CN109996122A - A kind of video recommendation method, device, server and storage medium - Google Patents
A kind of video recommendation method, device, server and storage medium Download PDFInfo
- Publication number
- CN109996122A CN109996122A CN201910295961.2A CN201910295961A CN109996122A CN 109996122 A CN109996122 A CN 109996122A CN 201910295961 A CN201910295961 A CN 201910295961A CN 109996122 A CN109996122 A CN 109996122A
- Authority
- CN
- China
- Prior art keywords
- video
- sequence
- client
- subsequence
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
- H04N21/251—Learning process for intelligent management, e.g. learning user preferences for recommending movies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
- H04N21/262—Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists
- H04N21/26258—Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists for generating a list of items to be played back in a given order, e.g. playlist, or scheduling item distribution according to such list
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
- H04N21/4667—Processing of monitored end-user data, e.g. trend analysis based on the log file of viewer selections
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
- H04N21/4668—Learning process for intelligent management, e.g. learning user preferences for recommending movies for recommending content, e.g. movies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/47—End-user applications
- H04N21/482—End-user interface for program selection
- H04N21/4826—End-user interface for program selection using recommendation lists, e.g. of programs or channels sorted out according to their score
Abstract
The embodiment of the invention provides a kind of video recommendation method, device, server and storage mediums, wherein, video recommendation method includes: to obtain the first video sequence, and first each video in video sequence feature, wherein, first video sequence are as follows: the sequence lined up by each first video interested of the client of video to be recommended by playing sequence, the first video interested are as follows: the client of video to be recommended played the interested video of user in video；The feature input of each video in first video sequence and the first video sequence prediction model trained in advance is subjected to video estimation, obtains the second video sequence；Recommend video to the client of video to be recommended according to the sequence of each video in the second video sequence.Video is recommended to client using technical solution provided in an embodiment of the present invention, the quantity of the video of user's viewing can be increased, to increase the duration that user uses Video Applications.
Description
Technical field
The present invention relates to Internet technical fields, more particularly to a kind of video recommendation method, device, server and storage
Medium.
Background technique
As mobile terminal is more more and more universal, all kinds of Video Applications become more and more popular in the terminal.Video Applications
Server can recommend a series of video to the client of Video Applications, after user opens Video Applications in the client,
Each video of server recommendation can successively be watched.For example, user can pass through after being mounted with short Video Applications in terminal
Each short-sighted frequency that the server of short Video Applications is recommended to the client of short Video Applications is successively watched in the operations such as sliding.
In the prior art, server is usually to method used by client recommendation video: server obtains client
Viewing record, recorded according to the viewing determine the corresponding user of the client to the degree of liking of different types of video, then
According to user to different types of video like degree to each video in server by user like degree from height to
The each video to have sorted is successively recommended client by low sequence, allow client according to server recommendation order according to
It is secondary to play each video.
However, inventor has found in the implementation of the present invention, at least there are the following problems for the prior art: due to service
Device be to each video by user like each video to have sorted is successively recommended into client after degree sorts from high to low
End, since this video sortord is more single, the video of position is substantially the video that user does not like after coming relatively,
After user, which has watched, comes the video compared with front position, Video Applications will be usually exited, reduce the video of user's viewing
Quantity, to reduce the duration that user uses Video Applications.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of video recommendation method, device, server and storage medium, to increase
The quantity for the video for adding user to watch, to increase the duration that user uses Video Applications.Specific technical solution is as follows:
In a first aspect, the embodiment of the invention provides a kind of video recommendation methods, which comprises
Obtain the feature of each video in the first video sequence and first video sequence, wherein first view
Frequency sequence are as follows: the sequence lined up by each first video interested of the client of video to be recommended by playing sequence, described first
Video interested are as follows: the client of the video to be recommended played the interested video of user in video；
By the feature input training in advance of each video in first video sequence and first video sequence
Prediction model carries out video estimation, obtains the second video sequence；Wherein, the prediction model is: according to Sample video sequence, institute
State the model that the feature of video in Sample video sequence is trained, wherein the video in the Sample video sequence are as follows:
Each client of server providing services played be arranged successively in video by playing sequence, the interested video of user；
Recommend video to the client of the video to be recommended according to the sequence of each video in second video sequence.
Optionally, the first video sequence of the acquisition, comprising:
Obtain the video playing record of the client of video to be recommended, wherein include in the video playing record: described
The playing sequence of each played video of the client of video to be recommended and the mark information of each played video, it is each
The mark information of played video is: the client of the video to be recommended played the operation of video progress according to user to this
It played the information of video addition to this；
According to the mark information of each played video, select user interested from each played video
Video, as the first video interested；
The described first video interested is ranked up by the playing sequence, obtains the first video sequence.
Optionally, training obtains the prediction model in the following way:
For each client of the server providing services, obtaining the client be played in video by playing sequence
The user being arranged successively video interested forms Sample video sequence by playing sequence；
Obtain the feature of each video in the Sample video sequence；
By in the first subsequence of the Sample video sequence and first subsequence each video feature input to
Training pattern carries out video estimation, obtains output sequence；Wherein, first subsequence is each height of Sample video sequence
Any subsequence in sequence；
It is that training benchmark adjusts institute according to the difference between the output sequence and the second subsequence with the second subsequence
State the model parameter to training pattern, realize to the training to training pattern, and using after training to training pattern as
Prediction model；Wherein, second subsequence are as follows: in the subsequence of the Sample video sequence, be located at first subsequence
Afterwards and the sequence adjacent with first subsequence.
Optionally, the mark information be the scoring of user, played duration ratio, whether by user's collection, whether
By at least one of user comment；Wherein, the duration ratio be video be played duration and the video total duration it
Than.
Optionally, the feature of the video, comprising: the total duration of video, the type of video, video suitable crowd in
It is at least one.
Second aspect, the embodiment of the present invention also provide a kind of all kinds of video recommendations devices, and described device includes:
Sequence obtaining unit, for obtaining the spy of each video in the first video sequence and first video sequence
Sign, wherein first video sequence are as follows: arranged by each first video interested of the client of video to be recommended by playing sequence
At sequence, first video interested are as follows: it is interested that the client of the video to be recommended played user in video
Video；
Sequence calculation sequence, for by the spy of each video in first video sequence and first video sequence
Sign input prediction model trained in advance carries out video estimation, obtains the second video sequence；Wherein, the prediction model is: root
The model being trained according to the feature of video in Sample video sequence, the Sample video sequence, wherein the sample view
Video in frequency sequence are as follows: being arranged successively in the played video of each client of server providing services by playing sequence,
The interested video of user；
Video recommendations unit, for the sequence according to each video in second video sequence to the video to be recommended
Client recommend video.
Optionally, the sequence obtaining unit, comprising:
Record obtains subelement, the video playing record of the client for obtaining video to be recommended, wherein the video
It plays in record and includes: the playing sequence of each played video of the client of the video to be recommended and each played
The mark information of the mark information of video, each played video is: the client of the video to be recommended is according to user to this
The operation that played video carries out played the information of video addition to this；
Video selection subelement, for the mark information according to each played video, from each described played
The interested video of user is selected in video, as the first video interested；
Video sorting subunit obtains for being ranked up by the playing sequence to the described first video interested
One video sequence.
Optionally, described device further include:
Training sequence obtaining unit obtains the client for being directed to each client of the server providing services
The user's video interested being arranged successively in played video by playing sequence forms Sample video sequence by playing sequence；
Training characteristics obtaining unit, for obtaining the feature of each video in the Sample video sequence；
Training sequence output unit, for by the first subsequence of the Sample video sequence and the first sub- sequence
The feature input of each video carries out video estimation to training pattern in column, obtains output sequence；Wherein, first subsequence is
Any subsequence in each subsequence of Sample video sequence；
Model parameter adjustment unit, for being training benchmark with the second subsequence, according to the output sequence and the second son
Difference between sequence, the adjustment model parameter to training pattern, is realized to the training to training pattern, and will instruction
After white silk to training pattern as prediction model；Wherein, second subsequence are as follows: the subsequence of the Sample video sequence
In, after first subsequence and the sequence adjacent with first subsequence.
Optionally, the mark information be the scoring of user, played duration ratio, whether by user's collection, whether
By at least one of user comment；Wherein, the duration ratio be video be played duration and the video total duration it
Than.
Optionally, the feature of the video, comprising: the total duration of video, the type of video, video suitable crowd in
It is at least one.
The third aspect, the embodiment of the invention provides a kind of server, including processor, communication interface, memory and logical
Believe bus, wherein the processor, the communication interface, the memory complete mutual communication by the bus；Institute
Memory is stated, for storing computer program；The processor is realized for executing the program stored on the memory
The described in any item method and steps of first aspect.
Fourth aspect, the embodiment of the invention provides a kind of computer readable storage medium, storage in the storage medium
There is computer program, first aspect described in any item steps are realized when the computer program is executed by processor.
5th aspect, the embodiment of the invention provides a kind of computer program products comprising instruction, when it is in computer
When upper operation, so that computer executes the described in any item video recommendation methods of first aspect.
In video recommendations scheme provided in an embodiment of the present invention, due to the client that the first video interested is video to be recommended
The interested video of user in played video is held, server obtains the first view that the first video interested is lined up by playing sequence
To get the sequence for seeing video that user prefers has been arrived after the feature of frequency sequence and the first video interested, by the first video
The feature input of each video prediction model trained in advance carries out obtained by video estimation in sequence and the first video sequence
The second video sequence, the sequence for seeing video also preferred for user, so as to by this user of the second video sequence compared with
The sequence for seeing video liked recommends each video to the client of video to be recommended, when what user was liked by oneself sees video
When sequence viewing video, the quantity of the video of user's viewing can be increased, to increase the duration that user uses Video Applications.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described.
Fig. 1 is a kind of flow diagram of video recommendation method provided in an embodiment of the present invention；
Fig. 2 is a kind of flow diagram that the method for the first video sequence is obtained in the embodiment of the present invention；
Fig. 3 is a kind of flow diagram of the method for training prediction model in the embodiment of the present invention；
Fig. 4 is a kind of structural schematic diagram of video recommendations device provided in an embodiment of the present invention；
Fig. 5 is a kind of structural schematic diagram of server provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description.
In order to increase the quantity for the video that user watches, to increase the duration that user uses Video Applications, the present invention is real
It applies example and provides a kind of video recommendation method, device, server and storage medium.
Video recommendation method is provided for the embodiments of the invention first below to be introduced.
It should be noted that video recommendation method provided by the embodiment of the present invention can be applied in Video Applications.
The executing subject of video recommendation method provided in an embodiment of the present invention can be server, as shown in Figure 1, this method
The following steps are included:
S110: the feature of each video in the first video sequence and the first video sequence is obtained, wherein the first video
Sequence are as follows: the sequence lined up by each first video interested of the client of video to be recommended by playing sequence, first is interested
Video are as follows: the client of video to be recommended played the interested video of user in video.
The client of above-mentioned video to be recommended, which played the interested video of user in video, may is that video to be recommended
Client played the video that user's scoring in video is higher than default scoring；Wherein, which can comment according to video
The full marks divided determine that the embodiment of the present invention does not limit specifically.For example, presetting scoring when the full marks to score video are 100 timesharing
It can be any score value in 60~90 points.When scoring of the user to a certain video is higher, illustrate that user more feels the video
Interest, user may prefer to this video.
The client of above-mentioned video to be recommended played the interested video of user in video, be also possible to: view to be recommended
The client of frequency played the video collected in video by user；Alternatively, the client of video to be recommended played quilt in video
The duration ratio of broadcasting is not less than the video of preset ratio, wherein duration ratio refers to the played duration of video and the video
The ratio between total duration, above-mentioned preset ratio for example can be any proportion not less than 50%, and the embodiment of the present invention does not limit specifically.
When user has collected the duration large percentage that a certain video or the video are played, it may also be said to which bright user is to the video
Interested, user may prefer to the video.
In embodiments of the present invention, it is emerging that the client that server can obtain video to be recommended played user's sense in video
The video of interest, lines up the first video sequence by playing sequence as the first video interested, and by each first video interested.
The feature of above-mentioned first video interested can be the type, total duration, suitable crowd of the first video interested
At least one of, it can also be that other features of video, the embodiment of the present invention do not limit specifically.The type of above-mentioned video is for example
It can be the types such as movement, plot, suspense, animation, ancient costume.
S120: by the feature input of each video in the first video sequence and the first video sequence prediction trained in advance
Model carries out video estimation, obtains the second video sequence；Wherein, above-mentioned prediction model is: being regarded according to Sample video sequence, sample
The model for being used to predict video that the feature of video is trained in frequency sequence, wherein the video in Sample video sequence
Are as follows: each client of server providing services played be arranged successively in video by playing sequence, the interested view of user
Frequently.
S130: recommend video to the client of video to be recommended according to the sequence of each video in the second video sequence.
Video recommendation method provided in an embodiment of the present invention, due to the client that the first video interested is video to be recommended
The interested video of user in played video, server obtain the first video that the first video interested is lined up by playing sequence
To get the sequence for seeing video that user prefers has been arrived after sequence and the feature of the first video interested, by the first video sequence
It is obtained to carry out video estimation for the feature input of each video prediction model trained in advance in column and the first video sequence
Second video sequence, the sequence for seeing video also preferred for user, so as to relatively be liked by this user of the second video sequence
The joyous sequence for seeing video recommends each video to the client of video to be recommended, when what user was liked by oneself sees the suitable of video
When sequence watches video, the quantity of the video of user's viewing can be increased, to increase the duration that user uses Video Applications.
It in one embodiment, can be according to the following steps as shown in Fig. 2, the first video sequence in step S110
S111~S113 is obtained:
S111: the video playing record of the client of video to be recommended is obtained, wherein wrap in above-mentioned video playing record
Contain: the playing sequence of each played video of the client of video to be recommended and the mark information of each played video,
The mark information of each played video is: the client of video to be recommended played the operation of video progress according to user to this
It played the information of video addition to this.
User may carry out some operations to the video of viewing, for example, skipping view in playing process when watching video
Frequently, the operations such as video, comment video are collected, whether user, which likes this, usually may determine that these operations of video from user
Video.After user operates video, client can add corresponding mark information to the video.According to user to view
The operation that frequency carries out, mark information can be the scoring of user, played duration ratio, whether by user's collection, whether by
At least one of user comment；Wherein, the duration ratio is the ratio between the total duration of duration and the video that video is played.
In a specific embodiment, the client of the available video to be recommended of server is default away from current time
Video playing record in period, above-mentioned preset time period can be not more than 30 days any time periods, such as above-mentioned
Preset time period can be ten days, periods, the embodiment of the present invention such as a week, five days do not limit specifically.When server obtains
Take be video to be recommended client away from current time preset time period video playing record when, can make predict mould
The second video sequence that type predicts more meets the sequence habit of the recent viewing video of user, further increases user's viewing
Video quantity and user use Video Applications duration.
S112: according to the mark information of each played video, select user interested from each played video
Video, as the first video interested.
The above-mentioned interested video of user, can be following at least one: user's scoring is higher than video, the quilt of default scoring
The video of user's collection, played duration ratio are not less than the video of preset ratio, the video crossed by user comment；Wherein,
Above-mentioned preset ratio for example can be any proportion not less than 50%, and the embodiment of the present invention does not limit specifically.
S113: the above-mentioned first video interested is ranked up by above-mentioned playing sequence, obtains the first video sequence.
Present embodiment determines the first video interested according to the mark information of each played video, since label is believed
The breath information according to added by the operation that user carries out each played video that is client, can make to determine in this way the
One video interested more meets the actual preferences of user, and the second video sequence for predicting prediction model more meets the reality of user
Border hobby, the quantity and user that further increase the video of user's viewing use the duration of Video Applications.
In one embodiment, as shown in figure 3, above-mentioned prediction can be obtained with the training of S150~S180 according to the following steps
Model:
S150: it for each client of server providing services, obtains suitable by playing in the played video of the client
The user that sequence is arranged successively video interested forms Sample video sequence by playing sequence.
In a specific embodiment, above-mentioned each client for server providing services, obtains the client
The user's video interested being arranged successively in played video by playing sequence, can realize by the following method: mention from server
For selecting preset quantity client to obtain for each client in selected client in each client of service
Obtaining the client played the user's video interested being arranged successively in video by playing sequence.Due to server providing services
Client terminal quantity is more, and preset quantity client is selected to carry out training pattern, it is possible to reduce the workload during model training,
Improve the training speed and efficiency of model.Specifically, server can be random from each client of server providing services
Select preset quantity client.
S160: the feature of each video in Sample video sequence is obtained.
S170: the feature of each video in the first subsequence and the first subsequence of Sample video sequence is inputted wait instruct
Practice model and carry out video estimation, obtain output sequence, wherein the first subsequence is appointing in each subsequence of Sample video sequence
One subsequence.
In a specific embodiment, the first subsequence can be with are as follows: in each subsequence of Sample video sequence, including
The subsequence of two videos.It is above-mentioned to training pattern can for shot and long term memory network (LongShortTerm Memory, referred to as
LSTM) model or production fight network (Generative AdversarialNetworks, abbreviation GAN) model, can also be with
It is other models for prediction.Wherein, LSTM model is more suitable for solving long sequence Dependence Problem, and LSTM model is for length
The predictablity rate that sequence relies on is higher, therefore, trains and make the second video predicted for the ease of treat training pattern
Sequence more meets the sequence habit that user watches video, and above-mentioned to training pattern can be LSTM model.
S180: being training benchmark with the second subsequence, according to the difference between output sequence and the second subsequence, adjustment to
The model parameter of training pattern realizes and treats the training of training pattern, and using after training to training pattern as prediction model；
Wherein, the second subsequence are as follows: in the subsequence of Sample video sequence, be located at after the first subsequence and adjacent with the first subsequence
Sequence.
In one embodiment, in each subsequence that the first subsequence is Sample video sequence, including two views
When the subsequence of frequency, the second subsequence can in Sample video sequence, it is after the first subsequence and with the first subsequence
An adjacent video.
In the present embodiment, when being had differences between the output sequence exported when training pattern and the second subsequence,
Illustrate that the model parameter to training pattern will lead to prediction result and the second subsequence is inconsistent, it can be according to output sequence and
Difference between two subsequences adjusts the model parameter to training pattern, so that prediction result is drawn close to the second subsequence.When to
When difference between the output sequence and the second subsequence of training pattern output is smaller, illustrate with the prediction obtained to training pattern
As a result also smaller with the second subsequence difference, at this point it is possible to using after training to training pattern as prediction model.
In one particular embodiment of the present invention, if to training pattern be LSTM model, acquisition it is interested by user
Video presses the Sample video sequence that playing sequence is formed are as follows: video 1- video 2- video 4- video 6- video 7- video 8- video 9,
First subsequence are as follows: video 1- video 2- video 4, by the first subsequence (video 1- video 2- video 4) and the spy of video 1
After sign, the feature of video 2, the feature of video 4 input LSTM model, obtained output sequence are as follows: video 7- video 6- video 9,
If the second subsequence are as follows: video 6- video 7- video 8- video 9, then can be according to output sequence (video 7- video 6- view
Frequently 9) the difference between this sequence, with the second subsequence (video 6- video 7- video 8- video 9) this sequence adjusts LSTM
The parameter of model realizes the training to LSTM model, and using the LSTM model after training as prediction model.Wherein, for
The process of LSTM model, adjusting parameter is self-regulating process, and the embodiment of the present invention repeats no more the process of adjusting parameter.
In one embodiment, before step S110, above-mentioned video recommendation method can with the following steps are included:
S190: judging that client is recorded with the presence or absence of video playing, if it has not, S1100 is executed, if it is, executing S110；
S1100: recommend each video to client according to preset video recommendations sequence.
Above-mentioned preset video recommendations sequence can be the video recommendations sequence of Server Default.When there is no views for client
When frequency plays record, recommends each video to client according to preset video recommendations sequence, make client that video may be implemented
Broadcasting.
Corresponding with above-mentioned video recommendation method, the embodiment of the invention also provides a kind of video recommendations devices, such as Fig. 4 institute
Show, which includes:
Sequence obtaining unit 210, for obtaining each video in the first video sequence and first video sequence
Feature, wherein first video sequence are as follows: playing sequence is pressed by each first video interested of the client of video to be recommended
The sequence lined up, first video interested are as follows: it is interested that the client of the video to be recommended played user in video
Video；
Sequence calculation sequence 220 is used for each video in first video sequence and first video sequence
The trained in advance prediction model of feature input carry out video estimation, obtain the second video sequence；Wherein, the prediction model
It is: the model being trained according to the feature of video in Sample video sequence, the Sample video sequence, wherein described
Video in Sample video sequence are as follows: each client of server providing services, which played, is successively arranged in video by playing sequence
Column, the interested video of user；
Video recommendations unit 230, for the sequence according to each video in second video sequence to described to be recommended
The client of video recommends video.
Video recommendations device provided in an embodiment of the present invention, due to the client that the first video interested is video to be recommended
The interested video of user in played video, server obtain the first video that the first video interested is lined up by playing sequence
To get the sequence for seeing video that user prefers has been arrived after sequence and the feature of the first video interested, by the first video sequence
It is obtained to carry out video estimation for the feature input of each video prediction model trained in advance in column and the first video sequence
Second video sequence, the sequence for seeing video also preferred for user, so as to relatively be liked by this user of the second video sequence
The joyous sequence for seeing video recommends each video to the client of video to be recommended, when what user was liked by oneself sees the suitable of video
When sequence watches video, the quantity of the video of user's viewing can be increased, to increase the duration that user uses Video Applications.
In one embodiment, the sequence obtaining unit 210 may include:
Record obtains subelement, the video playing record of the client for obtaining video to be recommended, wherein the video
It plays in record and includes: the playing sequence of each played video of the client of the video to be recommended and each played
The mark information of the mark information of video, each played video is: the client of the video to be recommended is according to user to this
The operation that played video carries out played the information of video addition to this；
Video selection subelement, for the mark information according to each played video, from each described played
The interested video of user is selected in video, as the first video interested；
Video sorting subunit obtains for being ranked up by the playing sequence to the described first video interested
One video sequence.
In one embodiment, described device can also include:
Training sequence obtaining unit obtains the client for being directed to each client of the server providing services
The user's video interested being arranged successively in played video by playing sequence forms Sample video sequence by playing sequence；
Training characteristics obtaining unit, for obtaining the feature of each video in the Sample video sequence；
Training sequence output unit, for by the first subsequence of the Sample video sequence and the first sub- sequence
The feature input of each video carries out video estimation to training pattern in column, obtains output sequence；Wherein, first subsequence is
Any subsequence in each subsequence of Sample video sequence；
Model parameter adjustment unit, for being training benchmark with the second subsequence, according to the output sequence and the second son
Difference between sequence, the adjustment model parameter to training pattern, is realized to the training to training pattern, and will instruction
After white silk to training pattern as prediction model；Wherein, second subsequence are as follows: the subsequence of the Sample video sequence
In, after first subsequence and the sequence adjacent with first subsequence.
In one embodiment, the mark information can for the scoring of user, played duration ratio, whether by
Whether user collects, by least one of user comment；Wherein, the duration ratio is the duration and the view that video is played
The ratio between total duration of frequency.
In one embodiment, the feature of the video may include: total duration, the type of video, video of video
At least one of suitable crowd.
The embodiment of the invention also provides a kind of servers, as shown in figure 5, including processor 301, communication interface 302, depositing
Reservoir 303 and communication bus 304, wherein processor 301, communication interface 302, memory 303 are completed by communication bus 304
Mutual communication,
Memory 303, for storing computer program；
Processor 301 when for executing the program stored on memory 303, realizes following steps:
Obtain the feature of each video in the first video sequence and first video sequence, wherein first view
Frequency sequence are as follows: the sequence lined up by each first video interested of the client of video to be recommended by playing sequence, described first
Video interested are as follows: the client of the video to be recommended played the interested video of user in video；
By the feature input training in advance of each video in first video sequence and first video sequence
Prediction model carries out video estimation, obtains the second video sequence；Wherein, the prediction model is: according to Sample video sequence, institute
State the model that the feature of video in Sample video sequence is trained, wherein the video in the Sample video sequence are as follows:
Each client of server providing services played be arranged successively in video by playing sequence, the interested video of user；
Recommend video to the client of the video to be recommended according to the sequence of each video in second video sequence.
Server provided in an embodiment of the present invention, the second obtained video sequence see the suitable of video for what user preferred
Sequence, the sequence for seeing video so as to prefer by the second video sequence this user are recommended to the client of video to be recommended
Each video can increase the number of the video of user's viewing when the sequence for seeing video that user is liked by oneself watches video
Amount, to increase the duration that user uses Video Applications.
The communication bus that above-mentioned server is mentioned can be Peripheral Component Interconnect standard (Peripheral Component
Interconnect, PCI) bus or expanding the industrial standard structure (Extended Industry Standard
Architecture, EISA) bus etc..The communication bus can be divided into address bus, data/address bus, control bus etc..For just
It is only indicated with a thick line in expression, figure, it is not intended that an only bus or a type of bus.Communication interface is used
Communication between above-mentioned electronic equipment and other equipment.
Memory may include random access memory (Random Access Memory, RAM), also may include non-easy
The property lost memory (Non-Volatile Memory, NVM), for example, at least a magnetic disk storage.Optionally, memory may be used also
To be storage device that at least one is located remotely from aforementioned processor.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit,
CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal
Processing, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing
It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete
Door or transistor logic, discrete hardware components.
In another embodiment provided by the invention, a kind of computer readable storage medium is additionally provided, which can
It reads to be stored with instruction in storage medium, when run on a computer, so that computer executes any institute in above-described embodiment
The video recommendation method stated.
In another embodiment provided by the invention, a kind of computer program product comprising instruction is additionally provided, when it
When running on computers, so that computer executes any video recommendation method in above-described embodiment.
For device/server/storage medium/program product embodiment, implement since it is substantially similar to method
Example, so being described relatively simple, the relevent part can refer to the partial explaination of embodiments of method.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all
Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention
It is interior.
Claims (12)
1. a kind of video recommendation method, which is characterized in that the described method includes:
Obtain the feature of each video in the first video sequence and first video sequence, wherein the first video sequence
Be classified as: the sequence lined up by each first video interested of the client of video to be recommended by playing sequence, first sense are emerging
Interesting video are as follows: the client of the video to be recommended played the interested video of user in video；
By the feature input of each video in first video sequence and first video sequence prediction trained in advance
Model carries out video estimation, obtains the second video sequence；Wherein, the prediction model is: according to Sample video sequence, the sample
The model that the feature of video is trained in this video sequence, wherein the video in the Sample video sequence are as follows: service
Each client that device provides service played be arranged successively in video by playing sequence, the interested video of user；
Recommend video to the client of the video to be recommended according to the sequence of each video in second video sequence.
2. the method according to claim 1, wherein the first video sequence of the acquisition, comprising:
Obtain the video playing record of the client of video to be recommended, wherein include in the video playing record: described wait push away
The playing sequence of each played video of the client of video and the mark information of each played video are recommended, it is each to have broadcast
The mark information for putting video is: the client of the video to be recommended played the operation of video progress to this to this according to user
The information of played video addition；
According to the mark information of each played video, the interested view of user is selected from each played video
Frequently, as the first video interested；
The described first video interested is ranked up by the playing sequence, obtains the first video sequence.
3. method according to claim 1 or 2, which is characterized in that training obtains the prediction model in the following way:
For each client of the server providing services, obtaining the client be played in video by playing sequence successively
The user of arrangement video interested forms Sample video sequence by playing sequence；
Obtain the feature of each video in the Sample video sequence；
The feature of each video in first subsequence of the Sample video sequence and first subsequence is inputted wait train
Model carries out video estimation, obtains output sequence；Wherein, first subsequence is each subsequence of Sample video sequence
In any subsequence；
Be training benchmark with the second subsequence, according to the difference between the output sequence and the second subsequence, adjustment it is described to
The model parameter of training pattern realizes to the training to training pattern, and using after training to training pattern as predicting
Model；Wherein, second subsequence are as follows: in the subsequence of the Sample video sequence, be located at first subsequence after and
The sequence adjacent with first subsequence.
4. according to the method described in claim 2, it is characterized in that, the mark information be user scoring, it is played when
Whether whether long ratio collected, by user by least one of user comment；Wherein, the duration ratio is that video is broadcast
The ratio between the total duration of the duration and the video put.
5. method according to claim 1 or 2, which is characterized in that the feature of the video, comprising: the total duration of video,
At least one of the type of video, the suitable crowd of video.
6. a kind of video recommendations device, which is characterized in that described device includes:
Sequence obtaining unit, for obtaining the feature of each video in the first video sequence and first video sequence,
In, first video sequence are as follows: lined up by each first video interested of the client of video to be recommended by playing sequence
Sequence, first video interested are as follows: the client of the video to be recommended played the interested video of user in video；
Sequence calculation sequence, for the feature of each video in first video sequence and first video sequence is defeated
Enter prediction model trained in advance and carry out video estimation, obtains the second video sequence；Wherein, the prediction model is: according to sample
The model that the feature of video is trained in this video sequence, the Sample video sequence, wherein the Sample video sequence
Video in column are as follows: each client of server providing services played be arranged successively in video by playing sequence, user
Interested video；
Video recommendations unit, for the sequence according to each video in second video sequence to the visitor of the video to be recommended
Recommend video in family end.
7. device according to claim 6, which is characterized in that the sequence obtaining unit, comprising:
Record obtains subelement, the video playing record of the client for obtaining video to be recommended, wherein the video playing
Include in record: the playing sequence and each played video of each played video of the client of the video to be recommended
Mark information, the mark information of each played video is: the client of the video to be recommended has broadcast this according to user
The operation for putting video progress played the information of video addition to this；
Video selection subelement, for the mark information according to each played video, from each played video
The interested video of middle selection user, as the first video interested；
Video sorting subunit obtains the first view for being ranked up by the playing sequence to the described first video interested
Frequency sequence.
8. device according to claim 6 or 7, which is characterized in that further include:
Training sequence obtaining unit obtains the client and has broadcast for being directed to each client of the server providing services
The user being arranged successively in video by playing sequence video interested is put, forms Sample video sequence by playing sequence；
Training characteristics obtaining unit, for obtaining the feature of each video in the Sample video sequence；
Training sequence output unit, for will be in the first subsequence of the Sample video sequence and first subsequence
The feature input of each video carries out video estimation to training pattern, obtains output sequence；Wherein, first subsequence is described
Any subsequence in each subsequence of Sample video sequence；
Model parameter adjustment unit, for being training benchmark with the second subsequence, according to the output sequence and the second subsequence
Between difference, the adjustment model parameter to training pattern realizes to the training to training pattern, and will be after training
To training pattern as prediction model；Wherein, second subsequence are as follows: in the subsequence of the Sample video sequence, position
After first subsequence and the sequence adjacent with first subsequence.
9. device according to claim 7, which is characterized in that the mark information be user scoring, it is played when
Whether whether long ratio collected, by user by least one of user comment；Wherein, the duration ratio is that video is broadcast
The ratio between the total duration of the duration and the video put.
10. device according to claim 6 or 7, which is characterized in that the feature of the video, comprising: video it is total when
At least one of length, the suitable crowd of the type of video, video.
11. a kind of server, which is characterized in that including processor, communication interface, memory and communication bus, wherein the place
Reason device, the communication interface, the memory complete mutual communication by the bus；
The memory, for storing computer program；
The processor realizes any method of claim 1-5 for executing the program stored on the memory
Step.
12. a kind of computer readable storage medium, which is characterized in that be stored with computer in the computer readable storage medium
Program realizes claim 1-5 any method and step when the computer program is executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910295961.2A CN109996122B (en) | 2019-04-12 | 2019-04-12 | Video recommendation method and device, server and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910295961.2A CN109996122B (en) | 2019-04-12 | 2019-04-12 | Video recommendation method and device, server and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109996122A true CN109996122A (en) | 2019-07-09 |
| CN109996122B CN109996122B (en) | 2021-07-20 |
Family
ID=67133512
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910295961.2A Active CN109996122B (en) | 2019-04-12 | 2019-04-12 | Video recommendation method and device, server and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109996122B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110825980A (en) * | 2019-11-05 | 2020-02-21 | 重庆邮电大学 | Microblog topic pushing method based on countermeasure generation network |
| CN111447239A (en) * | 2020-04-13 | 2020-07-24 | 北京字节跳动网络技术有限公司 | Video stream playing control method, device and storage medium |
| CN112507216A (en) * | 2020-12-01 | 2021-03-16 | 北京奇艺世纪科技有限公司 | Data object recommendation method, device, equipment and storage medium |
| CN113111217A (en) * | 2021-04-22 | 2021-07-13 | 北京达佳互联信息技术有限公司 | Training method of playing duration prediction model, video recommendation method and device |
| CN113609332A (en) * | 2021-08-11 | 2021-11-05 | 上海众源网络有限公司 | Method, system and device for recommending video live broadcast resources |
| CN114143612A (en) * | 2021-12-06 | 2022-03-04 | 北京达佳互联信息技术有限公司 | Video display method, video display device, electronic equipment, storage medium and program product |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120222058A1 (en) * | 2011-02-27 | 2012-08-30 | El Kaliouby Rana | Video recommendation based on affect |
| US20130021578A1 (en) * | 2011-07-20 | 2013-01-24 | Himax Technologies Limited | Learning-based visual attention prediction system and method thereof |
| CN103729363A (en) * | 2012-10-12 | 2014-04-16 | 上海聚力传媒技术有限公司 | Method and device for providing personalized video service through software and hardware environment of user |
| CN104735542A (en) * | 2015-03-30 | 2015-06-24 | 北京奇艺世纪科技有限公司 | Video play method and device |
| CN109104620A (en) * | 2018-07-26 | 2018-12-28 | 腾讯科技（深圳）有限公司 | A kind of short video recommendation method, device and readable medium |
-
2019
- 2019-04-12 CN CN201910295961.2A patent/CN109996122B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120222058A1 (en) * | 2011-02-27 | 2012-08-30 | El Kaliouby Rana | Video recommendation based on affect |
| US20130021578A1 (en) * | 2011-07-20 | 2013-01-24 | Himax Technologies Limited | Learning-based visual attention prediction system and method thereof |
| CN103729363A (en) * | 2012-10-12 | 2014-04-16 | 上海聚力传媒技术有限公司 | Method and device for providing personalized video service through software and hardware environment of user |
| CN104735542A (en) * | 2015-03-30 | 2015-06-24 | 北京奇艺世纪科技有限公司 | Video play method and device |
| CN109104620A (en) * | 2018-07-26 | 2018-12-28 | 腾讯科技（深圳）有限公司 | A kind of short video recommendation method, device and readable medium |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110825980A (en) * | 2019-11-05 | 2020-02-21 | 重庆邮电大学 | Microblog topic pushing method based on countermeasure generation network |
| CN110825980B (en) * | 2019-11-05 | 2022-07-01 | 重庆邮电大学 | Microblog topic pushing method based on countermeasure generation network |
| CN111447239A (en) * | 2020-04-13 | 2020-07-24 | 北京字节跳动网络技术有限公司 | Video stream playing control method, device and storage medium |
| CN111447239B (en) * | 2020-04-13 | 2023-07-04 | 抖音视界有限公司 | Video stream playing control method, device and storage medium |
| US11889135B2 (en) | 2020-04-13 | 2024-01-30 | Douyin Vision Co., Ltd. | Video stream playing control method and apparatus, and storage medium |
| CN112507216A (en) * | 2020-12-01 | 2021-03-16 | 北京奇艺世纪科技有限公司 | Data object recommendation method, device, equipment and storage medium |
| CN113111217A (en) * | 2021-04-22 | 2021-07-13 | 北京达佳互联信息技术有限公司 | Training method of playing duration prediction model, video recommendation method and device |
| CN113111217B (en) * | 2021-04-22 | 2024-02-27 | 北京达佳互联信息技术有限公司 | Training method of play duration prediction model, video recommendation method and device |
| CN113609332A (en) * | 2021-08-11 | 2021-11-05 | 上海众源网络有限公司 | Method, system and device for recommending video live broadcast resources |
| CN113609332B (en) * | 2021-08-11 | 2024-03-08 | 上海众源网络有限公司 | Video live broadcast resource recommendation method, system and device |
| CN114143612A (en) * | 2021-12-06 | 2022-03-04 | 北京达佳互联信息技术有限公司 | Video display method, video display device, electronic equipment, storage medium and program product |
| CN114143612B (en) * | 2021-12-06 | 2024-03-15 | 北京达佳互联信息技术有限公司 | Video display method, device, electronic equipment, storage medium and program product |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109996122B (en) | 2021-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109996122A (en) | A kind of video recommendation method, device, server and storage medium | |
| CN107038213B (en) | Video recommendation method and device | |
| CN111078939B (en) | Method, system and recording medium for extracting and providing highlight image in video content | |
| US20170228600A1 (en) | Analysis of video game videos for information extraction, content labeling, smart video editing/creation and highlights generation | |
| CN104182516B (en) | Information recommendation method and device and mobile terminal | |
| CN107172452A (en) | Direct broadcasting room recommends method and device | |
| CN107205178A (en) | Direct broadcasting room recommends method and device | |
| CN105095431A (en) | Method and device for pushing videos based on behavior information of user | |
| CN106131601A (en) | Video recommendation method and device | |
| CN107666623A (en) | The methods of exhibiting and device of broadcast information | |
| CN112866787B (en) | Bullet screen setting method, device and system | |
| CN105847985A (en) | Video recommendation method and device | |
| CN107818166A (en) | A kind of information recommends method, apparatus, server and system | |
| CN106294830A (en) | The recommendation method and device of multimedia resource | |
| JP2013033375A (en) | Information processing apparatus, information processing method, and program | |
| CN106294831B (en) | Information recommendation method and electronic equipment | |
| CN109151488B (en) | Method and system for recommending live broadcast room in real time according to user behaviors | |
| CN107846629A (en) | Recommend the method, apparatus and server of video to user | |
| CN109829064A (en) | Media resource shares and playing method and device, storage medium and electronic device | |
| CN104837034B (en) | A kind of information processing method, client and server | |
| CN106971006A (en) | The personalized push method and system of a kind of competitive sports information | |
| CN108228784A (en) | Video recommendation method and device, electronic equipment, storage medium, program | |
| US11468479B2 (en) | Methods and apparatus for generating information about portable device advertising | |
| CN109922357A (en) | The method and device of video recommendations | |
| US20130073364A1 (en) | Method of Downloading Contents |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |