according to the target characteristic value, and distributing the target content in a recommendation interface for displaying. Therefore, conversion and income of the recommended content in the pushing process are balanced, and due to the fact that multiple time windows are adopted and the data transmission mode is combined to carry out consideration on feature dimensions, the integrity of the portrayal of the object is guaranteed, the simulation degree of fluctuation of the interesting content of the user is improved, and the accuracy of the recommended content pushing is improved.

The terms "first," "second," "third," "fourth," and the like in the description and in the claims of the present application and in the drawings described above, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Moreover, the terms "comprises," "comprising," and "corresponding" and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

First, some nouns that may appear in the embodiments of the present application are explained.

rerank: the finger mixed arrangement or rearrangement is a stage of determining a final recommendation list and a display sequence thereof according to a candidate set generated in the recalling and sorting stages of the recommendation system.

Multi-objective optimization: two or more than two target functions are indicated, and the aim is to find a sorting method to enable a plurality of targets to achieve the overall optimization.

Heterogeneous cards: refers to different kinds of theme (item) cards with great differences in content, presentation form and profit value in a recommendation system.

User card preferences: the method refers to an image which is obtained by counting user behavior data and can depict the interest degree of a user in different cards.

It should be understood that the management method of recommended content provided by the present application may be applied to a system or a program that includes a management function of recommended content in a terminal device, for example, a content recommendation application, and specifically, the management system of recommended content may operate in a network architecture as shown in fig. 1, which is a network architecture diagram that the management system of recommended content operates, as can be seen from the figure, the management system of recommended content may provide a management process of recommended content with multiple information sources, that is, corresponding interactive data is generated for a server through an interactive operation on recommended content at a terminal side, so that the interactive data is parsed to obtain a corresponding object representation, so as to indicate a further content push process. Fig. 1 shows various terminal devices, which may be computer devices, and in an actual scenario, there may be more or fewer types of terminal devices participating in the process of managing recommended content, where the specific number and type depend on the actual scenario, and this is not limited herein, and in addition, fig. 1 shows one server, but in an actual scenario, there may also be participation of multiple servers, especially in a scenario of multi-objective model training interaction, where the specific number of servers depends on the actual scenario.

In this embodiment, the server may be an independent physical server, may also be a server cluster or a distributed system formed by a plurality of physical servers, and may also be a cloud server that provides basic cloud computing services such as cloud service, a cloud database, cloud computing, a cloud function, cloud storage, network service, cloud communication, middleware service, domain name service, security service, CDN, and a big data and artificial intelligence platform. The terminal may be, but is not limited to, a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and the like. The terminal and the server may be directly or indirectly connected through a wired or wireless communication manner, and the terminal and the server may be connected to form a block chain network, which is not limited herein.

It is understood that the above-mentioned management system for recommended content may be operated in a personal mobile terminal, for example: the application can be operated on a server as a content recommendation application, and can also be operated on a third-party device to provide management of recommended content so as to obtain a management processing result of the recommended content of the information source; the specific recommended content management system may be operated in the device in the form of a program, may also be operated as a system component in the device, and may also be operated as one of cloud service programs, and a specific operation mode is determined according to an actual scene, which is not limited herein.

With the rapid development of internet technology, people have higher and higher requirements for recommended content. The shuffling is a final stage in a recommendation system for recommending content, and can determine a recommendation list and a presentation order of the recommendation list, which are finally presented to a user, so that the potential of the shuffling is gradually emphasized in recent years.

Generally, different forms of recommended content shuffling can be performed based on policy rules, for example, different types of policies such as removing duplication, scattering results to increase diversity of recommended results, and interpolating a certain type of recommended results are performed, that is, different forms of recommended content are arranged by using a fixed arrangement rule.

In order to solve the above problems, the present application provides a recommended content management method, which manages recommended content by using a Machine Learning technology, where Machine Learning (ML) is a multi-domain cross subject, and relates to multiple subjects such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, and the like. The special research on how a computer simulates or realizes the learning behavior of human beings so as to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer. Machine learning is the core of artificial intelligence, is the fundamental approach for computers to have intelligence, and is applied to all fields of artificial intelligence. Machine learning and deep learning generally include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and formula learning.

Specifically, the method is applied to a flow framework of management of recommended content shown in fig. 2, and as shown in fig. 2, for a flow framework of management of recommended content provided in an embodiment of the present application, a user generates corresponding interactive data in a server through an interactive operation (clicking, downloading, staying, etc.) on the recommended content (heterogeneous card) at a terminal side, the server performs feature dimension analysis based on the interactive data to generate an object portrait, and counts the object portrait to obtain a feature value pushed by the user content, thereby performing ranking and interface distribution of the recommended content.

It can be understood that, because different optimization targets of the recommendation system may have a phenomenon of mutual restriction, one index may rise and the other index may drop obviously, and this phenomenon is particularly serious in this business scenario, because this scenario includes a plurality of heterogeneous cards (graphics, small video, short content, socialization, graphics, video, etc.), and individual cards have floating layer consumption, that is, the time length brought by a single click of a user, the revenue of an advertisement income is not only the revenue of watching the clicked content, but also the revenue brought by continuously watching different contents after clicking into the floating layer, and different types of cards have great difference in time length and revenue of the advertisement income.

Therefore, the multi-target mixed arrangement method is provided, the mutual influence of three targets of clicking, duration and advertising income can be balanced in the recommendation scene of various heterogeneous cards, and all indexes can be synchronously increased or the main target can be increased without falling other targets.

It can be understood that the method provided by the present application may be a program written as a processing logic in a hardware system, and may also be a recommended content management device, and the processing logic is implemented in an integrated or external manner. As an implementation manner, the recommended content management device determines interactive data corresponding to a target object by acquiring interactive operations of the target object on different forms of recommended content in a recommendation set within multiple time windows; then, counting object features and content features in the interactive data based on preset feature dimensions to obtain an object portrait, wherein the preset feature dimensions are set based on a time window, interactive operation and a data transmission mode corresponding to recommended content; evaluating a plurality of characteristics in the object portrait according to the multi-target model to obtain an interactive characteristic value set; further calling data corresponding to the operation items in the content characteristics to carry out statistics on accumulated profits so as to determine a benefit characteristic value corresponding to the recommended content; then, calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set to obtain a target characteristic value; and further determining target content for pushing in the recommendation set according to the target characteristic value, and distributing the target content in a recommendation interface for displaying. Therefore, conversion and income of the recommended content in the pushing process are balanced, and due to the fact that multiple time windows are adopted and the characteristic dimensions are considered in combination with a data transmission mode, the completeness of portrayal of the object is guaranteed, the simulation degree of fluctuation of the user interest content is improved, and the accuracy of the recommended content pushing is improved.

The scheme provided by the embodiment of the application relates to an artificial intelligence machine learning technology, and is specifically explained by the following embodiment:

with reference to the above flow architecture, a method for managing recommended content in the present application will be introduced below, please refer to fig. 3, where fig. 3 is a flow chart of a method for managing recommended content provided in an embodiment of the present application, where the method for managing recommended content may be executed by a terminal, may also be executed by a server, and may also be executed by both the terminal and the server, and the embodiment of the present application at least includes the following steps:

301. and acquiring interactive operation of the target object on different forms of recommended content in the recommendation set in a plurality of time windows to determine interactive data corresponding to the target object.

In this embodiment, the recommended content is a heterogeneous card, the content format of the heterogeneous card includes an image-text content, a video content or a link content, and may be specifically a media format such as an image-text, a small video, a short content, socialization, an attention image-text, an attention video, a jump link, and the like, for example, a scene shown in fig. 4, fig. 4 is a scene schematic diagram of the management method of the recommended content provided in the embodiment of the present application, the recommended content in different formats included in a recommendation interface A1 is shown in the scene schematic diagram, that is, the media formats of the contents 1 to 6 may be the same or different, the specific media format is determined according to an actual scene, and the recommended content is used for being displayed in the recommendation interface, and the recommendation interface may be a terminal application interface, so as to facilitate a further recommendation conversion process.

It is to be understood that the target object to which the recommended content is pushed may be a user, an account, or other representative object identifiers, and in the following embodiments, the target object is taken as an example of a target user, which is not limited herein.

Specifically, because the interest content of the user has volatility, data collection can be divided through multiple time periods in multiple time windows, namely, a recommendation cycle corresponding to a target object is obtained at first; then performing gradient configuration based on the recommendation cycle to determine a plurality of time windows; and further acquiring interactive operation of the recommended content in the recommendation set in a plurality of time windows to determine interactive data corresponding to the target object.

It can be understood that the time window is a basis for counting the interactive operations of the user, so as to reflect the preference characteristics of the user in different time length dimensions, and the preference characteristics can be used for selecting recommended content, so that the method of the embodiment can be applied to the user preferences in different time length dimensions, and the accuracy of selecting the recommended content is improved.

In one possible scenario, to characterize the long-term, short-term, real-time (different duration dimensions) card preferences of a user, the card representation is statistically populated with user behavior data from 3 time windows, respectively: and (1) portrait in year: counting data of one year before the current day; (2) month image: counting data of one month before the current day; (3) recent (session) portrait: and counting the current latest 50-time data of the user, thereby ensuring the comprehensiveness of the recommendation period corresponding to the target object and ensuring the accuracy of the content in different recommendation periods.

302. And counting the object characteristics and the content characteristics in the interactive data based on the preset characteristic dimension to obtain the object portrait.

In the embodiment, the preset feature dimension is set based on a time window, interactive operation and a data transmission mode corresponding to recommended content; specifically, basic data of the portrait is constructed for statistics of the object portrait, namely statistics of time periods, network states and consumed cards corresponding to consumption behaviors of each object in different time windows; in addition, the object feature is a feature corresponding to the target object, and the set of the features is an object portrait, for example: when the target object is a target account, the object characteristics are the relevant characteristics of the recommended content corresponding to the target account, so as to obtain account characteristics, and the corresponding object portrait can also be called as an account portrait; alternatively, when the target object is a target user, the object feature is a user feature, and the corresponding object representation may also be referred to as a user representation.

The time period is a time window which is divided into more fine-grained parts, so that the object portraits of the users in different time periods in different time windows are depicted. For example, if the time window is the last month, the time periods may include morning (9-11 am), afternoon (14-17 pm), and evening (19-22 pm) of each day in the month, so that the consumption habits of the user are counted based on the time periods, the preference of the user in a short period (month) is obtained, the comprehensiveness of the object portraits is improved, and the accuracy of selecting the subsequent recommended content is ensured.

Therefore, time interval division information indicated by the preset feature dimension can be determined according to historical data corresponding to the target object; then, determining a corresponding network state when the target object acquires recommended content within a statistical time period indicated by the time period division information, and determining a data transmission mode, for example, the network state is divided into WIFI, 4G, 3G, 2G and unknown network, so as to create portrait data in different data transmission modes; acquiring interactive operation of the target object on different types of recommended contents in a data transmission mode to determine object characteristics; then respectively counting exposure information and associated information corresponding to different types of recommended contents in a data transmission mode to determine the characteristics of the contents; further, the object feature and the content feature are integrated to obtain the object image.

The object portrait description mode considers that consumption habits of different people in different time periods and network states in a service scene are greatly different, so that the embodiment performs more detailed dimensional statistics on the time periods and the network states.

In one possible scenario, the time segment division (time segment division information): 1. hourly divisions into 4 segments per day: morning (2-5 am), morning (6-11 am), afternoon (12 am-18 am), evening (19 am-1 st day); 2. weekly divided into weekly and weekends: in the middle of the week (monday-friday, excluding friday nights), on weekends (friday nights, saturdays, sundays), wherein the consumption habits of the friday nights are special, so that the weekends are also divided, thereby improving the significance of the characteristics in each period of time, and facilitating the recommendation division among the contents.

Optionally, the time period is divided into 4 segments each day due to data sparsity, and the time period may be divided more finely. Specifically, the interaction frequency information of the target object for the recommended content may be determined according to the historical data corresponding to the target object; then extracting a hotspot time interval based on the interaction frequency information; and determining the hotspot time interval as the time interval division information indicated by the preset characteristic dimension. Therefore, time interval data with abundant interactive data is obtained, data statistics of data-free time intervals, such as late night time intervals, is avoided, and the validity of the data in the time intervals is guaranteed.

Optionally, in the statistical process of the data transmission manner, in consideration of different usage scenarios corresponding to different WiFi, and different corresponding recommendation contents, for example, recommendation contents are different between company WiFi and home WiFi, so the data transmission manner may be compared with the data transmission manner in the history data corresponding to the target object to obtain recommendation scenario information (determine whether the data transmission manner is a common scenario, such as a company); then, interactive operation of the target object on different types of recommended contents in a data transmission mode is respectively obtained based on the recommended scene information so as to determine object characteristics; and respectively counting exposure information and associated information corresponding to different types of recommended contents in a data transmission mode based on the recommended scene information to determine the content characteristics, so that the accuracy of the characteristics in different network scenes is ensured. For example, whether WiFi in the network status is a common WiFi, because the consumption habits of WiFi of some users in the company and WiFi in the home are different, so as to ensure the feature accuracy of the scene.

Illustratively, the object image is to count a time period, a network state and a consumed card corresponding to each consumption behavior of each user in a time window, and construct basic data of the image, for example, a partial month image of the user a includes:

and (3) network state: and (7) WIFI.

Time period: weekend evening.

Exposure (number): text 30, small video 10, short video 20, socialization 5, short content 15.

Click (number): teletext 5, small video 3, short video 6, socialization 0, short content 8.

Duration (sec): graph 200, small video 321, short video 680, socialization 0, short content 220.

Carrying (number): teletext 5, small video 26, short video 12, socialization 0, short content 31.

The object features include user preferences, and the content images include exposure numbers, click rates, driving numbers, consumption durations, and the like corresponding to recommended contents.

Specifically, for the exposure number, i.e. the exposure time of the card, in order to ensure the effectiveness of the exposure time, it can be set that the exposure is calculated only when 50% of the area of the card has to be exposed. In addition, the number of clicks of the number card is clicked, the card consumption duration can be set to be more than or equal to 6s before effective clicking is carried out, and therefore invalid clicks similar to the title party are filtered.

In addition, the click rate is directly obtained by the number of clicks/the number of exposures, but the click rate obtained directly by the number of clicks/the number of exposures is not believed because the number of card exposures of some users is small, and therefore, the calculation can be performed by using Wilson Score, and the formula is as follows:

r＝pos/n

wherein pos represents the number of clicks; n represents the number of exposures; z represents a confidence parameter and can take the value of 2, namely 95% confidence.

For the click preference in the object portrait, the preference of the user for a certain type of cards is drawn immediately, because the click rate of all the cards of some users is higher or lower, the Wilson Score formula is also used for calculation, and pos represents the click number of the certain type of cards in the formula; n represents all the number of clicks.

For the driving number, namely the click conversion corresponding to the card, but considering that some cards have floating layers, namely jump to the associated content, for the image-text card without the floating layer, the driving number is the click number; the number of the video cards with the floating layers is the total number of the videos consumed in the floating layers.

Counting the total consumption time of the user on each card according to the consumption time; and for the video card with the floating layer, the accumulated consumption time length in the floating layer is included.

For the single click driving coefficient, namely the driving number/the click number are calculated, the method is used for depicting the advertisement income brought by the single click of the user, and the advertisement income is positively correlated with the upglide times in the floating layer.

The consumption duration of a single click is calculated through the consumption duration/click number and is used for depicting the duration profit brought by the user through the single click.

Optionally, the description indicates a process of depicting the object portrait, but inaccurate description may occur for a user with less data, and in order to avoid the influence of the data on the content recommendation process, a crowd portrait may be used for substitution, that is, a crowd segmentation dimension is obtained first, and the crowd segmentation dimension includes age, gender, or liveness; then, carrying out interactive data statistics under different dimensions based on the crowd segmentation dimension to determine a crowd image; determining a confidence parameter corresponding to the target object; if the confidence parameter indicates that the target object is not trusted, replacing the object representation with a crowd representation.

In one possible scenario, a crowd representation is used in place of an object representation when the object representation fails to be captured or when the representation data is determined to be inconclusive in a policy. The statistical dimension of the crowd portrait is consistent with the object portrait, and all users are divided into different crowds according to different dimensions and are counted according to the crowds instead of counting the consumption behaviors of a single user. For example, the cut dimensions are as follows:

age: 0-6 years old, 7-12 years old, 13-15 years old, 16-18 years old, 19-22 years old, 23-25 years old, 26-30 years old, 31-40 years old, 41-50 years old, more than 50 years old.

Sex: male, female, unknown.

The activity degree: the number of clicks per the last 30 days of the user is divided into: no exposure, low activity, medium activity, high activity.

The effectiveness of the data of the portrait of each object is ensured through the alternative setting of the portrait of the crowd.

303. And evaluating a plurality of characteristics in the object portrait according to the multi-target model to obtain an interaction characteristic value set.

In this embodiment, the multi-target model is a scene having two or more than two target functions, for example, a scene shown in fig. 5, fig. 5 is a scene schematic diagram of another recommended content management method provided in this embodiment of the present application, where the target features, the content features, and the context features (floating layer relationships) are extracted through a feature layer to input the multi-target model, and a click prediction value and a duration prediction value are performed, that is, the multi-target model includes a click task and a duration task, which are respectively scored for reference in recommendation.

Further, after the click prediction value and the duration prediction value are calculated, the revenue needs to be balanced, that is, more duration and advertisement revenue are obtained. Because the card has a floating layer relation, the setting rule for the income is different, and the specific is as follows:

(1) Regarding a single recommended content.

In this embodiment, a single recommended content is a recommended content (card) without a floating layer or a jump relation, that is, a duration revenue corresponding to the recommended content and additional advertisement collection are directly calculated.

(2) Recommendation content regarding the presence of a floating layer.

In this embodiment, the main application scenario is a viewpoint information flow service in a content push application, and for heterogeneous cards (images, texts, small videos, and short video cards), an immersive video floating layer scenario that a short video card is clicked and then enters can be clicked, a user can slide up and consume more videos infinitely, and after consuming a few videos, an advertisement is inserted, and when the image and text card is clicked, the user only carries an image and text detail page of an advertisement, and cannot drive more consumption.

Therefore, in the present service scenario, there are natural great differences between the duration and the advertisement revenue of the graphics and text and the video card due to product forms, specifically as shown in fig. 6, fig. 6 is a scenario diagram of another recommended content management method provided in the embodiment of the present application, where the duration (seconds) consumed by a user clicking a certain card once and the distribution of the number of the drivers are shown in the scenario, and the horizontal axis is the quantiles. In the figure, a severe video user can consume nearly 1 hour by single click, and drive more than 100 video consumption, which is nearly 10 times of the time length of the image-text severe user by single click. In addition, fig. 7 is a scene diagram of another recommended content management method according to an embodiment of the present application, which shows that the advertisement revenue is tens of times of a single click of a graphic card from the perspective of the advertisement exposure. Therefore, when multi-objective optimization of click, duration and advertisement revenue is considered, a fusion strategy must be designed for consumption difference of the floating layer, as described in step 304.

In a possible scenario, as shown in fig. 8, fig. 8 is a scenario diagram of another method for managing recommended content provided in the embodiment of the present application, where the diagram shows that after clicking on recommended content B1, floating content B2 may be entered, and sliding or clicking operations may be performed in an interface corresponding to floating content B2 to adjust to more different floating contents, that is, associated contents, so as to count content features of the associated contents in recommended content B1, so as to improve relevance between the contents.

304. And calling data corresponding to the operation items in the content characteristics to carry out statistics on accumulated profits so as to determine a benefit characteristic value corresponding to the recommended content.

In this embodiment, for the statistical process of the accumulated revenue, that is, only the time length conversion and the advertisement revenue of the single recommended content are considered, and the floating layer content needs to consider the associated content.

Specifically, for the determination of the profit characteristic value of the floating layer content, the recommended content corresponding to the operation item in the content characteristic may be determined, and the associated content set of the recommended content corresponding to the operation item is determined; and then, carrying out accumulated profit statistics based on the associated content set to determine a profit characteristic value corresponding to the recommended content, thereby ensuring the accuracy of profit description of the floating-layer content.

Optionally, in order to ensure the effectiveness of the benefit description, when the user does not click the day, it is preferentially ensured that the click only sees the click task score; after the user clicks, determining click information corresponding to the associated content set by considering more acquired duration and advertisement revenue; specifically, if the click information reaches a preset condition (for example, the click number is greater than 5), performing statistics on accumulated revenue based on the associated content set to determine a revenue characteristic value corresponding to the recommended content; and when the click information does not reach the preset condition, taking the predicted value corresponding to the click information as the target characteristic value.

According to the method, aiming at the heterogeneous cards with huge time length and advertising income and income difference, the deep learning model and the income pre-estimation strategy are combined to perform multi-target sequencing, click, time length and advertising income indexes are improved, and therefore expectation among multiple targets is balanced.

Optionally, after the multi-objective model outputs the click and the duration task predicted value of each candidate item by the user, when the user does not click the day, in order to preferentially ensure that the click only sees the click task score, after the user clicks the day, the user considers obtaining more duration and advertisement revenue.

It can be understood that, for the difference problem of the floating layer card profit, in the fusion strategy, besides the multi-objective model is considered to output the click and duration task predicted value of each candidate item by the user, the duration and advertisement profit caused by single click of the user, which are calculated by the card portrait, are also considered. Because the advertising income is positively correlated with the number of times of sliding upwards in the floating layer, and ecpm of the advertisements to be consumed in the floating layer is unknown, the estimated advertising income is approximated by a driving coefficient of single click, and the finally obtained fusion score is only used for sequencing, so that the approximation is comparable without considering the difference in absolute value.

Optionally, a model can be trained independently for prediction of the profit of the floating layer card, the expected consumption profit of the user entering the floating layer by clicking once is predicted, and therefore the obtaining efficiency of the profit characteristic value is improved.

305. And calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set to obtain a target characteristic value.

In this embodiment, the target feature value may be calculated by weighting the revenue feature value and the predicted value included in the interaction feature value set.

Optionally, because the emphasis corresponding to different card contents is different, a targeted design may be performed, that is, the content type corresponding to the recommended content is determined first; then determining a corresponding characteristic weighting coefficient based on the content type; and then, carrying out weighted calculation on the income characteristic value and a predicted value contained in the interactive characteristic value set according to the characteristic weighting coefficient to obtain a target characteristic value, thereby improving the adaptability of the target characteristic value for recommending different forms of content.

306. And determining target content used for pushing in the recommendation set according to the target characteristic value, and distributing the target content in a recommendation interface for displaying.

In this embodiment, the recommended contents in the recommendation set may be sorted based on the target feature value, so as to obtain the target contents located in the front row, and the target contents are distributed in the recommendation interface for display, specifically, may be displayed in the middle, so as to be convenient for the user to perceive.

With the combination of the above embodiments, interactive data corresponding to a target object is determined by obtaining interactive operations of the target object on different forms of recommended content in a recommendation set within a plurality of time windows; then, counting object features and content features in the interactive data based on preset feature dimensions to obtain an object portrait, wherein the preset feature dimensions are set based on a time window, interactive operation and a data transmission mode corresponding to recommended content; evaluating a plurality of characteristics in the object portrait according to the multi-target model to obtain an interactive characteristic value set; further calling data corresponding to the operation items in the content characteristics to carry out statistics on accumulated profits so as to determine a benefit characteristic value corresponding to the recommended content; then, calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set to obtain a target characteristic value; and further determining target content for pushing in the recommendation set according to the target characteristic value, and distributing the target content in a recommendation interface for displaying. Therefore, conversion and income of the recommended content in the pushing process are balanced, and due to the fact that multiple time windows are adopted and the characteristic dimensions are considered in combination with a data transmission mode, the completeness of portrayal of the object is guaranteed, the simulation degree of fluctuation of the user interest content is improved, and the accuracy of the recommended content pushing is improved.

How to recommend recommended content, which may be media objects, such as videos, pictures, heterogeneous cards, and the like, based on the trained multi-objective model is described below with reference to fig. 9. Referring to fig. 9, fig. 9 is a scene schematic diagram of another recommended content management method provided in the embodiment of the present application, that is, a multi-target model includes a feature mapping layer, a feature extraction layer, a feature splicing layer, and a prediction layer. In practical application, when recommending media objects based on the multi-target model, the server acquires user data and content data of the media objects to be recommended, and then inputs the user data and the content data into the multi-target model.

And the multi-target model respectively performs characteristic mapping processing on the user data and the content data through a characteristic mapping layer to obtain characteristic vectors corresponding to the user data and the content data. Specifically, for example, the mapping process may be performed by a one-hot encoding method, or a pre-trained feature mapping model.

After the characteristic vectors corresponding to the user data and the content data are obtained, the characteristic extraction layer is used for extracting the characteristics of the obtained characteristic vectors to obtain the characteristic vectors of the media objects to be recommended. Specifically, the feature extraction layer is composed of a wide layer, a DNN layer and a shared NFM layer, so that the DNN layer can be used for performing implicit feature intersection on feature vectors of user data and content data to extract high-order feature vectors; the feature vectors of the user data and the content data are crossed in display features through an NFM layer, and the feature vectors are summed to obtain a multi-dimensional feature vector; and performing linear summation on the feature vectors of the user data and the content data based on the weight through the wide layer, and outputting the feature vectors with reduced dimensionality and the like.

After the characteristic vectors of the media objects to be recommended are obtained, vector splicing is carried out through the characteristic splicing layer, and splicing vectors are obtained. Therefore, interactive features are predicted through the prediction layer based on the splicing vector, and feature prediction results corresponding to the media objects to be recommended are obtained. Specifically, the prediction layer may be an artificial neural network model, and the feature prediction result is obtained by calling an activation function to predict the interaction feature; here, the prediction layer may belong to regression prediction and may also belong to classification prediction. When the prediction layer is in regression prediction, carrying out regression processing by calling a first activation function (such as a regression function) to predict and obtain a feature prediction result of each interactive feature; when the prediction layer is classified and predicted, a second activation function (such as a softmax classification function) is called for classification processing, and feature prediction results of the interactive features are obtained through prediction.

And recommending the media object to be recommended based on the feature prediction result output by the multi-target model. Specifically, the feature prediction result may be an estimated click rate, so that the media object to be recommended may be recommended based on the estimated click rate.

An exemplary application of the embodiments of the present invention in a practical application scenario will be described below.

In the application process of the multi-target model, the object characteristics and the content characteristics in the object portrayal depict the object portrayal of the preference degree of a user to various heterogeneous cards (pictures and texts, small videos, short contents, socialization, attention pictures and texts and attention videos) of the business under different time periods and network states, and the object portrayal is used as the characteristics of the model layer and used for the multi-target layer to estimate the card income.

Basically, based on the preference of the user to the content (classification, tag label, item), the preference of the user to different card types in different time periods and different context states is not carefully described, but the consumption habits of different groups of people in the business scene are very different, for example, although the users are severe users, some users only consume the image-text card in the daytime and consume the video card at night or on the weekend.

The multi-objective optimization is only based on the income of a single item, but floating layer consumption exists in individual cards in the scene, namely the income of the user caused by single click is not only the income of watching the clicked content, but also the income caused by continuously watching different contents after clicking into the floating layer, so that the multi-objective optimization effect is poor.

According to the embodiment, the object portrait depicting the preference degree of the user on various heterogeneous cards (pictures and texts, small videos, short contents, socialization, picture and text attention and video attention) of the business in different time periods and network states is constructed, the object portrait is used as the feature of a mixed model and used for multi-target income prediction, the multi-target model precision and the user experience are improved, and the problem that the card preference depicting of the user is inaccurate in different context states is solved.

According to the method, a deep learning model and a profit estimation strategy are designed for the heterogeneous cards with huge time and advertising income profit differences, multi-target optimization is carried out, click, time and advertising income indexes are improved, and the problem that the multi-target heterogeneous cards with huge profit differences are difficult to optimize is solved.

The following is described with reference to a revenue evaluation scenario. Referring to fig. 10, fig. 10 is a flowchart of another recommended content management method according to an embodiment of the present application, where the embodiment of the present application includes at least the following steps:

1001. whether the user clicks the current day.

In the embodiment, after the multi-objective model outputs the click and the duration task predicted value of each candidate item by the user, when the user does not click the day, in order to preferentially ensure that the click only sees the click task score, after the user clicks the day, the user considers obtaining more durations and advertisement revenue, and therefore the reliability of the revenue characteristic value is improved.

1002. And calculating a characteristic value corresponding to the click.

In this embodiment, if there is no click, that is, it is stated that the content has not yet been converted into revenue, the click number may be retained, and the target characteristic value may be described based on the predicted value corresponding to the click number, so as to improve the reliability of the target characteristic value.

1003. The revenue is estimated based on the object image.

In the embodiment, in the process of predicting the income, aiming at the problem of object portrait data unconfidibility, when the user is an extremely inactive user or the number of times of exposure of the user on a certain type of cards is extremely small, the crowd portrait is used for replacing the personal portrait.

1004. The total number of clicks of the object image is less than or equal to 5.

1005. The card data used in the object picture is the crowd picture data.

In this embodiment, the total number of clicks indicated in the object picture is extracted first; if the total number of clicks does not meet the click condition (the total number of clicks is less than or equal to 5), calling the crowd portrait to replace recommended contents of all types in the object portrait, wherein the specific numerical value is determined by the actual scene.

1006. And judging the confidence of each card of the user.

In this embodiment, since the push logics of different types of cards (recommended contents) are different, it is necessary to separately determine the confidence of each card.

1007. The exposure number of the object image card is less than or equal to 3.

1008. The object portrait type card data adopts crowd portrait data.

In the embodiment, firstly, recommended contents of various types indicated in the object portrait are determined; and then screening the recommended contents under each type based on the exposure conditions so as to replace the recommended contents (such as pictures and texts) under the types which do not meet the exposure conditions (the exposure number is less than or equal to 3) by adopting crowd pictures, wherein the specific numerical value is determined according to the actual scene.

1009. And calculating a target characteristic value.

In the embodiment, weighted calculation of the target characteristic value is performed based on the data after the confidence screening, so that an object portrait depicting preference degrees of a user to various heterogeneous cards (pictures and texts, small videos, short contents, socialization, attention pictures and texts and attention videos) of the business under different time periods and network states is constructed, the object portrait is used as a characteristic of a mixed model and used for multi-target income prediction, the precision of the multi-target model and the user experience are improved, and the problem that card preference depicting is inaccurate when the user is in different context states is solved.

In addition, a deep learning model and a profit estimation strategy are designed for heterogeneous cards with huge time length and advertising income and income differences, multi-target optimization is carried out, click, time length and advertising income indexes are improved, and the problem that the heterogeneous cards with huge income differences are difficult to optimize in multiple targets is solved.

In a possible scenario, content pushing performed by the embodiment is significantly improved in key indexes such as user click on a DAU, consumption duration, advertisement revenue and the like.

In another possible scenario, the interface distribution of the push content may be performed in a targeted manner, as shown in fig. 11, fig. 11 is a flowchart of another recommended content management method provided in the embodiment of the present application, and the embodiment of the present application at least includes the following steps:

1101. and determining a target characteristic value corresponding to the recommended content in the recommendation set.

In this embodiment, the calculation of the target feature value corresponding to the recommended content is described with reference to the embodiment shown in fig. 3, which is not described herein again.

1102. And updating the target characteristic value based on the target application type indicated by the recommendation interface to obtain the recommended characteristic value.

In this embodiment, since different target application types have preferences for different recommended contents, for example, the short video application is focused on pushing short videos, and the news application is focused on pushing graphics and texts, weighted update of the target feature value can be performed, so as to obtain the recommended feature value, thereby ensuring suitability of the recommended feature value for an application scene.

1103. And filling content according to the focus area corresponding to the target application type and the recommended characteristic value so as to push the recommended content.

In this embodiment, the target content is displayed in a focus area of the terminal application interface, where the focus area may be a center, an edge, or another position of the interface, and a specific position may be set according to a hot spot area corresponding to the type of the target application, for example, the center of the short video application is the focus area, and then the content with the highest recommendation characteristic value is filled in the focus area to push the recommended content, so that the adaptability between the content and the application is ensured, and the accuracy of the recommended content is further improved.

In order to better implement the above-mentioned aspects of the embodiments of the present application, the following also provides related apparatuses for implementing the above-mentioned aspects. Referring to fig. 12, fig. 12 is a schematic structural diagram of a management apparatus for recommending content according to an embodiment of the present application, where the management apparatus 1200 includes:

an obtaining unit 1201, configured to obtain interactive operations of a target object on different types of recommended content in a recommendation set within multiple time windows, so as to determine interactive data corresponding to the target object;

a counting unit 1202, configured to count object features and content features in the interactive data based on a preset feature dimension to obtain an object portrait, where the preset feature dimension is set based on the time window, the interactive operation, and a data transmission manner corresponding to the recommended content;

an evaluation unit 1203, configured to evaluate a plurality of features in the object image according to the multi-target model to obtain an interaction feature value set;

the statistical unit 1202 is further configured to invoke data corresponding to an operation item in the content feature to perform statistics on accumulated revenue, so as to determine a revenue feature value corresponding to the recommended content;

a management unit 1204, configured to perform calculation by combining the income feature value and a predicted value included in the interaction feature value set to obtain a target feature value;

the management unit 1204 is configured to determine, according to the target feature value, target content for pushing in the recommendation set, and distribute the target content in a recommendation interface for display.

Optionally, in some possible implementation manners of the present application, the obtaining unit 1201 is specifically configured to obtain a recommendation cycle corresponding to the target object;

the obtaining unit 1201 is specifically configured to perform gradient configuration based on the recommended period to determine a plurality of time windows;

the obtaining unit 1201 is specifically configured to obtain interactive operations of the recommended content in the recommendation set in a plurality of time windows, so as to determine the interactive data corresponding to the target object.

Optionally, in some possible implementation manners of the present application, the statistical unit 1202 is specifically configured to determine, according to historical data corresponding to the target object, time period division information indicated by the preset feature dimension;

the statistical unit 1202 is specifically configured to determine a corresponding network state when the target object acquires the recommended content within a statistical time period indicated by the time period division information, so as to determine the data transmission manner;

the statistical unit 1202 is specifically configured to obtain interactive operations of the target object on different types of recommended content in the data transmission manner, so as to determine the object characteristics;

the statistical unit 1202 is specifically configured to separately count exposure information and associated information corresponding to the different types of recommended content in the data transmission manner, so as to determine the content characteristics;

the statistical unit 1202 is specifically configured to integrate the object feature and the content feature to obtain the object portrait.

Optionally, in some possible implementation manners of the present application, the statistical unit 1202 is specifically configured to compare the data transmission manner with a data transmission manner in historical data corresponding to the target object to obtain recommended scene information;

the statistical unit 1202 is specifically configured to respectively obtain, based on the recommendation scene information, interactive operations of the target object on different types of recommended content in the data transmission manner, so as to determine the object features;

the counting unit 1202 is specifically configured to count exposure information and associated information corresponding to the different types of recommended content in the data transmission manner, respectively, based on the recommended scene information, so as to determine the content features.

Optionally, in some possible implementation manners of the present application, the statistical unit 1202 is specifically configured to determine, according to the history data corresponding to the target object, interaction frequency information of the target object for the recommended content;

the statistical unit 1202 is specifically configured to extract a hotspot time interval based on the interaction frequency information;

the statistical unit 1202 is specifically configured to determine the hot spot time interval as the time interval division information indicated by the preset feature dimension.

Optionally, in some possible implementation manners of the present application, the statistical unit 1202 is specifically configured to obtain a crowd segmentation dimension, where the crowd segmentation dimension includes an age, a gender, or an activity;

the statistical unit 1202 is specifically configured to perform interactive data statistics in different dimensions based on the crowd segmentation dimension to determine a crowd image;

the statistical unit 1202 is specifically configured to determine a confidence parameter corresponding to the target object;

the statistical unit 1202 is specifically configured to replace the object portrait with the crowd portrait if the confidence parameter indicates that the target object is not trusted.

Optionally, in some possible implementations of the present application, the statistical unit 1202 is specifically configured to determine recommended content corresponding to the operation item in the content feature, and determine an associated content set of the recommended content corresponding to the operation item;

the statistical unit 1202 is specifically configured to perform statistics on accumulated revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content.

Optionally, in some possible implementation manners of the present application, the management unit 1204 is specifically configured to determine click information corresponding to the associated content set;

the management unit 1204 is specifically configured to, if the click information meets a preset condition, perform statistics on accumulated revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content;

or;

the management unit 1204 is specifically configured to, if the click information does not meet the preset condition, take a predicted value corresponding to the click information as a target feature value.

Optionally, in some possible implementations of the present application, the management unit 1204 is specifically configured to extract a total number of clicks indicated in the object representation;

the management unit 1204 is specifically configured to invoke the crowd representation to replace recommended contents of all types in the object representation if the total number of clicks does not satisfy the click condition.

Optionally, in some possible implementations of the present application, the management unit 1204 is specifically configured to determine recommended contents in each type indicated in the object representation;

the management unit 1204 is specifically configured to filter the recommended content in each type based on the exposure condition, so as to replace the recommended content in the type that does not satisfy the exposure condition with the crowd representation.

Optionally, in some possible implementations of the present application, the management unit 1204 is specifically configured to determine a content type corresponding to the recommended content;

the management unit 1204 is specifically configured to determine a corresponding feature weighting coefficient based on the content type;

the management unit 1204 is specifically configured to perform weighted calculation on the income feature value and a predicted value included in the interaction feature value set according to the feature weighting coefficient, so as to obtain the target feature value.

Interactive data corresponding to the target object are determined by obtaining interactive operations of the target object on different forms of recommended contents in the recommendation set in a plurality of time windows; then, counting object features and content features in the interactive data based on preset feature dimensions to obtain an object portrait, wherein the preset feature dimensions are set based on a time window, interactive operation and a data transmission mode corresponding to recommended content; evaluating a plurality of characteristics in the object portrait according to the multi-target model to obtain an interactive characteristic value set; further calling data corresponding to the operation items in the content characteristics to carry out statistics on accumulated profits so as to determine a benefit characteristic value corresponding to the recommended content; then, calculating by combining the income characteristic value and a predicted value contained in the interactive characteristic value set to obtain a target characteristic value; and further determining target content used for pushing in the recommendation set according to the target characteristic value, and distributing the target content in a recommendation interface for displaying. Therefore, conversion and income of the recommended content in the pushing process are balanced, and due to the fact that multiple time windows are adopted and the characteristic dimensions are considered in combination with a data transmission mode, the completeness of portrayal of the object is guaranteed, the simulation degree of fluctuation of the user interest content is improved, and the accuracy of the recommended content pushing is improved.

An embodiment of the present application further provides a terminal device, as shown in fig. 13, which is a schematic structural diagram of another terminal device provided in the embodiment of the present application, and for convenience of description, only a portion related to the embodiment of the present application is shown, and details of the specific technology are not disclosed, please refer to a method portion in the embodiment of the present application. The terminal may be any terminal device including a mobile phone, a tablet computer, a Personal Digital Assistant (PDA), a point of sale (POS), a vehicle-mounted computer, and the like, taking the terminal as the mobile phone as an example:

fig. 13 is a block diagram illustrating a partial structure of a mobile phone related to a terminal provided in an embodiment of the present application. Referring to fig. 13, the handset includes: radio Frequency (RF) circuitry 1310, memory 1320, input unit 1330, display unit 1340, sensor 1350, audio circuitry 1360, wireless fidelity (WiFi) module 1370, processor 1380, and power supply 1390. Those skilled in the art will appreciate that the handset configuration shown in fig. 13 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.

The following specifically describes each component of the mobile phone with reference to fig. 13:

RF circuit 1310 may be used for receiving and transmitting signals during a message transmission or call, and in particular, for processing received downlink information of a base station by processor 1380; in addition, the data for designing uplink is transmitted to the base station. In general, 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, and the like. In addition, RF circuit 1310 may also communicate with networks and other devices via wireless communications. The wireless communication may use any communication standard or protocol, including but not limited to global system for mobile communications (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.

The memory 1320 may be used to store software programs and modules, and the processor 1380 executes various functional applications and data processing of the cellular phone by operating the software programs and modules stored in the memory 1320. The memory 1320 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, a phonebook, etc.) created according to the use of the cellular phone, and the like. Further, the memory 1320 may include high speed random access memory and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.

The input unit 1330 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the cellular phone. Specifically, the input unit 1330 may include a touch panel 1331 and other input devices 1332. Touch panel 1331, also referred to as a touch screen, can collect touch operations by a user on or near the touch panel 1331 (e.g., operations by a user on or near touch panel 1331 using any suitable object or accessory such as a finger, a stylus, etc., and spaced touch operations within a certain range on touch panel 1331), and drive corresponding connected devices according to a preset program. Alternatively, the touch panel 1331 may include two portions of a touch detection device and a touch controller. The touch detection device detects the touch direction of a user, detects a signal brought by touch operation and transmits the signal to the touch controller; the touch controller receives touch information from the touch sensing device, converts the touch information into touch point coordinates, and sends the touch point coordinates to the processor 1380, where the touch controller can receive and execute commands sent by the processor 1380. In addition, the touch panel 1331 may be implemented by various types, such as a resistive type, a capacitive type, an infrared ray, and a surface acoustic wave. The input unit 1330 may include other input devices 1332 in addition to the touch panel 1331. In particular, other input devices 1332 may include, but are not limited to, one or more of a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, and the like.

The display unit 1340 may be used to display information input by or provided to the user and various menus of the mobile phone. The display unit 1340 may include a display panel 1341, and optionally, the display panel 1341 may be configured in the form of a Liquid Crystal Display (LCD), an organic light-emitting diode (OLED), or the like. Further, touch panel 1331 can overlay display panel 1341, and when touch panel 1331 detects a touch operation on or near touch panel 1331, processor 1380 can be configured to determine the type of touch event, and processor 1380 can then provide a corresponding visual output on display panel 1341 based on the type of touch event. Although in fig. 13, the touch panel 1331 and the display panel 1341 are two independent components to implement the input and output functions of the mobile phone, in some embodiments, the touch panel 1331 and the display panel 1341 may be integrated to implement the input and output functions of the mobile phone.

The handset may also include at least one sensor 1350, such as light sensors, motion sensors, and other sensors. Specifically, the light sensor may include an ambient light sensor that adjusts the brightness of the display panel 1341 according to the brightness of ambient light, and a proximity sensor that turns off the display panel 1341 and/or the backlight when the mobile phone is moved to the ear. As one of the motion sensors, the accelerometer sensor can detect the magnitude of acceleration in each direction (generally, three axes), can detect the magnitude and direction of gravity when stationary, and can be used for applications of recognizing the gesture of the mobile phone (such as horizontal and vertical screen switching, related games, magnetometer gesture calibration), vibration recognition related functions (such as pedometer and tapping), and the like; as for other sensors such as a gyroscope, a barometer, a hygrometer, a thermometer, and an infrared sensor, which can be configured on the mobile phone, further description is omitted here.

Audio circuitry 1360, speaker 1361, microphone 1362 may provide an audio interface between the user and the cell phone. The audio circuit 1360 may transmit the electrical signal converted from the received audio data to the speaker 1361, and the electrical signal is converted into a sound signal by the speaker 1361 and output; on the other hand, the microphone 1362 converts the collected sound signal into an electric signal, converts the electric signal into audio data after being received by the audio circuit 1360, and then processes the audio data by the audio data output processor 1380, and then sends the audio data to, for example, another cellular phone via the RF circuit 1310, or outputs the audio data to the memory 1320 for further processing.

WiFi belongs to short-distance wireless transmission technology, and the mobile phone can help a user to receive and send e-mails, browse webpages, access streaming media and the like through the WiFi module 1370, and provides wireless broadband internet access for the user. Although fig. 13 shows the WiFi module 1370, it is understood that it does not belong to the essential constitution of the handset, and may be omitted entirely as needed within the scope not changing the essence of the invention.

The processor 1380 is a control center of the mobile phone, connects various parts of the entire mobile phone using various interfaces and lines, and performs various functions of the mobile phone and processes data by operating or executing software programs and/or modules stored in the memory 1320 and calling data stored in the memory 1320, thereby integrally monitoring the mobile phone. Optionally, processor 1380 may include one or more processing units; alternatively, processor 1380 may integrate an application processor, which handles primarily the operating system, user interface, and applications, and a modem processor, which handles primarily wireless communications. It will be appreciated that the modem processor described above may not be integrated within processor 1380.

The handset also includes a power supply 1390 (e.g., a battery) to provide power to the various components, which may optionally be logically coupled to the processor 1380 via a power management system to manage charging, discharging, and power consumption management via the power management system.

Although not shown, the mobile phone may further include a camera, a bluetooth module, etc., which will not be described herein.

In the embodiment of the present application, the processor 1380 included in the terminal further has a function of performing the respective steps of the page processing method as described above.

Referring to fig. 14, fig. 14 is a schematic structural diagram of a server according to an embodiment of the present invention, where the server 1400 may generate a relatively large difference due to different configurations or performances, and may include one or more Central Processing Units (CPUs) 1422 (e.g., one or more processors) and a memory 1432, and one or more storage media 1430 (e.g., one or more mass storage devices) storing an application 1442 or data 1444. Memory 1432 and storage medium 1430, among other things, may be transient or persistent storage. The program stored on storage medium 1430 may include one or more modules (not shown), each of which may include a sequence of instructions operating on a server. Still further, a central processor 1422 may be disposed in communication with storage medium 1430 for executing a series of instruction operations on storage medium 1430 on server 1400.

The 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 XTM, unixTM, linuxTM, freeBSDTM, etc.

The steps performed by the management apparatus in the above-described embodiment may be based on the server configuration shown in fig. 14.

In an embodiment of the present application, a computer-readable storage medium is further provided, where management instructions for recommended content are stored in the computer-readable storage medium, and when the management instructions are executed on a computer, the computer is caused to perform the steps performed by the management apparatus for recommended content in the method described in the foregoing embodiments shown in fig. 2 to 11.

Also provided in the embodiments of the present application is a computer program product including instructions for managing recommended content, which when run on a computer, cause the computer to perform the steps performed by the apparatus for managing recommended content in the methods described in the embodiments shown in fig. 2 to 11.

The embodiment of the present application further provides a management system for recommended content, where the management system for recommended content may include a management device for recommended content in the embodiment described in fig. 12, a terminal device in the embodiment described in fig. 13, or a server described in fig. 14.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one type of logical functional division, and other divisions may be realized in practice, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a recommended content management apparatus, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.

Claims

1. A method for managing recommended content, comprising:

the method comprises the steps that interactive operation of a target object on different forms of recommended contents in a recommendation set in a plurality of time windows is obtained, so that interactive data corresponding to the target object are determined;

2. The method of claim 1, wherein the obtaining interaction operations of the target object on different forms of recommended content in the recommendation set within a plurality of time windows to determine interaction data corresponding to the target object comprises:

acquiring a recommendation cycle corresponding to the target object;

3. The method of claim 1, wherein the counting object features and content features in the interaction data based on preset feature dimensions to obtain an object representation comprises:

acquiring interactive operation of the target object on different types of recommended content in the data transmission mode to determine the object characteristics;

and integrating the object characteristics and the content characteristics to obtain the object portrait.

4. The method according to claim 3, wherein the obtaining the interaction operations of the target object on different types of recommended content in the data transmission mode to determine the object characteristics comprises:

5. The method according to claim 3, wherein the determining, according to the historical data corresponding to the target object, time interval division information indicated by the preset feature dimension includes:

extracting hotspot time periods based on the interaction frequency information;

6. The method of claim 3, further comprising:

determining a confidence parameter corresponding to the target object;

7. The method according to claim 1, wherein the invoking data corresponding to the operation item in the content feature to perform statistics of accumulated revenue to determine a revenue feature value corresponding to the recommended content comprises:

8. The method of claim 7, wherein the performing statistics of accumulated revenue based on the associated content set to determine the revenue feature value corresponding to the recommended content comprises:

determining click information corresponding to the associated content set;

if the click information reaches a preset condition, carrying out statistics on accumulated revenue based on the associated content set so as to determine the revenue characteristic value corresponding to the recommended content;

or;

9. The method of claim 8, further comprising:

extracting the total number of clicks indicated in the object picture;

and if the total clicking number does not meet the clicking condition, calling the crowd portrait to replace recommended contents of all types in the object portrait.

10. The method of claim 9, further comprising:

and screening the recommended contents under each type based on exposure conditions so as to replace the recommended contents under the types which do not meet the exposure conditions by adopting the crowd portrait.

11. The method of claim 1, wherein said combining the revenue eigenvalue and the predicted value included in the interaction eigenvalue set to calculate a target eigenvalue comprises:

determining a content type corresponding to the recommended content;

determining a corresponding feature weighting factor based on the content type;

12. The method according to claim 1, wherein the recommended content is a heterogeneous card, the content form of the heterogeneous card includes image-text content, video content or link content, the recommended interface is a terminal application interface, and the target content is displayed in a focus area of the terminal application interface.

13. A management apparatus for recommended content, comprising:

the statistical unit is used for counting the object features and the content features in the interactive data based on preset feature dimensions to obtain an object portrait, wherein the preset feature dimensions are set based on the time window, the interactive operation and a data transmission mode corresponding to the recommended content;

14. A computer device, the computer device comprising a processor and a memory:

the memory is used for storing program codes; the processor is configured to execute the method for managing recommended content according to any one of claims 1 to 12 according to instructions in the program code.

15. A computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to execute the method of managing recommended content of any one of claims 1 to 12.