CN104834657A

CN104834657A - User behavior analysis method and servers

Info

Publication number: CN104834657A
Application number: CN201410429505.XA
Authority: CN
Inventors: 何伟杰
Original assignee: Tencent Technology Beijing Co Ltd
Current assignee: Tencent Technology Beijing Co Ltd
Priority date: 2014-08-27
Filing date: 2014-08-27
Publication date: 2015-08-12
Anticipated expiration: 2034-08-27
Also published as: CN104834657B

Abstract

The invention discloses a user behavior analysis method and servers. The method comprises the steps of after booting a real-time analysis server, receiving historical user behavior data sent by an offline analysis server according to a first analysis result, wherein the first analysis result is an analysis result formed according to a first preset analysis strategy before the real-time analysis server boots; acquiring current user behavior data corresponding to current user behavior, wherein both the historical user behavior data and current user behavior data comprise user operations and operation objects which the user operations point at; and according to the first preset analysis strategy, analyzing the historical user behavior data and current user behavior data so as to obtain a second analysis result, wherein both the first analysis result and the second analysis result comprise at least one set of the operation objects, one operation object corresponds to at least two pieces of multimedia information, and the information content of the multimedia information corresponding to the same operation object has time correlation.

Description

User behavior analysis method and server

Technical field

The present invention relates to field of information processing data analysis technique, particularly relate to a kind of user behavior analysis method and server.

Background technology

Present inventor, in the process realizing the embodiment of the present application technical scheme, at least finds to there is following technical matters in correlation technique:

Current user behavior analysis comprises two kinds of modes:

The first: off-line analysis; Centralized stores mass data during off-line analysis; Data processing is carried out to the user behavior of multiple user; The advantage had is: accessible data volume is large, can do the depth analysis of analysis logic complexity; But it is long that shortcoming is analytical cycle, the acquisition time delay of analysis result is large; But a lot of data dynamically changed, ageing is very important, and time delay is excessive may will cause analysis result can not reflect active user's behavior, thus cannot the accuracy of impact analysis result.

The second: real-time analysis, dynamic acquisition during real-time analysis also analyzes user behavior, and to consider the impact of active user's behavior fully, time delay is little.But the shortcoming of real-time analysis is; Usually carry out the server of performance analysis, hardware and software resource is all very valuable and limited, and being preserve or abandon to the raw data processed, intermediate result and established result, is the major issue that real-time analysis server faces.If do not preserve above-mentioned data, to analysis result be regained when server exception, obviously because data abandon, obviously cannot complete analysis; If preserve above-mentioned data, the resource of a large amount of real-time analysis server can be taken, real-time analysis server resource may be caused nervous, As time goes on finally can be, the data volume stored in real-time analysis server is increasing, the resource that real-time analysis server can be used for carrying out data processing is fewer and feweri, finally causes the analysis result of real-time analysis server also cannot to supply by Quick.Obviously real-time analysis server cannot carry out the depth analysis of analysis logic complexity usually simultaneously.In addition, easily there is exception, cause real-time analysis server usually all to there is the not high problem of reliability.

Along with the development of the communication technology and Internet telephony, increasing consumer's custom watches various types of multimedia messages from internet, as TV play, film and animation etc.; Internet provides service to a large amount of users, faced by the user behavior data of magnanimity; Adopt off-line analysis due to ductility at that time, obviously cannot meet the demand of user's free online multimedium information; Real-time analysis is adopted obviously to be faced with very large problem and the problem of poor stability to the storage of user behavior data; This contradiction, in the user behavior analysis process such as program request of multimedia messages, seems more outstanding.

Summary of the invention

In view of this, the embodiment of the present invention is expected to provide user behavior analysis method and real-time analysis server, with the problem that the time delay solving off-line analysis in prior art stability that is large and real-time analysis is not high.

For achieving the above object, technical scheme of the present invention is achieved in that

Embodiment of the present invention first aspect provides a kind of user behavior analysis method,

Described method comprises:

After real-time analysis server starts, receive the historic user behavioral data that off-line analysis server sends according to the first analysis result; Described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Obtain the active user's behavioral data corresponding to active user's behavior; Described historic user behavioral data and described active user's behavioral data include the operand of user operation and described user operation sensing;

According to described first presupposition analysis strategy, analyze described historic user behavioral data and described active user's behavioral data, obtain the second analysis result;

Wherein, described first analysis result and described second analysis result are the set comprising operand described at least one; A described operand correspondence at least two multimedia messagess; The information content corresponding to the multimedia messages of same described operand has temporal associativity.

Preferably,

Described foundation first presupposition analysis strategy, analyze described historic user behavioral data and described active user's behavioral data, obtain the second analysis result and comprise:

According to described first presupposition analysis strategy and operation behavior, determine the assay value that operand described in each is corresponding;

Sort to described assay value, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described second analysis result;

Wherein, described I be not less than 1 positive integer.

Preferably,

Described according to described first presupposition analysis strategy and operation behavior, determine that the analysis that operand described in each is corresponding comprises:

The assay value S that each operand is corresponding is determined according to following formulae discovery;

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described j is that jth analyzes dimensional parameter; Described J is the total number analyzing dimension; Described f _jj () is the analytic function of jth analysis dimension; Described w _jfor jth analyzes the weight of dimension;

Wherein, described J be not less than 1 positive number.

Preferably,

Described method also comprises:

Receive the modifying factor that described off-line analysis server sends; Wherein, described modifying factor is that described off-line analysis server is determined the analysis of historic user behavioral data based on the second presupposition analysis strategy;

According to the first analysis strategy described in described modifying factor correction.

Embodiment of the present invention second aspect provides a kind of user behavior analysis method, and described method comprises:

After real-time analysis server starts, off-line analysis server, according to the first analysis result formed according to the first presupposition analysis strategy, determines the historic user behavioral data sent to real-time analysis server; Wherein, described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy; Described historic user behavioral data comprises the operand of user operation and described user operation sensing;

Described historic user behavioral data is sent to described real-time analysis server;

Wherein, described historic user behavioral data forms the second analysis result for described real-time analysis server;

Described first analysis result and described second analysis result are the set comprising operand described at least one; A described operand correspondence at least two multimedia messagess; The information content corresponding to the multimedia messages of same described operand has temporal associativity.

Preferably,

Described method also comprises:

Before described real-time analysis server starts, described off-line analysis server forms described first analysis result according to the first presupposition analysis strategy;

Described off-line analysis server forms described first analysis result according to described first presupposition analysis strategy and comprises:

Sort to described assay value, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described first analysis result;

Wherein, described I be not less than 1 positive integer.

Preferably,

The assay value S that operand described in each is corresponding is determined according to following formulae discovery;

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

Preferably,

Described method also comprises:

Described off-line analysis server also analyzes user behavior data according to the second analysis strategy, obtains modifying factor;

Described modifying factor is sent to described real-time analysis server;

Described modifying factor is for revising described first analysis strategy.

The embodiment of the present invention third aspect provides a kind of real-time analysis server, and described server comprises:

Receiving element, for after real-time analysis server starts, receives the historic user behavioral data that off-line analysis server sends according to the first analysis result; Described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Acquiring unit, for obtaining the active user's behavioral data corresponding to active user's behavior; Described historic user behavioral data and described active user's behavioral data include the operand of user operation and described user operation sensing;

First analytic unit, for according to described first presupposition analysis strategy, analyzes described historic user behavioral data and described active user's behavioral data, obtains the second analysis result;

Preferably,

Described first analytic unit comprises:

Determination module, for according to described first presupposition analysis strategy and operation behavior, determines the assay value that operand described in each is corresponding;

Order module, for sorting to described assay value, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described second analysis result;

Wherein, described I be not less than 1 positive integer.

Preferably,

Described determination module, specifically for determining according to following formulae discovery the assay value S that each operand is corresponding;

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

Preferably,

Described receiving element, also for receiving the modifying factor that described off-line analysis server sends; Wherein, described modifying factor is that described off-line analysis server is determined the analysis of historic user behavioral data based on the second presupposition analysis strategy;

Described server also comprises:

Amending unit, for according to the first analysis strategy described in described modifying factor correction.

Embodiment of the present invention fourth aspect provides a kind of off-line analysis server, and described server comprises:

Determining unit, for after real-time analysis server starts, according to the first analysis result formed according to the first presupposition analysis strategy, determines the historic user behavioral data sent to real-time analysis server; Described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Transmitting element, for sending described historic user behavioral data to described real-time analysis server;

Preferably,

Described server also comprises the second analytic unit;

Described second analytic unit, for according to described first presupposition analysis strategy and operation behavior, determines the assay value that operand described in each is corresponding; And described assay value is sorted, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described first analysis result;

Wherein, described I be not less than 1 positive integer.

Preferably,

Described second analytic unit, specifically for determining according to following formulae discovery the assay value S that operand described in each is corresponding;

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described j is that jth analyzes dimensional parameter; Described J is the total number analyzing dimension; Described f _jj () analyzes the analytic function of dimension for corresponding to jth; Described w _jfor jth analyzes the weight of dimension;

Wherein, described J be not less than 1 positive number.

Preferably,

Described second analytic unit, also for also analyzing user behavior data according to the second analysis strategy, obtains modifying factor;

Institute's transmitting element, for sending to described real-time analysis server by described modifying factor;

Described modifying factor is for revising described first analysis strategy.

User behavior analysis method described in the embodiment of the present invention and server, real-time analysis server receives the historic user behavioral data corresponding to the first analysis result from off-line analysis server, and form the second analysis result according to historical behavior data and active user's behavioral data, the first analysis strategy, have the following advantages:

First, adopt real-time analysis server to form described second analysis result, there is time delay little, the advantage that speed is high;

Secondly, receive historic user behavioral data from off-line analysis server, real-time analysis server is without the need to storing data after closedown, thus the data volume that real-time analysis server stores is little; And can obtain data from off-line analysis server when real-time analysis server is abnormal simultaneously, thus stability is high;

Again, the historic user behavioral data corresponding to the first analysis result is received from off-line analysis server, for receiving historic user behavioral data targetedly, the data interaction between off-line analysis server and real-time analysis server being reduced, the data volume that real-time analysis server stores when analyzing being reduced simultaneously;

Simultaneously, described first analysis result and described second analysis result are the set comprising at least one operand; And a described operand is to the multimedia messages that multiple information content should be had to have temporal associativity; By the method for described off-line analysis server and real-time analysis server binding analysis, be applied in the operation behavior analyses such as user on multimedia information program request, can solve user behavior analysis in existing multimedia messages on-demand process time ductility and stability between contradiction, relative to real-time analysis method completely, there are data and store advantage easily.

Accompanying drawing explanation

One of schematic flow sheet that Fig. 1 is the user behavior analysis method described in the embodiment of the present invention;

Fig. 2 is for analyzing the schematic flow sheet of formation second analysis result described in the embodiment of the present invention;

Fig. 3 is the schematic flow sheet two of the user behavior analysis method described in the embodiment of the present invention;

Fig. 4 is the schematic flow sheet three of the user behavior analysis method described in the embodiment of the present invention;

Fig. 5 is the schematic flow sheet four of the user behavior analysis method described in the embodiment of the present invention;

Fig. 6 is the schematic flow sheet five of the user behavior analysis method described in the embodiment of the present invention;

Fig. 7 is the schematic flow sheet six of the user behavior analysis method described in the embodiment of the present invention;

Fig. 8 is the schematic flow sheet seven of the user behavior analysis method described in the embodiment of the present invention;

Fig. 9 is the schematic flow sheet eight of the user behavior analysis method described in the embodiment of the present invention;

One of structural representation that Figure 10 is the real-time analysis server described in the embodiment of the present invention;

Figure 11 is the structural representation of the first analytic unit described in the embodiment of the present invention;

The structural representation two that Figure 12 is the real-time analysis server described in the embodiment of the present invention;

One of structural representation that Figure 13 is the off-line analysis server described in the embodiment of the present invention;

Figure 14 is the structural representation two of the off-line analysis server described in the embodiment of the present invention;

Figure 15 is the structural representation of the analytic system described in the embodiment of the present invention.

Embodiment

Below in conjunction with Figure of description and specific embodiment technical scheme of the present invention done and further elaborate.

Embodiment of the method one:

As shown in Figure 1, the present embodiment provides a kind of user behavior analysis method, and described method comprises:

Step S110: after real-time analysis server starts, receives the historic user behavioral data that off-line analysis server sends according to the first analysis result; Described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Step S120: obtain the active user's behavioral data corresponding to active user's behavior; Described historic user behavioral data and described active user's behavioral data include the operand of user operation and described user operation sensing;

Step S130: according to described first presupposition analysis strategy, analyzes described historic user behavioral data and described active user's behavioral data, obtains the second analysis result;

Described real-time analysis server is generally: after it gets described active user's behavioral data, can carry out user behavior analysis, to form analysis result immediately according to the first presupposition analysis strategy; And generally for and avoid in real-time analyzer, storing a large amount of data, after formation analysis result, the follow-up data that can not use will be dropped.

After described off-line analysis server is generally active user's behavioral data, user behavior analysis can not be carried out immediately according to the first presupposition analysis strategy, but described active user's behavioral data can be stored, within fixed time or analytical cycle, just carry out the analysis of described user behavior; And after usually forming analysis result, described user behavior data can not be abandoned immediately, in case follow-up abnormal time can be re-started the analysis of data by the data of storage.

Usual described real-time analysis server has multiple stage, forms a distributed real-time analyzer; The processing power of each described real-time analysis server, storage capacity more described off-line analysis server is weak, and described real-time analysis server may exist closedown and starting problem comparatively frequently.

Described off-line analysis server can have multiple stage, can form the off-line analysis cluster of a concentrated setting; Each described off-line analysis server has very strong processing power and database usually; Described database is for storing user behavior data corresponding to the various operation of user.

The startup of described real-time analysis server in the present embodiment comprises two kinds of situations:

The first: described real-time analysis server restarts after there is abnormal (as power down, systemic breakdown etc. are abnormal) at work, and this startup is abnormal startup;

The second: described real-time analysis server restarts after closing voluntarily under the operation of staff, this startup is normal startup.

After closing, described real-time analysis server will discard all user behavior datas, and such guarantee real-time analysis server need not store historic user behavioral data with a large amount of storage resources.But described real-time analysis server needs to continue user behavior analysis, and described real-time analysis server upon actuation, will receive described historic user behavioral data from offline service device in the present embodiment for this reason.And the historic user behavioral data received the historic user behavioral data of not all, but correspond to the historic user behavioral data of the first analysis result.

Described first analysis result is the same with described second analysis result, is all the analysis to user behavior; Described first analysis result is an analysis result early than described second analysis result; Based on the analysis result that user behavior data is formed before real-time analysis server starts; The data of real-time analysis server reception in the present embodiment, be only limitted to the data forming described first analysis result, not blindly all data stored in off-line analysis server are all sent to real-time analyzer, but select real-time analysis server needs for the formation of the historic user behavioral data of the second analysis result according to the first analysis result, thus greatly reducing the data of real-time analyzer reception and the data interaction amount between described real-time analysis server and described off-line analysis server.

Described user behavior can comprise all user operations that operand described in user carries out; Described user operation specifically can be as: the multimedia messages in operand as described in program request and stopping play as described in the operation such as multimedia messages.Described multimedia messages can be specifically TV play, animation, playlet or the film with multi-section.Described operand specifically can as TV play; A usual TV play comprises many collection; Each collection is a described multimedia messages; Many collection TV play in a TV play can form regular hour order according to the development of story of a play or opera content; The story of a play or opera content of usual i-th+1 collection is further developing on the basis of the i-th collection story of a play or opera content; Wherein said i be not less than 1 integer.

Described operand is animation also; A usual animation also can be divided into many collection; Each collection is equally also a described multimedia messages.The information content between the described animation of any two collection of same portion animation also has temporal sequencing; The one that this temporal sequencing is above-mentioned temporal associativity embodies.

Historic user behavioral data and active user's behavioral data are except the operand comprising user operation and user operation sensing; Also can comprise the information such as time point and user totem information that user performs this operation.

When described behavioral data comprises time point, be convenient to off-line analysis server and real-time analysis server, the general of data is selected; When described behavioral data comprises described user totem information, described off-line analysis server and the user behavior of real-time analysis server to each user is facilitated to analyze one by one.

Obtain the active user's behavioral data corresponding to active user's behavior described in step S120, comprise and directly obtain active user's behavioral data from client, described active user's behavioral data when also comprising from the user behavior data processing server of specifying.

In concrete implementation procedure; the client of user is after response user operation; the data of reflection user behavior are reported to user behavior data processing server; described user behavior data processing server can process data usually accordingly; the concrete data with different data format as different clients sent form the data with same form; re-send to described real-time analysis server, then now described step S120 is the active user's behavioral data receiving the transmission of described user behavior data processing server.

The present embodiment, when carrying out user behavior analysis, combines off-line analysis server and real-time analysis server; Real-time analysis server receives historic user behavioral data from off-line analysis server and in conjunction with active user's behavioral data, forms analysis result; Real-time analysis server is solved like this with regard to easy, when carrying out user behavior analysis, relative to existing off-line analysis method, there is time delay little, the advantage that efficiency is high, relative to existing real-time analysis method, has reliability high, the data volume stored is few and require low advantage to server hardware, and well can tackle the abnormal conditions of real-time analysis server.

In concrete implementation procedure, described method can also carry out subsequent analysis according to described second analysis result; As, carry out user's value analysis according to described second analysis result, so that after formation user value analysis result, for user improves its service applicable; For another example, according to described second analysis result for user recommends described operand etc.

As through as described in the execution of step S110 to step S140, after defining described second analysis result, find that user is very interested in TV play (TV play is the one of described multimedia messages) A, then now according to the second analysis result, when TV play A has update content, user TV play A can be pointed out timely to have renewal; Or according to the attribute information of TV play A, recommend to user the TV play B etc. having same alike result with described TV play A.Described attribute information can comprise described TV play type, as action movie, love film or swordsmen film etc.; Can also be that the information content has similarity, such as, be all the TV play etc. of prohibition of drug subject matter.

Comprehensively above-mentioned, the present embodiment is analyzed user behavior, defines the second analysis result comprising several operands; In the process analyzed, off-line analysis server and real-time analysis server are combined, there is the multiple advantages such as time delay is little, stability is high, and analyze for the multimedia messages performed user operation, solve in the on-demand process of existing multimedia messages, analysis data volume is large, and simple off-line analysis server or the real-time analysis server of relying on all is difficult to the problem providing analysis result easy in time.

Embodiment of the method two:

As shown in Figure 2, described step S130 comprises:

Step S131: according to described first presupposition analysis strategy and operation behavior, determine the assay value that operand described in each is corresponding;

Step S132: described assay value is sorted, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described second analysis result;

Wherein, described I be not less than 1 positive integer.

Describedly meet operand corresponding to the first pre-conditioned I described assay value, can be that assay value is greater than I the described operand analyzing threshold value, can also be I described operand maximum in ranking results.Select in the present embodiment for I described operand be the set that user makes earnest efforts the described multimedia messages watched most.

Comprehensively above-mentioned, method described in the present embodiment is for jointly analyzing the operand pointed by user behavior in conjunction with real-time analysis server and off-line analysis server, to show that user likes operating which operand, user likes as which TV play of program request, be convenient to the follow-up operand recommending user to like to user, not only have the advantage that time delay is little and degree of accuracy is high, can also be the hobby accurately knowing user, so that the follow-up hobby according to user provides corresponding service simultaneously.

Embodiment of the method three:

As shown in Figure 2, described step S130 comprises:

Wherein, described I be not less than 1 positive integer.

Described step S131 comprises:

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

In concrete implementation procedure, the function be different from described in the present embodiment also can be adopted to carry out cluster analysis to described operand and to obtain described assay value, concrete as based on the cluster analysis of distance or density clustering analysis; How described cluster analysis specifically carries out cluster analysis to described operand further illustrates with regard to no longer doing at this.

Below for method described in the present embodiment is applied in the example that user analyzes the operation behavior of the first operand (collection of drama C):

Described step S131 specifically can comprise:

Each acute corresponding assay value S is determined according to following formulae discovery;

S＝(N/M)*W1+(P/M)*W2+f(t,T)*W3+R*W4

Wherein, described N is the number of multimedia messages in viewed the first operand of user; Described M is the current renewal multimedia messages number of described first operand;

Described W1 is the first weight; Multimedia messages described in last in the first operand that described P is user's program request; Described W2 is the second weight; Described T is current time; Described t is between user's rearmost point sowing time; Described f (t, T) is the function of time; Described W3 the 3rd weight; Described R is the average performance level that described N multimedia messages is crossed in user's program request; Described W4 is the 4th weight;

Wherein, described formula S=(N/M) * W1+ (P/M) * W2+f (t, T) * W3+R*W4 can think above-mentioned a corresponding concrete formula.

Wherein, described first weights, the second weight, the 3rd weight and the 4th weight can pre-set.Described function f (t, T) can be a piecewise function, and concrete as when T-t is not more than 1 day, the value of described f (t, T) is 1; When the described span as T-t is 3 to 5 days, the value of described f (t, T) is 0.9; When the described span as T-t is 6 to 9 days, the value of described f (t, T) is 0.6; The like, the value of described T-t is larger, then the value of described f (t, T) is less.

In concrete implementation procedure, described first operation is asked for the assay value of object and is not limited to said method; Concrete as, suppose in above-mentioned formula, what the multimedia messages number of described user's program request was corresponding is the first analysis dimension; What last multimedia messages of the last program request of user was corresponding is the second analysis dimension; Corresponding between user's rearmost point sowing time is the 3rd analyze dimension; What described average performance level was corresponding is the 4th analysis dimension; Time then to user behavior determination assay value, only can adopt that above-mentioned four are analyzed in dimensions one, two or three analyze.

In concrete implementation procedure, described user behavior may also comprise user to the comment of play or and online friend carry out to this play discussion; Can also analyze according to information such as the key word of the comment content of user on multimedia information or comment frequencies when analyzing; Now, above-mentioned formula can be rewritten as follows:

S＝(N/M)*W1+(P/M)*W2+f(t,T)*W3+R*W4+F1*W5+K*W6

Wherein, described F1 is comment frequency, and described W5 is the 5th weight; Described W6 is the 6th weight.Described K is the value of corresponding key word, concrete like as comprised when comment content or the key word such as good-looking time, described K value can be not 0 integer, when specific implementation, the difference liking degree that can also characterize according to concrete key word, improves or reduces the value of described K; When the key word such as not like, ugly or disagreeable when comprising, described K value can be 0 or be negative value.In concrete implementation procedure, described first weight is all preferably positive number to the 6th weight.

Certainly, when specific implementation, using described comment frequency and can also comment on content separately as analysis foundation, also can be analyze dimension to be combined to one or more analysis dimensions that the 4th analyzes in dimension with first.

Present embodiments provide the behavior analysis method of multimedia messages described in a kind of user's program request, from four aspects, the play that acute behavior points to is chased after to user and analyze, can accurately confirm which user likes acute, is convenient to the follow-up lower play recommending it to like to user.

In the present embodiment, to select from multiple analysis dimension with predetermined funtcional relationship to determine the assay value of each operand, be convenient to follow-uply select to meet a first pre-conditioned I operand according to assay value; Relative to the above-mentioned cluster analysis mentioned, there is no the mutual interative computation between data, thus the advantages such as operation relation is simple, fast operation.

Embodiment of the method four:

As shown in Figure 3, described method also comprises:

Step S101: receive the modifying factor that described off-line analysis server sends; Wherein, described modifying factor is that described off-line analysis server is determined the analysis of historic user behavioral data based on the second presupposition analysis strategy;

Step S102: according to the first analysis strategy described in described modifying factor correction.

Because the analytic function of off-line analysis server is powerful, degree of depth excavation mass data being carried out to data can be crossed, described real-time analysis server also receives the analysis result that foundation second analysis strategy of off-line analysis server transmission is formed user's historical data in the present embodiment, and according to the modifying factor that this analysis result is formed; And correct described modifying factor in step s 102, to obtain the second analysis result more accurately.

Concrete as, when described multimedia messages is TV play, then described user behavior specifically can be user and chases after acute behavior; Described first analysis result and described second analysis result, for determining active user likes what is seen according to the first analysis strategy, can be revised each weight analyzing dimension and also can revise function corresponding to each dimensions.Such as user likes for a long time seeing and sees play at weekend, but finds that user watches acute A in program request every night recently; Break user's chasing after acute custom and then showing that user is delithted with this play in the past, even break and general chase after acute the custom, then utilizing formula

S=(N/M) * W1+ (P/M) * W2+f (t, T) * W3+R*W4 carries out assay value timing really, can in suitable correction described function f (t, T), to make the play of program request on weekdays, in the assay value of this dimension of program request time higher than the assay value of the play in program request at weekend.

In concrete implementation procedure, described off-line analysis server has more powerful processing capacity than real-time analysis server, usually also can carry out some depth analysis according to described second analysis strategy.These depth analysis are used for the depth rule of digging user behavior inherence, and these depth analysis can be lower to time-bounded requirement usually, analysis result are had to the analysis of certain influence.

It is concrete that as real-time analysis server and off-line analysis server, that the first presupposition analysis analysis of strategies can be utilized to go out active user is interested in which play, to point out user to watch.Described off-line analysis server simultaneously, and fetching is fixed time, interior user watches acute behavioral data, and then determines that user is to collection of drama type interested according to institute's behavioral data.Usual user to the interest of collection of drama type relative to user to concrete a certain portion acute interested be change be relatively slow, time-bounded requirement is lower.Obviously to collection of drama type interest analysis, the data volume of process will need data volume to be processed large to the acute perception in which concrete than user, and real-time analysis server possibly cannot complete, and now preferably be completed by off-line analysis server.Off-line analysis system, according to the analysis result of collection of drama type interest analysis, can adjust that it is current to the acute interested first presupposition analysis strategy in which to user, obtains more accurate first analysis result.

As in this period of time on June 30 to July 29, to obtaining user after acute a and the viewing behavior of acute b, to watch two acute favorable ratings be the same to user A, but off-line analysis server also likes the play seeing which kind to analyze to user, find that user A likes seeing American series; And now, if acute a is American series; Acute b is that English is acute; Reflect that user prefers acute a by described first analysis result; Specifically as by as described in sequence in the first analysis result represent, or described acute a is added the first analysis result and described acute b is got rid of the external expression at described first analysis result.

The present embodiment is on the basis of above-mentioned any embodiment, described real-time analysis server not only receives the historic user behavioral data sent according to the first analysis result from off-line analysis server, also receive the modifying factor that off-line analysis server sends, to adjust the first analysis strategy, again improve the degree of accuracy of the second analysis result.

Embodiment of the method five:

As shown in Figure 4, the present embodiment provides a kind of user behavior analysis method, and described method comprises:

Step S110: after real-time analysis server starts, receives the historic user behavioral data that off-line analysis server sends according to the first analysis result;

Step S120: obtain the active user's behavioral data corresponding to active user's behavior;

Step S130: according to the first presupposition analysis strategy, analyze described historic user behavioral data and described active user's behavioral data, obtain the second analysis result;

Step S140: described real-time analysis server also sends described second analysis result to described off-line analysis server;

When the formation time of the analysis result that described off-line analysis server is formed is more Zao than the formation time of described second analysis result, described second analysis result is used for as described first analysis result.

Usual described off-line analysis server and described real-time analysis server are all periodically form analysis result, the analytical cycle that described off-line analysis server is corresponding is the first analytical cycle, the analytical cycle of described real-time analysis service is the second analytical cycle, usual first analytical cycle is longer than the second analytical cycle, and concrete as the first analytical cycle is 24 hours, and the second analytical cycle is 15 minutes, off-line analysis server completes morning and once analyzes, and forms next analysis result by morning of second day, if real-time analysis server is because of system maintenance, in the same day 8:00 close, after 2 hours, described real-time analysis server opens again, now off-line analysis server does not form new analysis result, if the last analysis result directly formed with off-line analysis server is the first analysis result, do not consider causing the user behavior of middle 8 hours, under the circumstances, method described in the present embodiment also comprises: described real-time analysis server sends the step of described second analysis result to off-line analysis server, concrete can be periodic transmission, as 8 hours or 16 hours send once, usual employing periodically sends, it is long that it sends the second analytical cycle described in period ratio, to avoid frequently sending data to off-line analysis server.

If real-time analysis server have sent described second analysis result to off-line analysis server, then in said circumstances, the foundation of data that the second analysis result that real-time analysis server sends before being turned off will receive as real-time analysis server after 2 hours; Obvious in like this directly forming analysis result before 8 hours using off-line analysis server sends historic user behavioral data as the first analysis result to real-time analysis server, be obviously conducive to the raising of formation second analysis result degree of accuracy.

In the present embodiment, further define described real-time analysis server and send the second analysis result to described off-line analysis server before being turned off, because real-time analysis server is not closed, do not need to restart, then without the need to receiving data from described off-line analysis server, therefore described real-time analysis server only described second analysis result of transmission before it is closed can be limited, to reduce the data interaction between real-time analysis server and off-line analysis server; And the same method as described in a same embodiment, also can improve the accuracy of analysis of described second analysis result.

Institute step S140 is specially:

Described real-time analysis server is closed off-line analysis server described in forward direction at described real-time analysis server and is sent described second analysis result.

Embodiment of the method six:

As shown in Figure 5, the present embodiment provides a kind of user behavior analysis method, and described method comprises:

Step S210: after real-time analysis server starts, off-line analysis server, according to the first analysis result formed according to the first presupposition analysis strategy, determines the historic user behavioral data sent to real-time analysis server; Described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy; Described historic user behavioral data comprises the operand of user operation and described user operation sensing;

Step S220: send described historic user behavioral data to described real-time analysis server;

Described multimedia messages can be the information that animation, TV play or vaudeville etc. comprise sound and picture.

A described operand specifically comprises at least two described multimedia messagess; And the information content of same described multimedia messages has temporal associativity; The temporal associativity of this information content can be embodied in the dramatic progression situation as TV play, be also embodied in be vaudeville difference collection between priority broadcast order on.

What perform above-mentioned steps S210 to step S220 in the present embodiment is off-line analysis server; In specific implementation process, described off-line analysis server also receives the second analysis result that real-time analysis server is formed, and is not forming first analysis result that as off-line analysis server send described historic user behavioral data more late than the described second analysis result time; Adopt first analysis result self formed of this methods combining off-line analysis server, be conducive to the analytical precision improving real-time analyzer.

Embodiment of the method seven:

As shown in Figure 6, described method also comprises:

Step S200: before real-time analysis server starts, described off-line analysis server forms described first analysis result according to the first presupposition analysis strategy;

Described step S200 comprises:

According to described first presupposition analysis strategy and operation behavior, determine the assay value that operand described in each is corresponding; Sort to described assay value, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described first analysis result;

Wherein, described I be not less than 1 positive integer.

Described off-line analysis server will form the first analysis result according to the first analysis strategy voluntarily in the present embodiment, facilitate real-time analysis server when starting, according to analyzing formation first analysis result.

Off-line analysis server analyzes formation first analysis result voluntarily in the present embodiment, can reduce data interaction amount and the interaction times of off-line analysis server and real-time analysis server.

Embodiment of the method eight:

As shown in Figure 6, described method also comprises:

Described step S200 comprises:

Wherein, described I be not less than 1 positive integer.

Described according to described first presupposition analysis strategy and operation behavior, determine that the analysis that operand described in each is corresponding comprises: determine according to following formulae discovery the assay value S that operand described in each is corresponding;

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

Off-line analysis server described in the present embodiment and real-time analysis server adopt same analysis logic (i.e. described first analysis strategy) to form analysis result, specifically define in the present embodiment and according to the first analysis strategy, the operand that user operation is pointed to is analyzed, thus the analysis result obtained is the operand that user behavior points to.In concrete implementation procedure, can also to the operation behavior of user behavior itself, as the type of operation behavior, the relevance etc. between two operations is analyzed.

When described multimedia messages is TV play, described user behavior is for chasing after acute behavior; Below that method described in the present embodiment is applied to the concrete example that described user chases after acute behavior;

S＝(N/M)*W1+(P/M)*W2+f(t,T)*W3+R*W4

Sort according to described assay value, the play row choosing the individual described assay value of maximum I corresponding according to ranking results forms described first analysis result.

Adopt four to analyze the chase after acute behavior of dimension to user in the present embodiment to analyze, and obtain assay value, form the first analysis result according to assay value.Described first analysis result is the set wanting the collection of drama seen according to the current most probable of above-mentioned formula user, the concrete acute name etc. as 20 TV play.

Comprehensively above-mentioned, the present embodiment, on the basis of said method embodiment, has done further improvement, specifically defines to chase after acute behavior to user and analyze and concrete analytical approach, has the advantages such as the accurate and computation complexity of analysis result is low.

Embodiment of the method nine:

As shown in Figure 7, described method also comprises:

Step S201: described off-line analysis server also analyzes user behavior data according to the second analysis strategy, obtains modifying factor;

Step S202: described modifying factor is sent to described real-time analysis server;

Described modifying factor is for revising described first analysis strategy.

In concrete implementation procedure, described off-line analysis server is also by described first analysis strategy in off-line analysis server according to described modifying factor correction; W as described in specifically in formula as described in correction above-described embodiment _j.

Off-line analysis server described in the present embodiment not only can send historic user behavioral data according to the first analysis result to real-time analysis server, also transmission is used for the modifying factor revising the first analysis strategy, to revise described first analysis strategy, thus described real-time analysis server can be made to obtain more accurate second analysis result, be convenient to follow-uply carry out user's value analysis or recommend corresponding service to user.

Described off-line analysis server and described real-time analysis server incite somebody to action the first analysis strategy according to described modifying factor correction separately.

User behavior analysis method described in the present embodiment, the basis of above-described embodiment has been done further improvement, described off-line analysis server not only can be analyzed formation first analysis result and also will obtain the modifying factor of the first analysis strategy according to the second analysis strategy, to avoid described first analysis strategy Long-Time Service, do not carry out changing the problem that accurate analysis result can not be well provided.

Embodiment of the method ten:

As shown in Figure 8, the present embodiment provides a kind of user behavior analysis method, and described method comprises:

Step S310: off-line analysis server, when real-time analysis server starts, according to the first analysis result formed according to the first presupposition analysis strategy, determines the historic user behavioral data sent to real-time analysis server; Wherein, described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Step S320: off-line analysis server sends described historic user behavioral data to described real-time analysis server;

Step S330: real-time analysis server obtains the active user's behavioral data corresponding to active user's behavior; Wherein, described historic user behavioral data and described active user's behavioral data include the operand of user operation and described user operation sensing;

Step S340: real-time analysis server according to the first presupposition analysis strategy, is analyzed described historic user behavioral data and described for behavior current data, obtained the second analysis result.

In the present embodiment, analyze for behavior in conjunction with off-line analysis server and real-time analysis server, so both solve the problem that the time delay of off-line analysis server is large, also solve the inadequate problem of real-time analysis server stability simultaneously.The method of this off-line analysis server and real-time analysis server binding analysis is applied in the user behavior analysis of free online multimedium information of mass users, time delay and analysis result stability easily performance be obtained for very large lifting.

The step that described real-time analysis server performs in the present embodiment can see embodiment of the method one to embodiment of the method five; The step that described off-line analysis server performs can see embodiment of the method six to embodiment of the method nine, and detailed operation and concrete beneficial effect, see the corresponding part of above-described embodiment, just no longer do at this and further illustrate.

Embodiment of the method 11:

As shown in Figure 9, the present embodiment provides a kind of user behavior analysis method, and described method comprises:

Step S301: off-line analysis server is analyzed analysis strategy according to second and analyzed user behavior data, obtains modifying factor;

Step S302: described modifying factor is sent to real-time analysis server by off-line analysis server;

Step S303: real-time analysis server receives the modifying factor that described off-line analysis server sends;

Step S304: real-time analysis server is according to the first analysis strategy described in described modifying factor correction;

Step S305: off-line analysis server is according to the first analysis strategy described in described modifying factor correction;

Step S310: off-line analysis server, when real-time analysis server starts, according to the first analysis result formed according to the first presupposition analysis strategy, determines the historic user behavioral data sent to real-time analysis server;

Step S330: real-time analysis server obtains the active user's behavioral data corresponding to active user's behavior;

Step S340: real-time analysis server according to the first presupposition analysis strategy, is analyzed described historic user behavioral data and described for behavior current data, obtained the second analysis result;

Wherein, described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Described historic user behavioral data and described active user's behavioral data include the operand of user operation and described user operation sensing;

Real-time analysis server described in the present embodiment not only receives historic user behavioral data from described off-line analysis server and also receives modifying factor, and according to the first analysis strategy described in described modifying factor correction, to obtain the second analysis result more accurately; In concrete implementation procedure, the cycle that described off-line analysis server forms described modifying factor leads to and is greater than the cycle that described off-line analysis server forms described first analysis result, changes too frequently to avoid the first analysis strategy.Described in concrete implementation procedure, off-line analysis server self also can be stored in the first analysis strategy of off-line analysis server end according to described first analysis strategy correction; Thus more accurate analysis result can be obtained, be conducive to subsequent user value analysis or recommend to comprise the operand of at least two described multimedia messagess to user; And by analytical approach that off-line analysis server and real-time analysis server combine, being applied to operand comprises in the analytical applications scene of the user behavior of at least two multimedia messagess, effectively solve in existing user operation multimedia messages process, the problem such as the large or stability of time delay is inadequate.

Apparatus embodiments one:

As shown in Figure 10, the present embodiment provides a kind of real-time analysis server, and described server comprises:

Receiving element 110, for after real-time analysis server starts, receives the historic user behavioral data that off-line analysis server sends according to the first analysis result; Described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Acquiring unit 120, for obtaining the active user's behavioral data corresponding to active user's behavior; Described historic user behavioral data and described active user's behavioral data include the operand of user operation and described user operation sensing;

First analytic unit 130, for according to described first presupposition analysis strategy, analyzes described historic user behavioral data and described active user's behavioral data, obtains the second analysis result;

The concrete structure of described receiving element 110 can comprise wired or wireless communication interface, concrete as structures such as light communication interface, twisted-pair communication interface or receiving antennas.

The concrete structure of described acquiring unit 120 obtains the difference of the mode of described user behavior current data according to device and structure is different, user behavior current data as described in specifically receiving from user behavior data processing server as passed through, then the concrete structure of described acquiring unit 120 can be wired or wireless communication interface equally.

The concrete structure of described first analytic unit 130 can comprise processor and storage medium; Described storage medium stores computer-executable code, and described processor performs described computer-executable code and according to the first presupposition analysis strategy, can analyze described user behavior historical data and described user behavior current data, obtain the second analysis result.Described processor can be the structure that central processing unit, digital signal processor or microprocessor or programmable array etc. have processing capacity.

In concrete implementation procedure, described first analytic unit 130 can also be used for carrying out user's value analysis according to described second analysis result or recommending related service etc. to user.

The definition of the operand described in the present embodiment, user operation and described multimedia messages see the description of corresponding part in embodiment of the method one, just further can not illustrate at this.

Real-time analysis server described in the present embodiment is the method described in embodiment of the method one, provides hardware support, to be used in implementation method embodiment one technical scheme described arbitrarily, and same have the advantages such as the little or memory data output of time delay is few.

Apparatus embodiments two:

As shown in figure 11, described first analytic unit 130 comprises:

Determination module 131, for according to described first presupposition analysis strategy and operation behavior, determines the assay value that operand described in each is corresponding;

Order module 132, for sorting to described assay value, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described second analysis result;

Wherein, described I be not less than 1 positive integer.

The concrete structure of described determination module 131 and described order module 132 all may correspond in server; Described determination module 131 and described order module 132 accessible site correspond to same server or servers corresponding different respectively; When described determination module 131 and described order module 132 is integrated correspond to same server time, described server adopts time division multiplex or concurrent thread to process function corresponding to each module respectively.In addition, described order module 131 can also comprise comparer, for sorting to described assay value.

The present embodiment further defines the structure of described first analytic unit 130, and specifically to define be carry out analyzing and processing to operand in user behavior; The present embodiment is that any described method of embodiment of the method two provides hardware support, and same has the advantage that time delay is little and memory data output is little.

Apparatus embodiments three:

As shown in figure 11, described first analytic unit 130 comprises:

Wherein, described I be not less than 1 positive integer.

Described determination module 131, specifically for determining according to following formulae discovery the assay value S that each operand is corresponding;

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

The concrete structure of described determination module can comprise counter or have the processor of computing function, for obtaining described assay value from one or more analysis dimension, have algorithm simple, realize the easy and simple advantage of structure.

In concrete implementation procedure, described determination module is specifically for analyzing according to the acute behavior that chases after of following algorithm to user's free online multimedium information: wherein, the collection in a corresponding TV play of described multimedia messages;

S＝(N/M)*W1+(P/M)*W2+f(t,T)*W3+R*W4

Wherein, described formula S=(N/M) * W1+ (P/M) * W2+f (t, T) * W3+R*W4 is described formula an object lesson.

Apparatus embodiments four:

Described receiving element 110, also for receiving the modifying factor that described off-line analysis server sends; Wherein, described modifying factor is that described off-line analysis server is determined the analysis of historic user behavioral data based on the second presupposition analysis strategy;

As shown in figure 12, described server also comprises:

Amending unit 140, for according to the first analysis strategy described in described modifying factor correction.

The concrete structure of described amending unit 140 can comprise processor and storage medium; Described storage medium stores computer-executable code, and described processor performs described computer-executable code can according to the first analysis strategy described in modifying factor correction.Described processor can be the structure that central processing unit, digital signal processor or microprocessor or programmable array etc. have processing capacity.

The present embodiment is the further improvement on above-mentioned arbitrary equipment embodiment basis, to can be used in implementation method embodiment one to embodiment of the method four technical scheme described arbitrarily, same to have analysis result time delay little and obtain the stability advantages of higher of analysis result.

Apparatus embodiments five:

Described server also comprises transmitting element; Described transmitting element is used for sending described second analysis result to described off-line analysis server;

When described off-line analysis server does not form the analysis result more late than the formation time of described second analysis result, described second analysis result is used for as described first analysis result.

The concrete physical arrangement of described transmitting element 140 can be wired or wireless communication interface, concrete as structures such as light communication interface, twisted-pair communication interface or transmitting antennas.

Connected by Internet between described real-time analysis server and described off-line analysis server.

Real-time analysis server described in the present embodiment also comprises transmitting element 140 and sends the second analysis result to off-line analysis server, and when off-line analysis server does not form the analysis result more late than the formation time of described second analysis result, as the first analysis result, the degree of accuracy of the second analysis result effectively can be promoted like this.In concrete implementation procedure, described transmitting element 140 sends described second analysis result specifically for closing off-line analysis server described in forward direction at described real-time analysis server; Such energy reduces data volume mutual between described real-time analysis server and described off-line analysis server, saves energy consumption, reduces the busy degree of off-line analysis server.

Apparatus embodiments six:

As shown in figure 13, the present embodiment provides a kind of off-line analysis server, and described server comprises:

Determining unit 210, for after real-time analysis server starts, off-line analysis server, according to the first analysis result formed according to the first presupposition analysis strategy, determines the historic user behavioral data sent to real-time analysis server; Wherein, described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy;

Transmitting element 220, for sending described historic user behavioral data to described real-time analysis server;

The concrete structure of institute's determining unit 210 can comprise processor and storage medium; Described storage medium stores computer-executable code, and described processor performs described computer-executable code can determine according to the first analysis result the user behavior historical data treating to send to described real-time analysis server.Described processor can be the structure that central processing unit, digital signal processor or microprocessor or programmable array etc. have processing capacity.

The concrete structure of institute's transmitting element 220 can comprise wired or wireless communication interface, concrete as structures such as light communication interface, twisted-pair communication interface or receiving antennas.

The description of the behavioral data of historic user described in the present embodiment, active user's behavioral data, described user operation, operand and multimedia messages see corresponding part in embodiment of the method one, just no longer can be described in detail at this.

Off-line analysis server described in the present embodiment is the method described in embodiment of the method six, provides hardware support, to be used in implementation method embodiment six technical scheme described arbitrarily, and same have the advantages such as the little or memory data output of time delay is few; Off-line analysis server described in the present embodiment is used in the user behavior to all media information described in user's viewing or program request, is the operation behavior such as customer multi-media information on demand of magnanimity, improves and have the analytical structure that time delay is little and stability is high.

Apparatus embodiments seven:

As shown in figure 14, described server also comprises the second analytic unit 230 forming described first analysis according to the first presupposition analysis strategy;

Described second analytic unit 230, for according to described first presupposition analysis strategy and operation behavior, determines the assay value that operand described in each is corresponding; And described assay value is sorted, the operand meeting the first pre-conditioned I described assay value corresponding according to ranking results forms described first analysis result;

Wherein, described I be not less than 1 positive integer.

The concrete structure of the second analytic unit 230 can comprise processor and storage medium; Described storage medium stores computer-executable code, and described processor performs described computer-executable code can form the first analysis result according to the first analysis strategy; The first analysis strategy forming the second analysis result at the first analysis strategy described in concrete implementation procedure in serving with real-time analysis is consistent.Described processor can be the structure that central processing unit, digital signal processor or microprocessor or programmable array etc. have processing capacity.

Described off-line analysis server is also specially provided with the second analytic unit 230 for the formation of the first analysis result in the present embodiment, concrete realize hardware for the method described in the real-time example of method seven provides, decrease the data interaction between offline service device and real-time server.

Apparatus embodiments eight:

Wherein, described I be not less than 1 positive integer.

Described second analytic unit 230, specifically for determining according to following formulae discovery the assay value S that operand described in each is corresponding;

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

In concrete implementation procedure, described in can be specifically rewritten into following formula:

S＝(N/M)*W1+(P/M)*W2+f(t,T)*W3+R*W4

Described second analytic unit 230 is specifically determined the assay value of operand described in each according to above-mentioned formula and is formed described first analysis result.

Described second analytic unit 230 specifically can be the processor comprising counter or have computing function in the present embodiment, for sorting to assay value according to above-mentioned functional relation computational analysis value.

Server described in the present embodiment is the method described in embodiment of the method 12, provides support, and same has the low advantage of computation complexity.

Apparatus embodiments nine:

Described second analytic unit 230, also for also analyzing user behavior data according to the second analysis strategy, obtains modifying factor;

Institute's transmitting element 220, for sending to described real-time analysis server by described modifying factor;

Described modifying factor is for revising described first analysis strategy.

The second analytic unit 230 described in the present embodiment obtains modifying factor according to the second analysis strategy, and sent to described real-time analysis server by described transmitting element 220, in concrete implementation procedure, described second analytic unit 230 also corrects correcting according to described modifying factor the first analysis strategy stored therein.Comprehensively above-mentioned, the present embodiment provides a kind of off-line analysis server, can be used for any technology of method described in embodiment of the method nine

Apparatus embodiments ten:

Described modifying factor is for revising described first analysis strategy.

In concrete implementation procedure, described off-line analysis server also comprises receiving element; The second analysis result that described receiving element sends for receiving described real-time analysis server; The concrete structure of described receiving element comprises receiving interface, as the twisted-pair feeder receiving interface in wireline interface; The structures such as the receiving antenna in wave point.Described receiving element specifically for receiving described second analysis result, to reduce the data interaction between off-line analysis server and real-time analysis server before described real-time analysis server closedown; Wherein, described second analysis result is used for when described off-line analysis server does not form the analysis result more late than the formation time of described second analysis result, described second analysis result is considered as described first analysis result, as determining the foundation sending historic user data to real-time analysis server.

The present embodiment is the further improvement on the basis of apparatus embodiments six to apparatus embodiments nine, concrete realizes structure for any technical scheme of method described in embodiment of the method ten provides; The advantages such as the same time delay with analysis result formation is little.

Apparatus embodiments 11:

As shown in figure 15, the present embodiment provides an analytic system, and described analytic system comprises off-line analysis server 200, real-time analysis server 100 and user behavior data processing server 300; Described user behavior data processing server 300 is for receiving from client 400 and client 500 user behavior data reported, and the user behavior data received is formed the data of same data layout, be distributed to real-time analysis server 200 and off-line analysis server 100 respectively.

Described off-line analysis server 200 and real-time analysis server 100 also can the user behavior datas that report of direct reception client end.

In concrete implementation procedure, described off-line analysis server 200, when described real-time analysis server 100 starts, determines the historic user behavioral data sent to real-time analysis server 100 according to the first analysis result formed; And send described historic user behavioral data to real-time analysis server 200.Described first analysis result is before described real-time analysis server starts, the analysis result formed according to the first presupposition analysis strategy.

Described real-time analysis server 100, when starting, receives described historic user behavioral data from described off-line analysis server 200, and from active user's behavioral data that described user behavior data processing server 300 receives; Adopt the first presupposition analysis strategy to analyze described historic user behavioral data and active user's behavioral data again, form the second analysis result.Described first analysis result and described second analysis result are the set comprising operand described at least one; A described operand correspondence at least two multimedia messagess; The information content corresponding to the multimedia messages of same described operand has temporal associativity.

Described off-line analysis server 200, real-time analysis server 100, user behavior data processing server 300, client 400 and client 500 all can be connected by internet.Described Internet can be the hybrid network of wired network or wireless network or wired network and wireless network.

In several embodiments that the application provides, should be understood that disclosed equipment and method can realize by another way.Apparatus embodiments described above is only schematic, such as, the division of described unit, be only a kind of logic function to divide, actual can have other dividing mode when realizing, and as: multiple unit or assembly can be in conjunction with, maybe can be integrated into another server, or some features can be ignored, or do not perform.In addition, the coupling each other of shown or discussed each ingredient or direct-coupling or communication connection can be by some interfaces, and the indirect coupling of equipment or unit or communication connection can be electrical, machinery or other form.

The above-mentioned unit illustrated as separating component or can may not be and physically separates, and the parts as unit display can be or may not be physical location, namely can be positioned at a place, also can be distributed in multiple network element; Part or all of unit wherein can be selected according to the actual needs to realize the object of the present embodiment scheme.

In addition, each functional unit in various embodiments of the present invention can all be integrated in a processing module, also can be each unit individually as a unit, also can two or more unit in a unit integrated; Above-mentioned integrated unit both can adopt the form of hardware to realize, and the form that hardware also can be adopted to add SFU software functional unit realizes.

One of ordinary skill in the art will appreciate that: all or part of step realizing said method embodiment can have been come by the hardware that programmed instruction is relevant, aforesaid program can be stored in a computer read/write memory medium, this program, when performing, performs the step comprising said method embodiment; And aforesaid storage medium comprises: movable storage device, ROM (read-only memory) (ROI, Read-OJly IeIory), random access memory (RAI, RaJdoI Access IeIory), magnetic disc or CD etc. various can be program code stored medium.

The above; be only the specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, is anyly familiar with those skilled in the art in the technical scope that the present invention discloses; change can be expected easily or replace, all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of described claim.

Claims

1. a user behavior analysis method, is characterized in that,

Described method comprises:

2. method according to claim 1, is characterized in that,

Wherein, described I be not less than 1 positive integer.

3. method according to claim 2, is characterized in that,

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

4. the method according to any one of claims 1 to 3, is characterized in that,

Described method also comprises:

5. a user behavior analysis method, is characterized in that,

Described method comprises:

6. method according to claim 5, is characterized in that,

Described method also comprises:

Wherein, described I be not less than 1 positive integer.

7. method according to claim 6, is characterized in that,

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

8. the method according to any one of claim 5 to 7, is characterized in that,

Described method also comprises:

Described modifying factor is sent to described real-time analysis server;

Described modifying factor is for revising described first analysis strategy.

9. a real-time analysis server, is characterized in that,

Described server comprises:

10. server according to claim 9, is characterized in that,

Described first analytic unit comprises:

Wherein, described I be not less than 1 positive integer.

11. servers according to claim 10, is characterized in that,

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

12. servers according to any one of claim 9 to 11, is characterized in that,

Described server also comprises:

13. 1 kinds of off-line analysis servers, is characterized in that,

Described server comprises:

14. servers according to claim 13, is characterized in that,

Described server also comprises the second analytic unit;

Wherein, described I be not less than 1 positive integer.

15. servers according to claim 14, is characterized in that,

S = Σ_{j = 1}^{j = J} f_{j} (j) * w_{j}

Wherein, described J be not less than 1 positive number.

16., according to claim 13 to the server described in 15 any one, is characterized in that,

Described modifying factor is for revising described first analysis strategy.