CN114297417B

CN114297417B - Multimedia resource recommendation method and related device

Info

Publication number: CN114297417B
Application number: CN202111562839.0A
Authority: CN
Inventors: 廖一桥
Original assignee: Beijing Dajia Internet Information Technology Co Ltd
Current assignee: Beijing Dajia Internet Information Technology Co Ltd
Priority date: 2021-12-20
Filing date: 2021-12-20
Publication date: 2025-10-28
Anticipated expiration: 2041-12-20
Also published as: CN114297417A

Abstract

The present application discloses a multimedia resource recommendation method and related devices, which are used to solve the problems of complex, inefficient and poor accuracy in multimedia resource recommendation in related technologies. In this application, the user's recent historical behavior sequence is used to mine the user's expected recommendation degree for different multimedia resource types. The higher the expected recommendation degree, the more the user expects to obtain multimedia resources of this type. Therefore, this application can construct the user's target historical behavior sequence based on the type of multimedia resources the user expects to obtain, so that the target historical behavior sequence focuses more on the user's recent interests and the type of multimedia resources the user expects to obtain in the near future compared to the long-term behavior sequence of related technologies. In addition, in this application, there is no need to obtain the target historical behavior sequence for each resource type separately. The target historical behavior sequence can be calculated once based on one request. Therefore, the recommendation method of this application can simplify the operation and improve the recommendation efficiency compared to the existing technology.

Description

Multimedia resource recommendation method and related device

Technical Field

The present application relates to the field of multimedia information processing technologies, and in particular, to a multimedia resource recommendation method and a related device.

Background

In the training of the short video recommendation system model, more interesting contents can be recommended to the user based on the information of the user interest points contained in the user history behavior information, and the model learning is of guiding significance.

In the related art, after acquiring a multimedia resource to be recommended based on a user request, taking all historical behavior data of a user as input of a user behavior interest CTR model (SIM) based on retrieval, if candidate multimedia resources to be recommended comprise multiple types of multimedia resources, respectively generating a corresponding long-term behavior sequence for each type of multimedia resources to be recommended by a GSU (GENERAL SEARCH Unit) module as input of a ESU (Exact Search Unit) module, modeling the long-term behavior sequence based on a deep learning model of an attention mechanism to obtain evaluation scores of each candidate multimedia resource, and then performing sequencing recommendation on each multimedia resource to be recommended based on the evaluation scores. The method for recommending the multimedia resources in the related technology has complex and low-efficiency process and poor accuracy.

Disclosure of Invention

The application provides a multimedia resource recommendation method and a related device, which are used for solving the problems of complex and low-efficiency operation and poor accuracy when multimedia resource recommendation is performed in the related technology.

In a first aspect, the present application provides a multimedia resource recommendation method, the method comprising:

Responding to a multimedia resource recommendation request of a target account, and acquiring a recent historical behavior sequence of the target account, wherein the historical behavior in the recent historical behavior sequence represents the operation behavior of the target account for corresponding multimedia resources within a preset time period before the current time;

Determining the expected recommendation degree of the target account for a plurality of multimedia resource types in a first multimedia resource type set according to the recent historical behavior sequence, wherein the first multimedia resource type set is obtained based on multimedia resources corresponding to historical behaviors in the recent historical behavior sequence;

According to the expected recommendation degree of the target account for the multiple multimedia resource types in the first multimedia resource type set, determining the interest multimedia resource type of the target account;

Screening target historical behaviors matched with the interest multimedia resource type from the total historical behavior sequence of the target account to obtain a target historical behavior sequence corresponding to the target account;

according to the target historical behavior sequence, determining the interested degree of the target account to the multimedia resources to be recommended, and determining the target multimedia resources in the multimedia resources to be recommended according to the interested degree, wherein the target multimedia resources are used for being recommended to the target account.

In a possible implementation manner, before determining, according to the recent historical behavior sequence, expected recommendation degrees of the target account for the plurality of multimedia resource types in the first multimedia resource type set, respectively, the method further includes:

Acquiring multimedia resources corresponding to each history behavior in the recent history behavior sequence, and obtaining a first multimedia resource set;

classifying the first multimedia resource set to obtain a second multimedia resource type set;

Determining the interest degree of the target account on each media resource type in the second multimedia resource type set;

And screening the multimedia resource types with the interestingness lower than a preset interestingness threshold from the second multimedia resource type set to obtain the first multimedia resource type set.

In a possible implementation manner, the determining, according to the recent historical behavior sequence, the expected recommendation degree of the target account to the plurality of multimedia resource types in the first multimedia resource type set respectively specifically includes:

Respectively executing for each multimedia resource type in the first multimedia resource type set:

Determining the number n of the multimedia resources with the access time length higher than a time length threshold value in the multimedia resources included in the multimedia resource type, wherein n is a positive integer greater than or equal to 1;

and determining the expected recommendation degree of the target account on the multimedia resource type by adopting a relationship that the expected recommendation degree is in direct proportion to the n and in inverse proportion to the number of the multimedia resources included by the multimedia resource type.

In a possible implementation manner, the multimedia resource type includes multimedia resources including:

The accessed multimedia resources belonging to the multimedia resource type in the multimedia resources corresponding to the history behaviors of the recent history behavior sequence;

and/or the number of the groups of groups,

The associated multimedia resources of the accessed multimedia resources are multimedia resources synchronously recommended to the target account when the accessed multimedia resources are recommended, and the number of the accessed multimedia resources and the total number of the associated multimedia resources are not higher than an upper limit of the number.

In a possible implementation manner, before the determining, according to the target historical behavior sequence, the interest level of the target account to the recommended multimedia resource, the method further includes:

screening candidate multimedia resources belonging to the interest multimedia resource type from the multimedia resources to be recommended;

If the number of the candidate multimedia resources is lower than the preset number, screening candidate multimedia resources from the similar multimedia resource types of the interesting multimedia resource types until the total number of the candidate multimedia resources screened finally is not lower than the preset number;

And screening the finally screened candidate multimedia resources to obtain the multimedia resources to be recommended, which are finally used for determining the interest degree of the target account on the multimedia resources to be recommended.

In one possible implementation manner, if it is determined that the desired recommendation level distribution of the plurality of multimedia resource types meets a preset distribution, the step of determining the interesting multimedia resource types of the target account according to the desired recommendation levels of the target account for the plurality of multimedia resource types respectively is performed.

In a possible implementation manner, before the determining the interesting multimedia resource type of the target account according to the expected recommendation degrees of the target account to the plurality of multimedia resource types, the method further includes:

if it is determined that the expected recommendation degree distribution of the multiple multimedia resource types does not meet the preset distribution, performing cluster analysis on the multimedia resource types in the first multimedia resource type set to obtain a new first multimedia resource type set, and returning to execute the step of determining the expected recommendation degree of the target account on the multiple multimedia resource types in the first multimedia resource type set according to the recent historical behavior sequence.

In one possible implementation manner, the determining the interesting multimedia resource type of the target account according to the expected recommendation degree of the target account to the plurality of multimedia resource types in the first multimedia resource type set respectively specifically includes:

screening out the appointed number of multimedia resource types as the interest multimedia resource types of the target account according to the order of the expected recommendation degree from high to low, or

And screening out the multimedia resource types with the expected recommendation degree higher than the expected recommendation degree threshold value as the interest multimedia resource types of the target account according to the order of the expected recommendation degree from high to low.

In a possible implementation manner, the determining the interest level of the target account in each media resource type in the second set of media resource types specifically includes:

Determining the operating frequency of the target account on each multimedia resource type in the second multimedia resource type set based on the historical behaviors in the recent historical behavior sequence;

and determining the interestingness of each media resource type in the second multimedia resource type set based on the relation of the interestingness and the positive correlation of the operation frequency.

In one possible implementation, the recent historical behavior sequence includes:

A specified number of historical behaviors of the target account in a preset time period before the current time;

and/or the number of the groups of groups,

And historical behaviors of the target account in a preset time period before the current time.

In a second aspect, the present application provides a multimedia resource recommendation apparatus, the apparatus comprising:

A recent historical behavior sequence acquisition module configured to execute a multimedia resource recommendation request in response to a target account to acquire a recent historical behavior sequence of the target account, wherein the historical behaviors in the recent historical behavior sequence characterize the operation behaviors of the target account for corresponding multimedia resources within a preset time period before the current time;

The system comprises a target account, a recommendation degree determining module and a recommendation degree determining module, wherein the target account is configured to determine the recommendation degree of the target account to a plurality of multimedia resource types in a first multimedia resource type set according to the recent historical behavior sequence;

the interest multimedia resource type determining module is further configured to determine an interest multimedia resource type of the target account according to expected recommendation degrees of the target account for a plurality of multimedia resource types in the first multimedia resource type set respectively;

the target historical behavior screening module is configured to perform screening of target historical behaviors matched with the interesting multimedia resource types from the full-scale historical behavior sequences of the target accounts to obtain target historical behavior sequences corresponding to the target accounts;

The target multimedia resource determining module is further configured to determine the interest degree of the target account on the multimedia resources to be recommended according to the target historical behavior sequence, and determine the target multimedia resources in the multimedia resources to be recommended according to the interest degree, wherein the target multimedia resources are used for being recommended to the target account.

In one possible embodiment, the apparatus further comprises:

The first multimedia resource set obtaining module is configured to obtain multimedia resources corresponding to each history behavior in the recent history behavior sequence before the expected recommendation degree determining module determines the expected recommendation degree of the target account to the plurality of multimedia resource types in the first multimedia resource type set according to the recent history behavior sequence, so as to obtain a first multimedia resource set;

The classification module is configured to perform classification on the first multimedia resource set to obtain a second multimedia resource type set;

The interest-degree determining module is configured to determine the interest degree of the target account on each media resource type in the second multimedia resource type set;

And the first multimedia resource type set determining module is configured to execute screening of the multimedia resource types with the interestingness lower than a preset interestingness threshold value from the second multimedia resource type set to obtain the first multimedia resource type set.

In a possible implementation manner, the determining, according to the recent historical behavior sequence, an expected recommendation degree of the target account for a plurality of multimedia resource types in the first multimedia resource type set, respectively, is performed, and the expected recommendation degree determining module is specifically configured to perform:

and/or the number of the groups of groups,

In one possible embodiment, the apparatus further comprises:

The candidate multimedia resource screening module is configured to be executed to screen candidate multimedia resources belonging to the type of the interested multimedia resources from the multimedia resources to be recommended before the target multimedia resource determining module determines the interested degree of the multimedia resources to be recommended by the target account according to the target historical behavior sequence;

The resource supplementing module is configured to execute screening the candidate multimedia resources from the similar multimedia resource types of the interesting multimedia resource types if the number of the candidate multimedia resources is lower than a preset number, until the total number of the finally screened candidate multimedia resources is not lower than the preset number;

And the multimedia resource to be recommended determining module is configured to perform screening on the finally screened candidate multimedia resources to obtain the multimedia resources to be recommended which are finally used for determining the interest degree of the target account on the multimedia resources to be recommended.

In one possible embodiment, the apparatus further comprises:

and the expected recommendation degree distribution determining module is configured to determine that the expected recommendation degree distribution of the multiple multimedia resource types meets a preset distribution before the expected recommendation degree of the multiple multimedia resource types is determined by the interest multimedia resource type determining module according to the target account.

In one possible embodiment, the apparatus further comprises:

The clustering module is configured to perform cluster analysis on the multimedia resource types in the first multimedia resource type set to obtain a new first multimedia resource type set if the expected recommendation degree distribution determining module determines that the expected recommendation degree distribution of the multimedia resource types does not meet the preset distribution before the expected recommendation degree determining module determines the interest multimedia resource types of the target account according to the expected recommendation degrees of the target account for the multimedia resource types respectively;

And the iteration module is configured to return to the step of determining the expected recommendation degree of the target account to the plurality of multimedia resource types in the first multimedia resource type set according to the recent historical behavior sequence after the new first multimedia resource type set is obtained.

In a possible implementation manner, the step of determining the interesting multimedia resource type of the target account according to the expected recommendation degree of the target account to the plurality of multimedia resource types in the first multimedia resource type set respectively is performed, and the interesting multimedia resource type determining module is specifically configured to perform:

In a possible implementation manner, the determining the interest level of the target account for each media resource type in the second set of media resource types is performed, and the interest level determining module is specifically configured to perform:

and/or the number of the groups of groups,

In a third aspect, the present application also provides an electronic device, including:

A processor;

A memory for storing the processor-executable instructions;

wherein the processor is configured to execute the instructions to implement any of the multimedia asset recommendation methods as provided in the first aspect of the application.

In a fourth aspect, embodiments of the present application also provide a computer readable storage medium, which when executed by a processor of an electronic device, causes the electronic device to perform any of the multimedia asset recommendation methods as provided in the first aspect of the present application.

In a fifth aspect, an embodiment of the present application provides a computer program product comprising a computer program which, when executed by a processor, implements any of the multimedia resource recommendation methods as provided in the first aspect of the present application.

The technical scheme provided by the embodiment of the application at least has the following beneficial effects:

According to the method, the expected recommendation degree of the user for different multimedia resource types is mined by adopting the recent historical behavior sequence of the user, the higher the expected recommendation degree is, the more interested the user is in the type of multimedia resource, the more expected the type of multimedia resource is acquired, so that the method can mine the recent interest of the user and the interested degree of the user in each multimedia resource type based on the recent historical behavior sequence of the user, then the target historical behavior sequence of the user is constructed based on the multimedia resource types expected to be acquired by the user, and compared with the related long-term behavior sequence, the target historical behavior sequence is more focused on the recent interest of the user and the recently expected multimedia resource types of the user, and the accurate recommendation can be made for the user based on the target historical behavior sequence of the method. In addition, the method does not need to acquire the target historical behavior sequence for each multimedia resource type respectively, and calculates the target historical behavior sequence once based on a request, so that compared with the prior art, the recommendation method can simplify the operation and improve the recommendation efficiency.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. Additional features and advantages of the application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application. The objectives and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments of the present application will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic application scenario diagram of a multimedia resource recommendation method according to an embodiment of the present application;

FIG. 2 is a flowchart of training a multimedia resource recommendation model according to an embodiment of the present application;

Fig. 3 is a flowchart of a multimedia resource recommendation method according to an embodiment of the present application;

Fig. 4 is a flowchart illustrating a method for determining a first multimedia resource type set according to an embodiment of the present application;

FIG. 5 is a flowchart illustrating a method for determining a desired recommendation level according to an embodiment of the present application;

Fig. 6 is a flowchart of a method for determining a multimedia resource to be recommended according to an embodiment of the present application;

Fig. 7 is a block diagram of a multimedia resource recommendation device according to an embodiment of the present application;

fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

In order to enable a person skilled in the art to better understand the technical solutions of the present application, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the application described herein may be implemented in other sequences than those illustrated or otherwise described herein. The implementations described in the following exemplary examples do not represent all implementations consistent with the application. Rather, they are merely examples of apparatus and methods consistent with aspects of the application as detailed in the accompanying claims.

In addition, it should be noted that in the technical scheme of the application, the acquisition, storage, use, processing and the like of the data all conform to the relevant regulations of national laws and regulations.

In the following, some terms in the embodiments of the present application are explained for easy understanding by those skilled in the art.

(1) The term "plurality" in embodiments of the present application means two or more, and other adjectives are similar.

(2) "And/or" describes an association relationship of an association object, and indicates that there may be three relationships, for example, a and/or B, and may indicate that there are three cases of a alone, a and B together, and B alone. The character "/" generally indicates that the context-dependent object is an "or" relationship.

(3) The server is used for serving the terminal, the content of the service such as providing resources for the terminal and storing the terminal data, and the server corresponds to the application program installed on the terminal and operates in cooperation with the application program on the terminal.

(4) The terminal device may refer to APP (Application) of a software class or a client. The system has a visual display interface, can interact with a target account, corresponds to a server, and provides local service for clients. Applications for software classes, except some applications that only run locally, are typically installed on a common client terminal, and need to run in conjunction with a server. After the development of the internet, more commonly used application programs include, for example, short video applications, email clients when receiving email, and clients for instant messaging. For this type of application program, there is a need to have a corresponding server and service program in the network to provide a corresponding service, such as a database service, a configuration parameter service, etc., so that a specific communication connection needs to be established between the client terminal and the server terminal to ensure the normal operation of the application program.

In the related art, after acquiring a multimedia resource to be recommended based on a user request, taking all historical behavior data of a user as input of a user behavior interest CTR model (SIM) based on retrieval, if candidate multimedia resources to be recommended comprise multiple types of multimedia resources, respectively generating a corresponding long-term behavior sequence for each type of multimedia resources to be recommended by a GSU (GENERAL SEARCH Unit) module as input of a ESU (Exact Search Unit) module, modeling the long-term behavior sequence based on a deep learning model of an attention mechanism to obtain evaluation scores of each candidate multimedia resource, and then performing sequencing recommendation on each multimedia resource to be recommended based on the evaluation scores.

Therefore, in the related art, a long-term behavior sequence needs to be generated based on all the historical behaviors of the user and all the resources to be recommended for one request of the user. However, there may be hundreds or even thousands of candidate videos in the resource to be recommended corresponding to one user request, so that the process of building a long-term behavior sequence may need to be performed hundreds or thousands of times for one user request, and each long-term behavior sequence needs to be performed once for each candidate video in the GSU module, which results in high request frequency of the related art. Therefore, the related technology has complex and low-efficiency process of constructing the long-term behavior sequence, and further causes complex and low-efficiency recommendation process and poor accuracy.

In addition, the long-term behavior sequence generated by adopting all the historical behaviors of the user in the related technology focuses on the long-term historical interests of the user, so that the resources obtained through screening based on the long-term historical interests are often resources liked by the user in the past, but the currently liked resources cannot be met, and the current demands of the user, recommended to the user, cannot be met.

In view of the above, the present application provides a multimedia resource recommendation method and related apparatus, which are used to solve the problems of complex and low-efficiency operation and poor accuracy in performing multimedia resource recommendation in the related art.

The inventive concept of the present application can be summarized in that the embodiment of the present application firstly responds to a multimedia resource recommendation request of a target account, acquires a recent historical behavior sequence of the target account, then determines the expected recommendation degree of the target account for a plurality of multimedia resource types in a first multimedia resource type set according to the recent historical behavior sequence, simultaneously determines the interest multimedia resource types of the target account according to the expected recommendation degree of the target account for a plurality of multimedia resource types in the first multimedia resource type set, screens out a target historical behavior sequence matched with the interest multimedia resource types from the total historical behavior sequence of the target account, obtains the target historical behavior sequence corresponding to the target account, finally determines the interest degree of the target account for the multimedia resource to be recommended according to the target historical behavior sequence, determines the target multimedia resource in the multimedia resource to be recommended according to the interest degree, and recommends the target multimedia resource to the target account, thereby the recent behavior sequence of the target account is adopted to mine the expected recommendation degree of the user for a plurality of multimedia resource types in the first multimedia resource type set, the recent behavior sequence of the user is adopted to mine the expected degree of the user for the user to obtain the expected resource types of the user type based on the expected performance of the user type, the recent behavior sequence is based on the expected performance of the recent history behavior sequence of the user type, and the recent history behavior sequence is acquired for the recent recommendation of the user type is based on the expected user type of the recent user type of the user type, compared with the long-term behavior sequence of the related technology, the target historical behavior sequence is more focused on the recent interests of the user and the types of the multimedia resources which the user expects to obtain recently, and the target historical behavior sequence based on the application can better describe the requirements of the user to make accurate recommendation for the user. In addition, the method does not need to acquire the target historical behavior sequence for each multimedia resource type respectively, and calculates the target historical behavior sequence once based on a request, so that compared with the prior art, the recommendation method can simplify the operation and improve the recommendation efficiency.

After the design idea of the embodiment of the present application is introduced, some simple descriptions are made below for application scenarios applicable to the technical solution of the embodiment of the present application, and it should be noted that the application scenarios described below are only used for illustrating the embodiment of the present application and are not limiting. In the specific implementation, the technical scheme provided by the embodiment of the application can be flexibly applied according to actual needs.

Referring to fig. 1, an application scenario diagram of a multimedia resource recommendation method according to an embodiment of the present application is shown. The application scenario includes a plurality of terminal devices 101 (including terminal device 101-1, terminal device 101-2, terminal device 101-n.) and further includes a server 102. The terminal device 101 and the server 102 are connected through a wireless or wired network, and the server 102 provides multimedia resources for the terminal device 101 to display.

Terminal device 101 includes, but is not limited to, desktop computers, mobile phones, mobile computers, tablet computers, media players, smart wearable devices, smart televisions, and like electronic devices.

Server 102 may be a server, a server cluster formed by a plurality of servers, or a cloud computing center. The server 102 may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, basic cloud computing services such as big data and artificial intelligence platforms, and the like.

Of course, the method provided by the embodiment of the present application is not limited to the application scenario shown in fig. 1, but may be used in other possible application scenarios, and the embodiment of the present application is not limited. The functions that can be implemented by each device in the application scenario shown in fig. 1 will be described together in the following method embodiments, which are not described in detail herein.

In order to further explain the technical solution provided by the embodiments of the present application, the following details are described with reference to the accompanying drawings and the detailed description. Although embodiments of the present application provide the method operational steps shown in the following embodiments or figures, more or fewer operational steps may be included in the method, either on a routine or non-inventive basis. In steps where there is logically no necessary causal relationship, the execution order of the steps is not limited to the execution order provided by the embodiments of the present application.

It should be noted that, the resource recommendation method provided by the application is applicable to any network resource, such as a scene of short video, long video, network elements, commodities and the like, which need to be recommended. In addition, the user information required to implement resource recommendation is obtained through user authorization permissions.

The following describes some simple training procedures of the multimedia resource recommendation model applicable to the technical scheme of the embodiment of the application, so as to facilitate the understanding of the technical scheme provided by the embodiment of the application by the technical personnel in the field.

Referring to fig. 2, a flowchart of training a multimedia resource recommendation model according to an embodiment of the present application includes the following steps:

In step 201, training data is obtained to construct training samples of a multimedia asset recommendation model (i.e., the user behavioral interest CTR model described above).

In some embodiments, the training sample of the multimedia asset recommendation model includes target account features, multimedia asset features, contextual features, and behavior of the target account on the sample, such as praise, attention, long-time viewing of the sample, etc., and the training sample further includes a target historical behavior sequence of the user. The target historical behavior sequence in the present application is constructed based on the recent historical behavior sequence of the user (which will be described later herein).

The target account features may be target account ID, device ID, or other features that can characterize target account information, such as target account interest, target account age, etc. The target account feature may also be an average summed target account behavior sequence, such as an average summed last viewed multimedia asset ID sequence. The target account feature may also be a target account behavior sequence, for example, a target account history viewing multimedia resource ID sequence, a target account history viewing multimedia resource author ID sequence, a target account history viewing multimedia resource duration sequence, a target account history viewing multimedia resource sequence from the current time length, etc., which may be set according to the actual use situation, which is not limited by the embodiment of the present application.

The multimedia resource characteristics may be a multimedia resource ID, or other characteristics capable of representing multimedia resource information, for example, a multimedia resource age, a multimedia resource category, etc., which may be set according to an actual use situation, which is not limited by the embodiment of the present application.

In step 202, the structure and parameters of the ESU in the multimedia asset recommendation model, i.e., the training ESU module, are adjusted based on the training samples. This may be embodied as modeling the target historical behavior sequence obtained in step 201 based on the attention mechanism.

In step 203, training of the GSU module in the multimedia resource recommendation model is performed, so as to obtain a trained GSU module and an ESU module.

In step 204, multimedia asset recommendation is performed based on the trained multimedia asset model.

The application provides a multimedia resource recommendation method, which is mainly based on the process of training a multimedia resource recommendation model, adopts a recent historical behavior sequence of a user to mine the expected recommendation degree of the user for different multimedia resource types, and obtains the recently interested multimedia resource types of the user based on the expected recommendation degree of the user for the different multimedia resource types, thereby obtaining a sample historical behavior sequence of the user, namely obtaining a training sample in step 201. And training the multimedia asset recommendation model using steps 202 and 203 after the training samples are obtained. The method for recommending the multimedia resources mainly comprises the steps of acquiring a target historical behavior sequence of a user by using the same method as the method for acquiring the sample historical behavior sequence, and inputting the target historical behavior sequence into a trained multimedia resource recommendation model, so that the target multimedia resource is recommended to the user.

Referring to fig. 3, a flowchart of a multimedia resource recommendation method according to an embodiment of the present application is shown. As shown in fig. 3, the method may be implemented as the following steps:

In step 301, a recent historical behavior sequence of a target account is obtained in response to a multimedia resource recommendation request of the target account, wherein the historical behaviors in the recent historical behavior sequence represent the operation behaviors of the target account for corresponding multimedia resources within a preset time period before the current time.

For example, if the preset time period is set to 10 minutes, the historical behavior in the recent historical behavior sequence represents the operation behavior of the target account on the corresponding multimedia resource within 10 minutes before the current time, for example, the viewing duration or the praise number of each multimedia resource within 10 minutes is included in the recent historical behavior sequence.

In some embodiments, the multimedia resource recommendation request may be a refresh operation of the multimedia resource by the target account, or a search operation of the multimedia resource by the target account, which is not limited in detail in the present application.

In some embodiments, the target account may include a unique identifier of the target account, such as a target account ID, and may further include a device ID, which may be set according to an actual use situation, which is not limited by the embodiment of the present application.

In some embodiments, in response to a multimedia resource recommendation request of a target account, the resources in the multimedia resource library are screened through steps of vector recall, rough ranking and the like, and finally the multimedia resources to be recommended are obtained. The number of multimedia resources to be recommended is large, for example, the number of multimedia resources to be recommended may generally exceed about 1000.

In some embodiments, the recent historical behavior sequence includes a specified number of historical behaviors of the target account for a preset period of time prior to the current time and/or historical behaviors of the target account for a preset period of time prior to the current time. For example, taking a multimedia resource as a video as an example, the recent history behavior sequence may include a video ID sequence of the last 50 videos watched by the target account, a video author ID sequence of the last 50 videos watched by the target account, a video duration sequence of the last 50 videos watched by the target account, a video ID sequence of the last 50 videos watched by the target account for the length of the current time, etc., or may include a video ID sequence of the videos watched within the last 10 minutes of the target account, or a praise number of the videos watched within the last 10 minutes of the target account.

Thus, a reliable range can be determined for the recent historical behavior series acquired based on the multimedia resource recommendation request of the target account by setting a time period or a specified number, and the processing amount of the historical behavior can be reduced by limiting the number while the historical behavior is acquired, so that the recent hobbies of the user can be reflected.

The step of obtaining the recent historical behavior sequence of the target account needs to obtain the information of the target account in response to the multimedia resource recommendation request of the target account, so that any information of the target account is obtained after authorization approval in the application.

In step 302, according to the recent historical behavior sequence, expected recommendation degrees of the target account for the plurality of multimedia resource types in the first multimedia resource type set are determined, wherein the first multimedia resource type set is obtained based on multimedia resources corresponding to the historical behaviors in the recent historical behavior sequence.

Where the desired recommendation level refers to the level to which the interest of the target account in a certain multimedia asset type is not satisfied.

In one possible implementation manner, it may take a long time to determine the expected recommendation levels of all the multimedia resource types in the recent behavior sequence, so that the efficiency of determining the expected recommendation levels is low, so in order to improve the efficiency of determining the expected recommendation levels, in the embodiment of the present application, before determining the expected recommendation levels of the target account for the plurality of multimedia resource types in the first multimedia resource type set according to the recent historical behavior sequence, the range of the multimedia resource types may be narrowed, and may be specifically executed as steps as shown in fig. 4:

In step 401, a multimedia resource corresponding to each history behavior in a recent history behavior sequence is obtained, so as to obtain a first multimedia resource set.

In step 402, the first set of multimedia assets is classified to obtain a second set of multimedia asset types.

In step 403, the interest level of the target account in each media asset type in the second set of media asset types is determined.

In a possible implementation manner, the method for determining the interest level of the target account in each media resource type in the second multimedia resource type set in the embodiment of the application can be implemented by firstly determining the operation frequency of the target account in each media resource type in the second multimedia resource type set based on the historical behaviors in the recent historical behavior sequence, and then determining the interest level of each media resource type in the second multimedia resource type set based on the positive correlation relationship between the interest level and the operation frequency.

For example, according to the historical behaviors in the recent historical behavior sequence, it is determined that the multimedia resource types included in the second multimedia resource type set include game videos, food videos, advertisement videos and commodity videos, wherein the recommended game videos include 20, food videos include 15, advertisement videos include 3 and commodity videos include 5, the target account has 10 times of praying to the game videos, 15 times of praying to the food videos, 0 time of praying to the advertisement videos and 1 time of praying to the commodity videos, the target account has an operation frequency of 0.5 to the game videos, an operation frequency of 1 to the food videos, an operation frequency of 0 to the advertisement videos and an operation frequency of 0.2 to the commodity videos, and as the interest level is positively correlated with the operation frequency, that is, the higher the operation frequency is, the interest level of the target account has 0.5 to the game videos, the interest level of 1 to the commodity videos and the interest level of 0.2 to the commodity videos are determined.

Therefore, the interest degree of each multimedia resource type is determined through the operation frequency of each type of multimedia resource in the recent historical behavior sequence, the interest degree of the user on different types of multimedia resources can be mined from the user behavior, and the interest of the user on different types of resources can be accurately measured.

In step 404, multimedia resource types with interest levels lower than a preset interest level threshold are screened out from the second multimedia resource type set, so as to obtain a first multimedia resource type set.

By way of example, setting the interestingness threshold to be 0.5, in the above example, the interestingness of the target account to the game video is 0.5, the interestingness to the food is 1, the interestingness to the advertisement food is 0, and the interestingness to the commodity video is 0.2, then the multimedia resource types with the interestingness lower than 0.5 in the second multimedia resource type set can be filtered out, so as to obtain the game video and the food video, and the game video and the food video are used as the multimedia resource types in the first multimedia resource type set.

Therefore, the multimedia resource types higher than the interestingness threshold can be screened out by presetting the interestingness threshold, so that the range of the multimedia resource types can be narrowed, and the efficiency of determining the expected recommendation degree of the target account on the plurality of multimedia resource types in the first multimedia resource type set respectively can be improved.

In a possible implementation manner, after determining the first set of multimedia resource types, the expected recommendation degree of the target account for each of the plurality of multimedia resource types in the first set of multimedia resource types needs to be determined according to the recent historical behavior sequence of the target account, and in this embodiment of the present application, the steps shown in fig. 5 may be executed for each of the multimedia resource types in the first set of multimedia resource types respectively:

In step 501, the number n of multimedia resources with a history behavior access time length higher than a time length threshold is determined in the multimedia resources included in the multimedia resource type, where n is a positive integer greater than or equal to 1.

In a possible implementation manner, the multimedia resources included in the multimedia resource types in the embodiment of the present application specifically include the accessed multimedia resources belonging to the multimedia resource types in the multimedia resources corresponding to the history behaviors of the recent history behavior sequences, so that the multimedia resources provided by the user history behavior operation, that is, the accessed multimedia resources, can better represent the resources interested by the user.

In another embodiment, the multimedia resource type may further include an associated multimedia resource of the accessed multimedia resource, where the associated multimedia resource is a multimedia resource synchronously recommended to the target account when the accessed multimedia resource is recommended, and the number of the accessed multimedia resources and the total number of the associated multimedia resources are not higher than the upper number limit. That is, when the types of the multimedia resources which are accessed are insufficient, the associated multimedia resources which are synchronously recommended can be adopted for supplementing so as to expand the quantity of the multimedia resources, and when the subsequent recommendation based on the multimedia resources is carried out, the sufficient quantity of the multimedia resources are ensured to be used as supports, so that the recommendation efficiency is improved.

In step 502, a desired recommendation level for the multimedia asset type for the target account is determined using a relationship in which the desired recommendation level is proportional to n and inversely proportional to the number of multimedia assets included in the multimedia asset type.

In one possible implementation, the first multimedia resource type set includes a plurality of multimedia resource types, so that calculating the corresponding expected recommendation level for each multimedia resource type respectively can count historical behaviors in a recent behavior sequence of the target account, divide the historical behaviors of the target account for each multimedia resource type into positive samples and negative samples, and calculate the expected recommendation level of the target account for each multimedia resource type in a calculation manner of positive sample number/(positive sample number+negative sample number). The higher the positive sample duty cycle, the higher the desired recommendation level of the target account for the corresponding multimedia resource type, and the greater the demand of the target account for the multimedia resource type.

For example, if the multimedia resource types in the first multimedia resource type set include game videos and food products, 10 game videos and 5 food videos are recently recommended to the target account, only 6 game videos and 1 food video in the target account are recently accessed, and the access time length of 1 video in the game videos is higher than the time length threshold and the access time length of 1 video in the food videos is higher than the time length threshold.

The first assumption is that the multimedia resources included in the multimedia resource type include the accessed multimedia resources belonging to the multimedia resource type in the multimedia resources corresponding to the history behavior of the recent history behavior sequence, the number of multimedia resources of the game video is 6, that is, the total number of samples is 6, the number of positive samples is 1, the number of negative samples is 5, the calculated expected recommendation degree is 0.2, the number of multimedia resources of the food video is 1, that is, the total number of samples is 1, the number of positive samples is 1, the number of negative samples is 0, the calculated expected recommendation degree is 1, and obviously the expected recommendation degree of the target account to the multimedia resource type of the food video is higher.

The second assumption is that the multimedia resources included in the multimedia resource type include the accessed multimedia resources belonging to the multimedia resource type and the associated multimedia resources of the accessed multimedia resources in the multimedia resources corresponding to the history behaviors of the recent history behavior sequence. The number of multimedia resources of the game video is 10, that is, the total number of samples is 10, the number of positive samples is 1, the number of negative samples is 9, the calculated expected recommendation degree is 0.1, the number of multimedia resources of the food video is 5, that is, the total number of samples is 5, the number of positive samples is 1, the number of negative samples is 4, the calculated expected recommendation degree is 0.2, and obviously, the expected recommendation degree of the target account on the type of multimedia resources of the food video is higher.

Therefore, the method determines the expected recommendation degree of the target account for each multimedia resource type in the first multimedia resource type set, and can mine what type of multimedia resource in the multimedia resources recommended to the user is desired by the user, so that the recommendation accuracy is improved.

Therefore, the degree that the interest of each multimedia resource type of the target account is not satisfied can be determined through the target account to the expected recommendation degree of the plurality of multimedia resource types in the first multimedia resource type set respectively, and the recommended multimedia resource types are ensured to be in accordance with the recently interested multimedia resource types of the target account, so that the interest change problem of the target account in the modeling of the target historical behavior sequence is solved.

In step 303, the interesting multimedia resource type of the target account is determined according to the expected recommendation degree of the target account for the plurality of multimedia resource types in the first multimedia resource type set, respectively.

In a possible implementation manner, there may be no expected recommendation degree of the target account for a certain multimedia resource type, that is, there is no outstanding interest, which indicates that the most expected multimedia resource type of the target account is equal to the expected recommendation degree of the least expected multimedia resource type, so before determining the interest multimedia resource type of the target account according to the expected recommendation degree of the target account for the multiple multimedia resource types in the first multimedia resource type set, respectively, it is further required to determine that the expected recommendation degree distribution of the multiple multimedia resource types meets the preset distribution, that is, determine that there is outstanding interest.

The preset distribution may be represented by a gap between the expected recommendation levels, and a gap threshold may be set, where if the gap between the expected recommendation levels of the plurality of multimedia resource types is greater than the gap threshold, it is indicated that the expected recommendation level distribution of the plurality of multimedia resource types satisfies the preset distribution. The distribution condition can be represented as distribution concentration or distribution dispersion, a concentration threshold or a dispersion threshold is set, and if the expected recommendation degree distribution of the plurality of multimedia resource types is greater than the concentration threshold or the expected recommendation degree distribution of the plurality of multimedia resource types is less than the dispersion threshold, the expected recommendation degree distribution of the plurality of multimedia resource types is satisfied with the preset distribution.

In one possible implementation, if it is determined that the desired recommendation level distribution of the plurality of multimedia resource types meets the preset distribution, which indicates that there is a desired recommendation level of the target account for at least one multimedia resource type that is greater than the desired recommendation levels of the remaining multimedia resource types, the step of determining the interest multimedia resource types of the target account according to the desired recommendation levels of the target account for the plurality of multimedia resource types, respectively, is performed.

Therefore, the expected recommendation degree of the multimedia resource type can be determined to meet the requirement of determining the interest multimedia resource type of the target account by setting preset distribution, so that the interest multimedia resource type of the target account can be determined better.

In one possible implementation manner, if it is determined that the expected recommendation degree distribution of the plurality of multimedia resource types does not satisfy the preset distribution, which indicates that there may be no expected recommendation degree of the target account for a certain multimedia resource type, cluster analysis is performed on the multimedia resource types in the first multimedia resource type set to obtain a new first multimedia resource type set, and the step of determining, according to the recent historical behavior sequence, the expected recommendation degree of the target account for the plurality of multimedia resource types in the first multimedia resource type set respectively is performed. For example, if the expected recommendation degree of 1000 multimedia resource types does not meet the preset distribution, performing cluster analysis on the 1000 multimedia resource types, changing the 1000 multimedia resource types into 100 multimedia resource types after clustering, and determining the expected recommendation degree of each multimedia resource type based on the 100 multimedia resource types, thereby screening out the interest multimedia resource types of which the multiple multimedia resource types are used as target accounts.

In one possible implementation, the cluster analysis may merge the multimedia resource types adjacent to the desired recommendation level into one multimedia resource type based on a hierarchical clustered tree structure.

Therefore, through cluster analysis on the multimedia resource types which do not meet the preset distribution, the subdivision degree of the multimedia resource types can be reduced, so that the distribution of the multimedia resource types in a recent behavior sequence is concentrated as much as possible, the difference between the expected recommendation degrees of different multimedia resource types is as large as possible, the expected recommendation degrees of the multimedia resource types can be calculated better, and the interesting multimedia resource types of the target account can be determined better according to the expected recommendation degrees of the multimedia resource types.

In a possible implementation manner, after determining the expected recommendation degree of the target account for each multimedia resource type in the first multimedia resource type set, in the embodiment of the application, a specified number of multimedia resource types can be screened out as the interesting multimedia resource types of the target account according to the order of the expected recommendation degree from high to low, or the multimedia resource types with the expected recommendation degree higher than the expected recommendation degree threshold value can be screened out as the interesting multimedia resource types of the target account according to the order of the expected recommendation degree from high to low.

For example, 100 multimedia resource types may be ranked according to respective desired recommendation levels from high to low, and the designated number of the interested multimedia resource types is set to 10, and then the first 10 multimedia resource types in the ranking are selected as the interested multimedia resource types of the target account. The expected recommendation degree threshold value can be set to be 0.6, and the multimedia resource type with the expected recommendation degree being greater than or equal to the expected recommendation degree threshold value of 0.6 is selected as the interest multimedia resource type of the target account.

Therefore, the interested multimedia resource types of the target account can be screened out by a sorting or threshold setting method, the method can screen out the interested multimedia resource types of the user, and the screening method is simple, efficient and easy to implement.

In step 304, a target historical behavior matched with the type of the interested multimedia resource is screened from the full historical behavior sequence of the target account, and a target historical behavior sequence corresponding to the target account is obtained.

Wherein the full history behavior sequence includes all behavior sequences since the target account registration application, and the sequence lengths of different target accounts may be between hundreds and hundreds of thousands. The full-volume historical behavior sequence can comprise ID sequences of all the access multimedia resources of the target account, author ID sequences of all the access multimedia resources of the target account, duration sequences of all the access multimedia resources of the target account, sequences of all the access multimedia resources of the target account according to the current time length and can be set according to actual use conditions, and the embodiment of the application is not limited to the sequences.

In one possible implementation, the corresponding behavior sequence may be obtained from the full-scale historical behavior sequence of the target account according to the interest multimedia resource type of the target account obtained in step 303, so as to construct a new target historical behavior sequence.

In step 305, according to the target historical behavior sequence, determining the interest degree of the target account for the multimedia resources to be recommended, and determining the target multimedia resources in the multimedia resources to be recommended according to the interest degree, wherein the target multimedia resources are used for being recommended to the target account.

In a possible implementation manner, before determining the interest level of the target account to the multimedia resource to be recommended according to the target historical behavior sequence, the embodiment of the present application may further determine the multimedia resource to be recommended of the interest level of the target account to the multimedia resource to be recommended, and specifically may execute the steps as shown in fig. 6:

in step 601, candidate multimedia resources belonging to the type of the multimedia resource of interest are screened from the multimedia resources to be recommended.

In step 602, if the number of candidate multimedia resources is less than the preset number, candidate multimedia resources are selected from the similar multimedia resource types of the interesting multimedia resource types until the total number of the candidate multimedia resources finally selected is not less than the preset number.

In one possible implementation, a plurality of candidate multimedia resources corresponding to the interest multimedia resource type may be selected from the multimedia asset library based on the interest multimedia resource type of the target account obtained in step 303. The plurality of candidate multimedia resources may be from new to old multimedia resources or may be the most popular multimedia resources. When the number of the multimedia resources obtained by screening in the multimedia resource library does not meet the preset number, the requirement can be relaxed, and the multimedia resources corresponding to other multimedia resource types closest to the interesting multimedia resource type are screened until the number of the multimedia resources obtained by screening meets the preset number. The preset number can be thousands or tens of thousands, and can be set according to practical situations, which is not limited by the embodiment of the present application.

In step 603, the candidate multimedia resources that are finally screened out are screened to obtain the multimedia resources to be recommended that are finally used for determining the interest level of the target account for the multimedia resources to be recommended.

In a possible implementation manner, the candidate multimedia resources screened in the step 602 are used as a recall source, and funnel-type screening is performed through steps of coarse ranking, fine ranking and the like, and finally the multimedia resources to be recommended are screened. The number of the multimedia resources to be recommended which are screened out is generally about 1000, and the multimedia resources to be recommended can be set according to actual conditions, and the embodiment of the application is not limited to the number.

Therefore, the multimedia resources corresponding to the type of the interested multimedia resources which are most expected to be recommended by the target account can be guaranteed to be sufficiently supplied, and the problem of insufficient supply of the multimedia resources which are interested by the target account can be solved.

In a possible implementation manner, according to the target historical behavior sequence, the interest degree of the target account to the multimedia resource to be recommended is determined, and in the embodiment of the application, the target historical behavior sequence can be adopted to sort the multimedia resource to be recommended, so as to obtain the recommendation sequence of the multimedia resource to be recommended, and the interest degree of the target account to the multimedia resource to be recommended is greater when the recommendation sequence is higher.

For example, the sample history behavior sequence may be first obtained by using a method of obtaining the target history behavior sequence, and modeling the sample history behavior sequence based on a transducer (machine translation Attention mechanism) or a Multi-head Attention mechanism, etc., for example, an Attention mechanism based on QKV (query-key-value) may be used, a vector including a multimedia resource feature TARGET ITEM (target item) and other features is used as a query, and the sample history behavior sequence is used as a key and a value, so as to obtain a feature expression of the sample history behavior sequence. The estimated output corresponding to TARGET ITEM can be obtained by splicing the feature with other features through an MLP (Multi-Layer Perceptron, multi-Layer neural network). And training the multimedia resource recommendation model, namely estimating the score of the behavior of the target account to the multimedia resource sample to be recommended, inputting the model by using the target account characteristics, the multimedia resource characteristics, the context characteristics and the sample history behavior sequence, and estimating the score of the behavior of the target account to the multimedia resource sample to be recommended, such as praise, attention and the possibility of watching the multimedia resource sample to be recommended for a long time. And updating parameters of the multimedia resource recommendation model based on the estimated score of the behavior of the target account to be recommended multimedia resource sample and the behavior calculation loss function actually generated by the target account in the neural network training sample. And finally, combining the output target account with the estimated scores of various behaviors of the multimedia resource sample to be recommended by using a trained multimedia resource recommendation model based on an enstable sort formula to obtain a recommendation sequence of the multimedia resource to be recommended, and finally determining the interested degree of the target account to the multimedia resource to be recommended.

In one possible implementation, a target multimedia resource in the multimedia resources to be recommended is determined according to the interest degree, and the target multimedia resource is used for being recommended to a target account. According to the embodiment of the application, the multimedia resources to be recommended, which are in front of the recommendation sequence, can be recommended to the target account according to the recommendation sequence of the multimedia resources to be recommended, namely the target account is recommended to the target account with the multimedia resources to be recommended, which are of greater interest in the multimedia resources to be recommended.

Based on the description, the embodiment of the application firstly responds to a multimedia resource recommendation request of a target account, acquires a recent historical behavior sequence of the target account, then determines expected recommendation degrees of the target account for a plurality of multimedia resource types in a first multimedia resource type set according to the recent historical behavior sequence, simultaneously determines interesting multimedia resource types of the target account according to the expected recommendation degrees of the target account for the plurality of multimedia resource types in the first multimedia resource type set, screens out target historical behaviors matched with the interesting multimedia resource types from the overall historical behavior sequence of the target account to obtain a target historical behavior sequence corresponding to the target account, and finally determines the interesting degree of the target account for the multimedia resource to be recommended according to the target historical behavior sequence, determines the target multimedia resource in the multimedia resource to be recommended according to the interesting degree and recommends the target multimedia resource to the target account.

According to the method, the expected recommendation degree of the user on different multimedia resource types is mined by adopting the recent historical behavior sequence of the user, the higher the expected recommendation degree is, the more interested the user is in the type of multimedia resource, the more expected to acquire the type of multimedia resource is obtained, so that the method can mine the recent interest of the user and the interested degree of the user in each multimedia resource type based on the recent historical behavior sequence of the user, and then construct the target historical behavior sequence of the user based on the multimedia resource types expected to be acquired by the user, so that the target historical behavior sequence is more focused on the recent interest of the user and the recently expected multimedia resource types of the user compared with the related long-term behavior sequence, and the user can be accurately recommended by describing the user according to the target historical behavior sequence. In addition, the method does not need to acquire the target historical behavior sequence for each multimedia resource type respectively, and calculates the target historical behavior sequence once based on a request, so that compared with the prior art, the recommendation method can simplify the operation and improve the recommendation efficiency.

The embodiment of the application also provides a multimedia resource recommendation device based on the same inventive concept. Fig. 7 is a block diagram of a multimedia resource recommendation device according to an embodiment of the present application, and referring to fig. 7, the device includes a recent historical behavior sequence acquisition module 701, an expected recommendation degree determination module 702, an interest multimedia resource type determination module 703, a target historical behavior screening module 704, and a target multimedia resource determination module 705, where:

A recent history behavior sequence obtaining module 701 configured to obtain a recent history behavior sequence of a target account in response to a multimedia resource recommendation request of the target account, wherein the history behavior in the recent history behavior sequence characterizes an operation behavior of the target account for a corresponding multimedia resource within a preset period of time before a current time;

A desired recommendation level determining module 702 configured to determine, according to the recent historical behavior sequence, desired recommendation levels of the target account for a plurality of multimedia resource types in a first multimedia resource type set, respectively, where the first multimedia resource type set is obtained based on multimedia resources corresponding to historical behaviors in the recent historical behavior sequence;

The interesting multimedia resource type determining module 703 is further configured to perform determining an interesting multimedia resource type of the target account according to the expected recommendation degree of the target account for the plurality of multimedia resource types in the first multimedia resource type set, respectively;

A target historical behavior screening module 704, configured to perform screening of target historical behaviors matched with the interesting multimedia resource type from the total historical behavior sequence of the target account, so as to obtain a target historical behavior sequence corresponding to the target account;

The target multimedia resource determining module 705 is further configured to determine, according to the target historical behavior sequence, a degree of interest of the target account in the multimedia resource to be recommended, and determine, according to the degree of interest, a target multimedia resource in the multimedia resource to be recommended, where the target multimedia resource is used for being recommended to the target account.

In one possible embodiment, the apparatus further comprises:

the first multimedia resource set obtaining module is configured to obtain multimedia resources corresponding to each history behavior in the recent history behavior sequence before the expected recommendation degree determining module 702 determines the expected recommendation degrees of the target account to the plurality of multimedia resource types in the first multimedia resource type set according to the recent history behavior sequence, so as to obtain a first multimedia resource set;

In a possible implementation manner, the determining, according to the recent historical behavior sequence, the expected recommendation degree of the target account to the plurality of multimedia resource types in the first multimedia resource type set, respectively, is performed, and the expected recommendation degree determining module 702 is specifically configured to perform:

and/or the number of the groups of groups,

In one possible embodiment, the apparatus further comprises:

a candidate multimedia resource screening module configured to perform screening candidate multimedia resources belonging to the interest multimedia resource type from the multimedia resources to be recommended before the target multimedia resource determining module 705 determines the interest degree of the target account to the multimedia resources to be recommended according to the target historical behavior sequence;

In one possible embodiment, the apparatus further comprises:

the expected recommendation level distribution determining module is configured to determine that the expected recommendation level distribution of the multiple multimedia resource types meets a preset distribution before the interest multimedia resource type determining module 703 determines the interest multimedia resource types of the target account according to the expected recommendation levels of the multiple multimedia resource types of the target account.

In one possible embodiment, the apparatus further comprises:

A clustering module configured to perform, before the interesting multimedia resource type determining module 703 determines the interesting multimedia resource types of the target account according to the expected recommendation degrees of the target account for the plurality of multimedia resource types, if the expected recommendation degree distribution determining module determines that the expected recommendation degree distribution of the plurality of multimedia resource types does not satisfy the preset distribution, cluster analysis on the multimedia resource types in the first multimedia resource type set to obtain a new first multimedia resource type set;

In a possible implementation manner, the determining the interesting multimedia resource type of the target account according to the expected recommendation degree of the target account to the plurality of multimedia resource types in the first multimedia resource type set respectively is performed, and the interesting multimedia resource type determining module 703 is specifically configured to perform:

and/or the number of the groups of groups,

The multimedia resource recommendation device and the multimedia resource recommendation method provided by the embodiment of the application adopt the same inventive concept, can obtain the same beneficial effects, and are not described herein again.

After the multimedia resource recommendation method and the device of the exemplary embodiment of the application are introduced, an electronic device of the multimedia resource recommendation method provided by the embodiment of the application is introduced.

Those skilled in the art will appreciate that the various aspects of the application may be implemented as a system, method, or program product. Accordingly, aspects of the present application may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, micro-code, etc.) or an embodiment combining hardware and software aspects that may be referred to herein collectively as a "circuit," module "or" system.

In some possible embodiments, an electronic device according to the application may comprise at least one processor and at least one memory. Wherein the memory stores program code that, when executed by the processor, causes the processor to perform the multimedia asset recommendation method according to various exemplary embodiments of the present application described above in this specification. For example, the processor may perform steps as in a multimedia asset recommendation method.

An electronic device 800 according to such an embodiment of the application is described below with reference to fig. 8. The electronic device 800 shown in fig. 8 is merely an example and should not be construed as limiting the functionality and scope of use of embodiments of the present application.

As shown in fig. 8, the electronic device 800 is embodied in the form of a general-purpose electronic device. The components of electronic device 800 may include, but are not limited to, at least one processor 801 described above, at least one memory 802 described above, and a bus 803 connecting the various system components, including memory 802 and processor 801.

Bus 803 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a processor, and a local bus using any of a variety of bus architectures.

The memory 802 may include readable media in the form of volatile memory, such as Random Access Memory (RAM) 8021 and/or cache memory 8022, and may further include Read Only Memory (ROM) 8023.

The memory 802 may also include a program/utility 8025 having a set (at least one) of program modules 8024, such program modules 8024 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each or some combination of which may include an implementation of a network environment.

The electronic device 800 may also communicate with one or more external devices 804 (e.g., keyboard, pointing device, etc.), one or more devices that enable a user to interact with the electronic device 800, and/or any device (e.g., router, modem, etc.) that enables the electronic device 800 to communicate with one or more other electronic devices. Such communication may occur through an input/output (I/O) interface 805. Also, the electronic device 800 may communicate with one or more networks such as a Local Area Network (LAN), a Wide Area Network (WAN) and/or a public network, such as the Internet, through a network adapter 806. As shown, network adapter 806 communicates with other modules for electronic device 800 over bus 803. It should be appreciated that although not shown, other hardware and/or software modules may be used in connection with electronic device 800, including, but not limited to, microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, data backup storage systems, and the like.

In an exemplary embodiment, a computer readable storage medium is also provided, such as a memory 802, comprising instructions executable by the processor 801 to perform the above-described multimedia asset recommendation method. Alternatively, the storage medium may be a non-transitory computer readable storage medium, which may be, for example, ROM, random Access Memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, and the like.

In an exemplary embodiment, a computer program product is also provided, comprising a computer program which, when executed by the processor 801, implements any of the multimedia resource recommendation methods as provided by the present application.

In an exemplary embodiment, aspects of a multimedia asset recommendation method provided by the present application may also be implemented in the form of a program product comprising program code for causing a computer device to carry out the steps of the multimedia asset recommendation method according to the various exemplary embodiments of the present application as described herein above when the program product is run on the computer device.

The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of a readable storage medium include an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The program product for the multimedia asset recommendation method of embodiments of the present application may employ a portable compact disc read only memory (CD-ROM) and comprise program code and may be run on an electronic device. However, the program product of the present application is not limited thereto, and in this document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

The readable signal medium may include a data signal propagated in baseband or as part of a carrier wave with readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Program code for carrying out operations of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the consumer electronic device, partly on the consumer electronic device, as a stand-alone software package, partly on the consumer electronic device, partly on the remote electronic device, or entirely on the remote electronic device or server. In the case of remote electronic devices, the remote electronic device may be connected to the consumer electronic device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external electronic device (e.g., connected through the internet using an internet service provider).

It should be noted that although several units or sub-units of the apparatus are mentioned in the above detailed description, such a division is merely exemplary and not mandatory. Indeed, the features and functions of two or more of the elements described above may be embodied in one element in accordance with embodiments of the present application. Conversely, the features and functions of one unit described above may be further divided into a plurality of units to be embodied.

Furthermore, although the operations of the methods of the present application are depicted in the drawings in a particular order, this is not required or suggested that these operations must be performed in this particular order or that all of the illustrated operations must be performed in order to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step to perform, and/or one step decomposed into multiple steps to perform.

It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable image scaling device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable image scaling device, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable image scaling device to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable image scaling apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

1. A method for recommending multimedia resources, the method comprising:

The method comprises the steps of responding to a multimedia resource recommendation request of a target account, acquiring a recent historical behavior sequence of the target account, wherein the historical behaviors in the recent historical behavior sequence represent the operation behaviors of the target account for corresponding multimedia resources in a preset time period before the current time, and the recent historical behavior sequence comprises a specified number of historical behaviors of the target account in the preset time period before the current time and/or the historical behaviors of the target account in the preset time period before the current time;

determining the expected recommendation degree of the target account on the multimedia resource type by adopting a relationship that the expected recommendation degree is in direct proportion to the n and in inverse proportion to the number of the multimedia resources included in the multimedia resource type, wherein the first multimedia resource type set is obtained based on the multimedia resources corresponding to the historical behaviors in the recent historical behavior sequence;

2. The method of claim 1, wherein prior to determining the desired degree of recommendation by the target account for each of the plurality of multimedia asset types in the first set of multimedia asset types based on the recent historical behavior sequence, the method further comprises:

3. The method of claim 1, wherein the multimedia asset type comprises a multimedia asset comprising:

and/or the number of the groups of groups,

4. The method of claim 1, wherein prior to determining the level of interest of the target account in the multimedia asset to be recommended based on the target historical behavior sequence, the method further comprises:

5. The method according to claim 1, wherein the step of determining the interesting multimedia resource type of the target account according to the desired recommendation levels of the target account for the plurality of multimedia resource types, respectively, is performed if it is determined that the desired recommendation level distribution of the plurality of multimedia resource types satisfies a preset distribution.

6. The method of claim 5, wherein prior to determining the multimedia asset type of interest for the target account based on the desired recommendation levels for the plurality of multimedia asset types, respectively, for the target account, the method further comprises:

7. The method according to any one of claims 1-6, wherein the determining the interesting multimedia resource type of the target account according to the expected recommendation degree of the target account for the plurality of multimedia resource types in the first multimedia resource type set, respectively, specifically comprises:

8. The method according to claim 2, wherein determining the interest level of the target account in each media resource type in the second set of multimedia resource types comprises:

9. A multimedia asset recommendation device, the device comprising:

A recent history behavior sequence obtaining module configured to obtain a recent history behavior sequence of a target account in response to a multimedia resource recommendation request of the target account, wherein the history behavior in the recent history behavior sequence characterizes the operation behavior of the target account for corresponding multimedia resources in a preset time period before the current time;

A desired recommendation level determining module configured to perform, for each of the multimedia resource types in the first set of multimedia resource types, respectively:

10. The apparatus of claim 9, wherein the apparatus further comprises:

11. The apparatus of claim 9, wherein the multimedia asset type comprises multimedia assets comprising:

and/or the number of the groups of groups,

12. The apparatus of claim 9, wherein the apparatus further comprises:

13. The apparatus of claim 9, wherein the apparatus further comprises:

14. The apparatus of claim 13, wherein the apparatus further comprises:

15. The apparatus according to any of claims 9-14, wherein the determining the interesting multimedia resource type of the target account according to the desired degree of recommendation of the target account for the plurality of multimedia resource types in the first set of multimedia resource types, respectively, is performed, the interesting multimedia resource type determining module being specifically configured to perform:

16. The apparatus of claim 10, wherein the determining the interest level of the target account in each media asset type in the second set of multimedia asset types is performed, the interest level determination module being specifically configured to perform:

17. An electronic device, comprising:

A processor;

A memory for storing the processor-executable instructions;

Wherein the processor is configured to execute the instructions to implement the multimedia asset recommendation method of any of claims 1-8.

18. A computer readable storage medium, characterized in that instructions in the computer readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the multimedia resource recommendation method of any one of claims 1-8.

19. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the multimedia resource recommendation method of any one of claims 1-8.