CN110019945A - Video recommendation method and device, storage medium and electronic equipment - Google Patents
Video recommendation method and device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN110019945A CN110019945A CN201711454908.XA CN201711454908A CN110019945A CN 110019945 A CN110019945 A CN 110019945A CN 201711454908 A CN201711454908 A CN 201711454908A CN 110019945 A CN110019945 A CN 110019945A
- Authority
- CN
- China
- Prior art keywords
- video
- collection
- target
- user
- recommendation
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
- G06F16/735—Filtering based on additional data, e.g. user or group profiles
Abstract
The present invention provides a kind of video recommendation method, when detecting that user logs in video website, for currently logged on user, start it is multiple and different recall model, concentrated from the massive video set and determine its corresponding video collection；Then the video ID in each video collection is fused in target video set according to fixed integration percentage, and each of target video set video ID is ranked up；In conjunction with the corresponding video type of video ID each in the target video set, the video ID of setting number is chosen from the target video set by sequence, recommends login user.In method provided by the invention, it applies and multiple and different recalls model, so that each is recalled model and concentrates determining video collection from massive video, enrich the diversity of video selection, then after the video ID in determining each video collection being merged, it is ranked up selection, so that higher to the accuracy rate of user video recommendation.
Description
Technical field
The present invention relates to video field, in particular to a kind of video recommendation method and device, storage medium and electronic equipment.
Background technique
With the development of Information technology, each video software is applied and is given birth to, and largely enriches the daily life of people
It is living.Video software to promote the Experience Degree of user, and is proposed dependence while providing wide and full video resource for user
The individualized content in thousand people, thousand face that big data technology is realized is recommended, and recommends the interested video money of the user for each user
Source.At this stage, to the individualized content of user recommend by set it is a variety of it is different recall model realization, using recall model from
The interested video resource of user is chosen in the video of magnanimity, then recommends user.
Inventor, which passes through, to carry out existing individualized content recommendation process the study found that existing each video software is to use
During family provides individualized video recommendation, model is recalled only with one kind, in this way in recommendation process, can only be pushed away to user
A kind of video collection for recalling model generation is recommended, it is low for the recommendation accuracy rate of particular user.
Summary of the invention
The present invention provides a kind of video recommendation method, during solving the existing progress video recommendations to user, uses
Recall that model quantity is single, the low problem of caused video recommendations accuracy rate.
The present invention also provides a kind of video recommendations devices, to guarantee the realization and application of the above method in practice.
A kind of video recommendation method, which comprises
For currently logged on user, start it is multiple and different recall model, concentrated from the massive video set and determine it
Corresponding video collection；It include at least one video ID in each video collection, each video ID corresponds to institute
State a video of massive video concentration；
Each video ID in each video collection is fused to target video according to fixed integration percentage
In set；
The each video ID being fused in the target video set is ranked up；
In conjunction with the corresponding video type of video ID each in the target video set, selected from the target video set
The video ID for taking setting number, recommends the currently logged on user.
Above-mentioned method, optionally, the starting is multiple and different to recall model, concentrates from the massive video set true
Its fixed corresponding video collection, comprising:
Model is recalled for each of starting, distributes corresponding processing thread；
It controls each and recalls model, by its corresponding processing thread, concentrated from the massive video set true
Fixed corresponding video collection.
Above-mentioned method, optionally, each video ID by each video collection melts according to fixed
Composition and division in a proportion example is fused in target video set, comprising:
Determine the quantity of video ID in each video collection；
According to the quantity of video ID in each video collection, each video collection is calculated in the integration percentage
In corresponding accounting value；
From each video collection, according to its corresponding accounting value, the video ID for choosing respective numbers is fused to institute
State target video set.
Above-mentioned method, optionally, the described couple of each video ID being fused in the target video set are ranked up,
Include:
It determines in the target video set, the clicking rate of video corresponding to each video ID；
By the sequence of clicking rate from high to low, each of target video set video ID is ranked up.
Above-mentioned method, optionally, the corresponding video type of each video ID in target video set described in the combination,
The video ID that setting number is chosen from the target video set, recommends the currently logged on user, comprising:
From each video ID in the target video set, by its collating sequence, N number of video ID is chosen, the N is
The setting number, N are positive integer；Two distinct types of video is at least corresponded in N number of video ID.
Above-mentioned method, optionally, further includes:
During user checks the N number of recommendation video ID currently recommended, the refreshing instruction of real-time reception user；
When receiving the refreshing instruction of user, in the remaining video ID in the target video set, by remaining each
The collating sequence of a video ID chooses N number of video ID again, recommends the currently logged on user；It is described choose again it is N number of
Two distinct types of video is at least corresponded in video ID.
A kind of video recommendations device, described device include:
Start unit, for being directed to currently logged on user, start it is multiple and different recall model, from the magnanimity view set
Frequency, which is concentrated, determines its corresponding video collection；It include at least one video ID in each video collection, it is each described
Video ID corresponds to the video that the massive video is concentrated；
Integrated unit, for melting each video ID in each video collection according to fixed integration percentage
It is bonded in target video set；
Sequencing unit, for being ranked up to each video ID being fused in the target video set；
Recommendation unit is used in conjunction with the corresponding video type of video ID each in the target video set, from the mesh
The video ID for choosing setting number in video collection is marked, the currently logged on user is recommended.
Above-mentioned device, optionally, the start unit includes:
Subelement is distributed, for recalling model for each of starting, distributes corresponding processing thread；
It subelement is controlled, recalls model for controlling each, by its corresponding processing thread, set from described
Massive video, which is concentrated, determines corresponding video collection.
A kind of storage medium, the storage medium include the program of storage, wherein in described program operation described in control
Equipment where storage medium executes above-mentioned video recommendation method.
A kind of electronic equipment, including memory and one perhaps one of them or one of more than one program with
On program be stored in memory, and be configured to execute above-mentioned video recommendations side by one or more than one processor
Method.
Compared with prior art, the present invention includes the following advantages:
The present invention provides a kind of individualized video recommended method, when detecting that user logs in video website, for working as
Preceding login user, start it is multiple and different recall model, concentrated from the massive video set and determine its corresponding video
Set；Then the video ID in each video collection is fused to target video set according to fixed integration percentage
In, and each of target video set video ID is ranked up；In conjunction with each video ID in the target video set
Corresponding video type chooses the video ID of setting number from the target video set by sequence, recommends current login
User.In method provided by the invention, apply it is multiple and different recall model, so that each is recalled model from massive video
Concentrate determine video collection, enrich video selection diversity, then by the video ID in determining each video collection into
After row fusion, it is ranked up selection, so that higher to the accuracy rate of user video recommendation.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without any creative labor, it can also be obtained according to these attached drawings
His attached drawing.
Fig. 1 is a kind of method flow diagram of video recommendation method provided by the invention；
Fig. 2 is a kind of another method flow diagram of video recommendation method provided by the invention；
Fig. 3 is a kind of another method flow diagram of video recommendation method provided by the invention；
Fig. 4 is a kind of structural schematic diagram of video recommendations device provided by the invention；
Fig. 5 is a kind of another structural schematic diagram of video recommendations device provided by the invention；
Fig. 6 is the structural schematic diagram of a kind of electronic equipment provided by the 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, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The present invention can be used in numerous general or special purpose computing device environment or configurations.Such as: personal computer, service
Device computer, handheld device or portable device, laptop device, multi-processor device including any of the above devices or devices
Distributed computing environment etc..
The present invention provides a kind of video recommendation method, the method can be applied in video website or video APP
In processor, the recommendation process of the video recommendation method is executed by the processor, method flow diagram is as shown in Figure 1, packet
It includes:
S101: being directed to currently logged on user, start it is multiple and different recall model, concentrated from the massive video set true
Its fixed corresponding video collection；It include at least one video ID, each ID pairs of video in each video collection
A video for answering the massive video to concentrate；
In the embodiment of the present invention, for logging in video website or accessing each user of video APP, used for the login
Family starts multiple and different models of recalling, and each recalls model and is concentrated from the massive video set by its model rule really
Fixed corresponding video collection.Video ID is contained at least one in each video collection.
S102: each video ID in each video collection is fused to target according to fixed integration percentage
In video collection；
In the embodiment of the present invention, each recalls model after its corresponding video collection has been determined, determines each
The integration percentage of video ID in video collection, by the video ID in each video collection, according to the integration percentage of the determination,
It is fused in target video set.
S103: each video ID being fused in the target video set is ranked up；
It is each in the target video set to being fused to by preset ordering rule in method provided by the invention
A video ID is ranked up.
S104: in conjunction with the corresponding video type of video ID each in the target video set, from the target video collection
The video ID that setting number is chosen in conjunction, recommends the currently logged on user.
In method provided by the invention, by the sequence of sequence from the target video set by sequence, and combine
Video type corresponding to each video ID in the target video set chooses the video ID of setting number, recommends current
Login user.
Video recommendation method provided by the invention, has individualized feature, and the personalization refers to for different users
Recommend different video ID to it, user can watch video corresponding with the video ID by clicking each video ID.This
Invention logs in video website or the user of video APP in realizing individuation process to different user, for each, with
When family logs in, start it is multiple and different recall model, by its respective model rule from massive video library, determine and recommend video
Set, thus from multiple and different angles to realize to user carry out video recommendations.
Then, the video ID in determining each recommendation video collection is merged, since each model of recalling determines
Video collection in ordering rule it is different, so in method provided by the invention, by unified ordering rule, to fused
Each of target video set video ID is ranked up, and is adopted by high to low ordering rule by the interested degree of user
It is sorted with certain ordering rule.Then from each video ID by sequence, the video ID for choosing setting number is recommended
User, during recommendation, the video ID type for recommending user every time in order to prevent is single, in the present invention, regards from target
When choosing the video ID of setting number in frequency set, video type corresponding to each video ID is comprehensively considered, it is ensured that every time
The type coverage rate of video corresponding to the video ID of selection is high, and then further improves the consequently recommended recommendation to user and regard
Frequency meets the interest of user, improves the accuracy rate of video recommendations.
With reference to Fig. 2, showing in video recommendation method provided by the invention, the starting is multiple and different to recall model,
The detailed process for determining its corresponding video collection is concentrated from the massive video set, comprising steps of
S201: recalling model for each of starting, distributes corresponding processing thread；
S202: it controls each and recalls model by its corresponding processing thread, concentrated from the massive video set true
Fixed corresponding video collection.
In video recommendation method provided by the invention, model is recalled for each, recalling model is the video from magnanimity
The method of the middle selection interested candidate collection of user, and in method provided by the invention, it can be same for each login user
Shi Qidong it is multiple and different recall model, in order to improve each computational efficiency for recalling model, using distributed meter in the present invention
The mode of calculation recalls model for each and distributes a corresponding processing thread, while enabling multiple processing threads, so that
Each is recalled parallel the concentrating from the massive video set of model and determines corresponding video collection, and selection effect is improved
Rate.
With reference to Fig. 3, show in individualized video recommended method provided by the invention, it is described will be in each video collection
Video ID, according to fixed integration percentage, the process being fused in target video set specifically includes step:
S301: the quantity of video ID in each video collection is determined；
S302: it according to the quantity of video ID in each video collection, calculates each video collection and melts described
Corresponding accounting value in composition and division in a proportion example；
S303: from each video collection, according to its corresponding accounting value, the video ID fusion of respective numbers is chosen
To the target video set.
In method provided by the invention, multiple and different model selecting video set from massive video library is recalled starting
During, in order to guarantee to meet the needs of users from different angles, it is preferred that at least have one in each video collection
A video ID, for having chosen each video collection finished, in method provided by the invention, it is first determined each video
The quantity of video ID in set；Then the quantity according to video ID in each video collection, calculates each video collection pair
Accounting value in the integration percentage answered.
Such as: start 3 it is different recall model, have chosen three video collections A, B, C, wherein video ID in A
Number is 100, is 10 in B, is 50 in C.Fusion ratio is being calculated according to the quantity of video ID in each video collection
, can be preferred when example, all ID in three video collections are fused in target video set, then above three video
The integration percentage of set is 100:10:50.In recommended method provided by the invention, it is preferred that the target video set is initial
It is null set under state.In the present invention, a part of video ID can also be only selected to merge certain set, such as set A
It is middle to choose 50,10 are chosen in set B, choose 50 in C set, then the integration percentage of above three video collection is 50:
10:50.Method for choosing 50 in set A has oneself because recalling the video collection of model selection for each
Ordering rule, for each of set A video collection, it is preferred that can choose high preceding 50 videos of sorting position
ID。
In video recommendation method provided in an embodiment of the present invention, each video set can also be determined in the following ways
Close the corresponding accounting value in the integration percentage；For example, above three video collection A, B, C, wherein in A video ID number
It is 100, is 10 in B, be 50 in C, then, a kind of in A, B, C includes 160 video ID, and wherein the accounting of A is 5/8, B
Accounting be 1/16, C accounting be 5/16, by taking A as an example, 100*5/8 can be chosen from A by the accounting value of A accounting 5/8
Video ID is preferably rounded several views for having not been able to round numbers as a result, take the mode to round up in the embodiment of the present invention
Frequency ID.In the example, accounting value of the A in integration percentage 5/8:1/16:5/16 is 5/8.
It, can foundation when selecting video ID is fused to target video set from video collection in the embodiment of the present invention
User chooses the interest level of the corresponding video of video ID each in each video collection, can also be according to clicking rate
It is chosen.
It is above-mentioned to calculate the accounting value in the corresponding integration percentage of each video collection in method provided by the invention
When, model can be recalled in conjunction with each recalls characteristic, chooses ratio appropriate and is merged.
In video recommendation method provided by the invention, state to each video ID being fused in the target video set into
Row sequence process include:
It determines in the target video set, the clicking rate of video corresponding to each video ID；
By the sequence of clicking rate from high to low, each of target video set video ID is ranked up.
In method provided by the invention, for starting each recall model determine video collection internal referral knot
Fruit has the sequence of oneself, and the ordering rule inside each video collection finally determined is different, so for fused mesh
Mark each of video collection video ID, it is also necessary to unified sequence.
In method provided by the invention, the clicking rate CTR of the corresponding video of each video ID is preset by machine learning,
It is chosen by the sequence of clicking rate from high to low.
In video recommendation method provided by the invention, each video ID is corresponding in target video set described in the combination
Video type chooses the video ID of setting number from the target video set, recommends the mistake of the currently logged on user
Journey, comprising:
From each video ID in the target video set, by its collating sequence, N number of video ID is chosen, the N is
The setting number, N are positive integer；Two distinct types of video is at least corresponded in N number of video ID.
In method provided by the invention, in order to guarantee the diversity of recommendation results and solve the problems, such as the coverage rate of video, than
It, cannot all of a sort videos inside a combination if the every 10 video ID of a recommendation results are primary request combination
ID, such as corresponding video type are all science and technology videos.Coverage rate guarantees the video inside recommended candidate resource pool at least
It is exposed n times.Therefore, in the present invention, during choosing setting number from the target video set by sequence, in conjunction with
The corresponding video type of each video ID is chosen in target video set, in the recommendation results chosen each time, at least wraps
Containing two different video types.
Method provided by the invention, after process recommended to the user for the first time, further includes:
During user checks the N number of recommendation video ID currently recommended, the refreshing instruction of real-time reception user；
When receiving the refreshing instruction of user, in the remaining video ID in the target video set, by remaining each
The collating sequence of a video ID chooses N number of video ID again, recommends the currently logged on user；It is described choose again it is N number of
Two distinct types of video is at least corresponded in video ID.
Such as: finally in determining target video set, it is determined that 100 can recommend the video ID of user, due to
The limitation of the page layout of video website, may only preferably recommend every time 10 video ID of user in the page of website into
Row display.Then when user logs in video website, the implementation procedure of middle Fig. 1, recommends 10 video ID for user according to the present invention.
During user checks the 10 video ID recommended for the first time, if user is to recommending video to be refreshed, then receiving
When the refreshing instruction of user, in video collection in remaining 90 video ID, according to the sequence of clicking rate from high to low, choose
10 video ID, recommend user, refresh to 10 video ID of original recommendation.Again in this 10 video ID chosen
Two different video ID are included at least, equally to guarantee each video ID pushing away in recommendation process in target video set
Recommend coverage rate.
In the present invention, the corresponding user refreshing to the page is recommended each time presses above-mentioned implementation procedure from target video
Again the video ID that setting number is chosen in set is replaced the video ID in original recommendation page.
In the embodiment of the present invention, the video ID for recommending user can be carried out on the interface of user in conjunction with forms such as pictures
Display.
To sum up, individualized video recommended method provided by the invention, in the personalized recommendation system architecture design of its application
In, increase the concept of personalized recommendation engine, be similar to human brain, control recommendation service logic and each system component
Calling.The service logic being related in the present invention is primarily referred to as the data flow of recommender system, more compared to the prior art
Personalized recommendation system data flow is to recall model, sequence, directly generation recommendation results, in method provided by the invention, if
The personalized recommendation engine of meter increases by two new processes, proportion and concordance processing.That is the data flow of personalized recommendation engine
Cheng Weicong recalls model to proportion to sequence, preferably handles by concordance, generates last recommendation results.Also, in user
When being refreshed, recommendation results by newly generated recommendation results and before are merged.
In method provided by the invention, the model of recalling being related to is the selection interested time of user in the video of magnanimity
The method that selected works close.It may include that collaborative filtering, information filtering, the user's portrait filtered based on demographics and socialization are pushed away
Mode etc. is recommended, the problem of each way of recommendation can regard user from a different perspective as.It is each to recall model such as SVD+
+, ALS etc. recall the recommendation results of model generation, the proportion module in recommended engine merges its model recommendation results, i.e.,
Proportion realizes the combination for the method that different angle is seen, solves the method Single-issue of each recommendation results of each user.
In method provided by the invention, the control centre of entire recommended method is personalized recommendation engine.Recommended engine exists
During specific starting, the user behaviors log that the data source of required acquisition passes through real-time monitoring users is obtained.The present invention
In the method for offer, the program of personalized recommendation engine is constructed on the basis of realizing the streaming computing frame based on storm
Exploitation distributive type handle processing routine.In method provided by the invention, the ability of data is calculated and handled in extension
During, it can be realized by way of increasing the worker calculate node of storm.It ensure that personalized recommendation engine
Computing capability and response speed colleague, also ensured the horizontal extension of system, can handle the number of T rank or P rank
According to.
Personalized recommendation engine involved in the present invention, by monitoring users to the viewing log of video in video website,
The displaying log of user behaviors log, recommendation results, the distributive type that three independent storm are syncopated as in recommended engine calculate
Program module.In the present invention, in order to guarantee the consistency of recommending data, by programming, the log of a user is realized
It can only ensure that the consistency in module handled by one of task of a worker calculate node, intermodule
What consistency was guaranteed by CAS (the compare and swap) lock designed based on redis.
It is corresponding with video recommendation method shown in FIG. 1, the present invention also provides a kind of video recommendations device, for pair
The specific implementation of video recommendation method in Fig. 1, the video recommendations device provided in the embodiment of the present invention can be applied in video network
Stand or the processor of video APP in, structural schematic diagram as shown in figure 4,
Start unit 401, for being directed to currently logged on user, start it is multiple and different recall model, from the sea set
It measures and determines its corresponding video collection in video set；It include at least one video ID in each video collection, each
The video ID corresponds to the video that the massive video is concentrated；
Integrated unit 402, for by each video ID in each video collection, according to fixed fusion ratio
Example, is fused in target video set；
Sequencing unit 403, for being ranked up to each video ID being fused in the target video set；
Recommendation unit 404, in conjunction with the corresponding video type of video ID each in the target video set, from described
The video ID that setting number is chosen in target video set, recommends the currently logged on user.
In device provided by the invention, apply it is multiple and different recall model, so that each is recalled model from magnanimity
Video collection is determined in video set, the diversity of video selection is enriched, then by the video in determining each video collection
After ID is merged, it is ranked up selection, so that higher to the accuracy rate of user video recommendation.
With reference to Fig. 5, an a kind of detailed construction schematic diagram of recommendation apparatus provided by the invention, the start unit are shown
401 include:
Subelement 405 is distributed, for recalling model for each of starting, distributes corresponding processing thread；
Subelement 406 is controlled, recalls model by its corresponding processing thread, from the sea set for controlling each
It measures and determines corresponding video collection in video set.
The integrated unit 402 includes:
First determines subelement 407, for determining the quantity of video ID in each video collection；
Computation subunit 408 calculates each video for the quantity according to video ID in each video collection
It is integrated into corresponding accounting value in the integration percentage；
Fusion subelement 409, according to its corresponding accounting value, chooses respective numbers from each video collection
Video ID is fused to the target video set.
The sequencing unit 403 includes:
Second determines subelement 410, for determining in the target video set, video corresponding to each video ID
Clicking rate；
Sorting subunit 411, for pressing the sequence of clicking rate from high to low, to each of described target video set
Video ID is ranked up.
In prompting provided by the invention, in the recommendation unit 404, it is provided with
Subelement 412 is chosen, for by its collating sequence, selecting from each video ID in the target video set
Taking N number of video ID, the N is the setting number, and N is positive integer；Two kinds of different types are at least corresponded in N number of video ID
Video.
The embodiment of the invention also provides a kind of storage medium, the storage medium includes the program of storage, wherein in institute
It states the equipment where controlling the storage medium when program operation and executes above-mentioned video recommendation method, the method specifically includes:
For currently logged on user, start it is multiple and different recall model, concentrated from the massive video set and determine it
Corresponding video collection；It include at least one video ID in each video collection, each video ID corresponds to institute
State a video of massive video concentration；
Each video ID in each video collection is fused to target video according to fixed integration percentage
In set；
The each video ID being fused in the target video set is ranked up；
In conjunction with the corresponding video type of video ID each in the target video set, selected from the target video set
The video ID for taking setting number, recommends the currently logged on user.
Above-mentioned method, optionally, the starting is multiple and different to recall model, concentrates from the massive video set true
Its fixed corresponding video collection, comprising:
Model is recalled for each of starting, distributes corresponding processing thread；
It controls each and recalls model, by its corresponding processing thread, concentrated from the massive video set true
Fixed corresponding video collection.
Above-mentioned method, optionally, each video ID by each video collection melts according to fixed
Composition and division in a proportion example is fused in target video set, comprising:
Determine the quantity of video ID in each video collection；
According to the quantity of video ID in each video collection, each video collection is calculated in the integration percentage
In corresponding accounting value；
From each video collection, according to its corresponding accounting value, the video ID for choosing respective numbers is fused to institute
State target video set.
Above-mentioned method, optionally, the described couple of each video ID being fused in the target video set are ranked up,
Include:
It determines in the target video set, the clicking rate of video corresponding to each video ID；
By the sequence of clicking rate from high to low, each of target video set video ID is ranked up.
Above-mentioned method, optionally, the corresponding video type of each video ID in target video set described in the combination,
The video ID that setting number is chosen from the target video set, recommends the currently logged on user, comprising:
From each video ID in the target video set, by its collating sequence, N number of video ID is chosen, the N is
The setting number, N are positive integer；Two distinct types of video is at least corresponded in N number of video ID.
Above-mentioned method, optionally, further includes:
During user checks the N number of recommendation video ID currently recommended, the refreshing instruction of real-time reception user；
When receiving the refreshing instruction of user, in the remaining video ID in the target video set, by remaining each
The collating sequence of a video ID chooses N number of video ID again, recommends the currently logged on user；It is described choose again it is N number of
Two distinct types of video is at least corresponded in video ID.
The embodiment of the invention also provides a kind of electronic equipment, structural schematic diagram is as shown in fig. 6, specifically include memory
501 and one perhaps more than one 502 one of them or more than one program 502 of program be stored in memory 501
In, and be configured to by one or more than one processor 503 execute the one or more programs 502 include use
In the instruction performed the following operation:
For currently logged on user, start it is multiple and different recall model, concentrated from the massive video set and determine it
Corresponding video collection；It include at least one video ID in each video collection, each video ID corresponds to institute
State a video of massive video concentration；
Each video ID in each video collection is fused to target video according to fixed integration percentage
In set；
The each video ID being fused in the target video set is ranked up；
In conjunction with the corresponding video type of video ID each in the target video set, selected from the target video set
The video ID for taking setting number, recommends the currently logged on user.
It should be noted that all the embodiments in this specification are described in a progressive manner, each embodiment weight
Point explanation is the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.
For device class embodiment, since it is basically similar to the method embodiment, so being described relatively simple, related place ginseng
See the part explanation of embodiment of the method.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes that
A little elements, but also including other elements that are not explicitly listed, or further include for this process, method, article or
The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged
Except there is also other identical elements in the process, method, article or apparatus that includes the element.
For convenience of description, it is divided into various units when description apparatus above with function to describe respectively.Certainly, implementing this
The function of each unit can be realized in the same or multiple software and or hardware when invention.
As seen through the above description of the embodiments, those skilled in the art can be understood that the present invention can
It realizes by means of software and necessary general hardware platform.Based on this understanding, technical solution of the present invention essence
On in other words the part that contributes to existing technology can be embodied in the form of software products, the computer software product
It can store in storage medium, such as ROM/RAM, magnetic disk, CD, including some instructions are used so that a computer equipment
(can be personal computer, server or the network equipment etc.) executes the certain of each embodiment or embodiment of the invention
Method described in part.
A kind of video recommendation method provided by the present invention and device are described in detail above, it is used herein
A specific example illustrates the principle and implementation of the invention, and the above embodiments are only used to help understand originally
The method and its core concept of invention；At the same time, for those skilled in the art, according to the thought of the present invention, specific
There will be changes in embodiment and application range, in conclusion the content of the present specification should not be construed as to of the invention
Limitation.
Claims (10)
1. a kind of video recommendation method, which is characterized in that the described method includes:
For currently logged on user, start it is multiple and different recall model, concentrated from the massive video set and determine it respectively
Corresponding video collection；It include at least one video ID in each video collection, each video ID corresponds to the sea
Measure a video in video set；
Each video ID in each video collection is fused to target video set according to fixed integration percentage
In；
The each video ID being fused in the target video set is ranked up；
In conjunction with the corresponding video type of video ID each in the target video set, chooses and set from the target video set
Fixed number purpose video ID, recommends the currently logged on user.
2. the method according to claim 1, wherein the starting is multiple and different to recall model, from having set
Massive video concentrate determine its corresponding video collection, comprising:
Model is recalled for each of starting, distributes corresponding processing thread；
Control each and recall model, by its corresponding processing thread, concentrated from the massive video set determine with
Its corresponding video collection.
3. the method according to claim 1, wherein each video by each video collection
ID is fused in target video set according to fixed integration percentage, comprising:
Determine the quantity of video ID in each video collection；
According to the quantity of video ID in each video collection, it is right in the integration percentage to calculate each video collection
The accounting value answered；
From each video collection, according to its corresponding accounting value, the video ID for choosing respective numbers is fused to the mesh
Mark video collection.
4. the method according to claim 1, wherein described pair be fused to it is each in the target video set
Video ID is ranked up, comprising:
It determines in the target video set, the clicking rate of video corresponding to each video ID；
By the sequence of clicking rate from high to low, each of target video set video ID is ranked up.
5. the method according to claim 1, wherein each video ID in target video set described in the combination
Corresponding video type chooses the video ID of setting number from the target video set, recommends the current login and uses
Family, comprising:
From each video ID in the target video set, by its collating sequence, N number of video ID is chosen, the N is described
Number is set, N is positive integer；Two distinct types of video is at least corresponded in N number of video ID.
6. according to the method described in claim 5, it is characterized by further comprising:
During user checks the N number of recommendation video ID currently recommended, the refreshing instruction of real-time reception user；
When receiving the refreshing instruction of user, in the remaining video ID in the target video set, by remaining each view
The collating sequence of frequency ID chooses N number of video ID again, recommends the currently logged on user；Again the N number of video chosen
Two distinct types of video is at least corresponded in ID.
7. a kind of video recommendations device, which is characterized in that described device includes:
Start unit, for being directed to currently logged on user, start it is multiple and different recall model, from the massive video collection set
Its corresponding video collection of middle determination；It include at least one video ID, each video in each video collection
ID corresponds to the video that the massive video is concentrated；
Integrated unit, for each video ID in each video collection to be fused to according to fixed integration percentage
In target video set；
Sequencing unit, for being ranked up to each video ID being fused in the target video set；
Recommendation unit, for being regarded from the target in conjunction with the corresponding video type of video ID each in the target video set
The video ID that setting number is chosen in frequency set, recommends the currently logged on user.
8. device according to claim 7, which is characterized in that the start unit includes:
Subelement is distributed, for recalling model for each of starting, distributes corresponding processing thread；
Subelement is controlled, model is recalled for controlling each, by its corresponding processing thread, from the magnanimity set
Corresponding video collection is determined in video set.
9. a kind of storage medium, which is characterized in that the storage medium includes the program of storage, wherein run in described program
When control the equipment where the storage medium and execute video recommendation method as described in claim 1~6 any one.
10. a kind of electronic equipment, which is characterized in that including memory and one or more than one program, one of them
Perhaps more than one program is stored in memory and is configured to execute such as right by one or more than one processor
It is required that video recommendation method described in 1~6 any one.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711454908.XA CN110019945A (en) | 2017-12-28 | 2017-12-28 | Video recommendation method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711454908.XA CN110019945A (en) | 2017-12-28 | 2017-12-28 | Video recommendation method and device, storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110019945A true CN110019945A (en) | 2019-07-16 |
Family
ID=67187055
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711454908.XA Pending CN110019945A (en) | 2017-12-28 | 2017-12-28 | Video recommendation method and device, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110019945A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110633760A (en) * | 2019-09-25 | 2019-12-31 | 北京酷我科技有限公司 | Recommendation system integration strategy and recommendation system |
| CN110659388A (en) * | 2019-10-10 | 2020-01-07 | 北京奇艺世纪科技有限公司 | To-be-recommended information screening method and device, electronic equipment and storage medium |
| CN110750674A (en) * | 2019-10-09 | 2020-02-04 | 上海麦克风文化传媒有限公司 | User behavior sequence-based personalized recommendation system recall method, system, device and storage medium |
| CN111083561A (en) * | 2019-12-31 | 2020-04-28 | 深圳市商汤科技有限公司 | Video processing method, device, equipment and storage medium |
| CN111339416A (en) * | 2020-02-25 | 2020-06-26 | 咪咕文化科技有限公司 | Heat recall method and device, electronic equipment and storage medium |
| CN111859126A (en) * | 2020-07-09 | 2020-10-30 | 有半岛(北京)信息科技有限公司 | Recommended item determination method, device, equipment and storage medium |
| CN112417202A (en) * | 2020-09-04 | 2021-02-26 | 上海哔哩哔哩科技有限公司 | Content screening method and device |
| CN115474070A (en) * | 2022-08-10 | 2022-12-13 | 武汉斗鱼鱼乐网络科技有限公司 | Method, device, medium and equipment for displaying new content |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090100094A1 (en) * | 2007-10-15 | 2009-04-16 | Xavier Verdaguer | Recommendation system and method for multimedia content |
| CN104065981A (en) * | 2014-06-20 | 2014-09-24 | 海信集团有限公司 | Method and device for recommending videos |
| CN104239587A (en) * | 2014-10-17 | 2014-12-24 | 北京字节跳动网络技术有限公司 | Method and device for refreshing news list |
| CN104469430A (en) * | 2014-12-24 | 2015-03-25 | 武汉泰迪智慧科技有限公司 | Video recommending method and system based on context and group combination |
| CN106598741A (en) * | 2016-12-16 | 2017-04-26 | 飞狐信息技术（天津）有限公司 | Distributed type A/B test method, system and video recommendation system for personalized recommendation system |
-
2017
- 2017-12-28 CN CN201711454908.XA patent/CN110019945A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090100094A1 (en) * | 2007-10-15 | 2009-04-16 | Xavier Verdaguer | Recommendation system and method for multimedia content |
| CN104065981A (en) * | 2014-06-20 | 2014-09-24 | 海信集团有限公司 | Method and device for recommending videos |
| CN104239587A (en) * | 2014-10-17 | 2014-12-24 | 北京字节跳动网络技术有限公司 | Method and device for refreshing news list |
| CN104469430A (en) * | 2014-12-24 | 2015-03-25 | 武汉泰迪智慧科技有限公司 | Video recommending method and system based on context and group combination |
| CN106598741A (en) * | 2016-12-16 | 2017-04-26 | 飞狐信息技术（天津）有限公司 | Distributed type A/B test method, system and video recommendation system for personalized recommendation system |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110633760A (en) * | 2019-09-25 | 2019-12-31 | 北京酷我科技有限公司 | Recommendation system integration strategy and recommendation system |
| CN110633760B (en) * | 2019-09-25 | 2023-01-17 | 北京酷我科技有限公司 | Recommendation system integration method and recommendation system |
| CN110750674A (en) * | 2019-10-09 | 2020-02-04 | 上海麦克风文化传媒有限公司 | User behavior sequence-based personalized recommendation system recall method, system, device and storage medium |
| CN110750674B (en) * | 2019-10-09 | 2024-02-02 | 上海麦克风文化传媒有限公司 | Personalized recommendation system recall method, system, device and storage medium based on user behavior sequence |
| CN110659388A (en) * | 2019-10-10 | 2020-01-07 | 北京奇艺世纪科技有限公司 | To-be-recommended information screening method and device, electronic equipment and storage medium |
| CN110659388B (en) * | 2019-10-10 | 2022-07-22 | 北京奇艺世纪科技有限公司 | To-be-recommended information screening method and device, electronic equipment and storage medium |
| CN111083561B (en) * | 2019-12-31 | 2023-02-17 | 深圳市商汤科技有限公司 | Video processing method, device, equipment and storage medium |
| CN111083561A (en) * | 2019-12-31 | 2020-04-28 | 深圳市商汤科技有限公司 | Video processing method, device, equipment and storage medium |
| CN111339416A (en) * | 2020-02-25 | 2020-06-26 | 咪咕文化科技有限公司 | Heat recall method and device, electronic equipment and storage medium |
| CN111859126A (en) * | 2020-07-09 | 2020-10-30 | 有半岛(北京)信息科技有限公司 | Recommended item determination method, device, equipment and storage medium |
| WO2022007520A1 (en) * | 2020-07-09 | 2022-01-13 | 百果园技术(新加坡)有限公司 | Recommendation item determination method and apparatus, device, and storage medium |
| CN112417202A (en) * | 2020-09-04 | 2021-02-26 | 上海哔哩哔哩科技有限公司 | Content screening method and device |
| CN112417202B (en) * | 2020-09-04 | 2023-06-30 | 上海哔哩哔哩科技有限公司 | Content screening method and device |
| CN115474070A (en) * | 2022-08-10 | 2022-12-13 | 武汉斗鱼鱼乐网络科技有限公司 | Method, device, medium and equipment for displaying new content |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110019945A (en) | Video recommendation method and device, storage medium and electronic equipment | |
| Koch et al. | Optimising resource costs of cloud computing for education | |
| US9794211B2 (en) | Event-based comment grouping for content items | |
| US9406025B2 (en) | Touch prediction for visual displays | |
| US20150019469A1 (en) | Method of recommending items to a group of users | |
| CA2835324A1 (en) | Course recommendation system and method | |
| US11170059B2 (en) | Personalized content selection for time-constrained sessions | |
| CN103891245B (en) | The content detection of location-aware | |
| CN107408065A (en) | Supervision application loads | |
| Lesame | Vision and Practice: The South African Information Society Experience. | |
| Laghari et al. | Crowdsourcing platform for QoE evaluation for cloud multimedia services | |
| JP6067068B2 (en) | Service providing method and system for recommending postings in cooperation with app store and timeline | |
| Alvarez et al. | Using student surveys in determining the difficulty of programming assignments | |
| US10375184B2 (en) | Electronic content engagement determination | |
| CN111915383A (en) | Window-based article cold start recommendation method and device | |
| CN109241435A (en) | It is a kind of for digital cash transaction data push method, apparatus and system | |
| CN108959450A (en) | It is a kind of popular to comment on the system of determination, method, apparatus and readable medium | |
| Palacin et al. | Reframing e-participation for sustainable development | |
| Ma et al. | User preferences-aware recommendation for trustworthy cloud services based on fuzzy clustering | |
| US20140208261A1 (en) | Search engine optimization utilizing scrolling fixation | |
| CN107615789A (en) | The notice orientation participated in based on downstream user | |
| US11100424B2 (en) | Control system for learning and surfacing feature correlations | |
| Singh et al. | Apply AHP for resource allocation problem in cloud | |
| CN106911587A (en) | Trading volume control method and device | |
| Li et al. | Towards understanding the runtime configuration management of do-it-yourself content delivery network applications over public clouds |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190716 |
|
| RJ01 | Rejection of invention patent application after publication |