CN111143697B

CN111143697B - Content recommendation method and related device

Info

Publication number: CN111143697B
Application number: CN202010001701.2A
Authority: CN
Inventors: 宋天凡; 刘若尘; 沈琼烨
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2020-01-02
Filing date: 2020-01-02
Publication date: 2023-03-21
Anticipated expiration: 2040-01-02
Also published as: CN111143697A

Abstract

The application discloses a content recommendation method and a related device, which are used for recommending content by acquiring user figures of a plurality of users using target applications; then determining at least one first feature in the user representation; classifying the users according to the at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different; and further determining corresponding recommended content according to the user portrait of the target user and the user set to which the target user belongs. Therefore, content recommendation based on the user portrait is achieved, the process cannot be influenced by relevance among different contents, due to the stability of the user portrait, the recommendation range can be expanded conveniently based on the user portrait, and the flexibility and the application range of the content recommendation process are further improved.

Description

Content recommendation method and related device

Technical Field

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

Background

With the development of the related technologies of mobile terminals, more and more intelligent devices appear in the lives of people, wherein content recommendation through the intelligent devices is particularly prominent, however, due to the appearance of more and more contents, content matching cannot be performed on users one by one, and how to perform content recommendation on users becomes a problem.

Generally, content recommendation can be performed by using the similarity between contents, that is, the corresponding relationship between the pairwise similarity matrix among a plurality of contents and the behavior vector of the user is determined, so that the preference of the current user on all contents can be obtained, and prediction and recommendation of related contents can be performed.

However, in a scene with a complex content category, there may be no association, i.e. dissimilarity, between different contents; in this case, the recommendation process for a plurality of contents cannot be performed for contents of different categories, and the application range of content recommendation is affected.

Disclosure of Invention

In view of this, the present application provides a method for recommending content, which can effectively avoid inapplicability of recommendation due to differences between contents, and improve the application range of the content recommendation process.

A first aspect of the present application provides a content recommendation method, which may be applied to a system or a program including a content recommendation function in a terminal device, and specifically includes: obtaining a user representation of a plurality of users using a target application, the user representation determined based on a user's interaction record with a plurality of content;

determining at least one first feature in the user representation;

classifying users according to the at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different;

and determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs.

Optionally, in some possible implementation manners of the present application, the determining, according to the user image of the target user and the user set to which the target user belongs, the recommended content of the target user includes:

determining a recommended content set according to the user image of the target user and the user set to which the target user belongs, wherein the recommended content set comprises first content and second content;

respectively calculating first correlation matrixes of the first content and the second content based on the interaction records so as to obtain preference values according to a first formula;

and determining the recommended content of the target user according to the preference value.

Optionally, in some possible implementation manners of the present application, the determining, according to the user image of the target user and the user set to which the target user belongs, the recommended content of the target user includes: classifying the user portraits in the user set according to a second feature to obtain a first sub-set and a second sub-set, wherein the user portraits in the first sub-set have the second feature, the user portraits in the second sub-set do not have the second feature, and the second feature has a different dimension from the first feature in classification of the user portraits;

and if the ratio of the number of users corresponding to the first subset to the number of users corresponding to the user set meets a first condition, determining the recommended content of the target user based on the second characteristic.

Optionally, in some possible implementations of the present application, the determining the recommended content of the target user based on the second feature includes:

determining a third feature that the first subset and the second subset have in common, the third feature being different from the first feature in a dimension of classification of the user representation;

acquiring content corresponding to the second characteristic and content corresponding to the third characteristic;

respectively calculating a second correlation matrix of the content corresponding to the second characteristic and the content corresponding to the third characteristic based on the interaction record;

inputting the second correlation matrix into a second formula to obtain the preference value;

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

acquiring feature data in the interaction records, wherein the feature data is determined based on a behavior vector of a user;

and updating the preference value according to the characteristic data.

Optionally, in some possible implementation manners of the present application, the classifying users according to the at least one first feature to obtain at least two user sets includes:

clustering the users according to the at least one first characteristic to determine a central point;

and classifying according to the distance between the parameter of the user corresponding to the first characteristic and the central point to obtain at least two user sets.

Optionally, in some possible implementations of the present application, the determining at least one first feature in the user representation includes:

determining a feature set from the user representation;

obtaining the variance of each feature in the feature set by adopting principal component analysis;

determining at least one variance meeting a second condition to determine the corresponding first feature, wherein the preset condition is set based on the numerical value of the variance.

Optionally, in some possible implementations of the present application, the determining a feature set according to the user representation includes:

determining that the user portrayal corresponds to a plurality of fourth features under different contents;

acquiring missing values of a plurality of the fourth features;

screening a plurality of fourth features of which the missing values are smaller than a first threshold value to determine the feature set.

determining a deficiency parameter corresponding to the deficiency value;

determining a filling parameter of the missing parameter in a plurality of the fourth features;

and updating the missing parameters according to the filling parameters to determine the feature set.

Optionally, in some possible implementations of the present application, the screening a plurality of fourth features of which the missing value is smaller than the first threshold to determine the feature set includes:

screening a plurality of fourth features of which the missing values are smaller than a first threshold value;

obtaining a correlation degree among a plurality of fourth features;

combining the fourth features with the correlation degrees larger than a second threshold value to obtain fifth features;

determining the feature set according to the fifth feature.

Optionally, in some possible implementations of the present application, the merging the fourth features whose correlation degrees are greater than the second threshold to obtain the fifth feature includes:

determining a fourth characteristic of which the correlation degree is greater than a second threshold value, and acquiring a corresponding unit parameter;

and normalizing the unit parameters and combining the unit parameters to obtain a fifth characteristic.

Optionally, in some possible implementations of the application, the content is a game, and the first feature is used to indicate a game type.

A second aspect of the present application provides an apparatus for content recommendation, including: an acquisition unit to acquire a user representation of a plurality of users using a target application, the user representation determined based on a user's interaction record with a plurality of content;

a determination unit to determine at least one first feature in the user representation;

the classification unit is used for classifying users according to the at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different;

and the recommending unit is used for determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs.

Optionally, in some possible implementation manners of the present application, the recommending unit is specifically configured to determine a recommended content set according to the user image of the target user and a user set to which the target user belongs, where the recommended content set includes a first content and a second content;

the recommendation unit is specifically configured to calculate first correlation matrices of the first content and the second content respectively based on the interaction records, so as to obtain preference values according to a first formula;

and the recommending unit is specifically used for determining the recommended content of the target user according to the preference value.

Optionally, in some possible implementations of the present application, the recommending unit is specifically configured to classify the user portraits in the user set according to a second feature to obtain a first sub-set and a second sub-set, where the user portraits in the first sub-set have the second feature, the user portraits in the second sub-set do not have the second feature, and the second feature is different from the first feature in a classification dimension of the user portraits;

the recommending unit is specifically configured to determine the recommended content of the target user based on the second feature if a ratio of the number of users corresponding to the first subset to the number of users corresponding to the user set meets a first condition.

Optionally, in some possible implementations of the present application, the recommending unit is further configured to determine a third feature that the first subset and the second subset have in common, where the third feature is different from the first feature in a classification dimension of the user portrait;

the recommending unit is specifically configured to acquire the content corresponding to the second feature and the content corresponding to the third feature;

the recommendation unit is specifically configured to calculate, based on the interaction record, second correlation matrices of the content corresponding to the second feature and the content corresponding to the third feature respectively;

the recommending unit is specifically configured to input the second correlation matrix into a second formula to obtain the preference value;

Optionally, in some possible implementation manners of the present application, the recommending unit is further configured to obtain feature data in the interaction record, where the feature data is determined based on a behavior vector of a user;

the recommending unit is specifically configured to update the preference value according to the feature data.

Optionally, in some possible implementation manners of the present application, the classifying unit is specifically configured to cluster the users according to the at least one first feature to determine a central point;

the classification unit is specifically configured to classify the user according to a distance between a parameter corresponding to the first feature and the central point, so as to obtain at least two user sets.

Optionally, in some possible implementations of the present application, the determining unit is specifically configured to determine a feature set according to the user portrait;

the determining unit is specifically configured to obtain a variance of each feature in the feature set by using principal component analysis;

the determining unit is specifically configured to determine at least one variance that satisfies a second condition to determine the corresponding first feature, where the preset condition is set based on a magnitude of the variance.

Optionally, in some possible implementation manners of the present application, the determining unit is specifically configured to determine that the user icon corresponds to a plurality of fourth features under different contents;

the determining unit is specifically configured to acquire missing values of a plurality of the fourth features;

the determining unit is specifically configured to filter a plurality of fourth features of which the missing values are smaller than a first threshold value, so as to determine the feature set.

Optionally, in some possible implementation manners of the present application, the determining unit is further configured to determine a missing parameter corresponding to the missing value;

the determining unit is specifically configured to determine a filling parameter of the missing parameter in the plurality of fourth features;

the determining unit is specifically configured to update the missing parameter according to the padding parameter, so as to determine the feature set.

Optionally, in some possible implementations of the present application, the determining unit is specifically configured to filter a plurality of fourth features of which the missing values are smaller than a first threshold;

the determining unit is specifically configured to obtain a correlation between the plurality of fourth features;

the determining unit is specifically configured to combine fourth features with correlation degrees larger than a second threshold to obtain fifth features;

the determining unit is specifically configured to determine the feature set according to the fifth feature.

Optionally, in some possible implementation manners of the present application, the determining unit is specifically configured to determine a fourth feature that the correlation degree is greater than a second threshold, and obtain a corresponding unit parameter;

the determining unit is specifically configured to normalize and combine the unit parameters to obtain a fifth feature.

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 perform the method of content recommendation of the first aspect or any one of the first aspects 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 run on a computer, cause the computer to perform the method of content recommendation of any of the first aspect or the first aspect described above.

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

obtaining a user representation of a plurality of users using a target application, the user representation being determined based on a user's interaction record with a plurality of content; then determining at least one first feature in the user representation; classifying the users according to the at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different; and further determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs. Therefore, content recommendation based on the user portrait is achieved, the process cannot be influenced by relevance among different contents, due to the stability of the user portrait, the recommendation range can be expanded conveniently based on the user portrait, and the flexibility and the application range of the content recommendation process are further improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only 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 content recommendation system operates;

FIG. 2 is a scene diagram of a content recommendation method;

fig. 3 is a flowchart of content recommendation provided in an embodiment of the present application;

fig. 4 is a flowchart of a method for recommending content according to an embodiment of the present application;

FIG. 5 is a flowchart of another method for content recommendation provided by an embodiment of the present application;

fig. 6 is a schematic diagram illustrating a correspondence relationship between a feature indicator field and a feature indicator according to an embodiment of the present application;

fig. 7 is a schematic view of a content recommendation scenario provided in an embodiment of the present application;

fig. 8 is a schematic view of another scenario of content recommendation provided in an embodiment of the present application;

fig. 9 is a schematic view of another scenario of content recommendation provided in an embodiment of the present application;

fig. 10 is a schematic structural diagram of a content recommendation device according to an embodiment of the present application;

fig. 11 is a schematic structural diagram of a terminal device according to an embodiment of the present application.

Detailed Description

The embodiment of the application provides a content recommendation method and a related device, which can be applied to a system or a program containing a content recommendation function in terminal equipment, and can be used for acquiring user figures of a plurality of users using a target application, wherein the user figures are determined based on interaction records of the users and a plurality of contents; then determining at least one first feature in the user representation; classifying the users according to the at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different; and further determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs. Therefore, content recommendation based on the user portrait is achieved, the process cannot be influenced by relevance among different contents, due to the stability of the user portrait, the recommendation range can be expanded conveniently based on the user portrait, and the flexibility and the application range of the content recommendation process are further improved.

The terms "first," "second," "third," "fourth," and the like in the description and claims of this application and in the above-described drawings, 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. Furthermore, 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 expressly 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.

User portrait: the user information tagging is to abstract a tag set of a user by collecting and analyzing data of main information such as social attributes, living habits, consumption behaviors and the like of the user.

Preference degree: a digital representation of a user's likeability to a plurality of content.

And (3) interactive recording: the operation log of the user in the process of using the related program can comprise input content, output content or digital representation of specific operation process.

A behavior vector: the word vector is used for indicating the operation behavior of the user, and particularly is a vector in which words or phrases indicating the behavior of the user are mapped to real numbers.

It should be understood that the content recommendation method provided by the present application may be applied to a system or a program including a content recommendation function in a terminal device, for example, in an application program such as a webame game platform, a large number of different types of games exist in an application program background, however, a recommendation bit in a display interface of the application is limited, and these game contents cannot be all displayed, and a suitable game content needs to be selected to improve the interaction degree of a user; by the content recommendation method, the game content with higher user preference degree can be set at the corresponding recommendation position, the user can perform further interactive operation with high probability, and the activity of the user and a further interactive conversion process are improved. Specifically, the content recommendation system may operate in a network architecture as shown in fig. 1, which is a network architecture diagram of the content recommendation system, as can be seen from the diagram, the content recommendation system may provide content recommendation with a plurality of information sources, the terminal establishes a connection with the server through the network, further receives a plurality of contents sent by the server, and performs ranking recommendation according to the terminal itself, or performs recommendation according to the ranking sent by the server; it can be understood that, fig. 1 shows various terminal devices, in an actual scenario, there may be more or fewer types of terminal devices participating in the content recommendation process, and the specific number and type depend on the actual scenario, which 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-content application interaction, the specific number of servers depends on the actual scenario.

It should be noted that the content recommendation method provided in this embodiment may also be performed offline, that is, without the participation of a server, at this time, the terminal is locally connected with other terminals, and then a process of content recommendation between terminals is performed.

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

However, in a scene with a complex content category, there may be no association, i.e. dissimilarity, between different contents; at this time, the recommendation process of multiple contents cannot be performed on the contents with different categories, which affects the application range of content recommendation; as shown in fig. 2, which is a scene diagram of a content recommendation method, in the diagram, since a user 1 likes a content 1 and a content 3, and a user 3 likes the content 1, since the content 1 and the content 3 are both liked by the user 1, an association relationship is generated, so that the content 3 can be recommended to the user 1; however, if the user 3 does not like the content 1 or if the user 3 does not have history data for the contents 1 to 3, the above-described related recommendation process cannot be performed, which causes a problem that the cold start cannot be performed.

In order to solve the above problem, the present application provides a method for content recommendation, which is applied to the flow framework of content recommendation shown in fig. 3, as shown in fig. 3, for a flow framework of content recommendation provided in an embodiment of the present application, first collecting behavior vectors of relevant users from a server to generate a user portrait, and then extracting representative features in the user portrait to classify the users; and recommending the content according to the classified related characteristics in the user set, and displaying the content at the client.

It is understood that the method provided by the present application may be a program written as a processing logic in a hardware system, or may be a content recommendation device that implements the processing logic in an integrated or external manner. As one implementation, the content recommendation device obtains a user profile of a plurality of users using a target application, the user profile determined based on a record of user interactions with a plurality of content; then determining at least one first feature in the user representation; classifying the users according to the at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different; and further determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs. Therefore, content recommendation based on the user portrait is achieved, the process cannot be influenced by relevance among different contents, due to the stability of the user portrait, the recommendation range can be expanded conveniently based on the user portrait, and the flexibility and the application range of the content recommendation process are further improved.

With reference to the above flow architecture, the following describes a content recommendation method in the present application, please refer to fig. 4, where fig. 4 is a flow chart of a content recommendation method provided in an embodiment of the present application, and the content recommendation method may be applied to game content recommendation applications, for example: a wegame gaming platform; it can also be applied to the recommendation of song content, for example: QQ music; the embodiment of the application at least comprises the following steps:

401. a user representation of a plurality of users using a target application is obtained.

In this embodiment, the user representation is determined based on a record of user interaction with the plurality of content; the target application may be application software including a plurality of contents, such as games, music, or videos, and the plurality of contents may be different kinds of games, for example: paid games, free games, stand-alone games, network games, or the like; the plurality of contents may also be different styles or charging states of music, and the specific form depends on the actual scene, which is not limited herein.

In one possible scenario, the target application is a webame game platform, and the users are groups of users who have game downloads, clicks or purchases in the webame game platform. Specifically, the process of determining the user may be to invoke all users with accounts of the platform; or for users in a particular game category, such as: a stand-alone game user; the user data in other applications in the game indicated by the platform can also be called to perform content recommendation presetting on the users, and the specific mode is determined according to the actual scene.

It is understood that the user representation of the plurality of users may be historical information received from server statistics, or may be interaction information between a plurality of applications stored in the terminal itself. It should be noted that the process of acquiring the user portrait in the present embodiment is not limited to one type of application, and user portrait acquisition across categories of applications may be performed, for example: consumption data of a user in a game application, a music application and a video application and corresponding application tags are collected.

402. At least one first feature in the user representation is determined.

In this embodiment, the first feature may be a representative feature, for which most of the user images are related, but parameters corresponding to the representative feature are different in different user images, that is, the user images present a distribution with a large dispersion based on the representative feature; specifically, different representative characteristics are corresponded to different scenes; for example: in the application of the game platform, the user image generally comprises the game duration, but specific parameters of the game duration are different, some users may have the game duration as long as 8 hours, but some users may have the game duration of only half an hour, so that the user image forms a distribution with larger discreteness based on the characteristic of the game duration, and the game duration can be taken as a representative characteristic.

Alternatively, the first feature may be determined by using Principal Component Analysis (PCA), i.e., finding a feature that keeps the contribution to the parallax greatest in the user representation data set; the characteristics are used as the first characteristics to facilitate the following user classification, so that the users with similar characteristics are clustered together, the matching of the recommended content and the user set is conveniently judged, the problem of user screening one by one is avoided, and the efficiency of selecting a specific user group according to the recommended content in the follow-up process is improved.

It is to be understood that the screening condition of the above embodiment is the feature with the largest variance, and in some scenarios, a plurality of features with variances reaching the preset condition may be selected. Specifically, a feature set is determined according to the user portrait; then, acquiring the variance of each feature in the feature set by adopting principal component analysis; and then determining at least one variance meeting a preset condition to determine a plurality of corresponding features, wherein the preset condition is set based on the numerical value of the variance.

In one possible scenario, it is considered that the interaction records in the user image are obtained by the user through statistics in the historical data based on the behavior characteristics, but the data may be missing in the process of statistics, such as: the game duration of the user in 1 month per day is counted, but data of some dates are not counted possibly due to network fluctuation and other reasons, the problem of incomplete data is possibly caused, and further inaccuracy of variance calculation is caused. Specifically, first, a plurality of fourth features corresponding to different contents of the user image are determined, wherein the fourth features are feature data with data missing possibly; then acquiring a plurality of missing values of the fourth feature; and further screening a plurality of fourth features of which the missing values are smaller than the first threshold value to determine the feature set. For example, the fourth feature is a feature set with a missing value less than 50%.

Optionally, for the feature that the filtered missing value is smaller than the first threshold, there still exists a certain missing of data, for example: the missing value is 20%, which is less than the first threshold value 50%, but there is still a data missing situation in the data of the feature; at this time, in order to ensure the accuracy of the feature data, the missing item may be further filled. Specifically, a deficiency parameter corresponding to the deficiency value may be determined first; then determining filling parameters of the missing parameters in a plurality of fourth features; and updating the missing parameter according to the filling parameter to determine the feature set. For example: for the features with the missing values smaller than 5%, replacing missing parameters by using the mean value of filling parameters, and then determining a feature set; in one possible scenario, the time duration data of the stand-alone game in the user representation indicates that the data of the user in a week is 4, 6, 4, 6, and there is a day where data is not recorded, i.e. missing parameters, possibly due to network fluctuation, and at this time, the missing parameters may be padded with 5 if the average value calculated from the existing data is 5. Because the missing items of the feature data are filled, on one hand, the integrity of the feature data is ensured, and the user portrait indicated by the feature data is more accurate; on the other hand, for missing item filling of feature data, horizontal comparison between multiple users is facilitated, such as: after the users are classified based on the first features, similarity of feature data in specific user images among the users is further checked, and parameter optimization of an algorithm is facilitated.

Optionally, in the screening process of multiple features, there may be some features with similar parameter expressions, for example: login times, account number input times and the like; these similar features may be combined at this point. Specifically, the feature merging may be performed before the screening according to the missing condition of the feature data, or may be performed after the screening according to the missing condition of the feature data, and the specific order is not limited. Next, the feature data after being filtered according to the absence of the feature data, that is, the fourth feature will be described. Firstly, screening a plurality of fourth characteristics of which the missing values are smaller than a first threshold value; then obtaining the correlation degree among a plurality of fourth characteristics; then, combining the fourth features with the correlation degrees larger than a second threshold value to obtain fifth features; and then, the feature set is determined according to the fifth feature, namely, each feature item in the feature set describes the user portrait from different angles, so that the validity of data is ensured.

Optionally, there may be a case where the parameter units of the multiple features do not correspond to each other, for example: some features are described in units of times and other features are described in units of frequencies, which may have a computational impact on the process of variance statistics. Therefore, the unit parameters of each feature may be normalized, and specifically, for the fourth feature with the correlation degree greater than the second threshold in the foregoing embodiment, the corresponding unit parameter is obtained; then, the unit parameters are normalized and combined to obtain a fifth feature. Normalization is to convert unit parameters of each feature into relative value data through conversion of some algorithms, for example: transforming the parameters of each feature into relative values between 0 and 1; the normalized algorithm may adopt a standardized Standardization algorithm, a Standardization algorithm, or a MinMax Scale algorithm, and the specific algorithm is determined by an actual scene and is not limited herein.

It is to be understood that the above normalization process may be based on feature data after correlation degree combination, feature data after missing value screening, or initial feature data, and the specific feature data processing process may be performed based on any sort of the above missing value screening, similarity combination, and normalization processes, and the above order is only an example and is not limited.

403. And classifying the users according to the at least one first characteristic to obtain at least two user sets.

In the embodiment, the preference degree of each user set to the target application is different; that is, the preference degree of the multiple user sets for the target application can be presented in the form of gradient, for example: the user is divided into a core or a periphery according to the gradient of the preference degree, at this time, two classifications are taken as an example for description, specifically, more classifications may also be used, and the corresponding classification name is determined by a specific scenario, which is not limited herein.

It is understood that the classification result of the user set may be divided according to the relative distance between the center points after the user classification. Specifically, the users may be clustered by using a Kmeans algorithm to obtain at least one cluster center point, and then sorted according to the relative distance between the parameter of each user under the corresponding first characteristic and the center point to determine a specific user set, thereby implementing user classification.

404. And determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs.

In this embodiment, the recommended content may be determined based on features included in each user portrait in a user set to which the target user belongs, for example: the feature "single-machine game payment amount" is included in the user set to which the target user belongs, and at this time, the single-machine game can be recommended to the target user.

Optionally, if the corresponding recommended content includes multiple contents, for example: and recommending the stand-alone games, wherein the stand-alone games are multiple, and the preference values of multiple contents can be calculated at the moment, so that the sequence of the recommended contents is obtained. Specifically, a recommended content set is determined according to the user portrait of the target user and a user set to which the target user belongs, wherein the recommended content set comprises first content and second content, for example, any two games in a stand-alone game are selected; then, respectively calculating a first correlation matrix of the first content and the second content based on the interaction record so as to obtain a preference value according to a first formula; and then determining the recommended content of the target user according to the preference value. The first correlation matrix may be calculated according to an Otsuka-Ochiai coeffient formula, and further obtain a preference value, specifically, the calculation may be performed based on the formula:

wherein:

n (a ≈ B) = users who simultaneously prefer the first content and the second content;

n (a) = users who prefer the first content;

n (B) = a user who prefers the second content;

k is used to indicate a first correlation matrix of the first content and the second content.

Optionally, in some possible scenarios, a part of users in the user set to which the target user belongs includes the second feature, and another part does not include the second feature, and at this time, the determination of the related recommended content cannot be directly performed on the feature thereof, where the second feature is a distinguishing feature in the user set, for example, a part of users prefer a standalone game, and another part does not have a tag of the standalone game. Due to the similarity of user representations in the same user set, if a certain number of users including the second feature are reached, it can be considered that another part should also include the second feature. In particular, the user portraits in the user set may be classified according to the second feature to obtain the first subset and the second subset, for example: the second characteristic is stand-alone game, the first subset and the second subset are combined into a user set which prefers stand-alone game and a user set which does not mark preference stand-alone game; wherein the user representations in the first subset have the second feature and the user representations in the second subset do not have the second feature, and the second feature is different from the first feature in a classification dimension of the user representations; and when the ratio of the number of users corresponding to the first subset to the number of users corresponding to the user set meets a first condition, determining the recommended content of the target user based on the second characteristic. For example: if more than 50% of the users in the online game have the consumption records of the stand-alone game, the remaining users who do not have the consumption records of the stand-alone game are considered to be interested in the stand-alone game and are recommended.

In addition, the above-mentioned process of content recommendation according to the second feature may also relate to the recommendation order of different contents in the same type of application, for example, multiple content recommendations for the feature of a stand-alone game in a set of online game users. At this time, a third feature corresponding to the second subset may be determined, where the third feature is a content feature preferred by all users in the user set, for example: users in the plurality of user sets all prefer the online game; then acquiring the content corresponding to the second characteristic and the content corresponding to the third characteristic; respectively calculating a second correlation matrix of the content corresponding to the second characteristic and the content corresponding to the third characteristic based on the interaction record; inputting the second correlation matrix into a second formula to obtain the preference value; and then determining the recommended content of the target user according to the preference value, thereby further optimizing the preference value based on different dimensions. Wherein the first correlation matrix may be calculated according to the Otsuka-Ochiai coeffient formula.

Optionally, the data of the preference value may be weighted with some specific behavior vectors of the user, for example: the preference value may be updated by performing a weighted calculation on the purchase record of the specific content, discount information of the specific content, or the specific content that the user used. The specific weighting method depends on the action influence in the actual scene, and for example, a coefficient weighting of 1.5 times is performed on the preference value of the content having discount information, which is not limited herein. Through the weighted calculation of the user preference value, the influence of accidental factors on the recommended content can be better simulated, such as: and the occurrence of events such as discount and limit amount further improves the accuracy of content recommendation and the adaptability to different scenes.

Optionally, in the wegame platform, the recommended content is a recommended game; after the recommended content of the target user is determined, displaying tags, representative images or other elements which are relevant to the game in the recommended position of the platform or displaying the tags, the representative images or other elements in a pop-up window form; because the games are games with higher preference degrees acquired through the steps, after seeing the content recommendation, the user can further perform interactive operation with the webame game platform, so that the user activity of the webame game platform is improved.

In connection with the above embodiments, by obtaining a user representation of a plurality of users using a target application, the user representation is determined based on a user's interaction record with a plurality of content; then determining at least one first feature in the user representation; classifying the users according to the at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different; and further determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs. Therefore, content recommendation based on the user portrait is achieved, the process cannot be influenced by relevance among different contents, due to the stability of the user portrait, the recommendation range can be expanded conveniently based on the user portrait, and the flexibility and the application range of the content recommendation process are further improved.

In the foregoing embodiment, a content recommendation process is described, and in the following, a game application is described as a specific scenario, please refer to fig. 5, where fig. 5 is a flowchart of another content recommendation method provided in the embodiment of the present application, and the embodiment of the present application at least includes the following steps:

501. and acquiring user behavior data in the database.

In this embodiment, for a game scenario applied in this embodiment, behavior data of a user may be cross-platform data collection performed by a plurality of game platforms, may also be data collection of user historical operation records of a plurality of different games, and may also be a process of traversing relevant data in a database based on a certain keyword, for example: if the user needs to recommend the stand-alone game, the keywords related to the stand-alone game are traversed in the database, and corresponding data are collected. The database may be a local database, a database group connected through a network, or shared data of a cloud.

502. And counting the behavior data to obtain the user portrait.

In this embodiment, a further screening process may be performed on the user behavior data obtained in step 501, specifically, a feature set as shown in fig. 6 may be obtained after screening a missing value, fig. 6 is a schematic diagram of a corresponding relationship between a feature index field and a feature index provided in this embodiment of the present application, and further, a user portrait is digitized according to the feature index shown in fig. 6.

503. And extracting representative behavior characteristics.

In this embodiment, the representative behavior feature is a representative feature for indicating the behavior of the user, and the representative feature is a behavior feature related to most of the user images, for example: and (3) purchasing behavior: specifically, in the process of extracting the representative behavior features, feature correlations may be merged first, that is, the feature correlations are calculated, and the feature set may be simplified by the correlation matrix.

Further, unit parameter conversion can be performed in Jupitter notebook by using Python, and unit parameter distribution arrangement conversion is performed firstly, namely, the characteristics in different ranges are unified to the same range, so that the validity of the PCA result and the subsequent characteristic layering are not influenced by different characteristic value units. In this embodiment, a normalization method is adopted, that is, the distribution and correlation of the original data are not changed, wherein the normalization range is default range [0,1].

After the characteristics are screened, because the dimension discreteness of some characteristics in the characteristics is small, the influence on user classification is small, the calculation is convenient, the user hierarchy is better explained, and the representative dimension is selected for hierarchy. Therefore, the feature extraction process of the PCA is carried out next, and the feature which has the largest contribution to the variance in the data set is found; and uses these features for user stratification. Namely, some characteristics can express more variances by visually seeing the First PC and the Second PC. In one possible scenario, five features are available for subsequent user stratification (classification), respectively: game _ cnt (number of stand-alone game purchase activations), amt _ total (total amount of stand-alone game purchases), duration (total time length of stand-alone game), play _ length (active span of stand-alone game), last _ play _ date _ trans (last game date).

504. And carrying out user layering according to the representative behavior characteristics.

In this embodiment, according to the representative behavior characteristics determined in step 503, the users are first clustered by using a Kmeans algorithm,the number of Kmeans categories, K, was then confirmed by the elbow method in combination with the actual conditions. Wherein, due to W _K The smaller the difference between the data is, so that for different K, different central points and clusters to which the data points belong can be obtained after the Kmeans algorithm, and different metrics W can be obtained _K Can select W _K The number of inflection points 5 at which the fall tends to be steady is taken as the number of classifications.

Further, according to the data of the layered central point, namely the number of single-player game purchases, the total purchase amount of the single-player games, the total duration of the single-player games, the active span of the single-player games and the final game date; these 5 types of users are named by the relative distance of the user feature parameter from the center point from high to low: core, subcore, peripheral, latent, and silent.

505. And determining the preference value of the target user for different games.

In this embodiment, for the calculation of the preference value of the same type of game, for example, between a plurality of standalone games, the correlation matrix between the games, the correlation matrix M between the standalone and standalone games, may be calculated first _Single : otsuka-Ochiai coeffient calculations were performed using the number of purchasers, namely:

wherein:

n (A ≈ B) = users who purchase both A, B games simultaneously;

n (a) = user who purchased a game;

n (B) = user who purchased B money game;

k1 matrix M for indicating relevance between single game and single game _Single ；

Further, we get a vector for each user for the case of single-machine payment and perform vector normalization:

norm _1= [ float (i)/sum (raw) for i in raw ]// # i is the pay amount for each stand-alone game;

then, multiplying the correlation matrix of the stand-alone and stand-alone games with the normalized vectors to obtain the preference value of each user among a plurality of stand-alone games:

rank 1＝M _Single *Norm_1；

and further sorting according to the preference value rank1 to determine the recommended content.

In another possible scenario, where preference value calculation for different types of games is required, for example, between a plurality of stand-alone games and a network game, a correlation matrix between games, a stand-alone game and network game correlation matrix M, may be calculated first _Online : otsuka-Ochiai coeffient calculations were performed using the number of persons playing, namely:

wherein:

n (C ≈ D) = users who have played C, D two games simultaneously;

n (C) = users who have played a game of money C;

n (D) = users who have played a game of D money;

k2 matrix M for indicating relevance between single game and single game _Online ；

Further, for each user, a vector of the online game duration is obtained, and vector normalization is performed:

norm _2= [ float (i)/sum (raw) for i in raw ] # i is the game duration of each online game;

multiplying the normalized vector by the correlation matrix of the stand-alone game and the online game to obtain the preference value of each user between a plurality of stand-alone games and the online game,

rank2＝MOnline*Norm_2；

and further sorting according to the preference rank2 to determine the recommended content.

506. And recommending the content according to the preference value.

In this embodiment, performing content recommendation based on preference values may also take into account weighting of preference values of users for relevant operations of a specific game.

Optionally, the preference value of the user whose relevant characteristic parameter reaches a certain threshold value may be weighted. For example, for 10 games purchased on the platform, and the frequency of playing games logged on the platform reaches 2 times per week in the last two years, the user portrait defines the games as core users, since the exploration degree of the users on the platform games is higher, most of the games in the platform are already known and choose not to purchase, the ranking of new games in the original rank in 2019 in the preference sequence of the user games in the part is improved, and the initial preference score of the new games is multiplied by 1.5 to obtain an updated preference value. For this type of user, the recommendation method shown in fig. 7 may be adopted, as shown in fig. 7, the scene diagram of content recommendation provided in the embodiment of the present application is shown, a game A1 (content 1) after weighting is shown in the diagram, and a display area of the game A1 is a portion of a display interface with the largest proportion.

Alternatively, the Target Group Index (TGI) may also be calculated from historical discounted purchases to reflect the strength or weakness of the Target Group within a particular research range (game discount). Wherein TGI index = ratio of population having a certain characteristic in the target population/ratio of population having the same characteristic in the population; for example, the TGI for 9-fold game is equal to the purchase ratio of the full platform 9-fold game to the purchase ratio of the full platform discounted game. In one possible scenario, the preference degree of the user for 6-7 folds is calculated to be higher, so that the preference value of the game with 6-7 folds for the discount is multiplied by 1.5 to obtain a new preference value. For this type of game, a recommendation manner shown in fig. 8 can be adopted, and as shown in fig. 8, the method is another scene schematic diagram for content recommendation provided in the embodiment of the present application, and a game B1 (content 1-content 5) after weighting is shown in the diagram, where the game B1 after weighting includes a plurality of games, and is arranged in a certain order in a region corresponding to B1, and is played in a scrolling manner.

Optionally, in order to reduce the ineffective exposure, the games that the user has purchased may be sorted to the bottom, that is, the games that the user has purchased may be seen only at the end of the recommended waterfall stream after being placed as the end of the recommended sequence. For this type of game, a recommendation manner shown in fig. 9 may be adopted, and as shown in fig. 9, the method is another scene diagram for content recommendation provided in the embodiment of the present application, and a game C1 (content 1-content 5) after weighting is shown in the diagram, where the game C1 after weighting includes a plurality of games, and the games are sorted and put at the bottom because the games are purchased by the user, that is, are placed at the lowest end of the display interface.

Optionally, for the display modes shown in fig. 7 to 9, discount information may be added for pre-exposure, that is, the discount information is pushed to the user, so as to improve the purchase rate of the user.

By combining the embodiment, under the condition of low data characteristic quantity, the target user with the required characteristics is obtained by integrating the user portrait aggregation, and the business condition is combined, so that the ratio of activity and click after the user is interested in is increased; meanwhile, in the conversion effect, the expansion of game users can be carried out on different types of games, for example, the conversion of online game users is realized through a single user.

In order to better implement the above-mentioned solution of the embodiments of the present application, the following also provides a related apparatus for implementing the above-mentioned solution. Referring to fig. 10, fig. 10 is a schematic structural diagram of a content recommendation device according to an embodiment of the present application, where the content recommendation device 1000 includes:

an obtaining unit 1001 for obtaining user portrayal of a plurality of users using a target application, the user portrayal being determined based on interaction records of the users with a plurality of contents;

a determining unit 1002 for determining at least one first feature in the user representation;

a classifying unit 1003, configured to classify users according to the at least one first feature to obtain at least two user sets, where a preference degree of each user set to the target application is different;

a recommending unit 1004, configured to determine recommended content of a target user according to a user image of the target user and a user set to which the target user belongs.

Optionally, in some possible implementations of the present application, the recommending unit 1004 is specifically configured to determine a recommended content set according to the user image of the target user and a user set to which the target user belongs, where the recommended content set includes a first content and a second content;

the recommending unit 1004 is specifically configured to calculate first correlation matrices of the first content and the second content respectively based on the interaction records, so as to obtain preference values according to a first formula;

the recommending unit 1004 is specifically configured to determine the recommended content of the target user according to the preference value.

Optionally, in some possible implementations of the present application, the recommending unit 1004 is specifically configured to classify the user portraits in the user set according to a second feature to obtain a first sub-set and a second sub-set, where the user portraits in the first sub-set have the second feature, the user portraits in the second sub-set do not have the second feature, and the classification dimension of the user portraits by the second feature is different from that of the first feature;

the recommending unit 1004 is specifically configured to determine the recommended content of the target user based on the second feature if a ratio of the number of users corresponding to the first subset to the number of users corresponding to the user set satisfies a first condition.

Optionally, in some possible implementations of the present application, the recommending unit 1004 is further configured to determine a third feature that the first subset and the second subset have in common, where the third feature is different from the first feature in a classification dimension of the user portrait;

the recommending unit 1004 is specifically configured to obtain content corresponding to the second feature and content corresponding to the third feature;

the recommending unit 1004 is specifically configured to calculate, based on the interaction records, second correlation matrices of the content corresponding to the second feature and the content corresponding to the third feature respectively;

the recommending unit 1004 is specifically configured to input the second correlation matrix into a second formula to obtain the preference value;

Optionally, in some possible implementation manners of the present application, the recommending unit 1004 is further configured to obtain feature data in the interaction record, where the feature data is determined based on a behavior vector of a user;

the recommending unit 1004 is specifically configured to update the preference value according to the feature data.

Optionally, in some possible implementation manners of the present application, the classifying unit 1003 is specifically configured to cluster the users according to the at least one first feature to determine a central point;

the classifying unit 1003 is specifically configured to classify the user according to a distance between a parameter corresponding to the first feature and the central point, so as to obtain at least two user sets.

Optionally, in some possible implementations of the present application, the determining unit 1002 is specifically configured to determine a feature set according to the user portrait;

the determining unit 1002 is specifically configured to obtain a variance of each feature in the feature set by using principal component analysis;

the determining unit 1002 is specifically configured to determine at least one variance that meets a second condition to determine the corresponding first feature, where the preset condition is set based on a value of the variance.

Optionally, in some possible implementations of the present application, the determining unit 1002 is specifically configured to determine that the user image corresponds to a plurality of fourth features under different contents;

the determining unit 1002 is specifically configured to obtain missing values of a plurality of the fourth features;

the determining unit 1002 is specifically configured to filter a plurality of fourth features of which the missing values are smaller than the first threshold, so as to determine the feature set.

Optionally, in some possible implementation manners of the present application, the determining unit 1002 is further configured to determine a missing parameter corresponding to the missing value;

the determining unit 1002 is specifically configured to determine a filling parameter of the missing parameter in the plurality of fourth features;

the determining unit 1002 is specifically configured to update the missing parameter according to the padding parameter, so as to determine the feature set.

Optionally, in some possible implementations of the present application, the determining unit 1002 is specifically configured to filter a plurality of fourth features of which the missing values are smaller than the first threshold;

the determining unit 1002 is specifically configured to obtain correlation degrees among a plurality of fourth features;

the determining unit 1002 is specifically configured to combine the fourth features with the correlation degree greater than the second threshold to obtain a fifth feature;

the determining unit 1002 is specifically configured to determine the feature set according to the fifth feature.

Optionally, in some possible implementation manners of the present application, the determining unit 1002 is specifically configured to determine a fourth feature that the correlation degree is greater than a second threshold, and obtain a corresponding unit parameter;

the determining unit 1002 is specifically configured to normalize and combine the unit parameters to obtain a fifth feature.

An embodiment of the present application further provides a terminal device, as shown in fig. 11, 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. 11 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. 11, the cellular phone includes: radio Frequency (RF) circuitry 1110, memory 1120, input unit 1130, display unit 1140, sensors 1150, audio circuitry 1160, wireless fidelity (WiFi) module 1170, processor 1180, and power supply 1190. Those skilled in the art will appreciate that the handset configuration shown in fig. 11 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 describes each component of the mobile phone in detail with reference to fig. 11:

RF circuit 1110 may be used for receiving and transmitting signals during a message transmission or call, and in particular, for receiving downlink messages from a base station and then processing the received downlink messages to processor 1180; in addition, the data for designing uplink is transmitted to the base station. In general, RF circuit 1110 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, the RF circuitry 1110 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 1120 may be used to store software programs and modules, and the processor 1180 may execute various functional applications and data processing of the mobile phone by operating the software programs and modules stored in the memory 1120. The memory 1120 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 1120 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 1130 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 1130 may include a touch panel 1131 and other input devices 1132. The touch panel 1131, also referred to as a touch screen, can collect touch operations of a user on or near the touch panel 1131 (for example, operations of the user on or near the touch panel 1131 using any suitable object or accessory such as a finger, a stylus pen, etc., and a range of touch operations on the touch panel 1131 in an interval), and drive the corresponding connection device according to a preset program. Alternatively, the touch panel 1131 may include two parts, namely 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, sends the touch point coordinates to the processor 1180, and can receive and execute commands sent by the processor 1180. In addition, the touch panel 1131 can be implemented by using various types, such as resistive, capacitive, infrared, and surface acoustic wave. The input unit 1130 may include other input devices 1132 in addition to the touch panel 1131. In particular, other input devices 1132 may include, but are not limited to, one or more of a physical keyboard, function keys (e.g., volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, and the like.

The display unit 1140 may be used to display information input by the user or information provided to the user and various menus of the cellular phone. The display unit 1140 may include a display panel 1141, and optionally, the display panel 1141 may be configured in the form of a Liquid Crystal Display (LCD), an organic light-emitting diode (OLED), or the like. Further, touch panel 1131 may cover display panel 1141, and when touch operation is detected on or near touch panel 1131, the touch operation is transmitted to processor 1180 to determine the type of touch event, and then processor 1180 provides corresponding visual output on display panel 1141 according to the type of touch event. Although in fig. 11, the touch panel 1131 and the display panel 1141 are two independent components to implement the input and output functions of the mobile phone, in some embodiments, the touch panel 1131 and the display panel 1141 may be integrated to implement the input and output functions of the mobile phone.

The handset may also include at least one sensor 1150, such as a light sensor, motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display panel 1141 according to the brightness of ambient light, and the proximity sensor may turn off the display panel 1141 and/or the backlight when the mobile phone moves 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 1160, speakers 1161, and microphone 1162 may provide an audio interface between a user and a handset. The audio circuit 1160 may transmit the electrical signal converted from the received audio data to the speaker 1161, and convert the electrical signal into a sound signal for output by the speaker 1161; on the other hand, the microphone 1162 converts the collected sound signal into an electrical signal, which is received by the audio circuit 1160 and converted into audio data, which is then processed by the audio data output processor 1180 and then sent to, for example, another cellular phone via the RF circuit 1110, or the audio data is output to the memory 1120 for further processing.

WiFi belongs to short-distance wireless transmission technology, and the cell phone can help a user to receive and send e-mails, browse webpages, access streaming media and the like through the WiFi module 1170, and provides wireless broadband internet access for the user. Although fig. 11 shows the WiFi module 1170, it is understood that it does not belong to the essential constitution of the handset, and can be omitted entirely as needed within the scope not changing the essence of the invention.

The processor 1180 is a control center of the mobile phone, and is connected to various parts of the whole mobile phone through various interfaces and lines, and executes various functions of the mobile phone and processes data by operating or executing software programs and/or modules stored in the memory 1120 and calling data stored in the memory 1120, thereby performing overall monitoring of the mobile phone. Optionally, processor 1180 may include one or more processing units; optionally, the processor 1180 may integrate an application processor and a modem processor, wherein the application processor mainly handles operating systems, user interfaces, application programs, and the like, and the modem processor mainly handles wireless communications. It will be appreciated that the modem processor described above may not be integrated within processor 1180.

The mobile phone further includes a power supply 1190 (e.g., a battery) for supplying power to each component, and optionally, the power supply may be logically connected to the processor 1180 through a power management system, so that the power management system implements functions of managing charging, discharging, and power consumption.

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

In the embodiment of the present application, the processor 1180 included in the terminal further has a function of executing the steps of the page processing method.

In an embodiment of the present application, a computer-readable storage medium is further provided, where content recommendation instructions are stored in the computer-readable storage medium, and when the computer-readable storage medium is executed on a computer, the computer is caused to perform the steps performed by the content recommendation device in the method described in the foregoing embodiments shown in fig. 3 to 9.

Also provided in the embodiments of the present application is a computer program product including content recommendation instructions, which when run on a computer, causes the computer to perform the steps performed by the content recommendation apparatus in the methods described in the embodiments of fig. 3 to 9.

An embodiment of the present application further provides a content recommendation system, where the content recommendation system may include the content recommendation device in the embodiment described in fig. 10 or the terminal device described in fig. 11.

It can be clearly understood by those skilled in the art that, for convenience and simplicity 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 logical division, and other divisions may be realized in practice, for example, a plurality of 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 position, or may be distributed on multiple 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 content recommendation device, 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 to illustrate the technical solutions of the present application, and not to limit 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 content recommendation, comprising:

obtaining a user profile for a plurality of users using a target application, the user profile determined based on a user's interaction record with a plurality of content;

determining a feature set from the user representation;

determining at least one of the variances that satisfies a second condition to determine a corresponding first feature, the second condition being set based on a numerical magnitude of the variance;

classifying users according to at least one first characteristic to obtain at least two user sets, wherein the preference degree of each user set to the target application is different;

2. The method of claim 1, wherein the determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs comprises:

respectively calculating a first correlation matrix of the first content and the second content based on the interaction record to obtain a preference value according to a first formula;

3. The method of claim 1, wherein the determining the recommended content of the target user according to the user image of the target user and the user set to which the target user belongs comprises:

classifying the user portraits in the user set according to a second feature to obtain a first sub-set and a second sub-set, wherein the user portraits in the first sub-set have the second feature, the user portraits in the second sub-set do not have the second feature, and the second feature has a different dimension from the first feature in classification of the user portraits;

4. The method of claim 3, wherein the determining the recommended content of the target user based on the second characteristic comprises:

determining a third feature that the user figures of the first subset and the user figures of the second subset have in common, the third feature being different from the first feature in a dimension of classification of the user figures;

respectively calculating a second correlation matrix of the content corresponding to the second characteristic and the content corresponding to the third characteristic based on the interaction records;

inputting the second correlation matrix into a second formula to obtain a preference value;

5. The method of claim 4, further comprising:

and updating the preference value according to the characteristic data.

6. The method according to any of claims 1-5, wherein the classifying users according to the at least one first feature to obtain at least two user sets comprises:

7. The method of claim 1, wherein determining a set of features from the user representation comprises:

determining that the user portrayal corresponds to a plurality of fourth characteristics under different contents;

acquiring missing values of a plurality of the fourth features;

and screening a plurality of fourth features of which the missing values are smaller than a first threshold value to determine the feature set.

8. The method of claim 7, further comprising:

determining a missing parameter corresponding to the missing value;

9. The method of claim 7, wherein the screening a plurality of fourth features having the missing value less than a first threshold to determine the feature set comprises:

obtaining a correlation degree among a plurality of fourth features;

determining the feature set according to the fifth feature.

10. The method according to claim 9, wherein the combining the fourth features with the correlation degree larger than the second threshold value to obtain the fifth features comprises:

determining a fourth characteristic with the correlation degree larger than a second threshold value, and acquiring a corresponding unit parameter;

11. The method of any of claims 1-5, wherein the content is a game and the first characteristic is indicative of a game type.

12. An apparatus for content recommendation, comprising:

an acquisition unit to acquire user figures of a plurality of users using a target application, the user figures determined based on a user's interaction records with a plurality of contents;

the recommendation unit is used for determining the recommendation content of the target user according to the user image of the target user and the user set to which the target user belongs;

the determining unit is specifically configured to determine a feature set according to the user portrait; obtaining the variance of each feature in the feature set by adopting principal component analysis; determining at least one of the variances that satisfies a second condition to determine the corresponding first feature, the second condition being set based on a numerical magnitude of the variance.

13. 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 perform the method of content recommendation of any of claims 1 to 11 according to instructions in the program code.

14. A computer-readable storage medium having stored therein instructions which, when run on a computer, cause the computer to perform the method of content recommendation of any of the preceding claims 1 to 11.