CN104954873A - Intelligent television video customizing method and intelligent television video customizing system - Google Patents
Intelligent television video customizing method and intelligent television video customizing system Download PDFInfo
- Publication number
- CN104954873A CN104954873A CN201410115778.7A CN201410115778A CN104954873A CN 104954873 A CN104954873 A CN 104954873A CN 201410115778 A CN201410115778 A CN 201410115778A CN 104954873 A CN104954873 A CN 104954873A
- Authority
- CN
- China
- Prior art keywords
- video
- information
- user
- video information
- particle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The invention discloses an intelligent television video customizing method and an intelligent television video customizing system. The intelligent television video customizing method comprises the steps of A. receiving user video customizing requirement information by a middle layer, storing the information and transmitting the information to a cloud-end server; B. collecting corresponding video information by the cloud-end server according to the received information, and returning the video information to the middle layer; C. receiving the video information by the middle layer, performing kernel space mapping and cluster analysis on the video information by means of a multi-kernel particle swarm clustering algorithm for obtaining a clustering result which comprises a plurality of kinds of video information, comparing the clustering result with the video customizing requirement information, thereby obtaining a most matched kind of video information; and D. displaying the obtained kind of video information. The video data processing algorithm which is used by the intelligent television video customizing system can effectively process high-dimensional nonlinear big data and can remarkably improve clustering precision and reduces time complexity so that an intelligent television can supply individualized services in a targeted manner, so that the intelligent television is more intelligent and user-friendly.
Description
Technical field
The present invention relates to technical field of information processing, particularly relate to a kind of intelligent television video method for customizing and system.
Background technology
The fast development of the Internet has very large impact for the development of whole intelligent television industry, and there is following advantage the Internet: use the user of the Internet many; Fast by the speed of the Internet obtaining information and transmission of information; The Internet can realize resource-sharing across time and space, can complete real-time, interactive; The Internet has feature that is personalized and hommization; Be impartial to for each user in the Internet; Unnecessary expense can be saved by the Internet.Advantage how by means of the Internet under the impact of this brute force attack in the Internet promotes the direction developing into intelligent television development of intelligent television, but make the more personalized hommization of intelligent television by the Internet at present, also there is a lot of deficiency in the technology realizing real-time, interactive and resource-sharing, user cannot be met for the requirement experiencing satisfaction and autonomous control ability
Therefore, prior art has yet to be improved and developed.
Summary of the invention
In view of above-mentioned the deficiencies in the prior art, the object of the present invention is to provide a kind of intelligent television video method for customizing and system, be intended to solve the real-time, interactive resource run in current intelligent television evolution, hommization information processing, the problems such as user's subjective initiative.
Technical scheme of the present invention is as follows:
A kind of intelligent television video method for customizing, wherein, said method comprising the steps of:
A, intermediate layer receive user video customized demand information, are sent to cloud server after this information being preserved;
This video information according to the received corresponding video information of user video customized demand information search, and is back to intermediate layer by B, cloud server;
C, intermediate layer receive the video information returned, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping and cluster analysis to this video information, obtain the cluster result including multiple category video information, this cluster result and described user video customized demand information are contrasted, obtains the category video information of mating most with described user video customized demand information;
D, obtained category video information to be shown at application interface.
Described intelligent television video method for customizing, wherein, by obtained category video information after application interface shows, is sent to described cloud server by this classification video information and saves as the characteristic behavior collection of this user.
Described intelligent television video method for customizing, wherein, also comprises after described step D:
Receive the suggestion feedback information of user for shown category video information, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping to this suggestion feedback information and cluster analysis obtains result set, the characteristic behavior collection of this user this result set and cloud server preserved is compared, obtain the characteristic information of user, utilize this characteristic information to upgrade the characteristic behavior collection of user.
Described intelligent television video method for customizing, wherein, also comprises before described steps A:
Intermediate layer receives the log-on message of user, and this log-on message is sent to cloud server;
Cloud server searches the user characteristics behavior collection whether existing and conform to this log-on message, if exist, is then shown at application interface by found user characteristics behavior collection.
Described intelligent television video method for customizing, wherein, also comprises before described steps A:
Intermediate layer obtains device id, and sends it to cloud server;
Cloud server searches the Family characteristics behavior collection whether existing and conform to this device id, if there is Family characteristics behavior collection and do not find the user characteristics behavior collection of active user, is then shown at application interface by found Family characteristics behavior collection.
Described intelligent television video method for customizing, wherein, described step C is specially:
C1, intermediate layer receive the video information returned, and set up multinuclear nuclear matrix and utilize it to carry out nuclear space mapping to described video information, obtaining nuclear space video information;
C2, obtained nuclear space video information is input in particle cluster algorithm as particle, finds optimal particle by iteration and fitness evaluation;
C3, obtained optimal particle to be input in k-means clustering algorithm as initial cluster center, to utilize it to upgrade optimal particle, obtain the cluster result including multiple category video information;
C4, this cluster result and described user video customized demand information to be contrasted, obtain the category video information of mating most with described user video customized demand information.
Described intelligent television video method for customizing, wherein, set up multinuclear nuclear matrix in described step C1 and be specially:
Employing expression formula is
multinuclear nuclear matrix, wherein
for multiple different kernel function,
for the linear combination coefficient of kernel function, when given
individual different kernel function, then to having
individual different nuclear matrix
, herein in conjunction with the knowwhy of Semidefinite Programming, use Semidefinite Programming interior point method to solve multinuclear nuclear matrix, obtain the optimum linearity combination coefficient of nuclear matrix, will by optimum linearity combination coefficient
individual nuclear matrix is added, and obtains more excellent multinuclear nuclear matrix
.
Described intelligent television video method for customizing, wherein, described step C2 is specially:
C21, obtained nuclear space video information to be input in particle cluster algorithm as particle, and to obtain N number of primary group by the fitness calculating each particle for N time;
C22, upgraded the desired positions of each particle by the current fitness that contrasts each particle and the fitness of desired positions that lives through thereof, the fitness of the desired positions lived through by the current fitness and colony contrasting each particle upgrades the overall desired positions of each particle;
C23, according to the particle cluster algorithm adjustment speed of particle and position, calculate Colony fitness variance and also judge whether it is less than predetermined threshold value, if so, then find optimal particle and perform step C3
Described intelligent television video method for customizing, wherein, described step C3 is specially:
C31, using the initial cluster center of obtained optimal particle as k-means clustering algorithm, and determine the clustering of corresponding optimal particle according to most adjacent principle;
C32, calculate new cluster centre according to determined clustering, and replace initial cluster center with it;
C33, judge cluster centre whether also in change or whether reach maximum iteration time, if cluster centre no longer changes or reaches maximum iteration time, then perform step C34;
C34, judge that whether the fitness of new cluster centre is due to previous particle, if so, then upgrades optimal particle, obtain the cluster result including multiple category video information.
A kind of intelligent television video custom-built system, wherein, described system comprises:
Intermediate layer, for receiving user video customized demand information, and is sent to cloud server after this information being preserved; And accept the video information that cloud server returns according to user video customized demand information, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping and cluster analysis to this video information, obtain the cluster result including multiple category video information, this cluster result and described user video customized demand information are contrasted, obtains the category video information of mating most with described user video customized demand information;
Cloud server, for according to the received corresponding video information of user video customized demand information search, and is back to intermediate layer by this video information;
Application interface display module, for showing obtained category video information.
Beneficial effect: the invention provides a kind of intelligent television video method for customizing and system, this system can realize user and freely control and real-time, interactive, have good ageing, and the video data Processing Algorithm that system uses effectively can process the large-scale data of high dimensional nonlinear, to be significantly improved clustering precision compared to its tool of traditional algorithm, the effect, the method for the present invention that reduce time complexity make intelligent television provide personalized service more targetedly, make its more intelligent and hommization.Also the interconnected networking development of intelligent television has been promoted.
Accompanying drawing explanation
Fig. 1 is intelligent television video method for customizing flow chart in the specific embodiment of the invention.
Fig. 2 is the schematic diagram of intelligent television video custom-built system application interface in the specific embodiment of the invention.
Fig. 3 is the schematic diagram of intelligent television video custom-built system user login interface in the specific embodiment of the invention.
Fig. 4 is the schematic diagram of intelligent television video custom-built system user register interface in the specific embodiment of the invention.
Fig. 5 is the concrete grammar flow chart of step S300 in Fig. 1.
Fig. 6 is the concrete grammar flow chart of step S320 in Fig. 5.
Fig. 7 is the concrete grammar flow chart of step S330 in Fig. 5.
Fig. 8 is multinuclear population clustering algorithm flow chart in the specific embodiment of the invention.
Fig. 9 is the theory diagram of intelligent television video custom-built system in the specific embodiment of the invention.
Embodiment
The invention provides a kind of intelligent television video method for customizing and system, for making object of the present invention, technical scheme and effect clearly, clearly, the present invention is described in more detail below.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
A kind of intelligent television video method for customizing as shown in Figure 1, wherein, said method comprising the steps of:
S100, intermediate layer receive user video customized demand information, are sent to cloud server after this information being preserved.
This video information according to the received corresponding video information of user video customized demand information search, and is back to intermediate layer by S200, cloud server.
Intermediate layer is also referred to as application service layer, and specially the realization of department's service logic, cloud server stores all video informations, such as, can comprise the video information of Tengxun, Sohu, the strange skill of love, excellent current all video website such as extremely.The complete video information that video is corresponding specifically can comprise: source video sequence, video name, video format, video profile, video director, video cast, video playback address, video area, video type etc., video information data of the present invention is all that a complete video information is converted into one group of data, the coding display of this data acquisition, finally can be undertaken resolving by coding obtaining the complete video information of this group.
S300, intermediate layer receive the video information returned, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping and cluster analysis to this video information, obtain the cluster result including multiple category video information, this cluster result and described user video customized demand information are contrasted, obtains the category video information of mating most with described user video customized demand information.
In the present embodiment, described multinuclear population clustering algorithm be merged the algorithm comprising multinuclear nuclear matrix algorithm, particle cluster algorithm and k-means clustering algorithm general designation (hereinafter can introduce in detail these three kinds calculate ratio juris and the present embodiment organically blend for these three kinds of algorithms after embody rule).It is worth mentioning that, the most similar cluster result of a lot of feature is just obtained by multinuclear population clustering algorithm, need by mating with user video customized demand information, category video information that is the most similar with user video customized demand information, that mate most could be determined, this classification video information is returned to user.
S400, obtained category video information to be shown at application interface.
Further, by obtained category video information after application interface shows, this classification video information is sent to described cloud server and saves as the characteristic behavior collection of this user.
Therefore, also comprise before described step S100:
Intermediate layer receives the log-on message of user, and this log-on message is sent to cloud server;
Cloud server searches the user characteristics behavior collection whether existing and conform to this log-on message, if exist, is then shown at application interface by found user characteristics behavior collection.If cloud server does not find corresponding user characteristics behavior collection, after then user enters system, system can show the high video information of the program request amount of some hot topics automatically to user, if user loses interest in these video informations, custom video information entry can be entered, namely perform step S100.
In addition, after logging in system by user, before intermediate layer receives user video customized demand information, further comprising the steps of:
Intermediate layer obtains device id, and sends it to cloud server;
Cloud server searches the Family characteristics behavior collection whether existing and conform to this device id, if there is Family characteristics behavior collection and do not find the user characteristics behavior collection of active user, is then shown at application interface by found Family characteristics behavior collection.
Therefore, in video custom-built system of the present invention, ID is registered to should user characteristics behavior collection by user, by the device id of intelligent television to using the feature of colony by intelligent television, i.e. corresponding family behavioural characteristic collection, described user characteristics behavior collection is the video preference information set for individual subscriber, and described family behavioural characteristic collection is the video preference information set for all members of family bound with this device id.
Concrete application interface can as shown in Figure 2, and system application interface comprises login button, custom video information entry button and feedback opinion entrance, enters login interface, as shown in Figure 3 by clicking login button.The user used first needs first to register, and after clicking new user's registration of log-in interface, enter register interface, as shown in Figure 4, the information of registration has: device id, user name, password.Wherein device id is by the direct automatic acquisition of video device interface, obtains device id mainly in order to identify that user uses the characteristic of family.Automatically log in after having registered and turn back to the first interface of application, the local display user's information of the now login at first interface.After logging in first, the logging status of system meeting recording user, facilitates user to use later.After user enters application, first three time to enter application and automatically shows the high video information of the program request amount of some hot topics to user, if user loses interest in these video informations, can according to the customization requirement of oneself, click custom video information entry, enter custom video information interface, input the video customized demand information of oneself, preferably, user video customized demand information is except generally common video searching information, further comprises other demand informations (such as plot requirement, picture texture, personage's setting etc.) of user
The present embodiment utilizes the processing procedure to video information of multinuclear population clustering algorithm to be briefly first by the concept of kernel function and the feature of kernel function, multiple different kernel function is carried out linear combination, be combined to form multinuclear nuclear matrix, the knowledge of Semidefinite Programming is used to solve the linear combination coefficient obtaining multinuclear nuclear matrix, and substituted into by this linear combination coefficient in the multinuclear nuclear matrix be combined to form, and then obtain the multinuclear nuclear matrix after solving.Use multinuclear nuclear matrix to carry out nuclear space mapping to data message, obtain the data set of nuclear space, use multinuclear population clustering algorithm to carry out cluster to nuclear space data set, obtain different category combinations.After obtaining different classifications, require to return to the corresponding category video information of user according to the customization of user.Improve and return the accuracy of video and the satisfaction of user.
In preferred embodiment as shown in Figure 5, described step S300 is specially:
S310, intermediate layer receive the video information returned, and set up multinuclear nuclear matrix and utilize it to carry out nuclear space mapping to described video information, obtaining nuclear space video information.
The video information general data returned by cloud server is huge, and classification is complicated, and these video informations have higher-dimension and nonlinear feature (namely data complexity is very high).Multinuclear population clustering algorithm of the present invention uses conventional kernel function gaussian kernel function, Polynomial kernel function and perceptron kernel function are carried out nuclear matrix linear combination and are obtained multinuclear nuclear matrix, because nuclear space can by the data of high dimensional nonlinear by mapping the dimension reducing data, make it become linear separability simultaneously, and each kernel function has the feature of oneself, the Data distribution8 obtained after mapping is also different, therefore, the multiple kernel function advantage of this nuclear matrix set, linear combination becomes more excellent kernel function, in specific embodiment, set up multinuclear nuclear matrix to be specially:
Employing expression formula is
multinuclear nuclear matrix, wherein
for multiple different kernel function,
for the linear combination coefficient of kernel function, when given
individual different kernel function, then to having
individual different nuclear matrix
, herein in conjunction with the knowwhy of Semidefinite Programming, use Semidefinite Programming interior point method to solve multinuclear nuclear matrix, obtain the optimum linearity combination coefficient of nuclear matrix, will by optimum linearity combination coefficient
individual nuclear matrix is added, and obtains more excellent multinuclear nuclear matrix
.
Then use multinuclear nuclear matrix obtained above to carry out nuclear space mapping to video information, obtain nuclear space video information.Again in conjunction with the knowledge of theorem in Euclid space range formula, apply it on nuclear space, obtain the core distance of nuclear space video set at nuclear space.In a word, achieving dimensionality reduction and making video information data become linear separability by multinuclear nuclear matrix.
S320, obtained nuclear space video information is input in particle cluster algorithm as particle, finds optimal particle by iteration and fitness evaluation.
S330, obtained optimal particle to be input in k-means clustering algorithm as initial cluster center, to utilize it to upgrade optimal particle, obtain the cluster result including multiple category video information.
Particle cluster algorithm is a kind of evolvement method based on swarm intelligence.Each potential solution of problem is the particle in search volume, and each particle has its corresponding speed, position and a fitness determined by target function, and algorithm evaluates the quality of particle by fitness.Algorithm first initialization a group random particles, then finds optimal solution by iteration.In each iteration, particle upgrades oneself by tracking two " extreme values ": the optimal solution being particle itself and finding, i.e. an individual extreme value
; Another is the optimal solution that whole population finds at present, is referred to as global extremum
.Particle finds above after two extreme values, just upgrades oneself speed and position according to formula (1) and formula (2):
（1）
（2）
Wherein,
the speed of current particle,
it is the current location of particle.
,
it is the dimension of current spatial.
,
the random number between [0,1],
with
for Studying factors, usually get
.
weight coefficient, generally value between 0.1 to 0.9.If
carrying out and linearly reduce with algorithm iteration, will significantly improve convergence of algorithm performance.If
for maximum weighted coefficient,
for minimum weight coefficient,
for current iteration number of times,
for algorithm iteration total degree, then have,
（3）
Use k-means clustering algorithm to optimize particle cluster algorithm, need the convergence opportunity determining particle cluster algorithm, research finds the state can following the tracks of population according to the overall variation of all particle fitness, and whether evaluation algorithm restrains.
If the number of particles of population is
,
be
the fitness of individual particle,
for the average fitness that population is current, the Colony fitness variance of population
be defined as follows:
（4）
Fitness variance can reflect " convergence " degree of all particles in population.
less, then population is tending towards convergence; If
be zero, then colony's fitness is almost identical, and particle swarm optimization algorithm is absorbed in Premature Convergence or reaches global convergence; Otherwise the different population of fitness is then in the random search stage.
The present embodiment processes data together with k-means clustering algorithm by using particle cluster algorithm, the incision one of k-means clustering algorithm is the Premature Convergence avoiding particle cluster algorithm, be exactly that particle cluster algorithm has Global treatment in addition, the process of local needs to introduce k-means clustering algorithm and determines to also have particle cluster algorithm can determine the initial center of k-means clustering algorithm.If be used alone particle cluster algorithm to enter convergence too early and cause result inaccuracy; The selection being used alone k-means clustering algorithm initial center is not necessarily very accurate, causes result also accurate not, therefore makes respective advantage evade the defect of the other side by the combination of two algorithms.
S340, this cluster result and described user video customized demand information to be contrasted, obtain the category video information of mating most with described user video customized demand information.
In preferred embodiment as shown in Figure 6, described step S320 is specially:
S321, obtained nuclear space video information to be input in particle cluster algorithm as particle, and to obtain N number of primary group by the fitness calculating each particle for N time.
S322, upgraded the desired positions of each particle by the current fitness that contrasts each particle and the fitness of desired positions that lives through thereof, the fitness of the desired positions lived through by the current fitness and colony contrasting each particle upgrades the overall desired positions of each particle.
S323, according to the particle cluster algorithm adjustment speed of particle and position, calculate Colony fitness variance and also judge whether it is less than predetermined threshold value, if so, then find optimal particle and perform step S300.
In preferred embodiment as shown in Figure 7, described step S330 is specially:
S331, using the initial cluster center of obtained optimal particle as k-means clustering algorithm, and determine the clustering of corresponding optimal particle according to most adjacent principle;
S332, calculate new cluster centre according to determined clustering, and replace initial cluster center with it;
S333, judge cluster centre whether also in change or whether reach maximum iteration time, if cluster centre no longer changes or reaches maximum iteration time, then perform step S334;
S334, judge that whether the fitness of new cluster centre is due to previous particle, if so, then upgrades optimal particle, obtain the cluster result including multiple category video information.
The multinuclear population clustering algorithm flow process of the present embodiment as shown in Figure 8, its step is as follows:
The video information that S1, reception cloud server return.
S2, assembly multinuclear nuclear matrix, carry out dimension-reduction treatment by nuclear space mapping pair video information, obtain nuclear space video information.
S3, nuclear space video information are input in particle cluster algorithm as particle, calculated the fitness of each particle, obtain N number of primary group by N time.Be first a certain class by each particle random division, as initial clustering, and calculate all kinds of cluster centres, position encoded as primary, calculate the fitness of particle, simultaneously as the personal best particle of particle, and the speed of random initializtion particle.Repeatedly carry out
secondary, symbiosis becomes
individual primary group.
S4, the contrast fitness of particle and its fitness of desired positions of living through, the more desired positions of new particle.
S5, the contrast fitness of particle and its fitness of desired positions of living through, the more overall desired positions of new particle.
S6, according to the particle cluster algorithm adjustment position of particle and speed.
Whether S7, colony's fitness method are less than defined threshold, if so, then perform step S8, if not, then return step S6.
Judge whether current particle group arrives convergence state, if Colony fitness variance according to above-mentioned formula (4)
be less than defined threshold, then particle be further processed.
S8, obtain optimal particle, use k-means clustering algorithm to be optimized.Select
individual optimal particle carries out K-means local search algorithm, avoids entering Premature Convergence trap.
S9, using optimal particle as initial cluster center, according to arest neighbors rule, determine should the clustering of particle.
S10, calculate new cluster centre according to clustering, replace initial cluster center.Because K-means clustering algorithm has stronger local search ability, the convergence rate therefore introducing the particle cluster algorithm after the optimization of K-means clustering algorithm can improve greatly.
S11, whether reach termination condition (cluster centre whether no longer change or whether reach maximum iteration time), if so, then perform step S12, if not, then perform and return step S10.
S12, contrast the fitness of new cluster centre.
S13, whether optimum.Whether the fitness judging new cluster centre is optimized, if so, then performs step S14, if not, then returns step S10.Judge the fitness of new cluster centre, if more excellent than previous particle, then more new particle perform step S14, terminates algorithm.
S14, algorithm terminate, and obtain optimal particle.
In preferred embodiment, also comprise after described step S400:
Receive the suggestion feedback information of user for shown category video information, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping to this suggestion feedback information and cluster analysis obtains result set, the characteristic behavior collection of this user this result set and cloud server preserved is compared, obtain the characteristic information of user, utilize this characteristic information to upgrade the characteristic behavior collection of user.And the characteristic behavior collection of user after upgrading is saved in cloud server, so that it is preliminary reference when this user returns video information as cloud server again, make video customization flow process more and more perfect.
Concrete, user obtains the video information that server end returns, field feedback entrance is clicked after browsing, enter feedback opinion interface, can by personal experience's suggestion (such as to the satisfaction returning video information, description of contents need be improved, demand for oneself is marked to each point of return data, scoring reason etc.) feed back to intermediate layer by suggestion feedback column, intermediate layer utilizes multinuclear population clustering algorithm to process this suggestion feedback information equally, and utilize the result set obtained to upgrade the characteristic behavior collection of user, thus realize the real-time, interactive of user and system.By processing the feedback opinion information of user in real time, and carrying out analyzing and processing for feedback opinion information, summing up the characteristic behavior of user and the global feature behavior of subscriber household, system can be made more targetedly for user provides Video service.
A kind of intelligent television video custom-built system as shown in Figure 9, wherein, described system comprises:
Intermediate layer 100, for receiving user video customized demand information, and is sent to cloud server after this information being preserved; And accept the video information that cloud server returns according to user video customized demand information, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping and cluster analysis to this video information, obtain the cluster result including multiple category video information, this cluster result and described user video customized demand information are contrasted, obtain the category video information of mating most with described user video customized demand information, specifically as described in step S100 and S300.
Cloud server 200, for according to the received corresponding video information of user video customized demand information search, and is back to intermediate layer, specifically as described in step S200 by this video information.
Application interface display module 300, for showing obtained category video information, specifically as described in step S400.
The present embodiment provides a kind of intelligent television video method for customizing and system, this system can realize user and freely control and real-time, interactive, have good ageing, and the video data Processing Algorithm that system uses effectively can process the large-scale data of high dimensional nonlinear, to be significantly improved clustering precision compared to its tool of traditional algorithm, the effect, the method for the present invention that reduce time complexity make intelligent television provide personalized service more targetedly, make its more intelligent and hommization.Also the interconnected networking development of intelligent television has been promoted.
Should be understood that, application of the present invention is not limited to above-mentioned citing, for those of ordinary skills, can be improved according to the above description or convert, and all these improve and convert the protection range that all should belong to claims of the present invention.
Claims (10)
1. an intelligent television video method for customizing, is characterized in that, said method comprising the steps of:
A, intermediate layer receive user video customized demand information, are sent to cloud server after this information being preserved;
This video information according to the received corresponding video information of user video customized demand information search, and is back to intermediate layer by B, cloud server;
C, intermediate layer receive the video information returned, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping and cluster analysis to this video information, obtain the cluster result including multiple category video information, this cluster result and described user video customized demand information are contrasted, obtains the category video information of mating most with described user video customized demand information;
D, obtained category video information to be shown at application interface.
2. intelligent television video method for customizing according to claim 1, it is characterized in that, by obtained category video information after application interface shows, this classification video information is sent to described cloud server and saves as the characteristic behavior collection of this user.
3. intelligent television video method for customizing according to claim 2, is characterized in that, also comprise after described step D:
Receive the suggestion feedback information of user for shown category video information, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping to this suggestion feedback information and cluster analysis obtains result set, the characteristic behavior collection of this user this result set and cloud server preserved is compared, obtain the characteristic information of user, utilize this characteristic information to upgrade the characteristic behavior collection of user.
4. intelligent television video method for customizing according to claim 2, is characterized in that, also comprise before described steps A:
Intermediate layer receives the log-on message of user, and this log-on message is sent to cloud server;
Cloud server searches the user characteristics behavior collection whether existing and conform to this log-on message, if exist, is then shown at application interface by found user characteristics behavior collection.
5. intelligent television video method for customizing according to claim 4, is characterized in that, also comprise before described steps A:
Intermediate layer obtains device id, and sends it to cloud server;
Cloud server searches the Family characteristics behavior collection whether existing and conform to this device id, if there is Family characteristics behavior collection and do not find the user characteristics behavior collection of active user, is then shown at application interface by found Family characteristics behavior collection.
6. intelligent television video method for customizing according to claim 1, it is characterized in that, described step C is specially:
C1, intermediate layer receive the video information returned, and set up multinuclear nuclear matrix and utilize it to carry out nuclear space mapping to described video information, obtaining nuclear space video information;
C2, obtained nuclear space video information is input in particle cluster algorithm as particle, selects optimal particle by iteration and fitness evaluation;
C3, selected optimal particle to be input in k-means clustering algorithm as initial cluster center, to utilize it to upgrade optimal particle, obtain the cluster result including multiple category video information;
C4, this cluster result and described user video customized demand information to be contrasted, obtain the category video information of mating most with described user video customized demand information.
7. intelligent television video method for customizing according to claim 6, is characterized in that, sets up multinuclear nuclear matrix and be specially in described step C1:
Employing expression formula is
multinuclear nuclear matrix, wherein
for multiple different kernel function,
for the linear combination coefficient of kernel function, when given
individual different kernel function, then to having
individual different nuclear matrix
, will by optimum linearity combination coefficient
individual nuclear matrix is added, and obtains more excellent multinuclear nuclear matrix
.
8. intelligent television video method for customizing according to claim 6, it is characterized in that, described step C2 is specially:
C21, obtained nuclear space video information to be input in particle cluster algorithm as particle, and to obtain N number of primary group by the fitness calculating each particle for N time;
C22, upgraded the desired positions of each particle by the current fitness that contrasts each particle and the fitness of desired positions that lives through thereof, the fitness of the desired positions lived through by the current fitness and colony contrasting each particle upgrades the overall desired positions of each particle;
C23, according to the particle cluster algorithm adjustment speed of particle and position, calculate Colony fitness variance and also judge whether it is less than predetermined threshold value, if so, then select optimal particle and perform step C3.
9. intelligent television video method for customizing according to claim 6, it is characterized in that, described step C3 is specially:
C31, using the initial cluster center of selected optimal particle as k-means clustering algorithm, and determine the clustering of corresponding optimal particle according to arest neighbors rule;
C32, calculate new cluster centre according to determined clustering, and replace initial cluster center with it;
C33, judge cluster centre whether also in change or whether reach maximum iteration time, if cluster centre no longer changes or reaches maximum iteration time, then perform step C34;
C34, judge whether the fitness of new cluster centre is better than previous particle, if so, then upgrades optimal particle, obtain the cluster result including multiple category video information.
10. an intelligent television video custom-built system, is characterized in that, described system comprises:
Intermediate layer, for receiving user video customized demand information, and is sent to cloud server after this information being preserved; And accept the video information that cloud server returns according to user video customized demand information, multinuclear population clustering algorithm is utilized to carry out nuclear space mapping and cluster analysis to this video information, obtain the cluster result including multiple category video information, this cluster result and described user video customized demand information are contrasted, obtains the category video information of mating most with described user video customized demand information;
Cloud server, for according to the received corresponding video information of user video customized demand information search, and is back to intermediate layer by this video information;
Application interface display module, for showing obtained category video information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410115778.7A CN104954873B (en) | 2014-03-26 | 2014-03-26 | A kind of smart television video method for customizing and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410115778.7A CN104954873B (en) | 2014-03-26 | 2014-03-26 | A kind of smart television video method for customizing and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104954873A true CN104954873A (en) | 2015-09-30 |
| CN104954873B CN104954873B (en) | 2018-10-26 |
Family
ID=54169159
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410115778.7A Active CN104954873B (en) | 2014-03-26 | 2014-03-26 | A kind of smart television video method for customizing and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104954873B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105956113A (en) * | 2016-05-05 | 2016-09-21 | 南京邮电大学 | High-dimensional clustering method of video data mining on the basis of particle swarm optimization |
| CN106570014A (en) * | 2015-10-09 | 2017-04-19 | 阿里巴巴集团控股有限公司 | Method and device for determining home attribute information of user |
| CN108133121A (en) * | 2018-02-24 | 2018-06-08 | 北京科技大学 | The method of piezoelectric transducer port equivalent admittance circuit parameter estimation |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030097196A1 (en) * | 2001-11-13 | 2003-05-22 | Koninklijke Philips Electronics N.V. | Method and apparatus for generating a stereotypical profile for recommending items of interest using item-based clustering |
| CN101634996A (en) * | 2009-08-13 | 2010-01-27 | 浙江大学 | Individualized video sequencing method based on comprehensive consideration |
| US20110295892A1 (en) * | 2010-05-25 | 2011-12-01 | General Electric Company | System and method for web mining and clustering |
| US20120246161A1 (en) * | 2011-03-24 | 2012-09-27 | Kabushiki Kaisha Toshiba | Apparatus and method for recommending information, and non-transitory computer readable medium thereof |
| CN102780920A (en) * | 2011-07-05 | 2012-11-14 | 上海奂讯通信安装工程有限公司 | Television program recommending method and system |
| CN102999493A (en) * | 2011-09-08 | 2013-03-27 | 百度在线网络技术（北京）有限公司 | Method and device for achieving video resource recommendation |
| CN103220555A (en) * | 2013-03-27 | 2013-07-24 | 深圳创维数字技术股份有限公司 | Method, device and system for classifying digital television users |
| EP2677758A1 (en) * | 2012-06-19 | 2013-12-25 | Thomson Licensing | Mind opening content recommending system |
| CN103544206A (en) * | 2013-07-16 | 2014-01-29 | Tcl集团股份有限公司 | Method and system for achieving individualized recommendations |
-
2014
- 2014-03-26 CN CN201410115778.7A patent/CN104954873B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030097196A1 (en) * | 2001-11-13 | 2003-05-22 | Koninklijke Philips Electronics N.V. | Method and apparatus for generating a stereotypical profile for recommending items of interest using item-based clustering |
| CN101634996A (en) * | 2009-08-13 | 2010-01-27 | 浙江大学 | Individualized video sequencing method based on comprehensive consideration |
| US20110295892A1 (en) * | 2010-05-25 | 2011-12-01 | General Electric Company | System and method for web mining and clustering |
| US20120246161A1 (en) * | 2011-03-24 | 2012-09-27 | Kabushiki Kaisha Toshiba | Apparatus and method for recommending information, and non-transitory computer readable medium thereof |
| CN102780920A (en) * | 2011-07-05 | 2012-11-14 | 上海奂讯通信安装工程有限公司 | Television program recommending method and system |
| CN102999493A (en) * | 2011-09-08 | 2013-03-27 | 百度在线网络技术（北京）有限公司 | Method and device for achieving video resource recommendation |
| EP2677758A1 (en) * | 2012-06-19 | 2013-12-25 | Thomson Licensing | Mind opening content recommending system |
| CN103220555A (en) * | 2013-03-27 | 2013-07-24 | 深圳创维数字技术股份有限公司 | Method, device and system for classifying digital television users |
| CN103544206A (en) * | 2013-07-16 | 2014-01-29 | Tcl集团股份有限公司 | Method and system for achieving individualized recommendations |
Non-Patent Citations (1)
| Title |
|---|
| 陶新民: "一种改进的粒子群和K均值混合聚类算法", 《电子与信息学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570014A (en) * | 2015-10-09 | 2017-04-19 | 阿里巴巴集团控股有限公司 | Method and device for determining home attribute information of user |
| CN105956113A (en) * | 2016-05-05 | 2016-09-21 | 南京邮电大学 | High-dimensional clustering method of video data mining on the basis of particle swarm optimization |
| CN105956113B (en) * | 2016-05-05 | 2019-05-31 | 南京邮电大学 | Video data digging High Dimensional Clustering Analysis method based on particle group optimizing |
| CN108133121A (en) * | 2018-02-24 | 2018-06-08 | 北京科技大学 | The method of piezoelectric transducer port equivalent admittance circuit parameter estimation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104954873B (en) | 2018-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020094060A1 (en) | Recommendation method and apparatus | |
| WO2020135535A1 (en) | Recommendation model training method and related apparatus | |
| US9864951B1 (en) | Randomized latent feature learning | |
| CN113468227B (en) | Information recommendation method, system, equipment and storage medium based on graph neural network | |
| US11068285B2 (en) | Machine-learning models applied to interaction data for determining interaction goals and facilitating experience-based modifications to interface elements in online environments | |
| US10324937B2 (en) | Using combined coefficients for viral action optimization in an on-line social network | |
| CN108470075A (en) | A kind of socialization recommendation method of sequencing-oriented prediction | |
| US20150278218A1 (en) | Method and system to determine a category score of a social network member | |
| CA3062119A1 (en) | Method and device for setting sample weight, and electronic apparatus | |
| CN110990624B (en) | Video recommendation method, device, equipment and storage medium | |
| CN113360777B (en) | Content recommendation model training method, content recommendation method and related equipment | |
| CN116244513B (en) | Random group POI recommendation method, system, equipment and storage medium | |
| US20160086086A1 (en) | Multi-media content-recommender system that learns how to elicit user preferences | |
| CN111797320A (en) | Data processing method, device, equipment and storage medium | |
| CN110909258B (en) | Information recommendation method, device, equipment and storage medium | |
| CN104954873A (en) | Intelligent television video customizing method and intelligent television video customizing system | |
| US20150278836A1 (en) | Method and system to determine member profiles for off-line targeting | |
| CN111369324B (en) | Target information determining method, device, equipment and readable storage medium | |
| CN117435819A (en) | Method, system and storage medium for recommending interest points through space-time dynamic perception | |
| CN109190040A (en) | Personalized recommendation method and device based on coevolution | |
| CN111861674B (en) | Product recommendation method and system | |
| CN104636489B (en) | The treating method and apparatus of attribute data is described | |
| CN115129945A (en) | Graph structure contrast learning method, equipment and computer storage medium | |
| CN114596108A (en) | Object recommendation method and device, electronic equipment and storage medium | |
| CN117556149B (en) | Resource pushing method, device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |