CN114297419A

CN114297419A - Method, apparatus, device, medium and program product for predicting multimedia object

Info

Publication number: CN114297419A
Application number: CN202111666248.8A
Authority: CN
Inventors: 宛言; 高梓尧
Original assignee: Beijing Zhuoyue Lexiang Network Technology Co ltd
Current assignee: Beijing Dajia Internet Information Technology Co Ltd
Priority date: 2021-12-31
Filing date: 2021-12-31
Publication date: 2022-04-08
Anticipated expiration: 2041-12-31

Abstract

The present disclosure relates to a method, apparatus, device, medium, and program product for predicting a multimedia object. The method comprises the following steps: acquiring at least one first user behavior data of a target multimedia object in a first time period, wherein the at least one first user behavior data comprises user behavior data beyond the playing amount of the target multimedia object; predicting a first probability distribution of an amount of play of the target multimedia object over a second time period according to the at least one first user behavior data, wherein the second time period follows the first time period; predicting a probability that the target multimedia object belongs to at least one preset category according to the first probability distribution.

Description

Method, apparatus, device, medium and program product for predicting multimedia object

Technical Field

The present disclosure relates to the field of multimedia technologies, and in particular, to a method, an apparatus, a device, a medium, and a program product for predicting a multimedia object.

Background

Compared with characters and pictures, the expressive force of multimedia objects such as videos and audios is stronger. For example, the video combines the auditory information and the visual information, so that the content is rich, the expressive force is strong, and the intuitiveness is good, and the video is popular in various social media. With the acceleration of life rhythm, the fragmented information acquisition mode of the short video is gradually loved by people, more and more people are willing to share the short video shot by themselves on the short video platform and share the short video which is interesting by themselves to other people. For a multimedia object platform (such as a video platform or an audio platform), predicting a huge number of multimedia objects is of great significance in saving the bandwidth of a content distribution network and the like.

Disclosure of Invention

The present disclosure provides a prediction scheme for multimedia objects.

According to an aspect of the present disclosure, there is provided a method for predicting a multimedia object, including:

acquiring at least one first user behavior data of a target multimedia object in a first time period, wherein the at least one first user behavior data comprises user behavior data beyond the playing amount of the target multimedia object;

predicting a first probability distribution of an amount of play of the target multimedia object over a second time period according to the at least one first user behavior data, wherein the second time period follows the first time period;

predicting a probability that the target multimedia object belongs to at least one preset category according to the first probability distribution.

In the disclosed embodiment, by obtaining at least one first user behavior data of a target multimedia object in a first time period, wherein the at least one first user behavior data includes user behavior data outside the playing amount of the target multimedia object, predicting a first probability distribution of the playing amount of the target multimedia object in a second time period according to the at least one first user behavior data, and predicting a probability that the target multimedia object belongs to at least one preset category according to the first probability distribution, thereby predicting a probability distribution of the playing amount of the target multimedia object in a certain period in the future based on the user behavior data outside the playing amount of the target multimedia object in the certain period in the past, and predicting a probability that the target multimedia object belongs to at least one preset category based on the probability distribution of the playing amount of the target multimedia object in the certain period in the future, therefore, the probability that the target multimedia object belongs to the preset category can be accurately predicted, the future performance of the target multimedia object can be predicted based on the generated user behavior data, the delay is low, and therefore the multimedia object platform is facilitated to more effectively distribute the multimedia object, and the bandwidth of a content distribution network is facilitated to be saved.

In one possible implementation, the first user behavior data includes at least two;

predicting, based on the at least one first user behavior data, a first probability distribution of an amount of play of the target multimedia object over a second time period, comprising:

for any one first user behavior data in at least two first user behavior data, predicting a second probability distribution of the playing amount of the target multimedia object in a second time period according to the first user behavior data;

predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data.

In this implementation, by using at least two types of first user behavior data, for any one of the at least two types of first user behavior data, a second probability distribution of a play amount of the target multimedia object in a second time period is predicted according to the first user behavior data, and a first probability distribution of the play amount of the target multimedia object in the second time period is predicted according to at least two second probability distributions that are in one-to-one correspondence with the at least two types of first user behavior data, thereby enabling a probability that the target multimedia object belongs to at least one preset category to be predicted more accurately.

In one possible implementation form of the method,

the predicting, for any one of the at least two first user behavior data, a second probability distribution of a play amount of the target multimedia object in a second time period according to the first user behavior data includes: for any one of at least two kinds of first user behavior data, adopting a probability distribution model corresponding to the first user behavior data in a pre-trained mixed probability distribution model to predict the first user behavior data to obtain a second probability distribution of the playing amount of the target multimedia object in a second time period, wherein the mixed probability distribution model comprises at least two probability distribution models which are in one-to-one correspondence with the at least two kinds of first user behavior data, and the at least two probability distribution models are respectively used for predicting the second probability distribution of the playing amount corresponding to the corresponding first user behavior data;

the predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data comprises: predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data and weights in one-to-one correspondence with the at least two probability distribution models.

In the implementation mode, the probability distribution model corresponding to at least two kinds of first user behavior data one to one is adopted to fit the corresponding probability distribution, so that the first user behavior data of a single value can be converted into the probability distribution, the trend of the playing amount can be more accurately represented, and richer features are provided for the probability prediction of the next layer.

In a possible implementation, the predicting, according to the first probability distribution, the probability that the target multimedia object belongs to at least one preset category includes:

performing full-connection processing on the first probability distribution to obtain a first full-connection processing result;

and activating the first full-connection processing result to obtain the probability that the target multimedia object belongs to at least one preset category.

In this implementation, by performing the full join processing and the activation processing on the first probability distribution, linear dimensional transformation and nonlinear fitting can be performed on the first probability distribution, and thus the probability that the target multimedia object belongs to at least one preset category can be accurately predicted.

In a possible implementation manner, before predicting the user behavior data by using a probability distribution model corresponding to the user behavior data in the pre-trained mixed probability distribution model, the method further includes:

acquiring at least two second user behavior data of a training object in a third time period, wherein the training object is a multimedia object, the at least two second user behavior data at least comprise user behavior data except for the playing amount of the training object, and the third time period is before the first time period;

determining a real category to which the training object belongs;

for any one second user behavior data in the at least two second user behavior data, adopting a probability distribution model corresponding to the second user behavior data in the mixed probability distribution model to predict the second user behavior data to obtain a third probability distribution of the playing amount of the training object in a fourth time period, wherein the fourth time period is after the third time period;

predicting a fourth probability distribution of the playing amount of the training object in the fourth time period according to at least two third probability distributions in one-to-one correspondence with the at least two second user behavior data and weights in one-to-one correspondence with the at least two probability distribution models;

performing full-connection processing on the fourth probability distribution to obtain a second full-connection processing result;

activating the second full-connection processing result to obtain the probability that the training object belongs to at least one preset category;

and updating the weight and the parameters of the full-connection processing according to the real category and the probability that the training object belongs to at least one preset category.

By adopting the implementation mode, the prediction model can learn the capability of predicting the category to which the multimedia object belongs.

In one possible implementation, the determining the real category to which the training object belongs includes:

obtaining a second amount of play of the training object over the fourth time period and a third amount of play of a set of designated multimedia objects over the fourth time period;

and determining the real category of the training object according to the ratio of the second playing amount to the third playing amount.

In the implementation mode, the second playing amount of the training object in the fourth time period and the third playing amount of the designated multimedia object set in the fourth time period are obtained, and the real category to which the training object belongs is determined according to the ratio of the second playing amount to the third playing amount, so that accurate labeling data can be automatically obtained without manually labeling the category of the training object.

activating the first full-connection processing result to obtain a first probability that the target multimedia object belongs to at least one preset category;

acquiring a first playing amount of the target multimedia object in a first time period;

predicting a second probability that the target multimedia object belongs to the at least one preset category according to the first playing amount;

and determining the probability that the target multimedia object belongs to at least one preset category according to the first probability and the second probability.

In this implementation, in addition to determining the first probability according to at least one first user behavior data except the playing amount of the target multimedia object in the first time period, the second probability is predicted according to the playing amount of the target multimedia object in the first time period, and the probability that the target multimedia object belongs to at least one preset category can be determined according to the weighted sum of the first probability and the second probability, so that the accuracy of predicting the probability that the target multimedia object belongs to at least one preset category can be further improved by performing prediction in combination with the playing amount.

According to an aspect of the present disclosure, there is provided a prediction apparatus of a multimedia object, including:

the first obtaining module is used for obtaining at least one first user behavior data of a target multimedia object in a first time period, wherein the at least one first user behavior data comprises user behavior data beyond the playing amount of the target multimedia object;

a first prediction module for predicting a first probability distribution of an amount of play of the target multimedia object over a second time period, wherein the second time period is subsequent to the first time period, based on the at least one first user behavior data;

and the second prediction module is used for predicting the probability that the target multimedia object belongs to at least one preset category according to the first probability distribution.

the first prediction module is to:

In one possible implementation, the first prediction module is configured to:

for any one of at least two kinds of first user behavior data, adopting a probability distribution model corresponding to the first user behavior data in a pre-trained mixed probability distribution model to predict the first user behavior data to obtain a second probability distribution of the playing amount of the target multimedia object in a second time period, wherein the mixed probability distribution model comprises at least two probability distribution models which are in one-to-one correspondence with the at least two kinds of first user behavior data, and the at least two probability distribution models are respectively used for predicting the second probability distribution of the playing amount corresponding to the corresponding first user behavior data;

predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data and weights in one-to-one correspondence with the at least two probability distribution models.

In one possible implementation, the second prediction module is configured to:

In one possible implementation, the apparatus further includes:

a second obtaining module, configured to obtain at least two types of second user behavior data of a training object in a third time period, where the training object is a multimedia object, the at least two types of second user behavior data at least include user behavior data other than a play amount of the training object, and the third time period is before the first time period;

a determining module, configured to determine a real category to which the training object belongs;

a third prediction module, configured to predict, for any one of the at least two second user behavior data, the second user behavior data by using a probability distribution model corresponding to the second user behavior data in the mixed probability distribution model, so as to obtain a third probability distribution of a playback volume of the training object in a fourth time period, where the fourth time period is after the third time period;

a fourth prediction module, configured to predict a fourth probability distribution of the playing amount of the training object in the fourth time period according to at least two third probability distributions that are in one-to-one correspondence with the at least two second user behavior data and weights that are in one-to-one correspondence with the at least two probability distribution models;

the full-connection module is used for performing full-connection processing on the fourth probability distribution to obtain a second full-connection processing result;

the activation module is used for activating the second full-connection processing result to obtain the probability that the training object belongs to at least one preset category;

and the updating module is used for updating the weight and the parameters of the full-connection processing according to the real category and the probability that the training object belongs to at least one preset category.

In one possible implementation, the determining module is configured to:

In one possible implementation, the second prediction module is configured to:

According to an aspect of the present disclosure, there is provided an electronic device including: one or more processors; a memory for storing executable instructions; wherein the one or more processors are configured to invoke the memory-stored executable instructions to perform the above-described method.

According to an aspect of the present disclosure, there is provided a computer readable storage medium having stored thereon computer program instructions which, when executed by a processor, implement the above-described method.

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.

Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the accompanying drawings.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure.

Fig. 1 shows a flowchart of a prediction method of a multimedia object provided by an embodiment of the present disclosure.

Fig. 2 is a schematic diagram illustrating an application scenario of a prediction method for a multimedia object according to an embodiment of the present disclosure.

Fig. 3 is a schematic diagram illustrating a report corresponding to a short video with the highest probability of belonging to the S class in the method for predicting a multimedia object according to the embodiment of the present disclosure.

Fig. 4 shows a block diagram of a prediction apparatus for a multimedia object provided by an embodiment of the present disclosure.

Fig. 5 illustrates a block diagram of an electronic device 800 provided by an embodiment of the disclosure.

Fig. 6 illustrates a block diagram of another electronic device 1900 provided by an embodiment of the disclosure.

Detailed Description

Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. In the drawings, like reference numbers can indicate functionally identical or similar elements. While the various aspects of the embodiments are presented in drawings, the drawings are not necessarily drawn to scale unless specifically indicated.

The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and/or B, may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the term "at least one" herein means any one of a plurality or any combination of at least two of a plurality, for example, including at least one of A, B, C, and may mean including any one or more elements selected from the group consisting of A, B and C.

Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements and circuits that are well known to those skilled in the art have not been described in detail so as not to obscure the present disclosure.

In the related art, it is generally determined to which category a multimedia object belongs, based on posterior data of the multimedia object. For example, after the multimedia object is released, a period of time (e.g., 3-7 days) is waited, the daily playing amount of the multimedia object in the period of time is counted, and the peak value of the playing amount of the multimedia object in the period of time is determined. And calculating the ratio of the peak value to the total playing amount of all the multimedia objects on the multimedia object platform, and determining which ratio interval the ratio falls into in advance so as to determine which category the multimedia object belongs to. For example, if the ratio is greater than 0.5%, the multimedia object is determined to belong to the S class, if the ratio is greater than 0.1% and less than or equal to 0.5%, the multimedia object is determined to belong to the a class, and so on.

In this way, since the data for determining to which category the multimedia object belongs is a posteriori data, it is often necessary to wait for more than 3-7 days, and thus the delay is high. Moreover, this method often requires manual recalibration of the ratio interval at intervals (e.g., monthly), which is tedious and less accurate.

To solve the technical problem similar to the above, the embodiments of the present disclosure provide a method, an apparatus, a device, a medium, and a program product for predicting a multimedia object by obtaining at least one first user behavior data of a target multimedia object in a first time period, wherein the at least one first user behavior data includes user behavior data other than a play amount of the target multimedia object, predicting a first probability distribution of the play amount of the target multimedia object in a second time period according to the at least one first user behavior data, and predicting a probability that the target multimedia object belongs to at least one preset category according to the first probability distribution, thereby predicting a probability distribution of the play amount of the target multimedia object in a future certain period based on the user behavior data other than the play amount of the target multimedia object in the past certain period, and predicting the probability that the target multimedia object belongs to at least one preset category based on the probability distribution of the playing amount of the target multimedia object in the future within the period of time, so that the probability that the target multimedia object belongs to the preset category can be accurately predicted, and the embodiment of the disclosure can predict the future performance of the target multimedia object based on the generated user behavior data, and has low delay, thereby being beneficial to more effectively distributing the multimedia object by a multimedia object platform and saving the bandwidth of a content distribution network.

The following describes a method for predicting a multimedia object according to an embodiment of the present disclosure in detail with reference to the accompanying drawings.

Fig. 1 shows a flowchart of a prediction method of a multimedia object provided by an embodiment of the present disclosure. In a possible implementation, the prediction method of the multimedia object may be performed by a terminal device or a server or other electronic devices. The terminal device may be a User Equipment (UE), a mobile device, a User terminal, a cellular phone, a cordless phone, a Personal Digital Assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, or a wearable device. In some possible implementations, the method of predicting the multimedia object may be implemented by a processor calling computer readable instructions stored in a memory. As shown in fig. 1, the prediction method of the multimedia object includes steps S11 through S13.

In step S11, at least one first user behavior data of the target multimedia object in the first time period is obtained, wherein the at least one first user behavior data includes user behavior data beyond the playing amount of the target multimedia object.

In step S12, a first probability distribution of the play-out amount of the target multimedia object over a second time period is predicted based on the at least one first user behavior data, wherein the second time period follows the first time period.

In step S13, a probability that the target multimedia object belongs to at least one preset category is predicted according to the first probability distribution.

In the disclosed embodiment, the target multimedia object may be any multimedia object that needs to be predicted. The multimedia object may be a video, an audio, or the like, wherein the video may be a short video or a long video. For example, each video on the video platform may be respectively used as a target multimedia object to predict the probability that each video on the video platform belongs to at least one preset category. For another example, each video meeting the preset condition on the video platform may be respectively used as the target multimedia object to predict the probability that each video meeting the preset condition on the video platform belongs to at least one preset category. The preset condition may include at least one of a preset issuing time condition, a preset definition condition, a preset duration condition, and the like.

The first time period may represent a certain time period before the current time. The first time period may or may not include the current time. The time span of the first time period may be a first preset duration. For example, the first preset time period may be 1 hour, half hour, 2 hours, 6 hours, 12 hours, 24 hours, and so on.

In the embodiment of the present disclosure, the user behavior may represent a behavior of the user with respect to the multimedia object, and the user behavior data may be data generated based on the behavior of the user with respect to the multimedia object. For example, the user behavior may include at least one of play, like, share, forward, comment, favorite, search, and the like, and accordingly, the user behavior data may include at least one of play data, like data, share data, forward data, comment data, favorite data, search data, and the like. For example, the user behavior data may include at least one of a play amount, an approval amount, a share amount, a forward amount, a review amount, a collection amount, a search amount, and the like.

The first user behavior data may represent user behavior data of the target multimedia object over a first time period. The first user behavior data may include one or more than two types of user behavior data, and at least one type of user behavior data includes user behavior data other than the play amount of the target multimedia object.

In one possible implementation, the first user behavior data includes at least two types of user behavior data. In this implementation manner, the at least two types of first user behavior data may include the playing amount of the target multimedia object, or may not include the playing amount of the target multimedia object. For example, the at least two types of first user behavior data may include at least two types of praise amount, share amount, forward amount, comment amount, collection amount, search amount, play amount, and the like of the target multimedia object in the first time period. In one example, the first user behavior data may include 4 types of user behavior data, which are the amount of approval, the amount of review, the amount of search, and the amount of forwarding, respectively, for the target multimedia object in the first time period.

In another possible implementation, the first user behavior data includes only one type of user behavior data. In this implementation, the first user behavior data is user behavior data other than the playback volume of the target multimedia object. For example, the first user behavior data may be one of an amount of approval, an amount of sharing, an amount of forwarding, an amount of review, an amount of collection, and an amount of search of the target multimedia object in the first time period.

The second time period may represent a certain time period after the current time. That is, the second time period may represent a certain time period in the future. The time span of the second time period may be a second preset duration. For example, the second predetermined period of time may be 1 hour, half hour, 2 hours, 6 hours, 12 hours, 24 hours, and so forth.

In an embodiment of the disclosure, the first probability distribution may represent a probability distribution of an amount of play of the target multimedia object over the second time period, and the first probability distribution is predicted from at least one first user behavior data. The first probability distribution may be a discrete probability distribution or a continuous probability distribution. In one possible implementation manner, the first probability distribution may include a probability that the play amount of the target multimedia object in the second time period belongs to each of a plurality of preset play amount intervals. In another possible implementation, the first probability distribution may be a gaussian distribution.

In a possible implementation manner, a plurality of play amount intervals may be divided in advance, where the number of play amount intervals is greater than or equal to the number of preset categories. For example, the playback amount per hour may be divided into 10 playback amount sections, which are [0,0], (0,10], (10,100], (100,1000], (1000,5000], (5000,10000], (10000,20000], (20000,50000], (50000,100000) and (100000, + ∞). these 10 playback amount sections, z to U (0, …,9), may be represented by a variable z.

In one possible implementation, the first user behavior data includes at least two; predicting, based on the at least one first user behavior data, a first probability distribution of an amount of play of the target multimedia object over a second time period, comprising: for any one first user behavior data in at least two first user behavior data, predicting a second probability distribution of the playing amount of the target multimedia object in a second time period according to the first user behavior data; predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data.

In this implementation, the second probability distribution may represent a probability distribution of a play amount of the target multimedia object in the second time period, and the second probability distribution is predicted according to any one of the first user behavior data. The second probability distribution corresponds one-to-one to the first user behavior data. For example, the at least one first user behavior data includes an amount of approval, an amount of review, an amount of search, and an amount of forwarding for the target multimedia object during the first time period, a first second probability distribution of the amount of play of the target multimedia object in a second time period may be predicted based on the amount of praise of the target multimedia object in the first time period, predicting a second probability distribution of a second item of the playing amount of the target multimedia object in a second time period according to the comment amount of the target multimedia object in the first time period, predicting a third probability distribution of the playing amount of the target multimedia object in the second time period according to the searching amount of the target multimedia object in the first time period, and predicting a fourth item second probability distribution of the playing amount of the target multimedia object in the second time period according to the forwarding amount of the target multimedia object in the first time period. Wherein the second probability distribution may be a discrete probability distribution or a continuous probability distribution. As an example of this implementation, the second probability distribution may include a probability that the play amount of the target multimedia object in the second time period belongs to each of a plurality of preset play amount intervals. For example, the first item of second probability distribution may represent a probability that the play amount of the target multimedia object in the second time period belongs to each of a plurality of preset play amount intervals, which is determined according to the praise amount of the target multimedia object in the first time period; the second item of second probability distribution can represent the probability that the playing amount of the target multimedia object in the second time period belongs to each playing amount interval in a plurality of preset playing amount intervals, wherein the probability is determined according to the comment amount of the target multimedia object in the first time period; the third item of second probability distribution can represent the probability that the playing amount of the target multimedia object in the second time period belongs to each playing amount interval in the preset multiple playing amount intervals, wherein the probability is determined according to the searching amount of the target multimedia object in the first time period; the fourth second probability distribution may represent a probability that the play amount of the target multimedia object in the second time period belongs to each of the preset play amount intervals, which is determined according to the forwarding amount of the target multimedia object in the first time period. As another example of this implementation, the second probability distribution may be a gaussian distribution.

In this implementation, after at least two second probability distributions that correspond to the at least two first user behavior data in a one-to-one manner are predicted, a first probability distribution of the playing amount of the target multimedia object in the second time period is predicted according to the at least two second probability distributions. For example, after a first item second probability distribution, a second item second probability distribution, a third item second probability distribution and a fourth item second probability distribution of the playing amount of the target multimedia object in a second time period are respectively predicted according to the praise amount, the comment amount, the search amount and the forwarding amount of the target multimedia object in a first time period, a first probability distribution of the playing amount of the target multimedia object in the second time period is predicted according to the first item second probability distribution, the second item second probability distribution, the third item second probability distribution and the fourth item second probability distribution.

As an example of this implementation, the predicting, for any one of the at least two first user behavior data, a second probability distribution of a playing amount of the target multimedia object in a second time period according to the first user behavior data includes: for any one of at least two kinds of first user behavior data, adopting a probability distribution model corresponding to the first user behavior data in a pre-trained mixed probability distribution model to predict the first user behavior data to obtain a second probability distribution of the playing amount of the target multimedia object in a second time period, wherein the mixed probability distribution model comprises at least two probability distribution models which are in one-to-one correspondence with the at least two kinds of first user behavior data, and the at least two probability distribution models are respectively used for predicting the second probability distribution of the playing amount corresponding to the corresponding first user behavior data; the predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data comprises: predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data and weights in one-to-one correspondence with the at least two probability distribution models.

In one example, the Mixed probability distribution Model is a Gaussian Mixed Model (GMM) including at least two Gaussian models in one-to-one correspondence with the at least two first user behavior data. For example, the gaussian mixture model may include a gaussian model corresponding to the praise amount, a gaussian model corresponding to the comment amount, a gaussian model corresponding to the search amount, and a gaussian model corresponding to the forwarding amount.

In one example, the first probability distribution may be denoted as p (z | x), and p (z | x) may include probabilities that the play amount of the target multimedia object in the second time period belongs to each of preset 10 play amount intervals. Wherein x may represent at least one first user behavior data of the target multimedia object over a first time period. For example, x ═ x₁,x₂,x₃,x₄]Wherein x is₁Representing the amount of approval, x, of the target multimedia object during the first time period₂Representing the amount of review, x, of the target multimedia object over a first time period₃Representing the search volume, x, of the target multimedia object over a first period of time₄Representing the amount of forwarding of the target multimedia object over the first time period.

In one example, x_i～N(μ_i,σ_i) I.e. x_iObey mean value of mu_iStandard deviation of σ_iI is more than or equal to 1 and less than or equal to 4. To simplify the expression, θ may be adopted_iA parameter, θ, representing a Gaussian distribution of the ith first user behavior data_i＝[μ_i,σ_i]. The parameters of the gaussian distribution of any kind of user behavior data can be obtained by fitting according to the user behavior data of a specified multimedia object set in a certain specified time period in the past.

Wherein the number of multimedia objects in the specified set of multimedia objects may be plural. For example, the set of designated multimedia objects may be composed of all training objects, i.e. the set of designated multimedia objects may comprise all training objects. As another example, a given set of multimedia objects may include all multimedia objects on a multimedia object platform. As another example, the set of designated multimedia objects may include all multimedia objects on the multimedia object platform that satisfy the predetermined condition. As another example, a specified set of multimedia objects may be generated from a plurality of multimedia objects selected by a user.

According to the praise amounts of the multimedia objects in the appointed multimedia object set in the appointed time period, the mean value mu of the Gaussian distribution of the praise amounts can be calculated₁And standard deviation σ₁(ii) a According to the evaluation quantity of each multimedia object in the appointed multimedia object set in the appointed time period, the mean value mu of the Gaussian distribution of the evaluation quantity can be calculated₂And standard deviation σ₂(ii) a According to the search amount of each multimedia object in the designated multimedia object set in the designated time period, the mean value mu of the Gaussian distribution of the search amount can be calculated₃And standard deviation σ₃(ii) a According to the forwarding amount of each multimedia object in the designated multimedia object set in the designated time period, the mean value mu of the Gaussian distribution of the forwarding amount can be calculated₄And standard deviation σ₄。

First user behavior data x_iIs a probability density function f (x) of the gaussian distribution_i) Can be determined using equation 1:

first user behavior data x_iThe corresponding second probability distribution can be noted as φ (z | θ)_i). Combining the first user behavior data x_iSubstituting the value of (A) into equation 1, first user behavior data x can be obtained_iCorresponding gaussian distribution. By calculating first user behavior data x_iThe corresponding area under the curve of the gaussian distribution can obtain the probability that the playing amount of the target multimedia object in the second time period belongs to each playing amount interval of the preset multiple playing amount intervals, that is, the first user behavior data x can be obtained_iA corresponding second probability distribution.

In one example, the first probability distribution p (z | x) may be determined using equation 2:

wherein alpha is_iRepresenting first user behavior data x_iThe corresponding weight.

In this example, by fitting the corresponding probability distribution by using a probability distribution model corresponding to at least two types of first user behavior data one to one, the first user behavior data of a single value can be converted into the probability distribution, so that the trend of the play amount can be more accurately represented, and richer features are provided for the probability prediction of the next layer.

As another example of the implementation manner, the predicting, for any one of the at least two first user behavior data, a second probability distribution of a play amount of the target multimedia object in a second time period according to the first user behavior data includes: for any one first user behavior data in at least two first user behavior data, adopting a probability distribution function corresponding to the first user behavior data to predict the first user behavior data to obtain a second probability distribution of the playing amount of the target multimedia object in a second time period; the predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data comprises: predicting a first probability distribution of the playing amount of the target multimedia object in the second time period according to at least two second probability distributions in one-to-one correspondence with the at least two first user behavior data and weights in one-to-one correspondence with the at least two probability distribution models.

In the embodiment of the present disclosure, the number of the preset categories may be one or more than two. For example, the preset categories include 4 categories, which are distributed as S-category, a-category, B-category, and C-category. The probability that the target multimedia object belongs to at least one preset category may be predicted only according to the first probability distribution, or may be predicted according to the first probability distribution and other information, which is not limited herein. In the case where the number of the preset categories is two or more, the probability that the target multimedia object belongs to each of the two or more preset categories may be predicted according to the first probability distribution.

In one example, the probability that the target multimedia object belongs to at least one preset category may be denoted as p (y | x). In case the number of preset categories is at least two, p (y | x) may include a probability that the target multimedia object belongs to each of the at least two preset categories. In one example, p (y | x) may be determined using equation 3:

p (y | x) ═ p (z | x) × p (y | z) formula 3.

In a possible implementation, the predicting, according to the first probability distribution, the probability that the target multimedia object belongs to at least one preset category includes: performing full-connection processing on the first probability distribution to obtain a first full-connection processing result; and activating the first full-connection processing result to obtain the probability that the target multimedia object belongs to at least one preset category.

In this implementation, the first full-join processing result may represent a result of performing full-join processing on the first probability distribution. As an example of this implementation, the first probability distribution may be fully-connected through a fully-connected layer, resulting in a first fully-connected processing result. The parameters of the full-connection process may include a weight matrix and an offset term.

As an example of this implementation, in the case where the number of preset categories is two or more, the activation processing may be performed using a softmax function; in the case where the number of preset categories is one, a sigmoid function may be employed for the activation process. Of course, other activation functions may be used to perform activation processing on the full connection processing result, which is not limited herein.

In one example, the probability p (y | x) that the target multimedia object belongs to at least one preset category may be determined using equation 4:

p (y | x) ═ softmax (W × p (z | x) + b) formula 4,

where W denotes a weight matrix of the full join process, b denotes an offset term of the full join process, p (z | x) is a matrix of 10 × 1, W is a matrix of 4 × 10, and b is a matrix of 4 × 1.

In a possible implementation manner, before predicting the user behavior data by using a probability distribution model corresponding to the user behavior data in the pre-trained mixed probability distribution model, the method further includes: acquiring at least two second user behavior data of a training object in a third time period, wherein the training object is a multimedia object, the at least two second user behavior data at least comprise user behavior data except for the playing amount of the training object, and the third time period is before the first time period; determining a real category to which the training object belongs; for any one second user behavior data in the at least two second user behavior data, adopting a probability distribution model corresponding to the second user behavior data in the mixed probability distribution model to predict the second user behavior data to obtain a third probability distribution of the playing amount of the training object in a fourth time period, wherein the fourth time period is after the third time period; predicting a fourth probability distribution of the playing amount of the training object in the fourth time period according to at least two third probability distributions in one-to-one correspondence with the at least two second user behavior data and weights in one-to-one correspondence with the at least two probability distribution models; performing full-connection processing on the fourth probability distribution to obtain a second full-connection processing result; activating the second full-connection processing result to obtain the probability that the training object belongs to at least one preset category; and updating the weight and the parameters of the full-connection processing according to the real category and the probability that the training object belongs to at least one preset category.

In this implementation, the training objects may represent multimedia objects used to train the predictive model. The training objects may be of the same type as the target multimedia object. For example, the training object and the target multimedia object may both be short videos. As another example, the training object and the target multimedia object may both be long videos. As another example, the training object and the target multimedia object may both be audio. The number of training subjects may be plural. For example, the prediction model may be trained using 100 training objects with a true class of S, 100 training objects with a true class of a, 100 training objects with a true class of B, and 100 training objects with a true class of C.

The third time period and the fourth time period are both time periods prior to the current time, and the third time period is prior to the fourth time period. The time span of the third time period may be a third preset duration. For example, the third predetermined period of time may be 1 hour, half hour, 2 hours, 6 hours, 12 hours, 24 hours, and so forth. The third preset duration may be equal to the first preset duration. For example, the third preset time period and the first preset time period may both be 1 hour. The time span of the fourth time period may be a fourth preset time period. For example, the fourth preset time period may be 1 hour, half hour, 2 hours, 6 hours, 12 hours, 24 hours, and so on. The fourth preset duration may be equal to the second preset duration. For example, the fourth preset time period and the second preset time period may both be 1 hour.

The second user behavior data may represent user behavior data of the training subject over a third time period. The second user activity data may be of the same data type as the first user activity data. For example, the first user behavior data includes the amount of approval, the amount of review, the amount of search, and the amount of forwarding of the target multimedia object in the first time period, and the second user behavior data includes the amount of approval, the amount of review, the amount of search, and the amount of forwarding of the training object in the third time period.

The third probability distribution may represent a probability distribution of the amount of play of the training subject over the fourth time period, and the third probability distribution is predicted from any one of the second user behavior data. The third probability distribution corresponds to the second user behavior data one to one. For example, the at least two types of second user behavior data include an amount of approval, an amount of review, an amount of search, and an amount of forwarding of the training object in the third time period, a first third probability distribution of an amount of playback of the training object in the fourth time period may be predicted according to the amount of approval of the training object in the third time period, a second third probability distribution of an amount of playback of the training object in the fourth time period may be predicted according to the amount of review of the training object in the third time period, a third probability distribution of an amount of playback of the training object in the fourth time period may be predicted according to the amount of search of the training object in the third time period, and a fourth third probability distribution of an amount of playback of the training object in the fourth time period may be predicted according to the amount of forwarding of the training object in the third time period. Wherein the third probability distribution may be a discrete probability distribution or a continuous probability distribution. As an example of this implementation, the third probability distribution may include a probability that the play amount of the training subject in the fourth time period belongs to each of the preset plurality of play amount intervals. For example, the first third probability distribution may represent a probability that the play amount of the training object in the fourth time period belongs to each of a plurality of preset play amount intervals, which is determined according to the praise amount of the training object in the third time period; the second item of third probability distribution may represent a probability that the play amount of the training object in the fourth time period belongs to each of the preset multiple play amount intervals, which is determined according to the comment amount of the training object in the third time period; the third probability distribution may represent a probability that the playing amount of the training object in the fourth time period belongs to each of the preset playing amount intervals, which is determined according to the search amount of the training object in the third time period; the fourth third probability distribution may represent a probability that the playback amount of the training object in the fourth time period belongs to each of the preset multiple playback amount intervals, which is determined according to the forwarding amount of the training object in the third time period. As another example of this implementation, the third probability distribution may be a gaussian distribution.

In this implementation, after at least two third probability distributions that are in one-to-one correspondence with the at least two second user behavior data are predicted, a fourth probability distribution of the playing amount of the training object in the fourth time period is predicted according to the at least two third probability distributions. For example, after a first item third probability distribution, a second item third probability distribution, a third item third probability distribution and a fourth item third probability distribution of the playing amount of the training object in a fourth time period are respectively predicted according to the praise amount, the comment amount, the search amount and the forwarding amount of the training object in the third time period, a fourth probability distribution of the playing amount of the training object in the fourth time period is predicted according to the first item third probability distribution, the second item third probability distribution, the third item third probability distribution and the fourth item third probability distribution.

After the fourth probability distribution is obtained, the probability that the training object belongs to at least one preset category can be obtained by performing full connection processing and activation processing on the fourth probability distribution.

In this implementation, the predictive model may include a mixed probability distribution model, a fusion layer, a fully-connected layer, and an activation function layer. The output of the mixed probability distribution model is connected with the input of the full connection layer, and the output of the full connection layer is connected with the input of the activation function layer. The parameters to be trained of the mixed probability distribution model comprise the weights, and the parameters to be trained of the full connection layer can comprise a weight matrix and bias items. Wherein the weight and the parameter of the full connection process may be initialized randomly.

In one example, the value of the loss function L may be determined using equation 5:

where j denotes the jth preset category and m denotes the number of preset categories, e.g. m 4, p_jRepresenting the predicted probability of belonging to the jth preset category, g_jIndicating a true value, g, belonging to the jth preset category_jIs equal to 0 or 1, and

for example, g₁Indicating true values, g, belonging to class S₂Indicating true values, g, belonging to class A₃Indicating true values, g, belonging to class B₄Indicating a true value belonging to class C. If the real category to which the training object belongs is S type, g₁＝1，g₂＝g₃＝g₄0; if the real class to which the training object belongs is class A, g₂＝1，g₁＝g₃＝g₄0; if the real class to which the training object belongs is B class, g₃＝1，g₁＝g₂＝g₄0; if the real class to which the training object belongs is class C, g₄＝1，g₁＝g₂＝g₃＝0。

In one example, the weights and parameters of the full-connection process may be updated using a learning rate β. The model may be trained until convergence, or a preset number of iterations may be trained.

In this implementation, the parameters to be trained in the prediction model include only the weight of the probability distribution model and the parameters of the full-join process, and therefore, a prediction model with high accuracy can be trained even with a small number of training subjects (i.e., a small amount of training data). By adopting the implementation mode, the prediction model can learn the capability of predicting the category to which the multimedia object belongs.

As an example of this implementation, the determining the real category to which the training object belongs includes: obtaining a second amount of play of the training object over the fourth time period and a third amount of play of a set of designated multimedia objects over the fourth time period; and determining the real category of the training object according to the ratio of the second playing amount to the third playing amount.

For example, for any training object, if the ratio of the second playback volume to the third playback volume falls within the first ratio interval, it may be determined that the real category to which the training object belongs is S-class; if the ratio of the second playing amount to the third playing amount falls into a second ratio interval, determining that the real category to which the training object belongs is a category A; if the ratio of the second playing amount to the third playing amount falls into a third ratio interval, determining that the real category to which the training object belongs is B-type; if the ratio of the second playing amount to the third playing amount falls into the fourth ratio interval, it may be determined that the real category to which the training object belongs is class C. The lower boundary value of the first ratio interval is greater than the upper boundary value of the second ratio interval, the lower boundary value of the second ratio interval is greater than the upper boundary value of the third ratio interval, and the lower boundary value of the third ratio interval is greater than the upper boundary value of the fourth ratio interval.

In this example, by obtaining the second playback amount of the training object in the fourth time period and the third playback amount of the designated multimedia object set in the fourth time period, and determining the real category to which the training object belongs according to the ratio of the second playback amount to the third playback amount, accurate labeling data can be automatically obtained without manually labeling the category of the training object.

As another example of this implementation, the determining the real category to which the training object belongs includes: obtaining a fourth playback amount of the training object on a date to which the training object belongs during the fourth time period, and a fifth playback amount of the specified set of multimedia objects on the date to which the training object belongs during the fourth time period; and determining the real category to which the training object belongs according to the ratio of the fourth playing amount to the fifth playing amount. In this example, the time span of the fourth time period is less than 24 hours. For example, the time span of the fourth time period is 1 hour. In this example, the real category of the training object in each time period of the date to which the fourth time period belongs may be determined according to the play amount of the date to which the fourth time period belongs, so that the efficiency of determining the real category to which the training object belongs can be improved.

In another possible implementation manner, the first user behavior data only includes one type of user behavior data, and the first user behavior data is user behavior data other than the playing amount of the target multimedia object; predicting, based on the at least one first user behavior data, a first probability distribution of an amount of play of the target multimedia object over a second time period, comprising: and predicting the first user behavior data by adopting a preset probability distribution model to obtain a first probability distribution of the playing amount of the target multimedia object in a second time period.

As an example of this implementation, before the predicting the first user behavior data by using the preset probability distribution model, the method further includes: acquiring second user behavior data of a training object in a third time period, wherein the training object is a multimedia object, the second user behavior data is user behavior data except for the playing amount of the training object, and the third time period is before the first time period; determining a real category to which the training object belongs; predicting the second user behavior data by adopting the probability distribution model to obtain a third probability distribution of the playing amount of the training object in a fourth time period, wherein the fourth time period is after the third time period; performing full-connection processing on the third probability distribution to obtain a third full-connection processing result; activating the third full-connection processing result to obtain the probability that the training object belongs to at least one preset category; and updating the weight and the parameters of the full-connection processing according to the real category and the probability that the training object belongs to at least one preset category.

In another possible implementation manner, the predicting, according to the first probability distribution, the probability that the target multimedia object belongs to at least one preset category includes: performing full-connection processing on the first probability distribution to obtain a first full-connection processing result; activating the first full-connection processing result to obtain a first probability that the target multimedia object belongs to at least one preset category; acquiring a first playing amount of the target multimedia object in a first time period; predicting a second probability that the target multimedia object belongs to the at least one preset category according to the first playing amount; and determining the probability that the target multimedia object belongs to at least one preset category according to the first probability and the second probability. In this implementation, the at least one first user behavior data used to predict the first probability does not include a play amount. In this implementation, in addition to determining the first probability according to at least one first user behavior data except the playing amount of the target multimedia object in the first time period, the second probability is predicted according to the playing amount of the target multimedia object in the first time period, and the probability that the target multimedia object belongs to at least one preset category can be determined according to the weighted sum of the first probability and the second probability, so that the accuracy of predicting the probability that the target multimedia object belongs to at least one preset category can be further improved by performing prediction in combination with the playing amount.

In a possible implementation, after predicting the probability that the target multimedia object belongs to at least one preset category, the method further includes: and distributing the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to at least one preset category.

As an example of this implementation, the at least one preset category includes an S class; the allocating the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to at least one preset category includes: and allocating the bandwidth corresponding to the target multimedia object according to the probability that the target-to-media object belongs to the S class, wherein the bandwidth corresponding to the target multimedia object is positively correlated with the probability that the target multimedia object belongs to the S class. That is, the greater the probability that the target multimedia object belongs to the S class is, the greater the bandwidth corresponding to the target multimedia object is; the smaller the probability that the target multimedia object belongs to the S class is, the smaller the bandwidth corresponding to the target multimedia object is.

As another example of this implementation, the at least one category includes class C; the allocating the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to at least one preset category includes: and distributing the bandwidth corresponding to the target multimedia object according to the probability that the target pair media object belongs to the C class, wherein the bandwidth corresponding to the target multimedia object is inversely related to the probability that the target multimedia object belongs to the C class. That is, the higher the probability that the target multimedia object belongs to class C, the smaller the bandwidth corresponding to the target multimedia object; the smaller the probability that the target multimedia object belongs to the class C, the larger the bandwidth corresponding to the target multimedia object.

As another example of this implementation, the number of preset categories is greater than or equal to 2; the allocating the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to at least one preset category includes: determining the preset category corresponding to the maximum probability in the probabilities that the target multimedia object belongs to at least one preset category as the category to which the target multimedia object belongs; and distributing the bandwidth corresponding to the target multimedia object according to the category of the target multimedia object. For example, the at least one preset category includes S-class, a-class, B-class, and C-class; the probability that the target multimedia object belongs to the at least one preset category includes 4 probabilities, which are respectively: the probability that the target multimedia object belongs to class S, the probability that the target multimedia object belongs to class a, the probability that the target multimedia object belongs to class B, and the probability that the target multimedia object belongs to class C. If the probability that the target multimedia object belongs to the S class is the largest in the 4 probabilities, the class to which the target multimedia object belongs can be determined to be the S class; if the probability that the target multimedia object belongs to the class A is the highest among the 4 probabilities, the class to which the target multimedia object belongs can be determined to be the class A; if the probability that the target multimedia object belongs to the class B is the largest in the 4 probabilities, the class to which the target multimedia object belongs can be determined to be the class B; if the probability that the target multimedia object belongs to the class C is the highest among the 4 probabilities, it may be determined that the class to which the target multimedia object belongs is the class C. In this example, the bandwidth corresponding to class S is greater than the bandwidth corresponding to class a, the bandwidth corresponding to class a is greater than the bandwidth corresponding to class B, and the bandwidth corresponding to class B is greater than the bandwidth corresponding to class C. For example, if the multimedia object V₁Belonging to S class, multimedia object V₂Belongs to class A, then is assigned to multimedia object V₁Is greater than the bandwidth allocated to the multimedia object V₂The bandwidth of (c).

According to the implementation mode, the multimedia object platform is facilitated to distribute the multimedia objects more effectively, and therefore the bandwidth of the content distribution network is facilitated to be saved.

The following describes a method for predicting a multimedia object according to an embodiment of the present disclosure with a specific application scenario. In this application scenario, the prediction method of the multimedia object may be performed by a server corresponding to the short video platform.

Fig. 2 is a schematic diagram illustrating an application scenario of a prediction method for a multimedia object according to an embodiment of the present disclosure. In the example shown in fig. 2, the predictive model may include a GMM, a fusion layer, a fully-connected layer, and an activation function layer. The GMM may include 4 gaussian models, which are a gaussian model corresponding to the praise amount, a gaussian model corresponding to the comment amount, a gaussian model corresponding to the search amount, and a gaussian model corresponding to the forwarding amount. The gaussian models can each calculate a gaussian distribution using equation 1. The fusion layer can calculate the probability distribution using equation 2. The full-link layer may perform the full-link process using equation 4. The activation function layer may employ a softmax function.

In the application scenario, 100 training objects with a real class of S, 100 training objects with a real class of a, 100 training objects with a real class of B, and 100 training objects with a real class of C may be used to train the prediction model. The user behavior data adopted by the training prediction model can include the playing amount, the praise amount, the comment amount, the search amount and the forwarding amount of the training object. The playing amount can be used for determining the real category of the training object, and the praise amount, the comment amount, the search amount and the forwarding amount can be used as the input of the prediction model.

By adopting the window function, the praise amount, the comment amount, the search amount and the forwarding amount of each training object in the training object set in a specified time period can be obtained from the data source. According to the praise amount of the training object set in a specified time period, the mean value mu of the Gaussian distribution of the praise amount can be obtained through fitting₁And standard deviation σ₁(ii) a According to the evaluation quantity of the training object set in a specified time period, Gaussian of the evaluation quantity can be obtained through fittingMean value of distribution mu₂And standard deviation σ₂(ii) a According to the search quantity of the training object set in a specified time period, the mean value mu of the Gaussian distribution of the search quantity can be obtained through fitting₃And standard deviation σ₃(ii) a According to the forwarding amount of the training object set in a specified time period, the mean value mu of the Gaussian distribution of the forwarding amount can be obtained through fitting₄And standard deviation σ₄。

After the training of the prediction model is completed, the praise amount, the comment amount, the search amount and the forwarding amount of the target short video in the last hour can be obtained, the praise amount, the comment amount, the search amount and the forwarding amount of the target short video in the last hour are input into the prediction model, and the probability that the target short video belongs to the S class, the A class, the B class and the C class is predicted through the prediction model.

According to the probability that each target short video belongs to the S class, the target short videos can be sorted, and M short videos with the highest sorting order can be used as the selected video, wherein M is an integer greater than or equal to 1.

In one example, a report of the pick video may be sent to the operator's terminal device every hour. Fig. 3 is a schematic diagram illustrating a report corresponding to a short video with the highest probability of belonging to the S class in the method for predicting a multimedia object according to the embodiment of the present disclosure. In the example shown in fig. 3, the report may include a hash value of the short video, a probability that the short video belongs to class S, class a, and class B, statistics from which the probability that the short video belongs to class S, class a, and class B is calculated, and a gaussian distribution map of the playback volume of the short video.

In the application scene, high-quality short videos can be timely found and captured from a large number of short videos in the short video platform, and a large bandwidth is allocated to the high-quality short videos, so that the activity of the short video platform can be improved on the premise of saving the bandwidth of a content distribution network. Namely, an operation strategy can be provided for the short video platform, and the short video platform is helped to better monitor and distribute traffic. In addition, in the application scene, the probability that each short video in the short video platform belongs to the S type, the A type, the B type and the C type can be dynamically updated every hour by taking hours as granularity, and the timeliness is high.

It is understood that the above-mentioned method embodiments of the present disclosure can be combined with each other to form a combined embodiment without departing from the logic of the principle, which is limited by the space, and the detailed description of the present disclosure is omitted. Those skilled in the art will appreciate that in the above methods of the specific embodiments, the specific order of execution of the steps should be determined by their function and possibly their inherent logic.

In addition, the present disclosure also provides a multimedia object prediction apparatus, an electronic device, a computer-readable storage medium, and a program, which can be used to implement any one of the multimedia object prediction methods provided by the present disclosure, and corresponding technical solutions and technical effects can be referred to in corresponding descriptions of the method section and are not described again.

Fig. 4 shows a block diagram of a prediction apparatus for a multimedia object provided by an embodiment of the present disclosure. As shown in fig. 4, the apparatus for predicting a multimedia object includes:

a first obtaining module 41, configured to obtain at least one first user behavior data of a target multimedia object in a first time period, where the at least one first user behavior data includes user behavior data other than a play amount of the target multimedia object;

a first prediction module 42, configured to predict, according to the at least one first user behavior data, a first probability distribution of a play amount of the target multimedia object over a second time period, wherein the second time period is subsequent to the first time period;

a second prediction module 43, configured to predict, according to the first probability distribution, a probability that the target multimedia object belongs to at least one preset category.

the first prediction module 42 is configured to:

In one possible implementation, the first prediction module 42 is configured to:

In a possible implementation manner, the second prediction module 43 is configured to:

In one possible implementation, the apparatus further includes:

In one possible implementation, the determining module is configured to:

In some embodiments, functions or modules included in the apparatus provided in the embodiments of the present disclosure may be used to execute the method described in the above method embodiments, and specific implementations and technical effects thereof may refer to the description of the above method embodiments, which are not described herein again for brevity.

Embodiments of the present disclosure also provide a computer-readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the above-described method. The computer-readable storage medium may be a non-volatile computer-readable storage medium, or may be a volatile computer-readable storage medium.

Embodiments of the present disclosure also provide a computer program, which includes computer readable code, and when the computer readable code runs in an electronic device, a processor in the electronic device executes the above method.

The disclosed embodiments also provide a computer program product comprising computer readable code or a non-volatile computer readable storage medium carrying computer readable code, which when run in an electronic device, a processor in the electronic device performs the above method.

An embodiment of the present disclosure further provides an electronic device, including: one or more processors; a memory for storing executable instructions; wherein the one or more processors are configured to invoke the memory-stored executable instructions to perform the above-described method.

The electronic device may be provided as a terminal, server, or other form of device.

Fig. 5 illustrates a block diagram of an electronic device 800 provided by an embodiment of the disclosure. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, or the like terminal.

Referring to fig. 5, electronic device 800 may include one or more of the following components: processing component 802, memory 804, power component 806, multimedia component 808, audio component 810, input/output (I/O) interface 812, sensor component 814, and communication component 816.

The processing component 802 generally controls overall operation of the electronic device 800, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations. The processing components 802 may include one or more processors 820 to execute instructions to perform all or a portion of the steps of the methods described above. Further, the processing component 802 can include one or more modules that facilitate interaction between the processing component 802 and other components. For example, the processing component 802 can include a multimedia module to facilitate interaction between the multimedia component 808 and the processing component 802.

The memory 804 is configured to store various types of data to support operations at the electronic device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phonebook data, messages, pictures, videos, and so forth. The memory 804 may be implemented by any type or combination of volatile or non-volatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.

The power supply component 806 provides power to the various components of the electronic device 800. The power components 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the electronic device 800.

The multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and a user. In some embodiments, the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front facing camera and/or a rear facing camera. The front camera and/or the rear camera may receive external multimedia data when the electronic device 800 is in an operation mode, such as a shooting mode or a video mode. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

The audio component 810 is configured to output and/or input audio signals. For example, the audio component 810 includes a Microphone (MIC) configured to receive external audio signals when the electronic device 800 is in an operational mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signals may further be stored in the memory 804 or transmitted via the communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

The I/O interface 812 provides an interface between the processing component 802 and peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to: a home button, a volume button, a start button, and a lock button.

The sensor assembly 814 includes one or more sensors for providing various aspects of state assessment for the electronic device 800. For example, the sensor assembly 814 may detect an open/closed state of the electronic device 800, the relative positioning of components, such as a display and keypad of the electronic device 800, the sensor assembly 814 may also detect a change in the position of the electronic device 800 or a component of the electronic device 800, the presence or absence of user contact with the electronic device 800, orientation or acceleration/deceleration of the electronic device 800, and a change in the temperature of the electronic device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of a nearby object without any physical contact. The sensor assembly 814 may also include a light sensor, such as a Complementary Metal Oxide Semiconductor (CMOS) or Charge Coupled Device (CCD) image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 may access a wireless network based on a communication standard, such as a wireless network (Wi-Fi), a second generation mobile communication technology (2G), a third generation mobile communication technology (3G), a fourth generation mobile communication technology (4G), a long term evolution of universal mobile communication technology (LTE), a fifth generation mobile communication technology (5G), or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 further includes a Near Field Communication (NFC) module to facilitate short-range communications. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

In an exemplary embodiment, the electronic device 800 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic components for performing the above-described methods.

In an exemplary embodiment, a non-transitory computer-readable storage medium, such as the memory 804, is also provided that includes computer program instructions executable by the processor 820 of the electronic device 800 to perform the above-described methods.

Fig. 6 illustrates a block diagram of another electronic device 1900 provided by an embodiment of the disclosure. For example, the electronic device 1900 may be provided as a server. Referring to fig. 6, electronic device 1900 includes a processing component 1922 further including one or more processors and memory resources, represented by memory 1932, for storing instructions, e.g., applications, executable by processing component 1922. The application programs stored in memory 1932 may include one or more modules that each correspond to a set of instructions. Further, the processing component 1922 is configured to execute instructions to perform the above-described method.

The electronic device 1900 may also include a power component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input/output (I/O) interface 1958. The electronic device 1900 may operate based on an operating system, such as the Microsoft Server operating system (Windows Server), stored in the memory 1932^TM) Apple Inc. of the present application based on the graphic user interface operating System (Mac OS X)^TM) Multi-user, multi-process computer operating system (Unix)^TM) Free and open native code Unix-like operating System (Linux)^TM) Open native code Unix-like operating System (FreeBSD)^TM) Or the like.

In an exemplary embodiment, a non-transitory computer readable storage medium, such as the memory 1932, is also provided that includes computer program instructions executable by the processing component 1922 of the electronic device 1900 to perform the above-described methods.

The present disclosure may be systems, methods, and/or computer program products. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied thereon for causing a processor to implement various aspects of the present disclosure.

The computer readable storage medium may be a tangible device that can hold and store the instructions for use by the instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic memory device, a magnetic memory device, an optical memory device, an electromagnetic memory device, a semiconductor memory device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a Static Random Access Memory (SRAM), a portable compact disc read-only memory (CD-ROM), a Digital Versatile Disc (DVD), a memory stick, a floppy disk, a mechanical coding device, such as punch cards or in-groove projection structures having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media as used herein is not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (e.g., optical pulses through a fiber optic cable), or electrical signals transmitted through electrical wires.

The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device via a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. The network adapter card or network interface in each computing/processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in the respective computing/processing device.

The computer program instructions for carrying out operations of the present disclosure may be assembler instructions, Instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, the electronic circuitry that can execute the computer-readable program instructions implements aspects of the present disclosure by utilizing the state information of the computer-readable program instructions to personalize the electronic circuitry, such as a programmable logic circuit, a Field Programmable Gate Array (FPGA), or a Programmable Logic Array (PLA).

Various aspects of the present disclosure are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer-readable program instructions.

These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer-readable program instructions may also be stored in a computer-readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer-readable medium storing the instructions comprises an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.

The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The computer program product may be embodied in hardware, software or a combination thereof. In an alternative embodiment, the computer program product is embodied in a computer storage medium, and in another alternative embodiment, the computer program product is embodied in a Software product, such as a Software Development Kit (SDK), or the like.

Having described embodiments of the present disclosure, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen in order to best explain the principles of the embodiments, the practical application, or improvements made to the technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for predicting a multimedia object, comprising:

2. The method of claim 1, wherein the first user behavior data comprises at least two;

3. The method of claim 2,

4. The method according to any of claims 1 to 3, wherein said predicting, from said first probability distribution, the probability that said target multimedia object belongs to at least one preset category comprises:

5. The method of claim 3, wherein before predicting the user behavior data using a probability distribution model corresponding to the user behavior data in the pre-trained mixed probability distribution model, the method further comprises:

determining a real category to which the training object belongs;

6. The method of claim 5, wherein determining the true class to which the training object belongs comprises:

7. The method according to any of claims 1 to 3, wherein said predicting, from said first probability distribution, the probability that said target multimedia object belongs to at least one preset category comprises:

8. An apparatus for predicting a multimedia object, comprising:

9. An electronic device, comprising:

one or more processors;

a memory for storing executable instructions;

wherein the one or more processors are configured to invoke the memory-stored executable instructions to perform the method of any one of claims 1 to 7.

10. A computer readable storage medium having computer program instructions stored thereon, which when executed by a processor implement the method of any one of claims 1 to 7.

11. A computer program product comprising computer readable code or a non-transitory computer readable storage medium carrying computer readable code which, when run in an electronic device, causes a processor in the electronic device to perform the method of any of claims 1 to 7.