Multimedia resource recommends method and deviceTechnical field
The present invention relates to MultiMedia Field, particularly relate to a kind of multimedia resource and recommend method and device.
Background technology
Along with the development of Internet technology, (Over The Top refers to by interconnection OTT based on internet video businessNet provides a user with various application service), the shape such as IPTV (Interactive Personality TV, IPTV)Become quick growth trend.Meanwhile, the intelligent terminal such as intelligent television, mobile phone and panel computer has been popular universal, everyoneMay have multiple terminal video playback equipment, the multi-screen epoch have arrived.
In prior art, main method user being carried out to video recommendations is, in the viewing to user of the single clientBehavior is analyzed, and then the viewing behavior according to user carries out the recommendation of user individual video.Use above-mentioned recommendation method,User cannot be covered and watch behavior in all kinds terminal, be likely to result in recommendation results and be not suitable for.
Summary of the invention
Technical problem
In view of this, the technical problem to be solved in the present invention is to provide a kind of multimedia resource recommendation method, to be applicable toThe multimedia resource of all kinds terminal is recommended.
Solution
In order to solve above-mentioned technical problem, according to one embodiment of the invention, it is provided that a kind of multimedia resource recommendation sideMethod, including:
In the case of the request receiving the browsing objective resource that targeted customer initiates at target terminal, obtain with describedThe multimedia resource each to be recommended that target resource is relevant；
Type according to described target terminal and various terminal user behavior data in preset time period, obtain each instituteState multimedia resource characteristic of correspondence to be recommended；
According to each described multimedia resource characteristic of correspondence to be recommended, each described multimedia resource to be recommended is arrangedSequence, and generate multimedia resource recommendation information according to ranking results.
For said method, in a kind of possible implementation, according to type and the various terminal of described target terminalUser behavior data in preset time period, obtains each described multimedia resource characteristic of correspondence to be recommended, including:
First use relevant to each described multimedia resource to be recommended in preset time period is obtained from various terminalsFamily behavioral data；
Described first user behavioral data is carried out tagsort and form collator, the first user behavior after being arrangedData；
According to the type of described target terminal, the first user behavioral data after described arrangement obtains each described in wait to push awayRecommending multimedia resource characteristic of correspondence, described multimedia resource characteristic of correspondence to be recommended includes described multimedia resource to be recommendedCorresponding terminal feature and resource characteristic.
For said method, in a kind of possible implementation, according to type and the various terminal of described target terminalUser behavior data in preset time period, obtains each described multimedia resource characteristic of correspondence to be recommended, including:
Second user behavior data relevant to each multimedia resource in preset time period is obtained from various terminals；
Described second user behavior data is carried out tagsort and form collator, the second user behavior after being arrangedData；
Type according to described target terminal and each described multimedia resource to be recommended, the second user after described arrangementBehavioral data obtains each described multimedia resource characteristic of correspondence to be recommended, described multimedia resource characteristic of correspondence to be recommendedThe terminal feature corresponding including described multimedia resource to be recommended and resource characteristic.
For said method, in a kind of possible implementation, corresponding according to each described multimedia resource to be recommendedFeature, is ranked up each described multimedia resource to be recommended, including:
Following formula 1 and following formula 2 is used to calculate the click probability of each described multimedia resource to be recommended,
Wherein, i represents ith feature, aiRepresent the weight coefficient that ith feature is corresponding, xiRepresent that ith feature is correspondingEigenvalue, N represents the number of feature, and f (x) represents each feature characteristic of correspondence that each described multimedia resource to be recommended includesThe result that value and weight coefficient are sued for peace after seeking product again, score (v) represents the click that each described multimedia resource to be recommended is correspondingProbability；
According to the click probability that each described multimedia resource to be recommended is corresponding, each described multimedia resource to be recommended is carried outSequence, and according to ranking results, choose from each described multimedia resource to be recommended and meet pre-conditioned each alternative multimediaResource.
For said method, in a kind of possible implementation, generate multimedia resource recommendation according to ranking resultsBreath, including:
To each described alternative multimedia resource, use in following steps one or more deletes；
Described targeted customer is deleted the most browsed many in preset time period from each described alternative multimedia resourceMedia resource；
The multimedia resource exceeding preset number in same information channel is deleted from each described alternative multimedia resource；
The multimedia resource exceeding preset number in same interest tags is deleted from each described alternative multimedia resource.
In order to solve above-mentioned technical problem, according to another embodiment of the present invention, it is provided that a kind of multimedia resource is recommendedDevice, including:
Source obtaining module to be recommended, for receiving browsing objective resource that targeted customer initiates at target terminalIn the case of request, obtain each to be recommended multimedia resource relevant to described target resource；
Feature acquisition module, is connected with described source obtaining module to be recommended, for the type according to described target terminalWith various terminals user behavior data in preset time period, obtain each described multimedia resource characteristic of correspondence to be recommended；
Recommendation information generation module, is connected with described feature acquisition module, for providing according to each described multimedia to be recommendedSource characteristic of correspondence, is ranked up each described multimedia resource to be recommended, and pushes away according to ranking results generation multimedia resourceRecommend information.
For said apparatus, in a kind of possible implementation, described feature acquisition module, including:
First data capture unit is described to be recommended many with each for obtain from various terminals in preset time periodThe first user behavioral data that media resource is relevant；
First taxonomic revision unit, is connected with described first data capture unit, for described first user behavior numberFirst user behavioral data according to carrying out tagsort and form collator, after being arranged；
Fisrt feature acquiring unit, is connected with described first taxonomic revision unit, for the class according to described target terminalType, obtains each described multimedia resource characteristic of correspondence to be recommended the first user behavioral data after described arrangement, describedMultimedia resource characteristic of correspondence to be recommended includes terminal feature and the resource characteristic that described multimedia resource to be recommended is corresponding.
For said apparatus, in a kind of possible implementation, described feature acquisition module, including:
Second data capture unit, for obtain from various terminals in preset time period with each multimedia resource phaseThe second user behavior data closed；
Second taxonomic revision unit, is connected with described second data capture unit, for described second user behavior numberThe second user behavior data according to carrying out tagsort and form collator, after being arranged；
Second feature acquiring unit, is connected with described second taxonomic revision unit, for the class according to described target terminalType and each described multimedia resource to be recommended, obtain each described to be recommended many second user behavior data after described arrangementMedia resource characteristic of correspondence, described multimedia resource characteristic of correspondence to be recommended includes that described multimedia resource to be recommended is correspondingTerminal feature and resource characteristic.
For said apparatus, in a kind of possible implementation, described recommendation information generation module, including:
Click on probability calculation unit, for using following formula 1 and following formula 2 to calculate the point of each described multimedia resource to be recommendedHit probability,
Wherein, i represents ith feature, aiRepresent the weight coefficient that ith feature is corresponding, xiRepresent that ith feature is correspondingEigenvalue, N represents the number of feature, and f (x) represents each feature characteristic of correspondence that each described multimedia resource to be recommended includesThe result that value and weight coefficient are sued for peace after seeking product again, score (v) represents the click that each described multimedia resource to be recommended is correspondingProbability；
Alternative resource acquiring unit, is connected with described click probability calculation unit, for according to each described many matchmakers to be recommendedThe click probability that body resource is corresponding, is ranked up each described multimedia resource to be recommended, and according to ranking results, from each describedMultimedia resource to be recommended is chosen and meets pre-conditioned each alternative multimedia resource.
For said apparatus, in a kind of possible implementation, described recommendation information generation module, also include:
Deleting unit, for each described alternative multimedia resource, use in following steps one or more deletesRemove；
Described targeted customer is deleted the most browsed many in preset time period from each described alternative multimedia resourceMedia resource；
The multimedia resource exceeding preset number in same information channel is deleted from each described alternative multimedia resource；
The multimedia resource exceeding preset number in same interest tags is deleted from each described alternative multimedia resource.
Beneficial effect
The multimedia resource of the embodiment of the present invention recommends method, can obtain various terminal user in preset time periodBehavioral data, it is possible to user is at all kinds terminal navigation patterns in covering, such that it is able to according to the characteristic of current target terminal, exampleSuch as screen size, network condition and viewing scene etc., user is carried out the resource recommendation of personalization.
According to below with reference to the accompanying drawings detailed description of illustrative embodiments, the further feature of the present invention and aspect being becomeClear.
Accompanying drawing explanation
The accompanying drawing of the part comprising in the description and constituting description together illustrates the present invention's with descriptionExemplary embodiment, feature and aspect, and for explaining the principle of the present invention.
Fig. 1 illustrates the flow chart of multimedia resource recommendation method according to an embodiment of the invention；
Fig. 2 illustrates another flow chart of multimedia resource recommendation method according to an embodiment of the invention；
Fig. 3 illustrates another flow chart of multimedia resource recommendation method according to an embodiment of the invention；
Fig. 4 illustrates the schematic diagram obtaining user behavior data according to an embodiment of the invention；
Fig. 5 illustrates the structured flowchart of multimedia resource recommendation apparatus according to another embodiment of the present invention；
Fig. 6 illustrates the structured flowchart of multimedia resource recommendation apparatus according to another embodiment of the present invention.
Detailed description of the invention
Various exemplary embodiments, feature and the aspect of the present invention is described in detail below with reference to accompanying drawing.In accompanying drawing identicalReference represent the same or analogous element of function.Although the various aspects of embodiment shown in the drawings, but removeNon-specifically is pointed out, it is not necessary to accompanying drawing drawn to scale.
The most special word " exemplary " means " as example, embodiment or illustrative ".Here as " exemplary "Illustrated any embodiment should not necessarily be construed as preferred or advantageous over other embodiments.
It addition, in order to better illustrate the present invention, detailed description of the invention below gives numerous details.It will be appreciated by those skilled in the art that do not have some detail, the present invention equally implements.In some instances, forMethod well known to those skilled in the art, means, element and circuit are not described in detail, in order to highlight the purport of the present invention.
Embodiment 1
Fig. 1 illustrates the flow chart of multimedia resource recommendation method according to an embodiment of the invention.As it is shown in figure 1, these are manyMedia resource recommends method, mainly can include that step 101 is to step 103.
Step 101, in the case of the request receiving the browsing objective resource that targeted customer initiates at target terminal, obtainTake each to be recommended multimedia resource relevant to described target resource.
The terminal of the embodiment of the present invention can include various types of terminal unit that can browse through multimedia resource, such asMobile phone (such as Android phone, iPhone mobile phone), computer, panel computer, TV etc., be not construed as limiting it.Wherein, manyMedia resource (Multimedia), can include the various media formats such as such as text, sound, video and image.
As an example of the embodiment of the present invention, user A has initiated to broadcast in the video playback client of mobile phone terminalPut the request of " big fish Caulis et folium euphorbiae milii ".In the case of server receives this request, obtain relevant to " big fish Caulis et folium euphorbiae milii " respectively waiting and push awayRecommend multimedia resource.Wherein, user A can be the targeted customer described in step 101, and mobile phone terminal can be in step 101Described target terminal, " big fish Caulis et folium euphorbiae milii " can be the target resource described in step 101.
Wherein, multimedia resource to be recommended can include various types of target resource tool currently browsed with targeted customerThere is the multimedia resource of dependency.The embodiment of the present invention does not limit the concrete acquisition mode of multimedia resource to be recommended.With videoAs a example by, multimedia resource to be recommended can belong to same video channel with target video, such as, broadly fall into acrobatic fighting film；Can alsoSerial is belonged to, such as " hungry game 1 " and " hungry game 2 " with target video；It is also possible to have identical with target videoStarring actors, or there is stronger broadcasting dependency etc., this is not construed as limiting.
It should be noted that those skilled in the art are it should be understood that there is various ways in which in prior art and can realizeObtain each to be recommended multimedia resource relevant to target resource, this is not construed as limiting.
Step 102, according to the type of described target terminal and various terminal user behavior data in preset time period,Obtain each described multimedia resource characteristic of correspondence to be recommended.
Wherein, the type of target terminal can be mobile phone (such as Android phone, iPhone mobile phone), computer, flat boardAny one in computer, television terminal etc., is not construed as limiting it.The user behavior data of the embodiment of the present invention can include baseThe various actions made for multimedia resource in user and the data produced.The embodiment of the present invention does not limit the tool of user behaviorBody type, for example, it is possible to include that the navigation patterns of user, the comment behavior of user, the scoring behavior of user, the point of user praise topStep on behavior etc..
Step 103, according to each described multimedia resource characteristic of correspondence to be recommended, to each described multimedia resource to be recommendedIt is ranked up, and generates multimedia resource recommendation information according to ranking results.
In the embodiment of the present invention, multimedia resource characteristic of correspondence can include various types of multimedia resource feature,Such as terminal feature, behavior characteristics, resource characteristic etc., be not construed as limiting this.Wherein, terminal feature can be used to indicate that for clearLook at the situation of terminal of multimedia resource, such as terminal type, terminal screen size etc..Behavior characteristics can be used to indicate that userThe situation of the behavior that multimedia resource is made, such as watch duration, comment number, mark, push up step on number etc..Resource characteristic can be usedIn the situation of the attribute representing multimedia resource, such as information channel, interest tags etc..
Further, multimedia resource characteristic of correspondence can also include identification information (the such as title, numbering of featureDeng) content (such as eigenvalue) corresponding with signature identification information.Wherein, eigenvalue can be continuous numerical value, it is also possible to beDiscrete type numerical value, it is also possible to be non-numeric type.For example, terminal screen size (terminal feature) can be discrete type numerical value,Such as 240*320,320*480；Scoring (behavior characteristics) can be continuous numerical value, and its span can be [0,10]；MoneySource channel (resource characteristic) can may belong to movie channel or series channel with right and wrong numeric type, such as video resource.
Wherein, step 102 can include multiple implementation, for example:
Mode one, determine each to be recommended multimedia resource relevant to target resource in a step 101 after, in step 102In, user behavior number relevant to each multimedia resource to be recommended in preset time period can be obtained respectively from various terminalsAccording to.Again from the acquired user behavior data relevant to each multimedia resource to be recommended, extract each multimedia to be recommended moneySource characteristic of correspondence.
Mode two, each multimedia resource (whole many matchmakers in the multimedia resources database of such as certain website can be added up in advanceBody resource) user behavior data in preset time period in various terminals, or add up the most each many at set intervalsMedia resource is the user behavior data in preset time period in various terminals, and each multimedia resource pair acquired in extractionThe feature answered.
After determining the multimedia resource each to be recommended relevant to target resource in a step 101, in a step 102, permissibleFrom each multimedia resource characteristic of correspondence extracted before, filter out each multimedia resource characteristic of correspondence to be recommended.Wherein,The terminal that the step of counting user behavioral data can be specified by certain in advance performs, it is also possible to by communicating with each terminalServer (server of such as website) perform.
For above-mentioned mode one, in a kind of possible implementation, as in figure 2 it is shown, according to described target terminalType and various terminal user behavior data in preset time period, obtain the spy that each described multimedia resource to be recommended is correspondingLevy (step 102), may include that
Step 201, obtain from various terminals in preset time period relevant to each described multimedia resource to be recommendedFirst user behavioral data；
Step 202, described first user behavioral data is carried out tagsort and form collator, first after being arrangedUser behavior data；
Step 203, type according to described target terminal, obtain each the first user behavioral data after described arrangementDescribed multimedia resource characteristic of correspondence to be recommended, described multimedia resource characteristic of correspondence to be recommended includes described to be recommended manyTerminal feature that media resource is corresponding and resource characteristic.
In step 201, acquired first user behavioral data can include relevant to each multimedia resource to be recommendedVarious user behavior datas, the embodiment of the present invention do not limit obtain first user behavioral data detailed process.For example, it is possible toDirectly obtain the user behavior data relevant to each multimedia resource to be recommended from various terminals, then carry out user behavior dataIt is polymerized in the buffer.
Additionally, the embodiment of the present invention does not limit the concrete time span of preset time period, such as, it can be one month, halfThe moon, ten days etc., this is not construed as limiting.It is understood that select suitable preset time period, it is ensured that user's row of acquisitionFor data, there is preferable representativeness.
In step 202., the tagsort of the embodiment of the present invention can include user behavior data according to certain spyLevy the process carrying out classifying, such as, by user behavior data according to terminal type, terminal screen size, viewing duration, commentNumber, top are stepped on the features such as number and are classified, and are not construed as limiting this.The form collator of the embodiment of the present invention can include sortedUser behavior data carries out the process arranging and adding up according to preset format, such as, is arranged by sorted user behavior dataBecome the form such as form, list, this is not construed as limiting.
For example, as a example by video, the first user behavioral data after arrangement can be as shown in table 1:
Table 1
| Resource ID | Terminal type | Terminal screen size | Viewing duration | Comment number | Number is stepped on top |
| 1 | Mobile phone | 100*300 | 5minute | 2 | 1 |
| 1 | Computer | 800*600 | 10minute | 5 | 3 |
In step 203, the first user behavioral data after arranging can obtain each according to the type of target terminalDescribed multimedia resource characteristic of correspondence to be recommended.For example, it is that video 1, target terminal are at multimedia resource to be recommendedIn the case of mobile phone, the feature obtaining video 1 can include terminal type (mobile phone), terminal screen size (100*300), viewingNumber (1) is stepped on duration (5minute), comment number (2), top.
For above-mentioned mode two, in a kind of possible implementation, as it is shown on figure 3, according to described target terminalType and various terminal user behavior data in preset time period, obtain the spy that each described multimedia resource to be recommended is correspondingLevy (step 102), may include that
Step 301, second user relevant to each multimedia resource obtained from various terminals in preset time periodBehavioral data；
Step 302, described second user behavior data is carried out tagsort and form collator, second after being arrangedUser behavior data；
Step 303, according to the type of described target terminal and each described multimedia resource to be recommended, after described arrangementSecond user behavior data obtains each described multimedia resource characteristic of correspondence to be recommended, described multimedia resource pair to be recommendedThe feature answered includes terminal feature and the resource characteristic that described multimedia resource to be recommended is corresponding.
In step 301, the second acquired user behavior data can include and each multimedia resource (such as multimediaWhole multimedia resources in resources bank) relevant various user behavior datas, the embodiment of the present invention does not limit acquisition the second useThe detailed process of family behavioral data.Such as, as shown in Figure 4, with the user behavior data of the various terminal of real-time collecting, and will be able to receiveThe user behavior data real-time storage of collection is at such as HDFS (Hadoop Distributed File System, distributed documentSystem) on.Further, the user behavior data obtained from various terminals can be stored in after merging such as server, alsoCan store with each terminal or every kind of corresponding corresponding file of terminal, this is not limited.
The multimedia resource of the embodiment of the present invention recommends method, it is possible to collect the data of user's different terminals, and to userBehavior carry out classification and form go out storage, for other business provide data analysis support, analyze user on different terminalsBehavior characteristics.
In step 302, the second user behavior data is carried out in process and the step 201 of tagsort and form collatorFirst user behavioral data is carried out tagsort similar with the process of form collator, do not repeat them here.
For example, as a example by video, the second user behavior data after arrangement can be as shown in table 2:
Table 2
| Video ID | Terminal type | Terminal screen size | Viewing duration | Comment number | Number is stepped on top |
| 1 | Mobile phone | 100*300 | 5minute | 2 | 1 |
| 2 | Computer | 800*600 | 10minute | 5 | 3 |
| 2 | Mobile phone | 100*300 | 16minute | 3 | 3 |
| 3 | Mobile phone | 100*300 | 12minute | 2 | 5 |
In step 303, can be according to the type of target terminal and each multimedia resource to be recommended, second after arrangingUser behavior data obtains each multimedia resource characteristic of correspondence to be recommended.For example, at multimedia resource to be recommended it isIn the case of video 2, target terminal are computer, the feature obtaining video 2 can include terminal type (computer), terminal screen chiNumber (3) is stepped on very little (800*600), viewing duration (10minute), comment number (5), top.
In a kind of possible implementation, according to each described multimedia resource characteristic of correspondence to be recommended, to each describedMultimedia resource to be recommended is ranked up (step 103), may include that
Step 401, employing following formula 1 and following formula 2 calculate the click probability of each described multimedia resource to be recommended,
Wherein, i represents ith feature, aiRepresent the weight coefficient that ith feature is corresponding, xiRepresent that ith feature is correspondingEigenvalue, N represents the number of feature, and f (x) represents each feature characteristic of correspondence that each described multimedia resource to be recommended includesThe result that value and weight coefficient are sued for peace after seeking product again, score (v) represents the click that each described multimedia resource to be recommended is correspondingProbability；
Step 402, according to click probability corresponding to each described multimedia resource to be recommended, to each described multimedia to be recommendedResource is ranked up, and according to ranking results, choose from each described multimedia resource to be recommended meet pre-conditioned each standbySelect multimedia resource.
Wherein, multimedia resource to be recommended can include that multiple feature, each feature have characteristic of correspondence value and weightCoefficient.Feature characteristic of correspondence value can be processed by such as Feature Engineering and obtain, and weight coefficient corresponding to feature can pass throughSuch as model training obtains.The each feature characteristic of correspondence value corresponding according to each multimedia resource to be recommended and weight coefficient, canTo be calculated click probability corresponding to each multimedia resource to be recommended (clicked probability).In recommendation process, wait to push awayThe click probability recommending multimedia resource corresponding is the highest, represents that the clicked probability of this multimedia resource is the biggest, can preferentially be pushed awayRecommend.
Such as, multimedia resource to be recommended has N number of feature.Wherein, the 1st feature characteristic of correspondence value is x1, weight systemNumber is a1；2nd feature characteristic of correspondence value is x2, weight coefficient is a2；The rest may be inferred, and ith feature characteristic of correspondence value isxi, weight coefficient is ai.So, the click probability that this multimedia resource to be recommended is corresponding is score (v):
It should be noted that the original value of eigenvalue can be numeric type or nonumeric type, through the place of Feature EngineeringReason, can obtain being easy to the final value of the quantization that model calculates.In order to make it easy to understand, be exemplified below:
For scoring (continuous numerical value), can map that between [0,10]；For viewing duration (discrete type numberValue), 24 features can be mapped as, such as video is to be watched at 20, then corresponding 20 feature values are 1；ForTerminal type (nonumeric type), such as, can use 1-Android mobile phone, 2-iPhone mobile phone, 3-panel computer, 4-computer, 5-TV represents.
It is understood that the number of multimedia resource to be recommended may be much larger than the many matchmakers generated required for recommendation informationThe number of body resource.In this case it is necessary to multimedia resource to be recommended is screened, obtain for generating recommendation informationAlternative multimedia resource.Further, in step 402, for example, it is possible to will click on probability to exceed the to be recommended of certain numerical valueMultimedia resource alternately multimedia resource, it is also possible to using ranking multimedia resource to be recommended within the specific limits as standbySelect multimedia resource, this is not construed as limiting.
In a kind of possible implementation, generate multimedia resource recommendation information (step 103) according to ranking results, canTo include:
Step 403, to each described alternative multimedia resource, use in following steps one or more deletes；
Described targeted customer is deleted the most browsed many in preset time period from each described alternative multimedia resourceMedia resource；
The multimedia resource exceeding preset number in same information channel is deleted from each described alternative multimedia resource；
The multimedia resource exceeding preset number in same interest tags is deleted from each described alternative multimedia resource.
The information channel of the embodiment of the present invention can represent the channel belonging to multimedia resource.It is said that in general, multimedia moneyThe information channel in source seldom can be changed after determining.Such as, video channel can be TV play, film, variety show etc.；LittleSay that channel can be describing love affairs, passes through, indulge in U.S. etc..Interest tags can include the key word for representing multimedia resource.Such as,The interest tags of video can be make laughs, take a risk, landscape etc..Wherein, information channel and interest tags belong to multimedia resourceEssential information.
When user is carried out personalized recommendation, not only need to predict the interest resource of user, target may be it is also contemplated thatThe characteristic of terminal and the multiformity of resource.Therefore, it can multimedia resource be screened and controls.The embodiment of the present invention does not limitsThe most concrete fixed screening factor, for example, it is possible to browse situation according to targeted customer, can be according to the feature of alternative multimedia resourceDeng.Further, based on the screening principle set, screening alternative multimedia resource, final acquisition is used for generating recommendationEach multimedia resource of information.
The multimedia resource of the embodiment of the present invention recommends method, can obtain various terminal user in preset time periodBehavioral data, it is possible to user is at all kinds terminal navigation patterns in covering, such that it is able to according to the characteristic of current target terminal, exampleSuch as screen size, network condition and viewing scene etc., user is carried out the resource recommendation of personalization.
Embodiment 2
Fig. 5 illustrates the structured flowchart of multimedia resource recommendation apparatus according to another embodiment of the present invention.Fig. 5 may be used forMultimedia resource shown in service chart 1 to Fig. 3 recommends method.For convenience of description, illustrate only in Figure 5 and present invention enforcementThe part that example is relevant.
As it is shown in figure 5, this multimedia resource recommendation apparatus, mainly may include that source obtaining module 51 to be recommended, be used forIn the case of the request receiving the browsing objective resource that targeted customer initiates at target terminal, obtain and described target resourceRelevant multimedia resource each to be recommended.Feature acquisition module 53, is connected, for root with described source obtaining module 51 to be recommendedAccording to type and the various terminal user behavior data in preset time period of described target terminal, obtain each described to be recommended manyMedia resource characteristic of correspondence.Recommendation information generation module 55, is connected with described feature acquisition module 53, for according to each describedMultimedia resource characteristic of correspondence to be recommended, is ranked up each described multimedia resource to be recommended, and raw according to ranking resultsBecome multimedia resource recommendation information.Concrete principle and example may refer to the associated description of embodiment 1 and Fig. 1.
In a kind of possible implementation, described feature acquisition module, may include that the first data capture unit, useIn obtaining the first user behavior relevant to each described multimedia resource to be recommended in preset time period from various terminalsData；First taxonomic revision unit, is connected with described first data capture unit, for entering described first user behavioral dataRow tagsort and form collator, the first user behavioral data after being arranged；Fisrt feature acquiring unit, with described firstTaxonomic revision unit connects, for the type according to described target terminal, the first user behavioral data after described arrangementObtain each described multimedia resource characteristic of correspondence to be recommended, described multimedia resource characteristic of correspondence to be recommended include described in treatRecommend terminal feature and resource characteristic that multimedia resource is corresponding.Concrete principle and example may refer to embodiment 1 and Fig. 2Associated description.
In a kind of possible implementation, described feature acquisition module, may include that the second data capture unit, useIn obtaining second user behavior data relevant to each multimedia resource in preset time period from various terminals；Second pointClass arranges unit, is connected with described second data capture unit, for described second user behavior data is carried out tagsortAnd form collator, the second user behavior data after being arranged；Second feature acquiring unit, with described second taxonomic revision listUnit connects, for the type according to described target terminal and each described multimedia resource to be recommended, second after described arrangementObtaining each described multimedia resource characteristic of correspondence to be recommended in user behavior data, described multimedia resource to be recommended is correspondingFeature includes terminal feature and the resource characteristic that described multimedia resource to be recommended is corresponding.Concrete principle and example may refer to realityExecute the associated description of example 1 and Fig. 3.
In a kind of possible implementation, described recommendation information generation module, may include that click probability calculation listUnit, for use following formula 1 and following formula 2 to calculate the click probability of each described multimedia resource to be recommended,
Wherein, i represents ith feature, aiRepresent the weight coefficient that ith feature is corresponding, xiRepresent that ith feature is correspondingEigenvalue, N represents the number of feature, and f (x) represents each feature characteristic of correspondence that each described multimedia resource to be recommended includesThe result that value and weight coefficient are sued for peace after seeking product again, score (v) represents the click that each described multimedia resource to be recommended is correspondingProbability.Alternative resource acquiring unit, is connected with described click probability calculation unit, for providing according to each described multimedia to be recommendedThe click probability that source is corresponding, is ranked up each described multimedia resource to be recommended, and according to ranking results, waits to push away described in eachRecommend multimedia resource is chosen and meet pre-conditioned each alternative multimedia resource.Concrete principle and example may refer to embodimentThe associated description of 1.
In a kind of possible implementation, described recommendation information generation module, also include: delete unit, for respectivelyDescribed alternative multimedia resource, use in following steps one or more deletes；From each described alternative multimedia resourceThe multimedia resource that the described targeted customer of middle deletion is the most browsed in preset time period；From each described alternative multimedia resourceThe same information channel of middle deletion exceedes the multimedia resource of preset number；Delete same from each described alternative multimedia resourceInterest tags exceedes the multimedia resource of preset number.Concrete principle and example may refer to the associated description of embodiment 1.
The multimedia resource recommendation apparatus of the embodiment of the present invention, can obtain various terminal user in preset time periodBehavioral data, it is possible to user is at all kinds terminal navigation patterns in covering, such that it is able to according to the characteristic of current target terminal, exampleSuch as screen size, network condition and viewing scene etc., user is carried out the resource recommendation of personalization.
Embodiment 3
Fig. 6 shows the structured flowchart of a kind of multimedia resource recommendation apparatus of an alternative embodiment of the invention.DescribedMultimedia resource recommendation apparatus 1100 can be to possess the host server of computing capability, personal computer PC or portabilityPortable computer or terminal etc..Calculating node is not implemented and limits by the specific embodiment of the invention.
Described multimedia resource recommendation apparatus 1100 includes processor (processor) 1110, communication interface(Communications Interface) 1120, memorizer (memory) 1130 and bus 1140.Wherein, processor 1110,Communication interface 1120 and memorizer 1130 complete mutual communication by bus 1140.
Communication interface 1120 is used for and network device communications, and wherein the network equipment includes such as Virtual Machine Manager center, is total toEnjoy storage etc..
Processor 1110 is used for performing program.Processor 1110 is probably a central processor CPU, or special collectionBecome circuit ASIC (Application Specific Integrated Circuit), or be configured to implement the present inventionOne or more integrated circuits of embodiment.
Memorizer 1130 is used for depositing file.Memorizer 1130 may comprise high-speed RAM memorizer, it is also possible to also includes non-Volatile memory (non-volatile memory), for example, at least one disk memory.Memorizer 1130 can also be to depositMemory array.Memorizer 1130 is also possible to by piecemeal, and described piece can be by certain rule sets synthesis virtual volume.
In a kind of possible embodiment, said procedure can be the program code including computer-managed instruction.This journeySequence is particularly used in: realize the operation of each step in embodiment 1.
Those of ordinary skill in the art are it is to be appreciated that each exemplary cell in embodiment described herein and algorithmStep, it is possible to being implemented in combination in of electronic hardware or computer software and electronic hardware.These functions are actually with hardware alsoIt is that software form realizes, depends on application-specific and the design constraint of technical scheme.Professional and technical personnel can be forSpecific application selects different methods to realize described function, but this realization is it is not considered that exceed the model of the present inventionEnclose.
If using the form of computer software realize described function and as independent production marketing or use time, then existTo a certain extent it is believed that all or part of (part such as contributed prior art) of technical scheme isEmbody in form of a computer software product.This computer software product is generally stored inside the non-volatile of embodied on computer readableIn storage medium, including some instructions with so that computer equipment (can be that personal computer, server or network setStandby etc.) perform all or part of step of various embodiments of the present invention method.And aforesaid storage medium include USB flash disk, portable hard drive,Read only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magneticThe various medium that can store program code such as dish or CD.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and anyThose familiar with the art, in the technical scope that the invention discloses, can readily occur in change or replace, should containCover within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with described scope of the claims.