CN104834652A - Short message service strategy construction method and device thereof serving to social network - Google Patents

Abstract

Each embodiment way of the invention provides a short message service strategy construction method and device serving to a social network. The method comprises the following steps: selecting the characteristics of user angles and system angles, wherein each user angle and each user angle independently have multiple dimensionalities; training a data set of the characteristics of the user angles and system angles to construct a Bayesian probabilistic model; and on the basis of the Bayesian probabilistic model, generating an optimal short message service strategy by aiming at users in the social network. The optimal short message strategy can be implemented so as to improve a success rate of the short message service invitation registration and/ or recall of the social network.

Description

A kind of construction method and device thereof of serving the note strategy of social networks

Technical field

The embodiments of the present invention social networks, relates more specifically to a kind of construction method and device thereof of serving the note strategy of social networks.

Background technology

Along with the development in Web2.0 epoch, SNS has been dissolved in the life of people, and people get to know new old friend on social networks, share Life intravenous drip mutually.Current, smart mobile phone is more and more universal, and mobile Internet is more and more perfect.How making good use of the brand-new information that mobile Internet brings, and reasonably process the challenge that it brings simultaneously, is an increasingly important problem.

As everyone knows, the feature affecting the behaviors such as user's registration, login, activity, friend-making is too numerous to enumerate, and the participation or the liveness that how to improve the behaviors such as the registration of these users, login, activity, friend-making are the problems that social networks network operator is concerned about.

Therefore, need now the model that structure one is perfect, it can learn and extract crucial feature, thus best note invitation registration can be implemented based on this model and/or recall strategy, to improve participation or the liveness of social networks, and evolution in time can also automatically upgrade himself model, to improve the accuracy of model.

Summary of the invention

In view of above present situation, the present invention's object is at least to provide a kind of construction method and device of serving the note strategy of social networks, based on the method and device, best note strategy can be implemented, thus the note invitation registration improving social networks and/or the success ratio of recalling.

Dynamically updating of implementation model is at least also according to another object of the present invention.

According to a first aspect of the invention, provide a kind of construction method of serving the note strategy of social networks, comprising: the feature selecting user perspective and system perspective, wherein each user perspective and each system perspective all have multiple dimension; The data set of the feature of described user perspective and system perspective is trained, builds bayesian probability model; And for the user in social networks, based on described bayesian probability model, produce best described note strategy.

According to the preferred embodiment of the invention, described note strategy comprises at least one that note invitation registration strategy and note are recalled in strategy.

According to the preferred embodiment of the invention, when implementing described note invitation registration strategy, whether the feature of described user perspective comprises inviter's quantity, invites temperature, invites time gap and user to have by least one in the registered described social networks of other channels; And the feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence and send invite the time interval and frequency at least one.

According to the preferred embodiment of the invention, when implementing described note and recalling strategy, the feature of described user perspective comprises inviter's quantity, invites temperature, invites time gap, user whether have by the registered described social networks of other channels, recall before log in frequency and recall in front login time at least one; And the feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence, send invite the time interval and frequency, note push good friend different content and push good friend and pushed user intimate degree at least one.

According to the preferred embodiment of the invention, be also, based on the new samples collection obtained in time, judge whether to need to carry out automatic online renewal to described bayesian probability model.

According to the preferred embodiment of the invention, the step of described judgement comprises the algorithm based on Pearson's Chi-square Test, calculates chi-square value, and estimates based on described chi-square value the probability accepting or refuse original hypothesis.

According to a second aspect of the invention, provide a kind of construction device of serving the note strategy of social networks, comprise: feature selecting device, be configured to the feature selecting user perspective and system perspective, wherein each user perspective and each system perspective all have multiple dimension; Model training apparatus, is configured to train the data set of the feature of described user perspective and system perspective, builds bayesian probability model; And tactful output unit, be configured to for the user in social networks, based on described bayesian probability model, produce best described note strategy.

According to the preferred embodiment of the invention, being also, model modification device, being configured to the new samples collection based on obtaining in time, judge whether to need to carry out automatic online renewal to described bayesian probability model.

According to the preferred embodiment of the invention, be also, the step of described judgement comprises the algorithm based on Pearson's Chi-square Test, calculates chi-square value, and estimates based on described chi-square value the probability accepting or refuse original hypothesis.。

Accompanying drawing explanation

When reading the detailed description hereafter to exemplary embodiment by reference to the accompanying drawings, these and other object, feature and advantage will become apparent, in the accompanying drawings:

Fig. 1 shows the bayesian probability model implementing note invitation registration strategy according to the preferred embodiment of the invention;

Fig. 2 shows a kind of according to the preferred embodiment of the invention process flow diagram of serving the construction method of the note strategy of social networks;

Fig. 3 shows a kind of according to the preferred embodiment of the invention block scheme of serving the construction device of the note strategy of social networks; And

Fig. 4 shows the block scheme of illustrative computer/server that each embodiment of the present invention can realize wherein.

Embodiment

Process flow diagram in accompanying drawing and block diagram, illustrate according to the architectural framework in the cards of the device of various embodiments of the invention, method and computer program product, function and operation.In this, each square frame in process flow diagram or block diagram can represent a part for module, program segment or a code, and a part for described module, program segment or code comprises one or more executable instruction for realizing the logic function specified.Also it should be noted that at some as in the realization of replacing, the function marked in square frame also can be different from occurring in sequence of marking in accompanying drawing.Such as, in fact the square frame that two adjoining lands represent can perform substantially concurrently, and they also can perform by contrary order sometimes, and this determines according to involved function.Also it should be noted that, the combination of the square frame in each square frame in block diagram and/or process flow diagram and block diagram and/or process flow diagram, can realize by the special hardware based system of the function put rules into practice or operation, or can realize with the combination of specialized hardware and computer instruction.

Below for the social networks of Renren Network, how introduction learns and extracts the model of key feature, structure note strategy.People set up social networks by friend relation in Renren Network, and wherein have greatly Renren Network user, also can carry out oneself address list synchronous by mobile terminal, thus form a mobile social networking based on user communication record data.Further to invite user to be registered as example.User starts to isolate often most, affects TA and is invited by good friend thus add network and the factor producing doings then has a lot.From user perspective, depend on that TA is subject to the invitation of how many good friends, or the user itself inviting TA to add network there is how many popularity etc.And influence factor equally also comes from background system, such as, send to the opportunity of invitation, and the short message content etc. of inviting.

Present inventor, after deep data analysis and research, finds when implementing invitation registration strategy, selects the feature in following two angles to be useful especially:

One, the feature of user perspective:

1. inviter quantity N, such as, according to different invitation numbers, can design corresponding dimension is 4;

2. invite temperature M, such as, according to the invitation number of times of different inviter, can design corresponding dimension is 3;

3. invite time gap gap, such as weigh according to each invitation time gap so far, can design corresponding dimension is 7;

4. whether user has by the registered Renren Network flag of other channels, such as, design corresponding dimension for 2 with this;

Two, the feature of system perspective:

1. send to the time T of invitation, such as with five time periods in one day distinguish (morning, noon, afternoon, at dusk, night), can design corresponding dimension is 5;

2. the note official documents and correspondence W invited, such as, according to the official documents and correspondence of different-style and content, can design corresponding dimension is 20;

3. send the time interval of inviting and frequency D, such as can design corresponding dimension is 3;

Above-mentioned seven features can have 4*3*7*2*5*20*3=50400 kind to combine according to the dimension of above-mentioned design.

Certainly, to it will be understood by those skilled in the art that in addition to the above features or instead, other user perspective and the feature of system perspective are also possible.But, according to various embodiments of the present invention, preferably implement the feature cited by above-mentioned user perspective and system perspective, build the note Policy model of the application.In addition, other user perspective and the dimension of system perspective are also possible.

Fig. 1 shows the bayesian probability model implementing note invitation registration strategy according to the preferred embodiment of the invention.

According to various embodiments of the present invention, described bayesian probability model can be preferably naive Bayesian probability model.

As shown in Figure 1, wherein pass through the training sample data collection of four class user characteristicses and three type systematic features, train for different user feature (N, M, gap in a model, flag) user, which type of system invites strategy (T, W, D) should be adopted, thus the user behavior probability of " affirmative " can be maximized, thus interactive critical mode between different characteristic can be obtained in a model.

In the model trained, for one not at the new user i of above-mentioned data centralization, its user characteristics U can be analyzed _i=(n, m, gap _i, flag _i), by maximizing the behavior probability of its " affirmative ", from the model trained, filter out crucial system features S _i=(t, w, d).Then, apply the system features that obtains and carry out short message sending, thus the probability this user being accepted the invitation add good friend's network reaches maximum.

Key equation in above-mentioned bayesian probability model can be expressed from the next,

$\mathrm{Pr}\left(S_i\right|U_i)=\underset{x\∈X}{\mathrm{\Σ}}\mathrm{Pr}\left(S_i\right|U_x,U_i)\·\mathrm{Pr}\left(U_x\right|U_i)=\underset{x\∈X}{\mathrm{\Σ}}\frac{\mathrm{Pr}\left(U_i\right|S_i,U_x)}{\mathrm{Pr}\left(U_i\right|U_x)}\·\mathrm{Pr}\left(S_i\right|U_x)\·\mathrm{Pr}\left(U_x\right|U_i)$

Wherein Pr represents conditional probability, U _iand S _irepresent the user characteristics of user i and corresponding optimizer system feature, X is training sample set.

Be described above the model implementing note invitation registration strategy for user.But, do not log in the situation of social networks for registered user for a long time, be necessary that implementing note recalls strategy, with the liveness of adding users.This note is recalled to the model of strategy, with the model class of said short message invitation registration strategy seemingly, but can increase further and/or revise characteristic details wherein.Such as, the feature of user perspective can increase by two following features:

5. log in frequency before recalling, such as, can be designed to 4 dimensions;

6. recall front login time, such as, can relate to into 5 dimensions.

The feature of system perspective then can increase:

4. note pushes the different content of good friend, such as, can be divided into 3 dimensions of friend recommendation, the strange thing enlivening good friend, popular strange thing etc.;

5. the good friend pushed and the intimate degree of pushed user, such as, be designed to 4+3=7 dimension by common good friend's number and active degree etc.

Similarly, the feature of above-mentioned increase is only preferred in each embodiment of the application, but those skilled in the art can design the user characteristics and system features that increase other, and designs corresponding dimension.

Then, similarly based on training sample data collection, thus recall strategy for different users applies different notes, make note recall effects reach optimization.

It will be understood by those skilled in the art that the effect that current sample can have been obtained optimum invitation and recalled according to the parameter trained in bayesian probability model designed above the present invention.But along with the deduction of time, user group, use habit, network environment may occur to change always, master mould may need to carry out in real time checking the effect seen and whether can also obtain expectation to new sample.When effect occurs significantly to decline time, just need to carry out readjusting of model parameter.And manually ceaselessly test the manpower and materials of at substantial, the research therefore for the renewal of model parameter automatic online is also significant.

The dynamic update algorithm that the application creatively proposes based on Pearson's Chi-square Test (Pearson ' schi-squared test) realizes this function.Suppose at time t ₀, model M ₀the bayesian probability model that the parameter starting most to train out based on sample set is formed, at second time point t ₁=t ₀+ △ t, in time period t ₀→ t ₁in create the new sample do not observed.This sample is tested, observes it and whether substantially can also meet model M ₀in parameter estimation, namely Pr (S|U consistent is substantially obtained to new samples collection _x), basically, there is not large-scale change in yet i.e. registration rate/recall rate.

Therefore, original hypothesis H0 is proposed: the system property probability distribution that user property contains does not change (i.e. Pr (S|U _x) be consistent).So, following statistic form is had:

	U 0	U 1	......	U n
					t0	Pr 0(S|U 0)	Pr 0(S|U 1)	......	Pr 0(S|U n)
t1	Pr 1(S|U 0)	Pr 1(S|U 1)	......	Pr 1(S|U n)

According to the algorithm frame of Pearson's Chi-square Test, in order to calculate chi-square value, can in first computation sheet the expectation of each field as follows:

$E_{\mathrm{pq}}=\frac{\left({\mathrm{\Σ}}_{t=0}^1{\mathrm{Pr}}_t\left(S\right|U_q)\right)\·\left({\mathrm{\Σ}}_{k=0}^n{\mathrm{Pr}}_p\left(S\right|U_k)\right)}{{\mathrm{\Σ}}_{k=0}^n{\mathrm{\Σ}}_{t=0}^1{\mathrm{Pr}}_t\left(S\right|U_k)}$

Thus, chi-square value can be calculated as follows:

${\mathrm{\χ}}^2={\mathrm{\Σ}}_{j=0}^n{\mathrm{\Σ}}_{i=0}^1\frac{{({\mathrm{Pr}}_i\left(S\right|U_j)-E_{\mathrm{ij}})}^2}{E_{\mathrm{ij}}}$

In order to determine to accept original hypothesis or refusal original hypothesis H0, need to consider degree of freedom df=(2-1) * (n-1)=n-1 simultaneously.Then, set a level of significance according to actual needs, usually can be set as 0.05.Based on level of significance and degree of freedom, look into dividing value table and can obtain the threshold value p under the critical level of degree of freedom df and setting, the chi-square value calculated and this threshold value p are compared, decides to accept or refusal original hypothesis H0, can Renewal model be determined whether.If accepted, then without Renewal model parameter, i.e. M ₁=M ₀; If refusal hypothesis, then need to re-use bayesian probability model and estimate new parameter, generate new model M ₁.The automatic online iteration update algorithm proposed as follows:

1. utilize bayesian probability model to generate M _t, t is current point in time;

2. utilize model to carry out work, through the △ t time, at t ₁moment collection is positioned at t ₀to t ₁the new samples data produced;

3. utilize new data and master pattern to carry out Chi-square Test, if accept H0, then perform the 2nd step; If refusal H0, then perform the 1st step.

So far, describe the note invitation registration of serving social networks and recall the model building method of strategy and dynamically update technology based on Pearson's Chi-square Test.

Fig. 2 shows a kind of according to the preferred embodiment of the invention process flow diagram of serving the construction method of the note strategy of social networks.

Wherein, according to the preferred embodiment of the invention, described note strategy comprises at least one that note invitation registration strategy and note are recalled in strategy.

Can comprise the following steps according to method S200 of the present invention:

Step S210, select the feature of user perspective and system perspective, wherein each user perspective and each system perspective all have multiple dimension;

In order to obtain best note strategy, inventor, after deep data analysis and research, finds that the feature selecting following two angles is useful especially, namely

When implementing described note invitation registration strategy, whether the feature of described user perspective can preferably include inviter's quantity, invites temperature, invite time gap and user to have by least one in the registered described social networks of other channels; And the feature of described system perspective can comprise send to invitation time, invitation note official documents and correspondence and send invite the time interval and frequency at least one.

When implementing described note and recalling strategy, the feature of described user perspective can preferably include inviter's quantity, invites temperature, invites time gap, user whether have by the registered described social networks of other channels, recall before log in frequency and recall in front login time at least one; And the feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence, send invite the time interval and frequency, note push good friend different content and push good friend and pushed user intimate degree at least one.

Similarly, it will be understood by those skilled in the art that and can increase according to actual conditions or reduce the feature of above-mentioned user perspective and system perspective.

Further, can according to its corresponding dimension of characteristic Design of above-mentioned user perspective and system perspective.Preferably, described dimension is multiple dimension.

Then, step S220, trains the data set of the feature of described user perspective and system perspective, builds bayesian probability model;

This step needs the data set of the feature to user perspective and system perspective to divide, and as the training of sample data collection, thus obtain the desired bayesian probability model of the present invention.

According to various embodiments of the present invention, bayesian probability model of the present invention is preferably naive Bayesian probability model.

Step S330, for the user in social networks, based on described bayesian probability model, produces best described note strategy.

This step, such as the new user in social networks, using the input of its corresponding user characteristic data as the bayesian probability model of above-mentioned acquisition, thus obtains best note strategy.

Further, method according to the present invention is also, based on the new samples collection obtained in time, judges whether to need to carry out automatic online renewal to described bayesian probability model.

The step of this judgement can comprise the algorithm based on Pearson's Chi-square Test, calculates chi-square value, and estimates based on described chi-square value the probability accepting or refuse original hypothesis.

The advantage of this step is automatically to correct the bayesian probability model obtained in the past in time, thus can at any time perform note invitation registration with maximum probability or recall strategy.

Finally, step S340 terminates.

Be described above construction method of serving the note strategy of social networks of the present invention.The method advantageously provides best note invitation registration and recalls strategy, thus not only effectively reduces note cost but also obtain splendid tactful effect.In addition, dynamically update technology in time according to proposed by the invention, the real-time accuracy of the model that the application builds can be guaranteed.

Fig. 3 shows a kind of according to the preferred embodiment of the invention block scheme of serving the construction device of the note strategy of social networks.

The construction device 300 of this note strategy at least comprises feature selecting device 310, model training apparatus 320, and tactful output unit 330.Wherein,

Feature selecting device 310, be configured to the feature selecting user perspective and system perspective, wherein each user perspective and each system perspective all have multiple dimension;

Model training apparatus 320, is configured to train the data set of the feature of described user perspective and system perspective, builds bayesian probability model; And

Strategy output unit 330, is configured to for the user in social networks, based on described bayesian probability model, produces best described note strategy.

Similarly, described note strategy comprises at least one that note invitation registration strategy and note are recalled in strategy.

When implementing described note invitation registration strategy, whether the feature of described user perspective preferably includes inviter's quantity, invites temperature, invites time gap and user to have by least one in the registered described social networks of other channels; And the feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence and send invite the time interval and frequency at least one.

When implementing described note and recalling strategy, the feature of described user perspective preferably includes inviter's quantity, invites temperature, invites time gap, user whether have by the registered described social networks of other channels, recall before log in frequency and recall in front login time at least one; And the feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence, send invite the time interval and frequency, note push good friend different content and push good friend and pushed user intimate degree at least one.

Construction of strategy device 300 according to the present invention further preferably comprises model modification device, is configured to the new samples collection based on obtaining in time, and judging whether needs to carry out automatic online renewal to described bayesian probability model; And the step of described judgement comprises the algorithm based on Pearson's Chi-square Test, calculate chi-square value, and estimate based on described chi-square value the probability accepting or refuse original hypothesis.

Only simply describe construction method and the device thereof of the note strategy of serving social networks according to the preferred embodiment of the invention above, but it will be appreciated by those skilled in the art that said method and device are mutually corresponding, those skilled in the art implement the method that the application instructs can build corresponding device and the device situation that configures under the instruction of method under.

Fig. 4 shows the block diagram of illustrative computer/server that each embodiment of the present invention can realize wherein.The computer system/server 412 of Fig. 4 display is only an example, should not bring any restriction to the function of the embodiment of the present invention and usable range.

As shown in Figure 4, computer system/server 412 shows with the form of universal computing device.The assembly of computer system/server 412 can include but not limited to: one or more processor or processing unit 416, system storage 428, connects the bus 418 of different system assembly (comprising system storage 428 and processing unit 416).

Bus 418 represent in a few class bus structure one or more, comprise memory bus or Memory Controller, peripheral bus, AGP, processor or use any bus-structured local bus in multiple bus structure.For example, these architectures include but not limited to industry standard architecture (ISA) bus, MCA (MAC) bus, enhancement mode isa bus, VESA's (VESA) local bus and periphery component interconnection (PCI) bus.

Computer system/server 412 typically comprises various computing systems computer-readable recording medium.These media can be any usable mediums can accessed by computer system/server 412, comprise volatibility and non-volatile media, moveable and immovable medium.

System storage 428 can comprise the computer system-readable medium of volatile memory form, such as random access memory (RAM) 430 and/or cache memory 432.Computer system/server 412 may further include that other are removable/immovable, volatile/non-volatile computer system storage medium.Only as an example, storage system 34 may be used for reading and writing immovable, non-volatile magnetic media (Fig. 4 does not show, and is commonly referred to " hard disk drive ").Although not shown in Fig. 4, the disc driver that removable non-volatile magnetic disk (such as " floppy disk ") is read and write can be provided for, and to the CD drive that removable anonvolatile optical disk (such as CD-ROM, DVD-ROM or other light media) is read and write.In these cases, each driver can be connected with bus 418 by one or more data media interfaces.Storer 428 can comprise at least one program product, and this program product has one group of (such as at least one) program module, and these program modules are configured to the function performing various embodiments of the present invention.

There is the program/utility 440 of one group of (at least one) program module 442, can be stored in such as storer 428, such program module 442 comprises---but being not limited to---operating system, one or more application program, other program modules and routine data, may comprise the realization of network environment in each or certain combination in these examples.Function in program module 442 embodiment that execution is described in the invention usually and/or method.

Computer system/server 412 also can communicate with one or more external unit 414 (such as keyboard, sensing equipment, display 424 etc.), also can make with one or more devices communicating that user can be mutual with this computer system/server 412, and/or communicate with any equipment (such as network interface card, modulator-demodular unit etc.) making this computer system/server 412 can carry out communicating with other computing equipments one or more.This communication can be passed through I/O (I/O) interface 422 and carry out.Further, computer system/server 412 can also such as, be communicated by network adapter 420 and one or more network (such as LAN (Local Area Network) (LAN), wide area network (WAN) and/or public network, the Internet).As shown in the figure, network adapter 420 is by bus 418 other module communications with computer system/server 412.Be understood that, although not shown, other hardware and/or software module can be used in conjunction with computer system/server 412, include but not limited to: microcode, device driver, redundant processing unit, external disk drive array, RAID system, tape drive and data backup storage system etc.

Give instructions of the present invention for the object illustrated and describe, but it is not intended to be exhaustive or be limited to the invention of disclosed form.It may occur to persons skilled in the art that a lot of amendment and variant.It will be appreciated by those skilled in the art that the method and apparatus in embodiment of the present invention can realize with software, hardware, firmware or its combination.

Therefore; embodiment is to principle of the present invention, practical application are described better and enable the other staff in those skilled in the art understand following content and select and describe; namely; under the prerequisite not departing from spirit of the present invention, all modifications made and replacement all will fall in the scope of claims definition.

Claims (12)

1. serve a construction method for the note strategy of social networks, it is characterized in that, comprising:

Select the feature of user perspective and system perspective, wherein each user perspective and each system perspective all have multiple dimension;

The data set of the feature of described user perspective and system perspective is trained, builds bayesian probability model; And

For the user in social networks, based on described bayesian probability model, produce best described note strategy.

2. method according to claim 1, wherein,

Described note strategy comprises at least one that note invitation registration strategy and note are recalled in strategy.

3. method according to claim 2, wherein,

When implementing described note invitation registration strategy, whether the feature of described user perspective comprises inviter's quantity, invites temperature, invites time gap and user to have by least one in the registered described social networks of other channels; And

The feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence and send invite the time interval and frequency at least one.

4. method according to claim 2, wherein,

When implementing described note and recalling strategy, the feature of described user perspective comprises inviter's quantity, invites temperature, invites time gap, user whether have by the registered described social networks of other channels, recall before log in frequency and recall in front login time at least one; And

The feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence, send invite the time interval and frequency, note push good friend different content and push good friend and pushed user intimate degree at least one.

5. method according to claim 1, is also,

Based on the new samples collection obtained in time, judge whether to need to carry out automatic online renewal to described bayesian probability model.

6. method according to claim 5, is also,

The step of described judgement comprises the algorithm based on Pearson's Chi-square Test, calculates chi-square value, and estimates based on described chi-square value the probability accepting or refuse original hypothesis.

7. serve a construction device for the note strategy of social networks, it is characterized in that, comprising:

Feature selecting device, be configured to the feature selecting user perspective and system perspective, wherein each user perspective and each system perspective all have multiple dimension;

Model training apparatus, is configured to train the data set of the feature of described user perspective and system perspective, builds bayesian probability model; And

Strategy output unit, is configured to for the user in social networks, based on described bayesian probability model, produces best described note strategy.

8. device according to claim 7, wherein,

Described note strategy comprises at least one that note invitation registration strategy and note are recalled in strategy.

9. device according to claim 8, wherein,

When implementing described note invitation registration strategy, whether the feature of described user perspective comprises inviter's quantity, invites temperature, invites time gap and user to have by least one in the registered described social networks of other channels; And

The feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence and send invite the time interval and frequency at least one.

10. device according to claim 8, wherein,

When implementing described note and recalling strategy, the feature of described user perspective comprises inviter's quantity, invites temperature, invites time gap, user whether have by the registered described social networks of other channels, recall before log in frequency and recall in front login time at least one; And

The feature of described system perspective comprise send to invitation time, invitation note official documents and correspondence, send invite the time interval and frequency, note push good friend different content and push good friend and pushed user intimate degree at least one.

11. devices according to claim 7, are also,

Model modification device, is configured to the new samples collection based on obtaining in time, and judging whether needs to carry out automatic online renewal to described bayesian probability model.

12. devices according to claim 11, are also,

The step of described judgement comprises the algorithm based on Pearson's Chi-square Test, calculates chi-square value, and estimates based on described chi-square value the probability accepting or refuse original hypothesis.