Multimedia object prediction method, device, equipment, medium and program product

By calculating the acceleration of user behavior proportion of multimedia objects in different time periods and using neural network prediction, the problems of high delay and low accuracy of multimedia object type prediction in the prior art are solved, and accurate prediction and bandwidth of multimedia objects are achieved save.

CN114519112BActive Publication Date: 2024-11-22BEIJING DAJIA INTERNET INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210107935.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-01-28
Publication Date
2024-11-22
Estimated Expiration
2042-01-28

AI Technical Summary

Technical Problem

The prior art has high latency and low accuracy when predicting multimedia object types. Especially on short video platforms, it is difficult to effectively save the bandwidth of the content distribution network when predicting a large number of multimedia objects.

Method used

By obtaining the user behavior proportion data of the target multimedia object in different time periods, the acceleration of the user behavior proportion is calculated, and using pre-trained neural networks to predict it, accurate prediction of multimedia object types and future performance predictions are achieved.

Benefits of technology

Accurate prediction of multimedia object types is achieved, with low latency, and multimedia objects can be distributed more efficiently, thereby saving the bandwidth of the content distribution network.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114519112B_ABST
    Figure CN114519112B_ABST
Patent Text Reader

Abstract

The present disclosure relates to a method, device, equipment, medium and program product for predicting multimedia objects. The method comprises: obtaining first user behavior proportion data of a target media object in a first time period, and second user behavior proportion data of the target multimedia object in a second time period, wherein the second time period is before the first time period; determining a first acceleration of the user behavior proportion of the target multimedia object according to the first user behavior proportion data and the second user behavior proportion data; and predicting the probability that the target multimedia object belongs to a preset type according to the first acceleration.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the field of multimedia technology, and in particular to a prediction method, device, equipment, medium and program product for a multimedia object. Background Art

[0002] Compared with text and pictures, multimedia objects such as video and audio are more expressive. For example, video combines auditory information with visual information, has richer content, stronger expressiveness, and better intuitiveness, so it is widely popular on various social media platforms. With the accelerated pace of life, short videos, a fragmented way of obtaining information, have gradually become popular among people. More and more people are willing to share their own short videos on short video platforms, and share with others short videos that they find interesting or other worth watching. For multimedia object platforms (such as video platforms or audio platforms), predicting a large number of multimedia objects is of great significance in saving bandwidth of content distribution networks. Summary of the invention

[0003] The present disclosure provides a technical solution for predicting multimedia objects.

[0004] According to one aspect of the present disclosure, a method for predicting a multimedia object is provided, comprising:

[0005] Obtaining first user behavior ratio data of a target media object in a first time period, and second user behavior ratio data of the target multimedia object in a second time period, wherein the second time period is before the first time period;

[0006] Determining a first acceleration of a user behavior ratio of the target multimedia object according to the first user behavior ratio data and the second user behavior ratio data;

[0007] The probability that the target multimedia object belongs to a preset type is predicted according to the first acceleration.

[0008] By obtaining first user behavior proportion data of the target media object in a first time period, and second user behavior proportion data of the target multimedia object in a second time period, a first acceleration of the user behavior proportion of the target multimedia object is determined according to the first user behavior proportion data and the second user behavior proportion data, and based on the first acceleration, the probability that the target multimedia object belongs to a preset type is predicted. Therefore, based on the acceleration of the user behavior proportion of the target multimedia object, the probability that the target multimedia object belongs to a preset type can be accurately predicted. In addition, the embodiment of the present disclosure can predict the future performance of the target multimedia object based on the generated user behavior data with low latency, thereby helping the multimedia object platform to distribute multimedia objects more efficiently, thereby helping to save bandwidth of the content distribution network.

[0009] In a possible implementation, the first time period includes at least two first sub-time periods, the first user behavior proportion data includes at least two first sub-user behavior proportion data corresponding one-to-one to the at least two first sub-time periods, the second time period includes at least two second sub-time periods, and the second user behavior proportion data includes at least two second sub-user behavior proportion data corresponding one-to-one to the at least two second sub-time periods;

[0010] The obtaining of the first user behavior proportion data of the target media object in the first time period, and the second user behavior proportion data of the target multimedia object in the second time period, includes:

[0011] Determine the first user behavior proportion data of the target for the media object in the first time period according to the at least two first sub-user behavior proportion data;

[0012] The second user behavior proportion data of the target multimedia object in the second time period is determined according to the at least two items of second sub-user behavior proportion data.

[0013] In this implementation, the first user behavior proportion data is determined based on at least two first sub-user behavior proportion data corresponding one-to-one to at least two first sub-time periods, the second user behavior proportion data is determined based on at least two second sub-user behavior proportion data corresponding one-to-one to at least two second sub-time periods, and the first acceleration is determined based on the first user behavior proportion data and the second user behavior proportion data determined thereby, thereby being able to determine a smoother first acceleration, thereby being able to achieve a smoother type prediction for the target multimedia object.

[0014] In one possible implementation,

[0015] The determining, according to the at least two first sub-user behavior proportion data, the first user behavior proportion data of the target to the media object in the first time period comprises: determining, according to the at least two first sub-user behavior proportion data and at least two first weights corresponding to the at least two first sub-user behavior proportion data, the first user behavior proportion data of the target to the media object in the first time period;

[0016] Determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data includes: determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data and at least two second weights corresponding one-to-one to the at least two second sub-user behavior proportion data.

[0017] In this implementation, the first user behavior proportion data is determined based on at least two first sub-user behavior proportion data and at least two first weights corresponding one-to-one to the at least two first sub-user behavior proportion data, the second user behavior proportion data is determined based on at least two second sub-user behavior proportion data and at least two second weights corresponding one-to-one to the at least two second sub-user behavior proportion data, and the first acceleration is determined based on the first user behavior proportion data and the second user behavior proportion data determined thereby, and the first acceleration determined thereby can more accurately predict the probability that the target multimedia object belongs to a preset type.

[0018] In a possible implementation, the first weight corresponding to any one of the at least two first sub-user behavior proportion data is negatively correlated with the interval between the first sub-time period corresponding to the first sub-user behavior proportion data and the current time, and is the exponential smoothing coefficient corresponding to the first sub-time period.

[0019] By adopting the first weight in this implementation to determine the first user behavior proportion data, and determining the first acceleration based on the first user behavior proportion data determined thereby, the first acceleration determined thereby can further improve the accuracy of type prediction for the target multimedia object.

[0020] In a possible implementation, the second weight corresponding to any one of the at least two second sub-user behavior proportion data is positively correlated with the interval between the second sub-time period corresponding to the second sub-user behavior proportion data and the current time.

[0021] By adopting the second weight in this implementation to determine the second user behavior proportion data, and determining the first acceleration based on the second user behavior proportion data determined thereby, the first acceleration determined thereby can further improve the accuracy of type prediction for the target multimedia object.

[0022] In a possible implementation manner, predicting, according to the first acceleration, a probability that the target multimedia object belongs to a preset type includes:

[0023] The first acceleration is input into a pre-trained neural network, and the probability that the target multimedia object belongs to a preset type is predicted by the neural network.

[0024] In this implementation, the first acceleration is input into a pre-trained neural network, and the probability that the target multimedia object belongs to a preset type is predicted via the neural network, thereby improving the accuracy and speed of predicting the probability that the target multimedia object belongs to the preset type.

[0025] In a possible implementation, before inputting the first acceleration into a pre-trained neural network, the method further includes:

[0026] Acquire a training object set, wherein the training object set includes a plurality of training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used to indicate whether the training objects belong to the preset type;

[0027] The neural network is trained using the training object set.

[0028] In this implementation, a training object set is obtained, wherein the training object set includes multiple training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used to indicate whether the training objects belong to the preset type. The training object set is used to train the neural network, thereby enabling the neural network to learn the ability to predict the probability that the multimedia object belongs to the preset type.

[0029] According to one aspect of the present disclosure, there is provided a device for predicting a multimedia object, comprising:

[0030] An acquisition module, used to obtain first user behavior ratio data of a target media object in a first time period, and second user behavior ratio data of the target multimedia object in a second time period, wherein the second time period is before the first time period;

[0031] a determination module, configured to determine a first acceleration of a user behavior ratio of the target multimedia object according to the first user behavior ratio data and the second user behavior ratio data;

[0032] A prediction module is used to predict the probability that the target multimedia object belongs to a preset type according to the first acceleration.

[0033] In a possible implementation, the first time period includes at least two first sub-time periods, the first user behavior proportion data includes at least two first sub-user behavior proportion data corresponding one-to-one to the at least two first sub-time periods, the second time period includes at least two second sub-time periods, and the second user behavior proportion data includes at least two second sub-user behavior proportion data corresponding one-to-one to the at least two second sub-time periods;

[0034] The acquisition module is used for:

[0035] Determine the first user behavior proportion data of the target for the media object in the first time period according to the at least two first sub-user behavior proportion data;

[0036] The second user behavior proportion data of the target multimedia object in the second time period is determined according to the at least two items of second sub-user behavior proportion data.

[0037] In a possible implementation, the obtaining module is used to

[0038] Determine the first user behavior proportion data of the target for the media object in the first time period according to the at least two first sub-user behavior proportion data and at least two first weights corresponding to the at least two first sub-user behavior proportion data one by one;

[0039] The second user behavior proportion data of the target for the media object in the second time period is determined according to the at least two second sub-user behavior proportion data and at least two second weights corresponding to the at least two second sub-user behavior proportion data.

[0040] In a possible implementation, the first weight corresponding to any one of the at least two first sub-user behavior proportion data is negatively correlated with the interval between the first sub-time period corresponding to the first sub-user behavior proportion data and the current time, and is the exponential smoothing coefficient corresponding to the first sub-time period.

[0041] In a possible implementation, the second weight corresponding to any one of the at least two second sub-user behavior proportion data is positively correlated with the interval between the second sub-time period corresponding to the second sub-user behavior proportion data and the current time.

[0042] In a possible implementation, the prediction module is used to:

[0043] The first acceleration is input into a pre-trained neural network, and the probability that the target multimedia object belongs to a preset type is predicted by the neural network.

[0044] In a possible implementation manner, the device further includes:

[0045] An acquisition module, used for acquiring a training object set, wherein the training object set includes a plurality of training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used for indicating whether the training objects belong to the preset type;

[0046] A training module is used to train the neural network using the training object set.

[0047] According to one aspect of the present disclosure, an electronic device is provided, comprising: one or more processors; a memory for storing executable instructions; wherein the one or more processors are configured to call the executable instructions stored in the memory to execute the above method.

[0048] According to one aspect of the present disclosure, a computer-readable storage medium is provided, on which computer program instructions are stored, and the computer program instructions implement the above method when executed by a processor.

[0049] According to one aspect of the present disclosure, a computer program product is provided, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in an electronic device, a processor in the electronic device executes the above method.

[0050] In an embodiment of the present disclosure, by obtaining first user behavior proportion data of a target media object in a first time period, and second user behavior proportion data of a target multimedia object in a second time period, a first acceleration of the user behavior proportion of the target multimedia object is determined based on the first user behavior proportion data and the second user behavior proportion data, and based on the first acceleration, a probability that the target multimedia object belongs to a preset type is predicted. Thus, based on the acceleration of the user behavior proportion of the target multimedia object, the probability that the target multimedia object belongs to a preset type can be accurately predicted. In addition, the embodiment of the present disclosure can predict the future performance of the target multimedia object based on the generated user behavior data with lower latency, thereby helping the multimedia object platform to distribute multimedia objects more efficiently, thereby helping to save bandwidth of the content distribution network.

[0051] 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.

[0052] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used to illustrate the technical solutions of the present disclosure together with the specification.

[0054] Figure 1 A flowchart of a multimedia object prediction method provided by an embodiment of the present disclosure is shown.

[0055] Figure 2 A comparison diagram is shown between the acceleration determined by the multimedia object prediction method provided by the embodiment of the present disclosure and the acceleration determined according to equation 12 in the related art.

[0056] Figure 3 A comparison chart showing the acceleration determined by the multimedia object prediction method provided by an embodiment of the present disclosure and the user behavior proportion data of the multimedia object is shown.

[0057] Figure 4 A block diagram of a multimedia object prediction device provided by an embodiment of the present disclosure is shown.

[0058] Figure 5 A block diagram of another electronic device 1900 provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0059] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0060] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0061] The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set consisting of A, B, and C.

[0062] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure.

[0063] In the related art, it is usually determined which type a multimedia object belongs to based on the a posteriori data of the multimedia object. For example, after the multimedia object is released, wait for a period of time (for example, 3-7 days), count the daily playback volume of the multimedia object during this period, and then determine the peak value of the playback volume of the multimedia object during this period. Calculate the ratio of the peak value to the total playback volume of all multimedia objects on the multimedia object platform, determine which of the pre-set ratio intervals the ratio falls into, and thus determine which type the multimedia object belongs to. For example, if the ratio is greater than 0.5%, it is determined that the multimedia object belongs to Class S, if the ratio is greater than 0.1% and less than or equal to 0.5%, it is determined that the multimedia object belongs to Class A, and so on.

[0064] Taking short videos as an example, in related technologies, after a short video is released, it is usually necessary to wait for 3-7 days, count the number of views of the short video every day during this period, and then determine the peak value of the views of the short video during this period. The ratio of the peak value to the total number of views of all short videos on the short video platform is calculated, and the ratio is determined to determine which pre-set ratio range the ratio falls into, thereby determining which type the short video belongs to.

[0065] In this way, since the data used to determine the type of multimedia object is a posteriori data, it often takes more than 3-7 days to wait, so the delay is high. In addition, this method often requires manual recalibration of the ratio interval every once in a while (for example, every month), which is cumbersome and has low accuracy.

[0066] In order to solve technical problems similar to those described above, the embodiments of the present disclosure provide a method, apparatus, device, medium and program product for predicting multimedia objects, by obtaining first user behavior proportion data of a target media object in a first time period, and second user behavior proportion data of a target multimedia object in a second time period, determining a first acceleration of the user behavior proportion of the target multimedia object according to the first user behavior proportion data and the second user behavior proportion data, and predicting the probability that the target multimedia object belongs to a preset type according to the first acceleration, thereby being able to accurately predict the probability that the target multimedia object belongs to a preset type based on the acceleration of the user behavior proportion of the target multimedia object, and the embodiments of the present disclosure are able to predict the future performance of the target multimedia object based on the generated user behavior data, with low latency, thereby helping the multimedia object platform to distribute multimedia objects more efficiently, thereby helping to save bandwidth of the content distribution network.

[0067] The multimedia object prediction method provided by the embodiments of the present disclosure is described in detail below with reference to the accompanying drawings.

[0068] Figure 1 A flowchart of a multimedia object prediction method provided by an embodiment of the present disclosure is shown. In one possible implementation, the multimedia object prediction method may be executed by a terminal device or a server or other electronic device. The terminal device may be a user equipment (UE), a mobile device, a user terminal, a 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, etc. In some possible implementations, the multimedia object prediction method may be implemented by a processor calling computer-readable instructions stored in a memory. For example Figure 1 As shown, the multimedia object prediction method includes steps S11 to S13.

[0069] In step S11, first user behavior ratio data of a target media object in a first time period and second user behavior ratio data of the target multimedia object in a second time period are obtained, wherein the second time period is before the first time period.

[0070] In step S12, a first acceleration of the user behavior ratio of the target multimedia object is determined according to the first user behavior ratio data and the second user behavior ratio data.

[0071] In step S13, the probability that the target multimedia object belongs to a preset type is predicted according to the first acceleration.

[0072] In the disclosed embodiments, the target multimedia object may be any multimedia object that needs to be predicted. The multimedia object may be a video or an audio, etc., and the video may be a short video or a long video. For example, each video on a video platform may be used as a target multimedia object to predict the probability that each video on the video platform belongs to a preset type. For another example, each video on the video platform that meets the preset conditions may be used as a target multimedia object to predict the probability that each video on the video platform that meets the preset conditions belongs to a preset type. The preset conditions may include at least one of a preset release time condition, a preset clarity condition, a preset duration condition, etc.

[0073] The first time period may represent a time period before the current time. The first time period may include the current time 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 duration may be 6 hours, 12 hours, 18 hours, etc.

[0074] The second time period may represent a time period before the first time period. The second time period may be continuous with the first time period or may not be continuous with the first time period. The time span of the second time period may be a second preset duration. For example, the second preset duration may be 6 hours, 12 hours, 18 hours, etc. The second preset duration may be equal to or different from the first preset duration.

[0075] In the embodiments of the present disclosure, user behavior may represent user behavior with respect to multimedia objects, and user behavior data may be data generated based on user behavior with respect to multimedia objects. For example, user behavior may include at least one of playing, liking, sharing, forwarding, commenting, collecting, searching, etc., and accordingly, user behavior data may include at least one of playing data, liking data, sharing data, forwarding data, commenting data, collecting data, searching data, etc. For example, user behavior data may include at least one of the amount of playing, the amount of liking, the amount of sharing, the amount of forwarding, the amount of commenting, the amount of collecting, the amount of searching, etc.

[0076] The user behavior percentage data of the target multimedia object may represent the ratio data of the user behavior data of the target multimedia object to the user behavior data of the same type of the target multimedia object set. For example, the play volume percentage data of the target multimedia object may represent the ratio data of the play volume of the target multimedia object to the play volume of the target multimedia object set; the like volume percentage data of the target multimedia object may represent the ratio data of the like volume of the target multimedia object to the like volume of the target multimedia object set.

[0077] Among them, the target multimedia object set can represent a multimedia object set used to predict the probability that the target multimedia object belongs to a preset type. The number of multimedia objects in the target multimedia object set can be more than two. The target multimedia object set may include the target multimedia object or may not include the target multimedia object. For example, the target multimedia object set may include all videos on the video platform. In another example, the target multimedia object set may include all videos on the video platform that meet preset conditions. In another example, the target multimedia object set can be generated based on multiple videos selected by the user. In this example, the user can select multiple videos as the target multimedia object set, so that the multiple videos selected by the user can be used as a comparison benchmark to predict the probability that the target video belongs to the preset type.

[0078] The first user behavior proportion data may represent the user behavior proportion data of the target multimedia object in the first time period. The first user behavior proportion data may include one or more user behavior proportion data of the target multimedia object in the first time period. In a possible implementation, the first user behavior proportion data may include at least one of the following: the play volume proportion data, the like volume proportion data, the share volume proportion data, the forwarding volume proportion data, the comment volume proportion data, the collection volume proportion data, the search volume proportion data, etc. of the target multimedia object in the first time period.

[0079] The second user behavior proportion data may represent the user behavior proportion data of the target multimedia object in the second time period. The second user behavior proportion data may include one or more user behavior proportion data of the target multimedia object in the second time period. In a possible implementation, the second user behavior proportion data may include at least one of the play volume proportion data, like volume proportion data, share volume proportion data, forwarding volume proportion data, comment volume proportion data, collection volume proportion data, search volume proportion data, etc. of the target multimedia object in the second time period.

[0080] In an embodiment of the present disclosure, the data types of the first user behavior proportion data and the second user behavior proportion data may be the same. For example, the first user behavior proportion data may include the playback volume proportion data of the target multimedia object in the first time period, and the second user behavior proportion data may include the playback volume proportion data of the target multimedia object set in the second time period. For another example, the first user behavior proportion data may include the playback volume proportion data and the like volume proportion data of the target multimedia object in the first time period, and the second user behavior proportion data may include the playback volume proportion data and the like volume proportion data of the target multimedia object set in the second time period. Of course, in some possible implementations, the data types of the first user behavior proportion data and the second user behavior proportion data may also be different.

[0081] In an embodiment of the present disclosure, the first acceleration may represent the acceleration of the user behavior ratio of the target multimedia object. The first acceleration may be determined according to the difference information between the first user behavior ratio data and the second user behavior ratio data.

[0082] In one example, the first acceleration A(t) may be determined using Formula 1:

[0083] A(t)=v1(t)-v2(t) Formula 1,

[0084] Among them, v1(t) represents the first user behavior proportion data, and v2(t) represents the second user behavior proportion data.

[0085] In another example, the difference between the first user behavior proportion data and the second user behavior proportion data may be calculated, and the ratio of the difference to the third preset time length may be determined as the first acceleration.

[0086] In an embodiment of the present disclosure, the first time period may include one or more first sub-time periods, the first user behavior proportion data may include one or more first sub-user behavior proportion data, and the first sub-user behavior proportion data corresponds to the first sub-time period one-to-one, and the first sub-user behavior proportion data represents the user behavior proportion data of the target multimedia object in the corresponding first sub-time period. Among them, the first sub-time period represents a sub-time period of the first time period, that is, the first sub-time period belongs to the first time period. When the number of first sub-time periods is more than two, any two first sub-time periods do not overlap. When the number of first sub-time periods is more than two, the duration of each first sub-time period may be the same or different. For example, the duration of each first sub-time period is 1 hour, or the duration of each first sub-time period is half an hour, and so on.

[0087] The second time period may include one or more than two second sub-time periods, the second user behavior proportion data may include one or more than two second sub-user behavior proportion data, and the second sub-user behavior proportion data corresponds to the second sub-time period one by one, and the second sub-user behavior proportion data represents the user behavior proportion data of the target multimedia object in the corresponding second sub-time period. Among them, the second sub-time period represents a sub-time period of the second time period, that is, the second sub-time period belongs to the second time period. When the number of second sub-time periods is more than two, any two second sub-time periods do not overlap. When the number of second sub-time periods is more than two, the duration of each second sub-time period may be the same or different. For example, the duration of each second sub-time period is 1 hour, or the duration of each second sub-time period is half an hour, and so on.

[0088] The duration of the first sub-time period may be the same as or different from the duration of the second sub-time period. For example, the duration of the first sub-time period and the duration of the second sub-time period are both 1 hour.

[0089] In a possible implementation, the first time period includes at least two first sub-time periods, the first user behavior proportion data includes at least two first sub-user behavior proportion data corresponding to the at least two first sub-time periods, the second time period includes at least two second sub-time periods, the second user behavior proportion data includes at least two second sub-user behavior proportion data corresponding to the at least two second sub-time periods; obtaining the first user behavior proportion data of the target media object in the first time period, and the second user behavior proportion data of the target multimedia object in the second time period, includes: determining the first user behavior proportion data of the target media object in the first time period based on the at least two first sub-user behavior proportion data; determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data.

[0090] In an example, the first time period includes 6 first sub-time periods, the first user behavior proportion data includes 6 first sub-user behavior proportion data, the second time period includes 6 second sub-time periods, and the second user behavior proportion data includes 6 second sub-user behavior proportion data, wherein the first sub-time period can be recorded as ti, and the first sub-user behavior proportion data can be recorded as v p (ti), the second sub-time period can be recorded as t-11+i, and the second sub-user behavior ratio data can be recorded as v p (t-11+i), 0≤i≤5. In this example, the duration of the first sub-time period and the second sub-time period are both 1 hour.

[0091] In this implementation, the first user behavior proportion data is determined based on at least two first sub-user behavior proportion data corresponding one-to-one to at least two first sub-time periods, the second user behavior proportion data is determined based on at least two second sub-user behavior proportion data corresponding one-to-one to at least two second sub-time periods, and the first acceleration is determined based on the first user behavior proportion data and the second user behavior proportion data determined thereby, thereby being able to determine a smoother first acceleration, thereby being able to achieve a smoother type prediction for the target multimedia object.

[0092] As an example of this implementation method, determining the first user behavior proportion data of the target for the media object in the first time period based on the at least two first sub-user behavior proportion data includes: determining the first user behavior proportion data of the target for the media object in the first time period based on the at least two first sub-user behavior proportion data and at least two first weights corresponding one-to-one to the at least two first sub-user behavior proportion data; determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data includes: determining the second user behavior proportion data of the target for the media object in the second time period based on the at least two second sub-user behavior proportion data and at least two second weights corresponding one-to-one to the at least two second sub-user behavior proportion data.

[0093] In this example, the first weight may represent the weight corresponding to the first sub-user behavior proportion data, and the second weight may represent the weight corresponding to the second sub-user behavior proportion data. p The first weight corresponding to (ti) (i.e., the first weight corresponding to the first sub-time period ti) can be recorded as α t-i , the second sub-user behavior proportion data v p The second weight corresponding to (t-11+i) (i.e., the second weight corresponding to the second sub-time period t-11+i) can be recorded as α t-11+i .

[0094] In this example, the first user behavior proportion data is determined based on at least two first sub-user behavior proportion data and at least two first weights corresponding one-to-one to the at least two first sub-user behavior proportion data, the second user behavior proportion data is determined based on at least two second sub-user behavior proportion data and at least two second weights corresponding one-to-one to the at least two second sub-user behavior proportion data, and the first acceleration is determined based on the first user behavior proportion data and the second user behavior proportion data determined thereby, and the first acceleration determined thereby can more accurately predict the probability that the target multimedia object belongs to a preset type.

[0095] In one example, the first weight corresponding to any one of the at least two first sub-user behavior proportion data is negatively correlated with the interval length between the first sub-time period corresponding to the first sub-user behavior proportion data and the current time, and is the exponential smoothing coefficient corresponding to the first sub-time period. In this example, the first weight corresponding to the first sub-time period is negatively correlated with the interval length between the first sub-time period and the current time, that is, the smaller the interval length between the first sub-time period and the current time, the larger the first weight corresponding to the first sub-time period, and the larger the interval length between the first sub-time period and the current time, the smaller the first weight corresponding to the first sub-time period. In this example, the exponential smoothing coefficient can be recorded as The larger j is, the closer the first weights corresponding to different first sub-time periods are. For example, j=1. By using the first weight in this example to determine the first user behavior proportion data, and determining the first acceleration based on the first user behavior proportion data determined thereby, the first acceleration determined thereby can further improve the accuracy of type prediction for the target multimedia object.

[0096] In one example, the first user behavior proportion data v1(t) may be determined using Formula 2:

[0097]

[0098] In another example, the first weight may not be the exponential smoothing coefficient corresponding to the first sub-time period, as long as it is negatively correlated with the interval between the first sub-time period and the current time.

[0099] In one example, the second weight corresponding to any one of the at least two second sub-user behavior proportion data is positively correlated with the interval between the second sub-time period corresponding to the second sub-user behavior proportion data and the current time. In this example, the second weight corresponding to the second sub-time period is positively correlated with the interval between the second time period and the current time, that is, the smaller the interval between the second time period and the current time, the smaller the second weight corresponding to the second sub-time period, and the larger the interval between the second time period and the current time, the larger the second weight corresponding to the second sub-time period. By using the second weight in this example to determine the second user behavior proportion data, and determining the first acceleration based on the second user behavior proportion data determined thereby, the first acceleration determined thereby can further improve the accuracy of type prediction for the target multimedia object.

[0100] In one example, the second user behavior proportion data v2(t) may be determined using Formula 3:

[0101]

[0102] As another example of this implementation method, determining the first user behavior proportion data of the target for the media object in the first time period based on the at least two first sub-user behavior proportion data includes: determining the average value of the at least two first sub-user behavior proportion data as the first user behavior proportion data of the target for the media object in the first time period; determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data includes: determining the average value of the at least two second sub-user behavior proportion data as the second user behavior proportion data of the target for the media object in the second time period.

[0103] As another example of this implementation method, determining the first user behavior proportion data of the target for the media object in the first time period based on the at least two first sub-user behavior proportion data includes: determining the median of the at least two first sub-user behavior proportion data as the first user behavior proportion data of the target for the media object in the first time period; determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data includes: determining the median of the at least two second sub-user behavior proportion data as the second user behavior proportion data of the target for the media object in the second time period.

[0104] In the disclosed embodiment, the probability that the target multimedia object belongs to the preset type is predicted at least based on the first acceleration. That is, the probability that the target multimedia object belongs to the preset type can be predicted only based on the first acceleration, or the probability that the target multimedia object belongs to the preset type can be predicted in combination with the first acceleration and other information.

[0105] Among them, the preset type can be any type that can represent the future performance of the multimedia object. For example, the preset type can be any type that can represent the future popularity, future playback volume, etc. of the multimedia object. Among them, the number of preset types can be one or more than two. For example, the preset type can include a hit category. For another example, the preset type can include a hit category and a non-hit category. For another example, the preset type can include a hit category S, a hit category A, a category B, and a category C. For another example, the preset type can include a premium category and a non-premium category.

[0106] In one possible implementation, predicting the probability that the target multimedia object belongs to a preset type based on the first acceleration includes: inputting the first acceleration into a pre-trained neural network, and predicting the probability that the target multimedia object belongs to a preset type via the neural network. In this implementation, the neural network may adopt DNN (Deep Neural Networks), for example, a feedforward DNN may be adopted. Of course, the neural network may also adopt neural networks such as MLP (Multi-Layer Perceptron), which is not limited here. In this implementation, by inputting the first acceleration into a pre-trained neural network, and predicting the probability that the target multimedia object belongs to a preset type via the neural network, the accuracy and speed of predicting the probability that the target multimedia object belongs to a preset type can be improved.

[0107] As an example of this implementation, before the first acceleration is input into the pre-trained neural network, the method further includes: obtaining a training object set, wherein the training object set includes a plurality of training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used to indicate whether the training objects belong to the preset type; and using the training object set to train the neural network. In one example, user behavior data within a fourth preset time period in a multimedia platform can be used to train the neural network. For example, the fourth preset time period can be 6 months.

[0108] In this example, the training object set may include multiple training objects, and for each training object, the third user behavior proportion data of the training object in the sixth time period and the fourth user behavior proportion data of the training object in the seventh time period may be obtained respectively, wherein the seventh time period is before the sixth time period. The second acceleration of the user behavior proportion of the training object may be determined based on the third user behavior proportion data and the fourth user behavior proportion data. The second acceleration may be input into the neural network, and the probability that the training object belongs to a preset type may be predicted via the neural network. The neural network may be trained based on the true value of the type of the training object and the probability that the training object belongs to a preset type. Wherein, during the training process of the neural network, the input data of the neural network includes but is not limited to the second acceleration.

[0109] In this example, a training object set is obtained, wherein the training object set includes multiple training objects, the training objects are multimedia objects, and the training objects have annotation data, the annotation data is used to indicate whether the training objects belong to the preset type, and the training object set is used to train the neural network, thereby enabling the neural network to learn the ability to predict the probability that the multimedia object belongs to the preset type.

[0110] After the neural network training is completed, the neural network can be deployed in a real production environment to predict in real time the probability that a multimedia object belongs to a preset type, so as to capture popular multimedia objects in real time.

[0111] Of course, in other possible implementations, the first acceleration may not be processed by a neural network. For example, a pre-designed function may be used to process the first acceleration to obtain the probability that the target multimedia object belongs to a preset type.

[0112] In one example, the probability that the target multimedia object belongs to a preset type can be denoted as y, where y∈[0, 1].

[0113] In a possible implementation, the predicting the probability that the target multimedia object belongs to a preset type according to the first acceleration includes: predicting the probability that the target multimedia object belongs to a preset type according to at least one of first change trend information of the user behavior of the target multimedia object in a third time period, second change trend information of the user behavior of the target multimedia object set in a third time period, first difference information between the user behavior of the target multimedia object and the user behavior of the target multimedia object set, second difference information between the change trend of the user behavior of the target multimedia object and the user behavior of the target multimedia object set, and third difference information between the user behavior data of the target multimedia object in a fourth time period and a fifth time period, and the first acceleration. In this implementation, by combining at least one of the first change trend information, the second change trend information, the first difference information, the second difference information, and the third difference information, the accuracy of predicting the probability that the target multimedia object belongs to a preset type can be further improved.

[0114] The third time period may represent a time period before the current time. The third time period may include the current time or may not include the current time. The time span of the third time period may be a fifth preset duration. For example, the fifth preset duration may be 24 hours, 12 hours, 48 ​​hours, etc.

[0115] The first change trend information may be determined based on first user behavior data of the target multimedia object within the third time period, and the second change trend information may be determined based on second user behavior data of the target multimedia object set within the third time period.

[0116] The first user behavior data may represent user behavior data of the target multimedia object in the third time period. The first user behavior data may include one or more user behavior data. In an example, the first user behavior data may include at least one of the following: play data, like data, share data, forward data, comment data, collection data, etc. of the target multimedia object in the third time period.

[0117] The second user behavior data may represent user behavior data of the target multimedia object set in the third time period. The second user behavior data may include one or more user behavior data. In an example, the second user behavior data may include at least one of play data, like data, share data, forward data, comment data, collection data, etc. of the target multimedia object set in the third time period.

[0118] In this implementation, the data type of the first user behavior data and the second user behavior data may be the same. For example, the first user behavior data may include the playback data of the target multimedia object in the third time period, and the second user behavior data may include the playback data of the target multimedia object set in the third time period. For another example, the first user behavior data may include the playback data and like data of the target multimedia object in the third time period, and the second user behavior data may include the playback data and like data of the target multimedia object set in the third time period. Of course, in some possible implementations, the data types of the first user behavior data and the second user behavior data may also be different.

[0119] In this implementation, the first change trend information may represent the change trend information of the user behavior for the target multimedia object within the third time period. The second change trend information may represent the change trend information of the user behavior for the target multimedia object set within the third time period. The change trend information may be any information that can reflect the change trend. For example, the change trend information may be any information that can reflect an upward trend, a stable trend, a downward trend, an upward speed, a downward speed, etc.

[0120] In one example, the third time period includes a third sub-time period and at least one fourth sub-time period, and the fourth sub-time period is before the third sub-time period; the first user behavior data includes: first sub-user behavior data of the target multimedia object in the third sub-time period, and at least one second sub-user behavior data of the target multimedia object in at least one fourth sub-time period; the second user behavior data includes: third sub-user behavior data of the target multimedia object set in the third sub-time period, and at least one fourth sub-user behavior data of the target multimedia object set in at least one fourth sub-time period.

[0121] The number of fourth sub-time periods is one or more than two. The third sub-time period and at least one fourth sub-time period all belong to the third time period, the third sub-time period does not overlap with any fourth sub-time period in the at least one fourth sub-time period, and when the number of fourth sub-time periods is more than two, any two fourth sub-time periods do not overlap. The duration of the third sub-time period and each fourth sub-time period may be the same or different. For example, the duration of the third sub-time period and each fourth sub-time period is 1 hour, or the duration of the third sub-time period and each fourth sub-time period is half an hour, and so on.

[0122] In one example, the number of fourth sub-time periods is more than two, the third sub-time period and each fourth sub-time period are both 1 hour long, the third sub-time period can be recorded as the Tth hour, and the at least one fourth sub-time period can be recorded as the TDth hour to the T-1th hour, respectively, where D is an integer greater than or equal to 2. For example, D is equal to 24.

[0123] The first sub-user behavior data may represent the user behavior data of the target multimedia object in the third sub-time period, and the second sub-user behavior data may represent the user behavior data of the target multimedia object in the fourth sub-time period. The second sub-user behavior data corresponds to the fourth sub-time period one by one. For example, if D is equal to 24, the first user behavior data includes 1 first sub-user behavior data and 24 second sub-user behavior data. The 24 second sub-user behavior data correspond to the T-24th hour to the T-1st hour one by one, that is, the 24 second sub-user behavior data include: the user behavior data of the target multimedia object in the T-24th hour, the user behavior data of the target multimedia object in the T-23rd hour, ..., the user behavior data of the target multimedia object in the T-1st hour. For example, if the user behavior data includes the playback volume, the 24 second sub-user behavior data may include: the playback volume of the target multimedia object in the T-24th hour, the playback volume of the target multimedia object in the T-23rd hour, ..., the playback volume of the target multimedia object in the T-1st hour.

[0124] The third sub-user behavior data may represent the user behavior data of the target multimedia object set in the third sub-time period, and the fourth sub-user behavior data may represent the user behavior data of the target multimedia object set in the fourth sub-time period. The fourth sub-user behavior data corresponds to the fourth sub-time period one by one. For example, if D is equal to 24, the first user behavior data includes 1 item of the third sub-user behavior data and 24 items of the fourth sub-user behavior data. The 24 items of the fourth sub-user behavior data correspond to the T-24th hour to the T-1st hour one by one, that is, the 24 items of the fourth sub-user behavior data include: the user behavior data of the target multimedia object set in the T-24th hour, the user behavior data of the target multimedia object set in the T-23rd hour, ..., the user behavior data of the target multimedia object set in the T-1st hour. For example, if the user behavior data includes the playback volume, the 24 items of the fourth sub-user behavior data may include: the playback volume of the target multimedia object set in the T-24th hour, the playback volume of the target multimedia object set in the T-23rd hour, ..., the playback volume of the target multimedia object set in the T-1st hour.

[0125] In this example, based on the first sub-user behavior data of the target multimedia object in the third sub-time period and at least one second sub-user behavior data of the target multimedia object in at least one fourth sub-time period, the first change trend information of the user behavior of the target multimedia object in the third time period can be more accurately determined. Based on the third sub-user behavior data of the target multimedia object set in the third sub-time period and at least one fourth sub-user behavior data of the target multimedia object set in at least one fourth sub-time period, the second change trend information of the user behavior of the target multimedia object set in the third time period can be more accurately determined, thereby more accurately predicting the probability that the target multimedia object belongs to a preset type.

[0126] In one example, based on the first user behavior data, determining the first change trend information of the user behavior for the target multimedia object within the third time period includes: determining the fourth difference information between the first sub-user behavior data and at least one item of the second sub-user behavior data; based on the fourth difference information, determining the first change trend information of the user behavior for the target multimedia object within the third time period; based on the second user behavior data, determining the second change trend information of the user behavior for the target multimedia object set within the third time period includes: determining the fifth difference information between the third sub-user behavior data and at least one item of the fourth sub-user behavior data; based on the fifth difference information, determining the second change trend information of the user behavior for the target multimedia object set within the third time period.

[0127] In this example, the fourth difference information may represent the difference information between the first sub-user behavior data and at least one item of the second sub-user behavior data. The fourth difference information may be calculated by calculating the ratio, difference, logarithm and ratio of the first sub-user behavior data and at least one item of the second sub-user behavior data.

[0128] For example, the fourth difference information X1 may be determined using Formula 4:

[0129]

[0130] Where n represents the playback volume of the target multimedia object, n T represents the playback volume of the target multimedia object in hour T, n m Indicates the playback volume of the target multimedia object in the mth hour.

[0131] For another example, the fourth difference information X1 may be determined using Formula 5:

[0132]

[0133] For another example, the fourth difference information X1 may be determined using Formula 6:

[0134]

[0135] After determining the fourth difference information, the fourth difference information may be used as the first change trend information. Alternatively, the fourth difference information may be processed by a preset process to obtain the first change trend information. For example, the fourth difference information may be multiplied by a preset coefficient to obtain the first change trend information.

[0136] In this example, the fifth difference information may represent the difference information between the third sub-user behavior data and at least one item of the fourth sub-user behavior data. The fifth difference information may be calculated by calculating the ratio, difference, logarithm and ratio of the third sub-user behavior data and at least one item of the fourth sub-user behavior data.

[0137] For example, the fifth difference information X2 may be determined using Formula 7:

[0138]

[0139] Where N represents the playback volume of the target multimedia object set, N T represents the playback volume of the target multimedia object set at hour T, N m Indicates the playback volume of the target multimedia object set in the mth hour.

[0140] For another example, the fifth difference information X2 may be determined using Formula 8:

[0141]

[0142] For another example, the fifth difference information X2 may be determined using Formula 9:

[0143]

[0144] After determining the fifth difference information, the fifth difference information may be used as the second change trend information. Alternatively, the fifth difference information may be processed by a preset process to obtain the second change trend information. For example, the fifth difference information may be multiplied by a preset coefficient to obtain the second change trend information.

[0145] In the above example, by determining the fourth difference information between the first sub-user behavior data and at least one item of the second sub-user behavior data, determining the first change trend information of the user behavior for the target multimedia object within the third time period based on the fourth difference information, determining the fifth difference information between the third sub-user behavior data and at least one item of the fourth sub-user behavior data, and determining the second change trend information of the user behavior for the target multimedia object set within the third time period based on the fifth difference information, the first change trend information and the second change trend information can be determined more accurately, thereby more accurately predicting the probability that the target multimedia object belongs to a preset type.

[0146] In this implementation, the second change trend information may be used as a comparison benchmark, and the probability that the target multimedia object belongs to a preset type may be predicted according to the first change trend information and the second change trend information.

[0147] In this implementation, the first difference information may represent the difference between the user behavior for the target multimedia object and the user behavior for the target multimedia object set. The first difference information may be calculated by calculating the ratio, difference, logarithm and ratio of the first user behavior data and the second user behavior data.

[0148] In one example, the first difference information between the user behavior for the target multimedia object and the user behavior for the target multimedia object set can be determined based on the ratio of the first sub-user behavior data in the first user behavior data to the third sub-user behavior data in the second user behavior data. For example, the first difference information X3 can be determined using Formula 10:

[0149]

[0150] Among them, n T represents the playback volume of the target multimedia object in hour T, N T Indicates the playback volume of the target multimedia object set in the Tth hour.

[0151] In another example, when determining the first difference information, in addition to considering the first sub-user behavior data and the third sub-user behavior data, at least one item of the second sub-user behavior data and / or at least one item of the fourth sub-user behavior data may also be considered.

[0152] By determining first difference information between user behavior for a target multimedia object and user behavior for a target multimedia object set based on first user behavior data and second user behavior data, and predicting the probability that the target multimedia object belongs to a preset type in combination with the first difference information, the accuracy of predicting the probability that the target multimedia object belongs to a preset type can be further improved.

[0153] In this implementation, the second difference information may represent the difference between the change trend of the user behavior for the target multimedia object and the user behavior for the target multimedia object set. The second difference information may be calculated by calculating the ratio or difference between the first change trend information and the second change trend information.

[0154] For example, the second difference information X4 may be determined using formula 11:

[0155]

[0156] Among them, X1 represents the first change trend information, and X2 represents the second change trend information.

[0157] By determining second difference information based on first change trend information of user behavior for a target multimedia object and second change trend information of user behavior for a target multimedia object set, and predicting the probability that the target multimedia object belongs to a preset type in combination with the second difference information, the accuracy of predicting the probability that the target multimedia object belongs to a preset type can be further improved.

[0158] In this implementation, the third difference information may represent the difference information between the user behavior data of the target multimedia object in the fourth time period and the fifth time period, wherein the fifth time period is the time period before the fourth time period. The time spans of the fourth time period and the fifth time period may be the same or different. For example, the time spans of the fourth time period and the fifth time period are both 1 hour, or both are half an hour, and so on. In one example, the fourth time period may represent the latest time period, and the fifth time period may represent the previous time period of the fourth time period. For example, the fourth time period may be the tth hour, and the fifth time period may be the t-1th hour.

[0159] As an example of this implementation, the probability that the target multimedia object belongs to a preset type can be predicted based on the first change trend information, the second change trend information, the third difference information and the first acceleration. In one example, the first change trend information, the second change trend information, the third difference information and the first acceleration can be input into a pre-trained neural network, and the probability that the target multimedia object belongs to a preset type can be predicted via the neural network. The first change trend information, the second change trend information, the third difference information and the first acceleration are processed by the pre-trained neural network to obtain the probability that the target multimedia object belongs to the preset type, thereby improving the accuracy and speed of predicting the probability that the target multimedia object belongs to the preset type. In another example, a pre-designed function can be used to process the first change trend information, the second change trend information, the third difference information and the first acceleration to obtain the probability that the target multimedia object belongs to the preset type.

[0160] In one example, before inputting the first change trend information, the second change trend information, the third difference information and the first acceleration into a pre-trained neural network, the method further includes: obtaining a training object set, wherein the training object set includes a plurality of training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used to indicate whether the training objects belong to the preset type; and using the training object set to train the neural network.

[0161] In this example, the training object set may include multiple training objects, and for each training object, user behavior data of the training object from the RD hour to the R hour may be obtained respectively. For example, for each training object, the playback volume of the training object from the RD hour to the R hour may be obtained respectively.

[0162] For any training object in the training object set, the third change trend information of the user behavior from the RD hour to the R hour can be determined based on the user behavior data of the training object at the R hour and the user behavior data from the RD hour to the R-1 hour; the fourth change trend information of the user behavior from the RD hour to the R hour can be determined based on the user behavior data of the training object set at the R hour and the user behavior data from the RD hour to the R-1 hour. For example, the third change trend information of the user behavior from the RD hour to the R hour can be determined based on the playback volume of the training object at the R hour and the playback volume from the RD hour to the R-1 hour; the fourth change trend information of the user behavior from the RD hour to the R hour can be determined based on the playback volume of the training object set at the R hour and the playback volume from the RD hour to the R-1 hour.

[0163] According to the user behavior data of the training subject at the Rth hour and the user behavior data of the R-1th hour, sixth difference information between the user behavior data of the training subject at the Rth hour and the user behavior data of the R-1th hour can be obtained.

[0164] The third user behavior proportion data of the training object in the sixth time period and the fourth user behavior proportion data of the training object in the seventh time period may be obtained, wherein the seventh time period is before the sixth time period. The second acceleration of the user behavior proportion of the training object may be determined according to the third user behavior proportion data and the fourth user behavior proportion data.

[0165] The third change trend information, the fourth change trend information, the sixth difference information and the second acceleration can be input into the neural network, and the probability that the training object belongs to the preset type can be predicted by the neural network. The neural network can be trained according to the true value of the type of the training object and the probability that the training object belongs to the preset type.

[0166] As another example of the implementation, the probability that the target multimedia object belongs to a preset type can be predicted based on the first change trend information, the second change trend information, the first difference information, the second difference information, the third difference information and the first acceleration. In one example, the first change trend information, the second change trend information, the first difference information, the second difference information, the third difference information and the first acceleration can be input into a pre-trained neural network, and the probability that the target multimedia object belongs to a preset type can be predicted via the neural network. The first change trend information, the second change trend information, the first difference information, the second difference information, the third difference information and the first acceleration are processed by the pre-trained neural network to obtain the probability that the target multimedia object belongs to the preset type, thereby improving the accuracy and speed of predicting the probability that the target multimedia object belongs to the preset type. In another example, a pre-designed function can be used to process the first change trend information, the second change trend information, the first difference information, the second difference information, the third difference information and the first acceleration to obtain the probability that the target multimedia object belongs to the preset type.

[0167] In one example, before inputting the first change trend information, the second change trend information, the first difference information, the second difference information, the third difference information and the first acceleration into a pre-trained neural network, the method further includes: obtaining a training object set, wherein the training object set includes a plurality of training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used to indicate whether the training object belongs to the preset type; and training the neural network using the training object set.

[0168] In this example, the training object set may include multiple training objects, and for each training object, user behavior data of the training object from the RD hour to the R hour may be obtained respectively. For example, for each training object, the playback volume of the training object from the RD hour to the R hour may be obtained respectively.

[0169] For any training object in the training object set, the third change trend information of the user behavior from the RD hour to the R hour can be determined based on the user behavior data of the training object at the R hour and the user behavior data from the RD hour to the R-1 hour; the fourth change trend information of the user behavior from the RD hour to the R hour can be determined based on the user behavior data of the training object set at the R hour and the user behavior data from the RD hour to the R-1 hour. For example, the third change trend information of the user behavior from the RD hour to the R hour can be determined based on the playback volume of the training object at the R hour and the playback volume from the RD hour to the R-1 hour; the fourth change trend information of the user behavior from the RD hour to the R hour can be determined based on the playback volume of the training object set at the R hour and the playback volume from the RD hour to the R-1 hour.

[0170] According to the user behavior data of the training object at the Rth hour and the user behavior data from the RD hour to the R-1th hour, and the user behavior data of the training object set at the Rth hour and the user behavior data from the RD hour to the R-1th hour, the seventh difference information between the user behavior for the training object and the user behavior for the training object set can be determined. For example, according to the user behavior data of the training object at the Rth hour and the user behavior data of the training object set at the Rth hour, the seventh difference information between the user behavior for the training object and the user behavior for the training object set can be determined. For example, according to the playback volume of the training object at the Rth hour and the playback volume of the training object set at the Rth hour, the seventh difference information between the user behavior for the training object and the user behavior for the training object set can be determined. For example, the ratio of the playback volume of the training object at the Rth hour to the playback volume of the training object set at the Rth hour can be used as the seventh difference information.

[0171] According to the third change trend information and the fourth change trend information, the eighth difference information between the change trend of the user behavior for the training object and the user behavior for the training object set can be determined. For example, the ratio of the third change trend information to the fourth change trend information can be used as the eighth difference information.

[0172] According to the user behavior data of the training subject at the Rth hour and the user behavior data of the R-1th hour, sixth difference information between the user behavior data of the training subject at the Rth hour and the user behavior data of the R-1th hour can be obtained.

[0173] The third user behavior proportion data of the training object in the sixth time period and the fourth user behavior proportion data of the training object in the seventh time period may be obtained, wherein the seventh time period is before the sixth time period. The second acceleration of the user behavior proportion of the training object may be determined according to the third user behavior proportion data and the fourth user behavior proportion data.

[0174] The third change trend information, the fourth change trend information, the seventh difference information, the eighth difference information, the sixth difference information and the second acceleration can be input into the neural network, and the probability that the training object belongs to the preset type can be predicted by the neural network. The neural network can be trained according to the true value of the type of the training object and the probability that the training object belongs to the preset type.

[0175] In a possible implementation manner, after predicting the probability that the target multimedia object belongs to a preset type, the method further includes: allocating a bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to the preset type.

[0176] As an example of this implementation, the preset type includes class S; the allocating the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to the preset type includes: allocating the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to class S, wherein the bandwidth corresponding to the target multimedia object is positively correlated with the probability that the target multimedia object belongs to class S. That is, the greater the probability that the target multimedia object belongs to class S, the greater the bandwidth corresponding to the target multimedia object; the smaller the probability that the target multimedia object belongs to class S, the smaller the bandwidth corresponding to the target multimedia object.

[0177] As another example of this implementation, the preset type includes class C; the allocating the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to the preset type includes: allocating the bandwidth corresponding to the target multimedia object according to the probability that the target multimedia object belongs to class C, wherein the bandwidth corresponding to the target multimedia object is negatively correlated with the probability that the target multimedia object belongs to class C. That is, the greater 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 class C, the larger the bandwidth corresponding to the target multimedia object.

[0178] As another example of this implementation, the number of preset types 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 the preset type includes: determining the preset type corresponding to the largest probability among the probabilities that the target multimedia object belongs to the preset type as the type to which the target multimedia object belongs; allocating the bandwidth corresponding to the target multimedia object according to the type to which the target multimedia object belongs. For example, the preset types include Class S, Class A, Class B, and Class C; the probability that the target multimedia object belongs to the preset type includes 4 probabilities, namely: 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 among the four probabilities, then it can be determined that the type to which the target multimedia object belongs is the S class; if the probability that the target multimedia object belongs to the A class is the largest among the four probabilities, then it can be determined that the type to which the target multimedia object belongs is the A class; if the probability that the target multimedia object belongs to the B class is the largest among the four probabilities, then it can be determined that the type to which the target multimedia object belongs is the B class; if the probability that the target multimedia object belongs to the C class is the largest among the four probabilities, then it can be determined that the type to which the target multimedia object belongs is the C class. In this example, the bandwidth corresponding to the S class is greater than the bandwidth corresponding to the A class, the bandwidth corresponding to the A class is greater than the bandwidth corresponding to the B class, and the bandwidth corresponding to the B class is greater than the bandwidth corresponding to the C class. For example, if multimedia object V1 belongs to the S class and multimedia object V2 belongs to the A class, then the bandwidth allocated to multimedia object V1 is greater than the bandwidth allocated to multimedia object V2.

[0179] According to this implementation, the multimedia object platform is helped to distribute multimedia objects more efficiently, thereby helping to save bandwidth of the content distribution network.

[0180] The following is an explanation of the prediction method of multimedia objects provided by the embodiment of the present disclosure through a specific application scenario. In this application scenario, the prediction method of the multimedia object can be executed by a server corresponding to the short video platform. In this application scenario, the preset type is a hit video, and the server can respectively predict the probability of each short video on the short video platform that meets the preset conditions being a hit video.

[0181] For any target short video, the server can obtain the playback volume of the target short video from the t-11th hour to the tth hour, and the total playback volume of all short videos on the short video platform from the t-11th hour to the tth hour. According to the playback volume of the target short video from the t-11th hour to the tth hour, and the total playback volume, the playback volume ratio of the target short video from the t-11th hour to the tth hour can be determined. According to the playback volume ratio of the target short video from the t-5th hour to the tth hour, the first playback volume ratio v1(t) of the target short video can be determined by using Formula 2; according to the playback volume ratio of the target short video from the t-11th hour to the t-6th hour, the second playback volume ratio v2(t) of the target short video can be determined by using Formula 3. According to the first playback volume ratio v1(t) and the second playback volume ratio v2(t) of the target short video, the first acceleration of the target short video can be determined by using Formula 1.

[0182] For short video platforms, the above formula 7 can be used to determine X2 based on the number of views of each short video on the short video platform from hour TD to hour T. For the target short video, X1 can be calculated using formula 4, and X4 can be calculated using formula 11. For short videos with X4 greater than 1, X3 can be calculated using formula 10.

[0183] The third difference information may be calculated based on the number of views of the target short video at the t hour and the number of views at the t-1 hour.

[0184] By inputting the first acceleration, X1, X2, X3, X4 and the third difference information into the pre-trained neural network, the probability that the target short video is a hit video can be obtained.

[0185] The server may also obtain a hit video set based on the M short videos with the highest probability of being hit videos, where M is greater than 1, or, may obtain a hit video set based on short videos with a probability of being hit videos greater than or equal to a preset probability.

[0186] In this application scenario, the short video platform can timely discover and capture high-quality short videos from a large number of short videos and allocate a larger bandwidth to them, thereby improving the activity of the short video platform while saving the bandwidth of the content distribution network. In addition, in this application scenario, the probability of each short video on the short video platform being a hit video can be dynamically updated every hour with a high timeliness.

[0187] Figure 2 A comparison diagram is shown between the acceleration determined by the multimedia object prediction method provided by the embodiment of the present disclosure and the acceleration determined according to equation 12 in the related art.

[0188] A′(t)=v(t)-v(t-1) Formula 12,

[0189] exist Figure 2 In , the solid line represents the acceleration curve determined by the prediction method of the multimedia object provided by the embodiment of the present disclosure, and the dotted line represents the acceleration curve determined according to equation 12. Figure 2 It can be seen that, compared with the acceleration determined according to Formula 12, the acceleration determined by the multimedia object prediction method provided by the embodiment of the present disclosure is more stable and can more significantly capture the changes in the proportion of user behaviors of multimedia objects.

[0190] Figure 3 A comparison chart showing the acceleration determined by the multimedia object prediction method provided by the embodiment of the present disclosure and the user behavior proportion data of the multimedia object is shown. Figure 3 In the figure, the solid line represents the curve of acceleration determined by the prediction method of the multimedia object provided by the embodiment of the present disclosure, and the dotted line represents the curve of user behavior proportion data of the same multimedia object. Figure 3 The ordinate on the left is the ordinate corresponding to the solid line. Figure 3 The ordinate on the right is the ordinate corresponding to the dotted line. Figure 3 It can be seen that compared with the user behavior proportion data of the multimedia object, the acceleration determined by the multimedia object prediction method provided by the embodiment of the present disclosure can significantly capture the changes in the user behavior proportion of the multimedia object.

[0191] It can be understood that the above-mentioned various method embodiments mentioned in the present disclosure can be combined with each other to form a combined embodiment without violating the principle logic. Due to space limitations, the present disclosure will not repeat them. It can be understood by those skilled in the art that in the above-mentioned method of the specific implementation method, the specific execution order of each step should be determined according to its function and possible internal logic.

[0192] In addition, the present disclosure also provides a prediction device, an electronic device, a computer-readable storage medium, and a program for multimedia objects, all of which can be used to implement any multimedia object prediction method provided by the present disclosure. The corresponding technical solutions and technical effects can be found in the corresponding records in the method part and will not be repeated here.

[0193] Figure 4 FIG. 1 is a block diagram of a multimedia object prediction device provided by an embodiment of the present disclosure. Figure 4 As shown, the prediction device of the multimedia object includes:

[0194] An acquisition module 41 is used to obtain first user behavior ratio data of a target media object in a first time period, and second user behavior ratio data of the target multimedia object in a second time period, wherein the second time period is before the first time period;

[0195] A determination module 42, configured to determine a first acceleration of a user behavior ratio of the target multimedia object according to the first user behavior ratio data and the second user behavior ratio data;

[0196] The prediction module 43 is configured to predict, according to the first acceleration, the probability that the target multimedia object belongs to a preset type.

[0197] In a possible implementation, the first time period includes at least two first sub-time periods, the first user behavior proportion data includes at least two first sub-user behavior proportion data corresponding one-to-one to the at least two first sub-time periods, the second time period includes at least two second sub-time periods, and the second user behavior proportion data includes at least two second sub-user behavior proportion data corresponding one-to-one to the at least two second sub-time periods;

[0198] The obtaining module 41 is used for:

[0199] Determine the first user behavior proportion data of the target for the media object in the first time period according to the at least two first sub-user behavior proportion data;

[0200] The second user behavior proportion data of the target multimedia object in the second time period is determined according to the at least two items of second sub-user behavior proportion data.

[0201] In a possible implementation, the obtaining module 41 is used to

[0202] Determine the first user behavior proportion data of the target for the media object in the first time period according to the at least two first sub-user behavior proportion data and at least two first weights corresponding to the at least two first sub-user behavior proportion data one by one;

[0203] The second user behavior proportion data of the target for the media object in the second time period is determined according to the at least two second sub-user behavior proportion data and at least two second weights corresponding to the at least two second sub-user behavior proportion data.

[0204] In a possible implementation, the first weight corresponding to any one of the at least two first sub-user behavior proportion data is negatively correlated with the interval between the first sub-time period corresponding to the first sub-user behavior proportion data and the current time, and is the exponential smoothing coefficient corresponding to the first sub-time period.

[0205] In a possible implementation, the second weight corresponding to any one of the at least two second sub-user behavior proportion data is positively correlated with the interval between the second sub-time period corresponding to the second sub-user behavior proportion data and the current time.

[0206] In a possible implementation, the prediction module 43 is used to:

[0207] The first acceleration is input into a pre-trained neural network, and the probability that the target multimedia object belongs to a preset type is predicted by the neural network.

[0208] In a possible implementation manner, the device further includes:

[0209] An acquisition module, used for acquiring a training object set, wherein the training object set includes a plurality of training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used for indicating whether the training objects belong to the preset type;

[0210] A training module is used to train the neural network using the training object set.

[0211] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. Its specific implementation and technical effects can refer to the description of the above method embodiments. For the sake of brevity, they will not be repeated here.

[0212] The present disclosure also provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions implement the above method when executed by a processor. The computer-readable storage medium may be a non-volatile computer-readable storage medium, or may be a volatile computer-readable storage medium.

[0213] The embodiment of the present disclosure further provides a computer program, including a computer-readable code. When the computer-readable code is executed in an electronic device, a processor in the electronic device executes the above method.

[0214] The embodiments of the present disclosure also provide a computer program product, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in an electronic device, a processor in the electronic device executes the above method.

[0215] An embodiment of the present disclosure also provides an electronic device, comprising: one or more processors; and a memory for storing executable instructions; wherein the one or more processors are configured to call the executable instructions stored in the memory to execute the above method.

[0216] The electronic device may be provided as a terminal, a server, or a device in other forms.

[0217] Figure 5 FIG. 1 is a block diagram of another electronic device 1900 provided by an embodiment of the present disclosure. Figure 5 , the electronic device 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932 for storing instructions executable by the processing component 1922, such as an application. The application stored in the memory 1932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above method.

[0218] The electronic device 1900 may also include a power supply 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 stored in the memory 1932, such as a Microsoft Server operating system (Windows Server 2003). TM ), a graphical user interface operating system launched by Apple (Mac OSX TM ), a multi-user, multi-process computer operating system (Unix TM ), a free and open source Unix-like operating system (Linux TM ), an open source Unix-like operating system (FreeBSD TM ) or similar.

[0219] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 1932 including computer program instructions, which can be executed by the processing component 1922 of the electronic device 1900 to perform the above method.

[0220] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0221] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, 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 disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0222] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded 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 can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0223] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0224] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0225] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0226] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0227] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the specified logical function. In some alternative implementations, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or action, or can be implemented with a combination of special hardware and computer instructions.

[0228] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is embodied as a computer storage medium, and in another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK) and the like.

[0229] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.

[0230] If the technical solution of the embodiments of the present disclosure involves personal information, the product using the technical solution of the embodiments of the present disclosure has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the technical solution of the embodiments of the present disclosure involves sensitive personal information, the product using the technical solution of the embodiments of the present disclosure has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the requirement of "explicit consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to collect his or her personal information; or on the device for processing personal information, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

[0231] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for predicting a multimedia object, characterized in that: include: Obtaining first user behavior ratio data of a target multimedia object in a first time period, and second user behavior ratio data of the target multimedia object in a second time period, wherein the second time period is before the first time period; the user behavior ratio data of the target multimedia object represents the ratio data of the user behavior data of the target multimedia object to the user behavior data of the same type of a target multimedia object set; the target multimedia object set represents a multimedia object set used to predict the probability that the target multimedia object belongs to a preset type; the first user behavior ratio data represents the user behavior ratio data of the target multimedia object in the first time period, and the second user behavior ratio data represents the user behavior ratio data of the target multimedia object in the second time period; Determining a first acceleration of a user behavior ratio of the target multimedia object according to the first user behavior ratio data and the second user behavior ratio data; The probability that the target multimedia object belongs to a preset type is predicted according to the first acceleration.

2. The method according to claim 1, characterized in that The first time period includes at least two first sub-time periods, the first user behavior proportion data includes at least two first sub-user behavior proportion data corresponding to the at least two first sub-time periods one by one, the second time period includes at least two second sub-time periods, and the second user behavior proportion data includes at least two second sub-user behavior proportion data corresponding to the at least two second sub-time periods one by one; The obtaining of first user behavior proportion data of the target multimedia object in a first time period, and second user behavior proportion data of the target multimedia object in a second time period, includes: Determine the first user behavior proportion data of the target multimedia object in the first time period according to the at least two first sub-user behavior proportion data; According to the at least two items of second sub-user behavior proportion data, second user behavior proportion data of the target multimedia object in the second time period is determined.

3. The method according to claim 2, characterized in that The determining, according to the at least two first sub-user behavior proportion data, the first user behavior proportion data of the target multimedia object in the first time period comprises: determining, according to the at least two first sub-user behavior proportion data and at least two first weights corresponding to the at least two first sub-user behavior proportion data, the first user behavior proportion data of the target multimedia object in the first time period; Determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data includes: determining the second user behavior proportion data of the target multimedia object in the second time period based on the at least two second sub-user behavior proportion data and at least two second weights corresponding one-to-one to the at least two second sub-user behavior proportion data.

4. The method according to claim 3, characterized in that: The first weight corresponding to any one of the at least two first sub-user behavior proportion data is negatively correlated with the interval between the first sub-time period corresponding to the first sub-user behavior proportion data and the current time, and is the exponential smoothing coefficient corresponding to the first sub-time period.

5. The method according to claim 3 or 4, characterized in that: The second weight corresponding to any one of the at least two second sub-user behavior proportion data is positively correlated with the interval between the second sub-time period corresponding to the second sub-user behavior proportion data and the current time.

6. The method according to any one of claims 1 to 4, characterized in that The predicting, based on the first acceleration, the probability that the target multimedia object belongs to a preset type includes: The first acceleration is input into a pre-trained neural network, and the probability that the target multimedia object belongs to a preset type is predicted by the neural network.

7. The method according to claim 6, characterized in that Before inputting the first acceleration into the pre-trained neural network, the method further includes: Acquire a training object set, wherein the training object set includes a plurality of training objects, the training objects are multimedia objects, and the training objects have annotation data, and the annotation data is used to indicate whether the training objects belong to the preset type; The neural network is trained using the training object set.

8. A prediction device for a multimedia object, characterized in that: include: An acquisition module is used to obtain first user behavior ratio data of a target multimedia object in a first time period, and second user behavior ratio data of the target multimedia object in a second time period, wherein the second time period is before the first time period; the user behavior ratio data of the target multimedia object represents the ratio data of the user behavior data of the target multimedia object to the user behavior data of the same type of a target multimedia object set; the target multimedia object set represents a multimedia object set used to predict the probability that the target multimedia object belongs to a preset type; the first user behavior ratio data represents the user behavior ratio data of the target multimedia object in the first time period, and the second user behavior ratio data represents the user behavior ratio data of the target multimedia object in the second time period; a determination module, configured to determine a first acceleration of a user behavior ratio of the target multimedia object according to the first user behavior ratio data and the second user behavior ratio data; A prediction module is used to predict the probability that the target multimedia object belongs to a preset type according to the first acceleration.

9. An electronic device, characterized in that: include: one or more processors; a memory for storing executable instructions; The one or more processors are configured to call the executable instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that The invention comprises a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code, wherein when the computer-readable code is executed in an electronic device, a processor in the electronic device executes the method according to any one of claims 1 to 7.