CN103164481A - Recommendation method and system of video with largest rising trend - Google Patents
Recommendation method and system of video with largest rising trend Download PDFInfo
- Publication number
- CN103164481A CN103164481A CN 201110426170 CN201110426170A CN103164481A CN 103164481 A CN103164481 A CN 103164481A CN 201110426170 CN201110426170 CN 201110426170 CN 201110426170 A CN201110426170 A CN 201110426170A CN 103164481 A CN103164481 A CN 103164481A
- Authority
- CN
- China
- Prior art keywords
- video
- candidate
- past
- concern
- trend
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a recommendation method and a system of a video with a largest rising trend. The method comprises the steps: counting past concern conditions of all videos, and generating a plurality of candidate videos and corresponding performance values according to the past concern conditions of all the videos; accumulating the concern conditions, in an increment clicking data base, of the candidate videos at all past time, and generating a history general comment score of all past time of each candidate video; obtaining comprehensive concern rates of the candidate videos according to the history general comment scores and the performance values; obtaining future concern values of the candidate videos according to the performance values, a speed formula and an accelerated speed formula; and obtaining a trend score according to the future concern values and the comprehensive concern rates. The recommendation method does not need manual intervention, the video which has the largest rising trend and enjoys great popularity for a period in the future can be automatically analyzed and found out for the first time and be recommended.
Description
Technical field
The present invention relates to a kind of recommend method and system of maximum ascendant trend video.
Background technology
Along with the explosive increase of internet content and the fast development of video website, the user finds that interested video is more and more difficult from a large amount of irrelevant information.How to recommend popular video most important from multitude of video, usually way has two kinds at present: a kind of way is high-quality by the web editor manual markings or follows the closely-related video of current events focus, and be pushed to the homepage of website, but, this way is bothersome requires great effort, and often can not reflect user's real demand; Another kind of way is by obtaining the click volume of each video in machine system statistics the past period, is popular video with video marker the most popular with users, recommends the user and watches.
compare popular video, maximum ascendant trend video is that prediction is within ensuing a period of time, the video that the user may be most interested in, as shown in Figure 1, the video of maximum ascendant trend be in the past distance in now nearer a period of time user's click be the video (be not fire video for a long time) of the trend of obvious accelerated growth, and can keep burning hoter state within a period of time in future, yet, maximum ascendant trend video is not that the absolute video playback number of times by certain period sorts with regard to getable result, therefore how automatic analysis is a problem demanding prompt solution with excavating maximum ascendant trend video.
Summary of the invention
The object of the present invention is to provide a kind of recommend method and system of maximum ascendant trend video, the method and system can automatically analyze and excavate the video of maximum ascendant trend.
For addressing the above problem, the invention provides a kind of recommend method of maximum ascendant trend video, comprising:
Add up the concern situation in all video past and deposit in an increment click data storehouse;
Generate the performance value of some candidate's videos and correspondence and deposit in candidate's video database according to the concern situation in all video past;
In the described increment click data of accumulative total storehouse, the free concern situation of past of each candidate's video generates the in the past free historical general comment score of this candidate's video;
Obtain the comprehensive concern rate of this candidate's video according to the performance value of this candidate's video in the historical general comment score of each candidate's video and candidate's video database;
Obtain the following concern value of this candidate's video according to the performance value of each candidate's video and speed formula, Acceleration Formula;
Obtain the trend score of this candidate's video according to concern in the future value of each candidate's video and comprehensive concern rate, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video and deposit in a trend video database the trend score of all candidate's videos is ascending.
Further, in said method, the concern situation in all video past of described statistics comprises: add up respectively watching number of times, collection number of times and commenting on number of times of all each hours in videos past.
Further, in said method, generate the performance value of some candidate's videos and correspondence according to the concern situation in all video past, comprising:
Choose several different past Preset Time sections, with each video of same in the past Preset Time section watch number of times, collection number of times and comment number of times to be weighted summation with different ratios respectively generating the performance value, with descending sequence of performance value of all videos in same Preset Time section in the past;
Choose respectively several forward videos of performance value described in different past Preset Time sections as candidate's video according to the result of described sequence.
Further, in said method, watch number of times, collection number of times and comment number of times in the different past Preset Time section of each video are also sued for peace with 1,3 and 5 proportion weighted respectively.
Further, in said method, in accumulative total increment click data storehouse, the free concern situation of past of each candidate's video generates the in the past free historical general comment score of this candidate's video, comprising: described number of times, collection number of times and the comment number of times watched of each candidate's video generated the in the past free historical general comment score of this candidate's video by the different proportion weighted sum.
Further, in said method, with each video free watch in the past number of times, collection number of times and comment number of times respectively with 1,3 and 5 proportion weighted and summation.
Further, in said method, obtain the comprehensive concern rate of this candidate's video according to the performance value of this candidate's video in the historical general comment score of each candidate's video and candidate's video database, comprise: the performance value of the different past Preset Time section of each candidate's video is pressed different proportion weighting and summation, and the result of described summation is obtained described comprehensive concern rate divided by the historical general comment score of this candidate's video.
Further, in said method, the now nearer past Preset Time section of distance is pressed higher proportion weighted than the present past Preset Time section far away of distance.
Further, in said method, described performance value according to each candidate's video, speed formula eey and Acceleration Formula are obtained the following concern value of this candidate's video and are comprised:
Performance value, speed formula and the Acceleration Formula of the past Preset Time section different according to each candidate's video obtained the click speed of one of them past Preset Time section of each candidate's video and clicked acceleration;
According to the described click speed of each candidate's video with click concern in the future value that acceleration obtains the following Preset Time section of this candidate's video one.
Further, in said method, the trend of each candidate's video must be divided into concern in the future value with comprehensive concern rate gained on duty of this candidate's video.
According to another side of the present invention, a kind of commending system of maximum ascendant trend video is provided, comprising:
Increment click data storehouse is for the concern situation in all video past of storage;
Candidate's video database is used for storage candidate's video and corresponding performance value;
The trend video database is used for storage ascendant trend video;
Data statistics module, the concern situation that is used for adding up all video past;
Candidate's video generation module is used for generating some candidate's videos and corresponding performance value according to the concern situation of all videos of described increment click data storehouse;
Trend video generation module is used for analyzing the ascendant trend video according to the concern situation in described increment click data storehouse, the performance value of each candidate's video in described candidate's video database and speed formula, Acceleration Formula.
Further, in said system, described trend video generation module comprises:
The comprehensive unit of paying close attention to, the free concern situation of past that is used for each candidate's video of the described increment click data of accumulative total storehouse generates the in the past free historical general comment score of this candidate's video;
Comprehensive concern rate unit is for obtain the comprehensive concern rate of this candidate's video according to the performance value of historical general comment score and each candidate's video;
The following unit of paying close attention to is used for obtaining the following concern value of this candidate's video according to performance value, speed formula and the Acceleration Formula of each candidate's video;
The ascendant trend unit, be used for the trend score of obtaining this candidate's video according to concern in future value and the comprehensive concern rate of each candidate's video, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video the trend score of all candidate's videos is ascending.
Further, in said system, described data statistics module is added up respectively watching number of times, collection number of times and commenting on number of times of all each hours in videos past.
Further, in said system, described candidate's video generation module is chosen several different past Preset Time sections, described number of times, collection number of times and the comment number of times watched of each video of same in the past Preset Time section is weighted summation to generate the performance value with different ratios respectively, with same descending sequence of performance value of all videos of Preset Time section in the past, choose respectively several forward videos of performance value described in different past Preset Time sections as candidate's video according to the result of described sequence.
Further, in said system, described comprehensive concern unit generates the in the past free historical general comment score of this candidate's video with described number of times, collection number of times and the comment number of times watched of each candidate's video by the different proportion weighted sum.
Further, in said system, different proportion weighting and summation are pressed with the performance value of the different past Preset Time section of each candidate's video in described comprehensive concern rate unit, and the result of described summation must be got described comprehensive concern rate divided by historical general comment.
Further, in said system, described following performance value, speed formula and the Acceleration Formula of paying close attention to the unit past Preset Time section different according to each candidate's video obtained each candidate's video click speed and the click acceleration of one of them past Preset Time section in the past; Obtain this candidate's video in concern in the future of following Preset Time section value according to described click speed and the click acceleration of each candidate's video.
Further, in said system, pay close attention to the trend score that obtains this candidate's video with comprehensive concern rate on duty the future that described ascendant trend unit will each candidate's video.
compared with prior art, the present invention is by the concern situation in all video past of statistics, generate some candidate's videos and corresponding performance value according to the concern situation of all videos, and add up the in the past free historical general comment score of each candidate's video according to described concern situation, the comprehensive concern rate of obtaining this candidate's video according to performance value and the historical general comment score of each candidate's video again, again according to performance value and the speed formula of each candidate's video, Acceleration Formula is obtained the following concern value of this candidate's video, the trend score of obtaining this candidate's video according to concern in future value and the comprehensive concern rate of each candidate's video at last, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video the trend score of all candidate's videos is ascending, do not need manual intervention, can and excavate the wide maximum ascendant trend video popular with users of following a period of time most probable and recommend at very first time automatic analysis, rather than the click volume of the passive some videos of wait reaches the prostatitis that is discharged to the website and recommends, can predict better the video fashion trend, hold user's demand, thereby improving the user experiences, strengthen website viscosity.
in addition, the present invention generates the performance value by each video hour concern value is merged respectively by several different Preset Time sections in the past, with same descending sequence of performance value of all videos of Preset Time section in the past, choose respectively several forward videos of different performance values described in the Preset Time section in the past as candidate's video according to the result of described sequence, the concern situation that the video of different Preset Time sections of past is carried out merges statistical study, rather than only the concern situation of the video in certain hour is carried out statistical study, even thereby realized that the concern situation data of several hours have disappearance in the past, do not affect the analysis result of final maximum ascendant trend video yet.
In addition, with the difference of each candidate's video in the past the performance value of Preset Time section press the different proportion weighting and sue for peace, and the result of described summation is drawn comprehensive concern rate divided by historical general comment score, the now nearer past Preset Time section of distance is pressed higher proportion weighted than the present past Preset Time section far away of distance, thereby excavate maximum ascendant trend video in the video that can just begin in several hours in the past to prevail, rather than those have kept popular state video for a long time.
Description of drawings
Fig. 1 is that the accumulative total of a maximum ascendant trend video is play number curve figure;
Fig. 2 is the process flow diagram of recommend method of the maximum ascendant trend video of the embodiment of the present invention one;
Fig. 3 is the module diagram of commending system of the maximum ascendant trend video of the embodiment of the present invention two.
Embodiment
For above-mentioned purpose of the present invention, feature and advantage can be become apparent more, the present invention is further detailed explanation below in conjunction with the drawings and specific embodiments.
Embodiment one
As shown in Figure 1, the invention provides a kind of recommend method of maximum ascendant trend video, comprising:
Step S11 adds up the concern situation in all video past and deposits in an increment click data storehouse, generates some candidate's videos and corresponding performance value and deposits in candidate's video database according to the concern situation in all video past;
Step S12, totally in described increment click data storehouse, the free concern situation of past of each candidate's video generates the in the past free historical general comment score of this candidate's video;
Step S13 obtains the comprehensive concern rate of this candidate's video according to the performance value of this candidate's video in the historical general comment score of each candidate's video and candidate's video database;
Step S14 obtains the following concern value of this candidate's video according to the performance value of each candidate's video and speed formula, Acceleration Formula;
Step S15, obtain the trend score of this candidate's video according to concern in the future value of each candidate's video and comprehensive concern rate, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video and deposit in a trend video database the trend score of all candidate's videos is ascending.
Particularly, in step S11, the concern situation in all video past of described statistics comprises: add up respectively watching number of times, collection number of times and commenting on number of times of all each hours in videos past.Described concern situation according to all video past generates some candidate's videos and corresponding performance value comprises: choose several different past Preset Time sections, watch number of times, collection number of times and comment number of times to be weighted summation with different ratios respectively each video of same in the past Preset Time section described and generate the performance value, with same descending sequence of performance value of all videos of Preset Time section in the past, choose respectively several forward videos of performance value described in different past Preset Time sections as candidate's video according to the result of described sequence.Wherein, watch number of times, collection number of times and comment number of times in the different past Preset Time section of each video are also sued for peace with 1,3 and 5 proportion weighted respectively, described several past Preset Time sections for example comprise: past 12 hours, past 6 hours and 2 hours three time periods of past, the number of several videos that described performance value is forward is for example 500.
In step S12, in accumulative total described increment click data storehouse the free concern situation of past of each candidate's video generate this candidate's video in the past free historical general comment score refer to described number of times, collection number of times and the comment number of times watched of each candidate's video generated the in the past free historical general comment score of this candidate's video by the different proportion weighted sum, for example, with each video free watch in the past number of times, collection number of times and comment number of times respectively with 1,3 and 5 proportion weighted and summation.
In step S13, the performance value of the different past Preset Time section of each candidate's video is pressed different proportion weighting and summation, and the result of described summation is obtained described comprehensive concern rate divided by the historical general comment score of this candidate's video, wherein, distance now nearer past Preset Time section than the present past Preset Time section far away of distance by higher proportion weighted, because the performance value of the now nearer candidate's video of distance is more valuable than the performance value of the present nearer candidate's video of distance aspect the maximum ascendant trend video of judgement.
In step S14, at first the process of obtaining the following concern value of this candidate's video according to the performance value of each candidate's video and speed formula, Acceleration Formula comprise the steps:, the performance value of the past Preset Time section different according to each candidate's video and speed formula and Acceleration Formula obtain each candidate's video in the past one of them in the past the Preset Time section click speed and click acceleration; Then, obtain in concern in the future value step S15 of the following Preset Time section of this candidate's video one according to described click speed and the click acceleration of each candidate's video, the trend of each candidate's video must be divided into concern in the future value with comprehensive concern rate gained on duty of this candidate's video.
The below for one more specifically example said method is described in more detail:
Step S21, the watching number of times, collection number of times and comment number of times and deposit in an increment click data storehouse of each hour in past of adding up respectively all video past.
step S22, take out the concern situation of 12 hours all videos in the past from increment click data storehouse, chose over 12 hours, past 6 hours and 2 hours in the past three past Preset Time sections, the described number of times of watching with each video in past 12 hours, collection number of times and comment number of times are respectively with 1, 3 and 5 ratio is weighted summation and generates performance value (score (t=12)), the descending sequence of performance value with all videos in past 12 hours, and will pass by the described number of times of watching of each video of 6 hours, collection number of times and comment number of times are respectively with 1, the ratio of 3 and 5 ratio is weighted summation and generates performance value (score (t=6)), the descending sequence of performance value with all videos in past 6 hours, to pass by again the described number of times of watching of each video of 2 hours, collection number of times and comment number of times are respectively with 1, the ratio of 3 and 5 ratio is weighted summation and generates performance value (score (t=2)), the descending sequence of performance value with all videos in past 2 hours, then choose respectively 500 forward videos of different performance values described in the Preset Time section in the past as candidate's video according to the result of described sequence, namely obtained over respectively 2 hours, 6 hours, clicked 500 maximum videos in 12 hours, amount to 1500, add up to the repetition video in these 1500 videos, obtain candidate's list of videos.
More specifically, the purpose of obtaining score (t) is the video performance value marking of several time periods in the past, the performance value marking of t hour in the past is (score (t)) to specify certain video, here get t=2,6,12, score (t) watches, collects, comments on number of times weighted sum gained for accumulative total in its t hour.
Make playcount (t), favtcount (t), commentcount (t) represent that respectively the accumulative total of t hour video v is watched number of times, collection number of times, comment number of times in the past:
score(t)＝playcount(t)*1+favtcount(t)*3+commentcount(t)*5。
Like this, can obtain for each video in candidate list score (2), score (6), score (12).
Step S23 generates the in the past free historical general comment score score (past) of this candidate's video with described number of times, collection number of times and the comment number of times watched of each candidate's video by suing for peace with 1,3 and 5 proportion weighted respectively.
Detailed, the historical general comment score of one video (score (past)) is its passing free accumulative total clicks, collection number, comment number weighted sum gained, make playcount (past), favtcount (past), commentcount (past) represent respectively video all time cumulation numbers of clicks of past, the collection number of times, the comment number of times:
score(past)＝playcount(past)*1+favtcount(past)*3+commentcount(past)*5
Step S24, the performance value in past 12 hours, the past 6 hours of each candidate's video and 2 hours is in the past also sued for peace by 5,3,1 proportion weighted respectively, and the result of described summation must be got described comprehensive concern rate (score (new)) divided by the historical general comment of this candidate's video, here consider this video in the past several hours focus on the accounting of paying close attention to total amount, i.e. this video video for a long time that can not be fire.
more specifically, the weighted sum of the performance value score (t) of score (new) t of being over hour video and the ratio of historical general comment score score (past), if the score of a video (new) is higher, illustrate that this video just began to prevail in the past in several hours, this point is extremely important to excavating potential popular video, if a video has kept popular state for a long time, performance value score (past) of several hours also can be very high in the past for it, but meaning is little concerning the excavation purpose of maximum ascendant trend video.
score(new)＝score(t＝2)/score(past)*5+score(t＝6)/score(past)*3
+score(t＝12)/score(past)*1
Step S25, according to the click speed on each candidate's video 12 hours in the past, the performance value in 6 hours past and 2 hours in the past and initial velocity formula and Acceleration Formula obtain each the candidate's video time point of 12 hours in the past with click acceleration, and according to the described click speed of each candidate's video with click acceleration and obtain this candidate's video concern in future value score (future) in 3 hours futures.For example take the average broadcasting time in past 12 hours, 6 hours, 2 hours as the basis, according to speed and Acceleration Formula simulate certain video concern number of times of several hours next.
Wherein, the purpose of obtaining score (future) is to predict the pouplarity of video in 3 hours in the future, and video is from uploading to now in section, and pouplarity may be with a variety of situations: what have has a welcome always; After the hot topic of the change gradually that has, very fast quiet again; What have is always unknown to the public, but has become fire suddenly after being transferred to social network sites some day; The welcome cycle of some movie video can be very long, and the welcome time of news category video can be relatively concentrated.
But within a very little time cycle, for example 12 hours, the welcome situation of video mostly can be more single, i.e. this video council per hour shows clicks and continues rising, continuous decrease, remain unchanged or be in the turnover situation that rises to decline.Therefore, when can suppose 12nd hour in the past, the per hour number of clicks of video is that v is click speed, in these 12 hours the acceleration of click speed constant be a.
We can calculate v and a according to Acceleration Formula: s=v*t+0.5*a*t2 thus.
v*12+0.5*a*12*12＝score(12) (1)
(v+a*6)*6+0.5*a*6*6＝score(6) (2)
(v+a*10)*2+0.5*a*2*2＝score(2) (3)
(1) obtain (4): a=score (6)/18-score (12)/36 with (2); V=score (12)/4-score (6)/3
(2) obtain (5): a=score (2)/4-score (6)/12 with (3); V=score (6) * 11/12-score (2) * 9/4
The result of average above (4) and (5):
a＝score(2)/8-score(6)/72-score(12)/72
v＝score(12)/8+score(6)*7/24-score(2)*9/8
Video may be calculated in the performance value in 3 hours futures:
(v+a*12)*3+0.5*a*3*3。
The above a of substitution follows v,
Can get score (2) * 27/16+score (6) * 5/16-score (12) * 3/16
Therefore:
score(future)＝score(6)*5/16-score(12)*3/16+score(2)*27/16
Step S15, pay close attention to the trend score (score (trend)) of obtaining this candidate's video with comprehensive concern rate on duty future that will each candidate's video, and sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video and deposit in a trend video database the trend score of all candidate's videos is ascending.
More specifically, the trend score score (trend) of video is the comprehensive consideration of score (future) and score (new):
score(trend)＝score(future)*score(new)
To descending sequence of score (trend) of candidate's video, the list of fast video can obtain to rise.
The method that the present embodiment provides is by the welcome indexs such as clicks of all videos in the periodic analysis video website, do not need manual intervention, can automatically excavate the wide video popular with users of following a period of time most probable, find fashion trend in the very first time, rather than the click volume of the passive some videos of wait reaches the prostatitis that is discharged to the website and recommends, more can hold user's potential demand, have very high robustness.
Embodiment two
As shown in Figure 3, the present invention also provides a kind of commending system of maximum ascendant trend video, comprises increment click data storehouse 1, candidate's video database 2, trend video database 3, data statistics module 4, candidate's video generation module 5 and trend video generation module 6.
Described increment click data storehouse 1 is used for the concern situation in all video past of storage.
Described candidate's video database 2 is used for storage candidate's video and corresponding performance value.
Described trend video database 3 is used for storage ascendant trend video.
The concern situation that described data statistics module 4 is used for all video past of statistics deposits described increment click data storehouse in, and is concrete, described data statistics module add up respectively all videos in the past each hours watch number of times, collection number of times and comment number of times.
Described candidate's video generation module 5 is used for generating some candidate's videos and corresponding performance value according to the concern situation of all videos of described increment click data storehouse, concrete, described candidate's video generation module is chosen several different past Preset Time sections, watch number of times, collection number of times and comment number of times to be weighted summation with different ratios respectively each video of same in the past Preset Time section described and generate the performance value, with same descending sequence of performance value of all videos of Preset Time section in the past; Choose respectively several forward videos of performance value described in different past Preset Time sections as candidate's video according to the result of described sequence.
Described trend video generation module 6 is used for analyzing the ascendant trend video and depositing in described trend video database according to the concern situation in described increment click data storehouse, the performance value of each candidate's video in described candidate's video database and speed formula, Acceleration Formula.
Wherein, described trend video generation module 6 comprises comprehensive concern unit 61, comprehensively concern rate unit 62, following unit 63 and the ascendant trend unit 64 paid close attention to.
Described comprehensive concern unit 61 is used for totally, and the free concern situation of past of each candidate's video of described increment click data storehouse generates the in the past free historical general comment score of this candidate's video, concrete, described comprehensive concern unit 61 generates the in the past free historical general comment score of this candidate's video with described number of times, collection number of times and the comment number of times watched of each candidate's video by the different proportion weighted sum.
Described comprehensive concern rate unit 62 is used for obtaining according to the performance value of historical general comment score and each candidate's video the comprehensive concern rate of this candidate's video, concrete, different proportion weighting and summation are pressed with the performance value of the different past Preset Time section of each candidate's video in described comprehensive concern rate unit 62, and the result of described summation must be got described comprehensive concern rate divided by historical general comment, here consider this video in the past several hours focus on the accounting of paying close attention to total amount, i.e. this video video for a long time that can not be fire.
The described following unit 63 of paying close attention to is used for obtaining the following concern value of this candidate's video according to the performance value of each candidate's video and speed formula, Acceleration Formula, concrete, the described following unit 63 of paying close attention to obtains the click speed of one of them past Preset Time section of each candidate's video and clicks acceleration by speed formula and Acceleration Formula according to the performance value of each past Preset Time section of each candidate's video; According to the described click speed of each candidate's video with click concern in the future value that acceleration obtains the following Preset Time section of this candidate's video one, here according to speed and Acceleration Formula can simulate certain video concern number of times of several hours next.
Described ascendant trend unit 64 is for the trend score of obtaining this candidate's video according to concern in future value and the comprehensive concern rate of each candidate's video, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video the trend score of all candidate's videos is ascending, concrete, pay close attention to the trend score that obtains this candidate's video with comprehensive concern rate on duty the future that described ascendant trend unit 64 will each candidate's video.
compared with prior art, the present invention is by the concern situation in all video past of statistics, generate some candidate's videos and corresponding performance value according to the concern situation of all videos, and add up the in the past free historical general comment score of each candidate's video according to described concern situation, the comprehensive concern rate of obtaining this candidate's video according to performance value and the historical general comment score of each candidate's video, performance value and speed formula according to each candidate's video, Acceleration Formula is obtained the following concern value of this candidate's video, obtain the trend score of this candidate's video according to concern in the future value of each candidate's video and comprehensive concern rate, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video the trend score of all candidate's videos is ascending, do not need manual intervention, can and excavate the wide maximum ascendant trend video popular with users of following a period of time most probable and recommend at very first time automatic analysis, rather than the click volume of the passive some videos of wait reaches the prostatitis that is discharged to the website and recommends, can predict better the video fashion trend, hold user's demand, thereby improving the user experiences, strengthen website viscosity.
in addition, the present invention generates the performance value by each video hour concern value is merged respectively by several different Preset Time sections in the past, with same descending sequence of performance value of all videos of Preset Time section in the past, choose respectively several forward videos of different performance values described in the Preset Time section in the past as candidate's video according to the result of described sequence, the concern situation that the video of different Preset Time sections of past is carried out merges statistical study, rather than only the concern situation of the video in certain hour is carried out statistical study, even thereby realized that the concern situation data of several hours have disappearance in the past, do not affect the analysis result of final maximum ascendant trend video yet.
In addition, with the difference of each candidate's video in the past the performance value of Preset Time section press the different proportion weighting and sue for peace, and the result of described summation is drawn comprehensive concern rate divided by historical general comment score, the now nearer past Preset Time section of distance is pressed higher proportion weighted than the present past Preset Time section far away of distance, thereby excavate maximum ascendant trend video in the video that can just begin in several hours in the past to prevail, rather than those have kept popular state video for a long time.
Adopt the mode of going forward one by one to describe by each embodiment in this instructions, what each embodiment stressed is and the difference of other embodiment that between each embodiment, identical similar part is mutually referring to getting final product.For embodiment disclosed system, due to corresponding with the disclosed method of embodiment, so describe fairly simple, relevant part partly illustrates referring to method and gets final product.
The professional can also further recognize, unit and the algorithm steps of each example of describing in conjunction with embodiment disclosed herein, can realize with electronic hardware, computer software or combination both, for the interchangeability of hardware and software clearly is described, composition and the step of each example described in general manner according to function in the above description.These functions are carried out with hardware or software mode actually, depend on application-specific and the design constraint of technical scheme.The professional and technical personnel can specifically should be used for realizing described function with distinct methods to each, but this realization should not thought and exceeds scope of the present invention.
Obviously, those skilled in the art can carry out various changes and modification and not break away from the spirit and scope of the present invention invention.Like this, if within of the present invention these were revised and modification belongs to the scope of claim of the present invention and equivalent technologies thereof, the present invention also was intended to comprise these change and modification.
Claims (18)
1. the recommend method of a maximum ascendant trend video, is characterized in that, comprising:
Add up the concern situation in all video past and deposit in an increment click data storehouse;
Generate the performance value of some candidate's videos and correspondence and deposit in candidate's video database according to the concern situation in all video past;
In the described increment click data of accumulative total storehouse, the free concern situation of past of each candidate's video generates the in the past free historical general comment score of this candidate's video;
Obtain the comprehensive concern rate of this candidate's video according to the performance value of this candidate's video in the historical general comment score of each candidate's video and candidate's video database;
Obtain the following concern value of this candidate's video according to the performance value of each candidate's video and speed formula, Acceleration Formula;
Obtain the trend score of this candidate's video according to concern in the future value of each candidate's video and comprehensive concern rate, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video and deposit in a trend video database the trend score of all candidate's videos is ascending.
2. the recommend method of maximum ascendant trend video as claimed in claim 1, is characterized in that, the concern situation in all video past of described statistics comprises: add up respectively watching number of times, collection number of times and commenting on number of times of all each hours in videos past.
3. the recommend method of maximum ascendant trend video as claimed in claim 2, is characterized in that, generates the performance value of some candidate's videos and correspondence according to the concern situation in all video past, comprising:
Choose several different past Preset Time sections, with each video of same in the past Preset Time section watch number of times, collection number of times and comment number of times to be weighted summation with different ratios respectively generating the performance value, with descending sequence of performance value of all videos in same Preset Time section in the past;
Choose respectively several forward videos of performance value described in different past Preset Time sections as candidate's video according to the result of described sequence.
4. the recommend method of maximum ascendant trend video as claimed in claim 3, is characterized in that, watch number of times, collection number of times and comment number of times in the different past Preset Time section of each video are also sued for peace with 1,3 and 5 proportion weighted respectively.
5. the recommend method of maximum ascendant trend video as claimed in claim 3, it is characterized in that, in accumulative total increment click data storehouse, the free concern situation of past of each candidate's video generates the in the past free historical general comment score of this candidate's video, comprising: watch number of times, collection number of times and the comment number of times of each candidate's video are generated the in the past free historical general comment score of this candidate's video by the different proportion weighted sum.
6. the recommend method of maximum ascendant trend video as claimed in claim 5, is characterized in that, with each video free watch in the past number of times, collection number of times and comment number of times respectively with 1,3 and 5 proportion weighted and summation.
7. the recommend method of maximum ascendant trend video as claimed in claim 5, it is characterized in that, obtain the comprehensive concern rate of this candidate's video according to the performance value of this candidate's video in the historical general comment score of each candidate's video and candidate's video database, comprise: the performance value of the different past Preset Time section of each candidate's video is pressed different proportion weighting and summation, and the result of described summation is obtained described comprehensive concern rate divided by the historical general comment score of this candidate's video.
8. the recommend method of maximum ascendant trend video as claimed in claim 7, is characterized in that, the now nearer past Preset Time section of distance is pressed higher proportion weighted than the present past Preset Time section far away of distance.
9. the recommend method of maximum ascendant trend video as claimed in claim 7, is characterized in that, described performance value, speed formula and Acceleration Formula according to each candidate's video obtained the following concern value of this candidate's video and comprised:
Performance value, speed formula and the Acceleration Formula of the past Preset Time section different according to each candidate's video obtained the click speed of one of them past Preset Time section of each candidate's video and clicked acceleration;
According to the described click speed of each candidate's video with click concern in the future value that acceleration obtains the following Preset Time section of this candidate's video one.
10. the recommend method of maximum ascendant trend video as claimed in claim 9, is characterized in that, the trend of each candidate's video must be divided into concern in the future value with comprehensive concern rate gained on duty of this candidate's video.
11. the commending system of a maximum ascendant trend video is characterized in that, comprising:
Increment click data storehouse is for the concern situation in all video past of storage;
Candidate's video database is used for storage candidate's video and corresponding performance value;
The trend video database is used for storage ascendant trend video;
Data statistics module, the concern situation that is used for adding up all video past;
Candidate's video generation module is used for generating some candidate's videos and corresponding performance value according to the concern situation of all videos of described increment click data storehouse;
Trend video generation module is used for analyzing the ascendant trend video according to the concern situation in described increment click data storehouse, the performance value of each candidate's video in described candidate's video database and speed formula, Acceleration Formula.
12. the commending system of maximum ascendant trend video as claimed in claim 11 is characterized in that, described trend video generation module comprises:
The comprehensive unit of paying close attention to, the free concern situation of past that is used for each candidate's video of the described increment click data of accumulative total storehouse generates the in the past free historical general comment score of this candidate's video;
Comprehensive concern rate unit is for obtain the comprehensive concern rate of this candidate's video according to the performance value of historical general comment score and each candidate's video;
The following unit of paying close attention to is used for obtaining the following concern value of this candidate's video according to performance value, speed formula and the Acceleration Formula of each candidate's video;
The ascendant trend unit, be used for the trend score of obtaining this candidate's video according to concern in future value and the comprehensive concern rate of each candidate's video, sort and therefrom choose several forward candidate's videos of described trend score as the ascendant trend video the trend score of all candidate's videos is ascending.
13. the commending system of maximum ascendant trend video as claimed in claim 12 is characterized in that, described data statistics module is added up respectively watching number of times, collection number of times and commenting on number of times of all each hours in videos past.
14. the commending system of maximum ascendant trend video as claimed in claim 13, it is characterized in that, described candidate's video generation module is chosen several different past Preset Time sections, with the same described number of times of watching of each video of Preset Time section in the past, collection number of times and comment number of times are weighted summation to generate the performance value with different ratios respectively, with same descending sequence of performance value of all videos of Preset Time section in the past, choose respectively several forward videos of performance value described in different past Preset Time sections as candidate's video according to the result of described sequence.
15. the commending system of maximum ascendant trend video as claimed in claim 14, it is characterized in that, described comprehensive concern unit generates the in the past free historical general comment score of this candidate's video with described number of times, collection number of times and the comment number of times watched of each candidate's video by the different proportion weighted sum.
16. the commending system of maximum ascendant trend video as claimed in claim 15, it is characterized in that, different proportion weighting and summation are pressed with the performance value of the different past Preset Time section of each candidate's video in described comprehensive concern rate unit, and the result of described summation must be got described comprehensive concern rate divided by historical general comment.
17. the commending system of maximum ascendant trend video as claimed in claim 16, it is characterized in that, described following performance value, speed formula and the Acceleration Formula of paying close attention to the unit past Preset Time section different according to each candidate's video obtain each candidate's video in the past one of them in the past the Preset Time section click speed and click acceleration, according to the described click speed of each candidate's video with click acceleration and obtain this candidate's video in concern in the future of following Preset Time section value.
18. the commending system of maximum ascendant trend video as claimed in claim 17 is characterized in that, pays close attention to the trend score that obtains described candidate's video with comprehensive concern rate on duty the future that described ascendant trend unit will each candidate's video.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110426170 CN103164481A (en) | 2011-12-16 | 2011-12-16 | Recommendation method and system of video with largest rising trend |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110426170 CN103164481A (en) | 2011-12-16 | 2011-12-16 | Recommendation method and system of video with largest rising trend |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103164481A true CN103164481A (en) | 2013-06-19 |
Family
ID=48587575
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110426170 Pending CN103164481A (en) | 2011-12-16 | 2011-12-16 | Recommendation method and system of video with largest rising trend |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103164481A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015027380A1 (en) * | 2013-08-26 | 2015-03-05 | Google Inc. | In-market classification for online content item placement campaigns |
| CN105243105A (en) * | 2015-09-18 | 2016-01-13 | 广州酷狗计算机科技有限公司 | Content sorting method and apparatus |
| CN105635675A (en) * | 2015-12-29 | 2016-06-01 | 北京奇艺世纪科技有限公司 | Panorama playing method and device |
| CN105828181A (en) * | 2016-04-14 | 2016-08-03 | 青岛海信传媒网络技术有限公司 | Video sequencing method and device based on video on-demand volume |
| CN105872798A (en) * | 2015-12-07 | 2016-08-17 | 乐视网信息技术（北京）股份有限公司 | Implementation method and system for quickly playing video |
| CN106168980A (en) * | 2016-07-26 | 2016-11-30 | 合网络技术（北京）有限公司 | Multimedia resource recommends sort method and device |
| CN106227834A (en) * | 2016-07-26 | 2016-12-14 | 合网络技术（北京）有限公司 | The recommendation method and device of multimedia resource |
| CN106372124A (en) * | 2016-08-23 | 2017-02-01 | 达而观信息科技（上海）有限公司 | Method and system for automatically identifying potential video |
| CN106713952A (en) * | 2015-11-16 | 2017-05-24 | 北京国双科技有限公司 | Video processing method and device |
| CN106791962A (en) * | 2016-11-29 | 2017-05-31 | 聚好看科技股份有限公司 | The interest video of user determines method and video server |
| CN107330709A (en) * | 2016-04-29 | 2017-11-07 | 阿里巴巴集团控股有限公司 | Determine the method and device of destination object |
| CN108305097A (en) * | 2018-01-11 | 2018-07-20 | 湖南云连天地网络科技有限公司 | A kind of data processing method, equipment and client |
| CN109446419A (en) * | 2018-10-17 | 2019-03-08 | 武汉斗鱼网络科技有限公司 | A kind of method and device for recommending video |
| CN110427556A (en) * | 2019-07-30 | 2019-11-08 | 牡丹江师范学院 | A kind of film recommended method based on literature and art study |
| CN111460181A (en) * | 2020-04-01 | 2020-07-28 | 北京奇艺世纪科技有限公司 | Multimedia resource recommendation method, electronic device and storage medium |
| CN111510783A (en) * | 2020-04-26 | 2020-08-07 | 咪咕动漫有限公司 | Method, device, electronic equipment and storage medium for determining video exposure |
| CN112052394A (en) * | 2020-09-10 | 2020-12-08 | 上海梅斯医药科技有限公司 | Method, system, terminal device and storage medium for recommending professional content information |
| CN114519112A (en) * | 2022-01-28 | 2022-05-20 | 北京卓越乐享网络科技有限公司 | Method, apparatus, device, medium and program product for predicting multimedia object |
-
2011
- 2011-12-16 CN CN 201110426170 patent/CN103164481A/en active Pending
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015027380A1 (en) * | 2013-08-26 | 2015-03-05 | Google Inc. | In-market classification for online content item placement campaigns |
| CN105243105A (en) * | 2015-09-18 | 2016-01-13 | 广州酷狗计算机科技有限公司 | Content sorting method and apparatus |
| CN106713952A (en) * | 2015-11-16 | 2017-05-24 | 北京国双科技有限公司 | Video processing method and device |
| CN106713952B (en) * | 2015-11-16 | 2020-02-07 | 北京国双科技有限公司 | Video processing method and device |
| CN105872798A (en) * | 2015-12-07 | 2016-08-17 | 乐视网信息技术（北京）股份有限公司 | Implementation method and system for quickly playing video |
| CN105635675A (en) * | 2015-12-29 | 2016-06-01 | 北京奇艺世纪科技有限公司 | Panorama playing method and device |
| CN105828181A (en) * | 2016-04-14 | 2016-08-03 | 青岛海信传媒网络技术有限公司 | Video sequencing method and device based on video on-demand volume |
| CN105828181B (en) * | 2016-04-14 | 2019-06-25 | 聚好看科技股份有限公司 | Video sequencing method and device based on video on demand amount |
| CN107330709A (en) * | 2016-04-29 | 2017-11-07 | 阿里巴巴集团控股有限公司 | Determine the method and device of destination object |
| CN107330709B (en) * | 2016-04-29 | 2020-05-05 | 阿里巴巴集团控股有限公司 | Method and device for determining target object |
| CN106168980B (en) * | 2016-07-26 | 2020-07-28 | 阿里巴巴（中国）有限公司 | Multimedia resource recommendation sequencing method and device |
| CN106168980A (en) * | 2016-07-26 | 2016-11-30 | 合网络技术（北京）有限公司 | Multimedia resource recommends sort method and device |
| CN106227834B (en) * | 2016-07-26 | 2020-03-27 | 合一网络技术(北京)有限公司 | Multimedia resource recommendation method and device |
| CN106227834A (en) * | 2016-07-26 | 2016-12-14 | 合网络技术（北京）有限公司 | The recommendation method and device of multimedia resource |
| CN106372124A (en) * | 2016-08-23 | 2017-02-01 | 达而观信息科技（上海）有限公司 | Method and system for automatically identifying potential video |
| CN106791962A (en) * | 2016-11-29 | 2017-05-31 | 聚好看科技股份有限公司 | The interest video of user determines method and video server |
| CN108305097A (en) * | 2018-01-11 | 2018-07-20 | 湖南云连天地网络科技有限公司 | A kind of data processing method, equipment and client |
| CN108305097B (en) * | 2018-01-11 | 2021-03-30 | 湖南一起分类网络科技有限公司 | Data processing method, equipment and client |
| CN109446419B (en) * | 2018-10-17 | 2020-10-16 | 武汉斗鱼网络科技有限公司 | Method and device for recommending video |
| CN109446419A (en) * | 2018-10-17 | 2019-03-08 | 武汉斗鱼网络科技有限公司 | A kind of method and device for recommending video |
| CN110427556A (en) * | 2019-07-30 | 2019-11-08 | 牡丹江师范学院 | A kind of film recommended method based on literature and art study |
| CN110427556B (en) * | 2019-07-30 | 2022-10-11 | 牡丹江师范学院 | Film recommendation method based on literary and artistic learning |
| CN111460181A (en) * | 2020-04-01 | 2020-07-28 | 北京奇艺世纪科技有限公司 | Multimedia resource recommendation method, electronic device and storage medium |
| CN111510783A (en) * | 2020-04-26 | 2020-08-07 | 咪咕动漫有限公司 | Method, device, electronic equipment and storage medium for determining video exposure |
| CN111510783B (en) * | 2020-04-26 | 2022-06-03 | 咪咕动漫有限公司 | Method, device, electronic equipment and storage medium for determining video exposure |
| CN112052394A (en) * | 2020-09-10 | 2020-12-08 | 上海梅斯医药科技有限公司 | Method, system, terminal device and storage medium for recommending professional content information |
| CN112052394B (en) * | 2020-09-10 | 2023-11-21 | 上海梅斯医药科技有限公司 | Professional content information recommendation method, system, terminal equipment and storage medium |
| CN114519112A (en) * | 2022-01-28 | 2022-05-20 | 北京卓越乐享网络科技有限公司 | Method, apparatus, device, medium and program product for predicting multimedia object |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103164481A (en) | Recommendation method and system of video with largest rising trend | |
| US8413052B2 (en) | Bidded marketplace for applications | |
| Li et al. | On popularity prediction of videos shared in online social networks | |
| CN104699705B (en) | Information-pushing method, server and system | |
| JP5926823B2 (en) | Automatic content recommendation | |
| CN103167330A (en) | Method and system for audio/video recommendation | |
| US8825668B2 (en) | Method and apparatus for updating song playlists based on received user ratings | |
| CN106055617A (en) | Data pushing method and device | |
| US20120084155A1 (en) | Presentation of content based on utility | |
| CN111708901A (en) | Multimedia resource recommendation method and device, electronic equipment and storage medium | |
| KR20160096122A (en) | Methods and systems for creating a data-driven attribution model for assigning attribution credit to a plurality of events | |
| CN103136280A (en) | System and method of automatically searching hot point video | |
| CN107197368A (en) | Determine method and system of the user to multimedia content degree of concern | |
| JP2013033375A (en) | Information processing apparatus, information processing method, and program | |
| CN101271559A (en) | Cooperation recommending system based on user partial interest digging | |
| US10671927B1 (en) | Impression effect modeling for content items | |
| WO2009152576A1 (en) | Assessing ditigal content across a communications network | |
| CN106604137B (en) | Method and device for predicting video watching duration | |
| CN104216883A (en) | Video recommendation reason generating system and method | |
| CN103309894A (en) | User attribute-based search realization method and system | |
| CN111815375B (en) | User portrayal method and device in advertisement putting | |
| CN103002329B (en) | Device and method for predicting reach-frequency table data of network video advertisements | |
| CN102957949A (en) | Device and method for recommending video to user | |
| CN103297815A (en) | Advertisement display method and system | |
| CN105100164A (en) | Network service recommendation method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C05 | Deemed withdrawal (patent law before 1993) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130619 |