CN103984775A

CN103984775A - Friend recommending method and equipment

Info

Publication number: CN103984775A
Application number: CN201410246939.6A
Authority: CN
Inventors: 潘照明; 王洲天; 戴朝约; 蒋惟
Original assignee: Netease Hangzhou Network Co Ltd
Current assignee: Netease Hangzhou Network Co Ltd
Priority date: 2014-06-05
Filing date: 2014-06-05
Publication date: 2014-08-13
Also published as: CN110162717A; CN110162717B

Abstract

The embodiment of the invention provides a friend recommending method. The method includes the steps that friend-making conditions of a first user are obtained, and second users with basic information meeting the friend-making conditions are sought for; according to basic information of the first user and the basic information of the second users, matching degrees between the second users and the first user are calculated by using a matching model, wherein the matching model is established in advance according to basic information and historical interaction information amount of two corresponding historical matching users in a social network; according to the matching degrees between the second users and the first user, at least one third user is screened from the second users to be recommended to the first user. According to the method, recommended friends can be more accord with preferences of users, and the situation that friends cannot be recommended can be reduced in social networks. In addition, the embodiment of the invention provides friend recommending equipment.

Description

A kind of method and apparatus of commending friends

Technical field

Embodiments of the present invention relate to network information processing field, and more specifically, embodiments of the present invention relate to a kind of method and apparatus of commending friends.

Background technology

The embodiments of the present invention that be intended to for stating in claims this part provide background or context.Description is not herein because be included in just admit it is prior art in this part.

Along with the universal and development of internet, increasing people link up, get to know friend by social networks, or even get to know love and marriage object.In social networks, get to know new good friend for the ease of user, system can may interested other users be recommended as commending friends certain user to this user, so that this user can carry out information interaction with commending friends.For example, in social networks under the scenes such as love and marriage, each user can have the essential information that represents personal considerations, in the time that system need to be a certain user's commending friends, can be according to representing the make friends friend-making condition (as the condition of choosing spouse) of interest of this user, essential information is met to make friends other users of condition of this user and recommend this user.

For a certain user, conventionally in a social networks essential information can to meet make friends other numbers of users of condition of this user very many, be that this user is really interested and wherein often only there is a little part.In order to recommend more to meet other users of user preference to user, available technology adopting be, obtain with same or analogous at least one other user of this user basic information as similar user according to user's essential information, then from the good friend user of similar user preference, filter out essential information and meet make friends at least one good friend user of condition of this user and recommend this user.Like this, because the user preference that essential information is similar is similar, when prior art is a certain user's commending friends, by good friend user's screening scope being narrowed down to and the good friend user of same or analogous other user institute preferences of this user basic information, make more to meet for the good friend that this user recommends user's preference.

Summary of the invention

But, because the good friend who recommends for this user in prior art is all and the good friend user of the same or analogous similar user of this user basic information institute preference, in social networks, be sometimes difficult to find and the same or analogous similar user of this user basic information, although sometimes can find similar user, but good friend's number of users of similar user preference is also often very few, to such an extent as to being difficult to find essential information to meet the make friends good friend user of condition of this user, these situations all can cause recommending out good friend for this user.

Therefore in the prior art, make in order to recommend more to meet the good friend of user preference to user for the good friend that this user recommends to be all and the good friend user of the same or analogous similar user of this user basic information institute preference, make not have in social networks that this user's similar user or the good friend user of similar user preference easily cause when very few cannot be user's commending friends, this is very bothersome process.

For this reason, be starved of a kind of method and apparatus of improved commending friends, so that in the time recommending more to meet the good friend of user preference to user without screen commending friends from the good friend user of this user's similar user institute preference, thereby avoid not having this user's similar user or the good friend user of similar user preference cannot be the problem of user's commending friends when very few in social networks.

In the present context, embodiments of the present invention expect to provide a kind of method and apparatus of commending friends.

In the first aspect of embodiment of the present invention, a kind of method of commending friends is provided, comprising: obtain the friend-making condition of first user, and search essential information and meet each second user of described friend-making condition; According to the essential information of described first user and the second user's essential information described in each, utilize Matching Model, calculate the matching degree between the second user and described first user described in each, described Matching Model is to set up according to having history mutual information amount between the essential information of mutual corresponding two the historical match user of historical information and corresponding two historical match user in social networks in advance; According to the matching degree between the second user described in each and described first user, from the second user described in each, filter out at least one the 3rd user and recommend to described first user.

In the second aspect of embodiment of the present invention, a kind of method of commending friends is provided, comprising: obtain the friend-making condition of first user, and search essential information and meet each second user of described friend-making condition; According to the essential information of the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance; Preference according to described first user to the second user described in each filters out at least one the 3rd user and recommends to described first user from the second user described in each.

In the third aspect of embodiment of the present invention, a kind of equipment of commending friends is provided, comprising: friend-making Condition Matching module, for obtaining the friend-making condition of first user, and search essential information and meet each second user of described friend-making condition; Matching degree computing module, for according to the essential information of described first user and the second user's essential information described in each, utilize Matching Model, calculate the matching degree between the second user and described first user described in each, described Matching Model is to set up according to having history mutual information amount between the essential information of mutual corresponding two the historical match user of historical information and corresponding two historical match user in social networks in advance; The first recommending module for according to the matching degree between the second user described in each and described first user, filters out at least one the 3rd user and recommends to described first user from the second user described in each.

In the fourth aspect of embodiment of the present invention, a kind of equipment of commending friends is provided, comprising: friend-making Condition Matching module, for obtaining the friend-making condition of first user, and search essential information and meet each second user of described friend-making condition; Preference computing module, for the essential information according to the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance; The second recommending module for the preference to the second user described in each according to described first user, filters out at least one the 3rd user and recommends to described first user from the second user described in each.

According to the method and apparatus of the commending friends of embodiment of the present invention, in the time being first user commending friends, can be in each second user who meets first user friend-making condition, by the corresponding relation between essential information and both history mutual information amounts of corresponding two historical match user, essential information based on first user and each the second user's essential information is predicted and is represented the matching degree of first user to each the second user preference degree, and/or, by the corresponding relation between essential information and this history friend-making behavior of the user as the historical friend-making object of action of first user, essential information based on each the second user is predicted and is represented the preference of first user to each the second user preference degree, thereby can screen the 3rd user for recommending to first user according to the matching degree doping and/or preference.Therefore, when to user's commending friends, can be according to make friends behavior or there is the mutual user of historical information and mate the possibility that account of the history predicts that the good friend of current recommendation is accepted by user of history, this not only makes the good friend who recommends to this user more meet this user's preference, and without screen commending friends from the good friend user of this user's similar user institute preference, thereby in social networks, not having this user's similar user or the good friend user of similar user preference also can realize to this user's commending friends when very few, reduced significantly cannot commending friends situation occur, for user has brought better experience.

summary of the invention

The inventor finds, conventionally it is all very huge in social networks, meeting the make friends number of users of condition of a user, but wherein often only there is the little part can be real interested and be accepted as good friend by this user, in order to recommend more to meet the good friend of user preference to user, the mode that prior art adopts is only mainly to screen the good friend user who recommends this user in the good friend user of the similar user institute preference similar to this user in essential information.In this way, although prior art can make to recommend the preference that this user's good friend user more meets this user, but the good friend's number of users that can not have the similar similar good friend of essential information or a similar good friend this user causes very little cannot be to this user's commending friends, visible, in prior art, cause from the good friend user of this user's similar user institute preference, filtering out for the reason of this user's commending friends is mainly the good friend user who recommends to this user.Therefore, for fear of the problem that cannot recommend, just need to be in making to recommend more to meet the good friend of user preference to user, only avoid adopting from the user of this user's similar user institute preference screening to recommend this user's good friend.

Based on above-mentioned research, basic thought of the present invention is: when for first user commending friends, can be in each second user who meets first user friend-making condition, by the corresponding relation between essential information and both history mutual information amounts of corresponding two historical match user, essential information based on first user and each the second user's essential information is predicted and is represented the matching degree of first user to each the second user preference degree, and/or, by the corresponding relation between essential information and this history friend-making behavior of the user as the historical friend-making object of action of first user, essential information based on each the second user is predicted and is represented the preference of first user to each the second user preference degree, thereby can screen the 3rd user for recommending to first user according to the matching degree doping and/or preference.Like this, not only can make the good friend who recommends to this user more meet this user's preference, and without screen commending friends from the good friend user of this user's similar user institute preference, also can realize to this user's commending friends when very few thereby do not there is this user's similar user or the good friend user of similar user preference in social networks, reduced cannot commending friends situation occur.

After having introduced ultimate principle of the present invention, lower mask body is introduced various non-limiting embodiment of the present invention.

application scenarios overview

First with reference to figure 1, Fig. 1 is the framework schematic diagram of an exemplary application scene of embodiments of the present invention.Wherein, user by client 102 with provide the server 101 of social networking service to carry out alternately.It will be understood by those skilled in the art that the framework schematic diagram shown in Fig. 1 is only the example that embodiments of the present invention can be achieved therein.The scope of application of embodiment of the present invention is not subject to the restriction of this any aspect of framework.

It should be noted that, client 102 herein can be existing, that research and develop or in the future research and development, can be by any type of wired and/or wireless connections (for example, Wi-Fi, LAN, honeycomb, concentric cable etc.) with the mutual any client of server 101, include but not limited to: existing, research and develop or the smart mobile phone of research and development in the future, non intelligent mobile phone, panel computer, laptop PC, desktop personal computer, small-size computer, medium-size computer, mainframe computer etc.

It is also to be noted that, server 101 be herein only existing, research and develop or examples research and development, that the equipment of service resources can be provided to user in the future.Embodiments of the present invention are unrestricted in this regard.

Based on the framework shown in Fig. 1, under the first exemplary application scene, server 101 can obtain the friend-making condition of first user, and searches essential information and meet each second user of described friend-making condition.Then, server 101 can be according to the essential information of described first user and the second user's essential information described in each, utilize Matching Model, calculate the matching degree between the second user and described first user described in each, described Matching Model is to set up according to having history mutual information amount between the essential information of mutual corresponding two the historical match user of historical information and corresponding two historical match user in social networks in advance.In addition after, server 101 can be according to the matching degree between the second user described in each and described first user, filters out at least one the 3rd user and recommend to described first user from the second user described in each.

Based on the framework shown in Fig. 1, under the second exemplary application scene, server 101 can obtain the friend-making condition of first user, and searches essential information and meet each second user of described friend-making condition.Then, server 101 can be according to the essential information of the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance.In addition after, server 101 can be according to described first user the preference to the second user described in each, from the second user described in each, filter out at least one the 3rd user and recommend to described first user.

Be understandable that, in application scenarios of the present invention, although herein and below by the action description of embodiment of the present invention for to be carried out by server 101, these actions also can be carried out by client 102, can certainly part by client 102 carry out, part carried out by server 101.The present invention is unrestricted aspect executive agent, as long as carried out the disclosed action of embodiment of the present invention.

illustrative methods

Below in conjunction with the application scenarios of Fig. 1, with reference to figure 2～5, the method for commending friends according to exemplary embodiment of the invention is described.It should be noted that above-mentioned application scenarios is only to illustrate for the ease of understanding spirit of the present invention and principle, embodiments of the present invention are unrestricted in this regard.On the contrary, embodiments of the present invention can be applied to applicable any scene.

Referring to Fig. 2, show the process flow diagram of method one embodiment of commending friends in the present invention.In the present embodiment, for example specifically can comprise the steps:

Step 201, obtain the friend-making condition of first user, and search essential information and meet each second user of described friend-making condition.

Arbitrary user in social networks, this user can arrange the essential information that represents its personal considerations on the one hand, this user can arrange the friend-making condition that its good friend's personal considerations need to meet according to demands of individuals on the other hand, and other users can check this user's essential information and friend-making condition, carry out friend-making behavior to determine whether with this user.For example, in marriage and making friend's social networks, friend-making condition can be user's the condition of choosing spouse.And for example, in the social networks of linking up alumnus, friend-making condition can be user's school, institute etc.Wherein, friend-making condition for example can comprise the restrictive conditions such as sex, area, age, height, income, educational background.

In the present embodiment, for to first user commending friends, can obtain the friend-making condition that first user sets in advance, then go to mate the friend-making condition of first user with the essential information of other users in social networks, thereby find out essential information and meet the second user of first user friend-making condition.Wherein, the friend-making condition of first user for representing the requirement of first user to its good friend's individual subscriber situation, for example, can comprise the conditions such as location, academic situation, working condition, personality; The second user's essential information is for representing the second user's personal considerations, for example, can comprise the second user's the information such as name, location, academic situation, working condition, personality.

It should be noted that, because the friend-making condition of first user often only can embody demands of individuals roughly, and the second user's essential information also often only can embody personal considerations roughly, therefore, often to meet the second number of users of first user friend-making condition very many for essential information, but many second users real interested good friend user that is not first user.For this reason, after finding the second user, also need to enter subsequent step, to continue screening from the second user, thereby more met the good friend user of first user preference.

Step 202, according to the essential information of described first user and the second user's essential information described in each, utilize Matching Model, calculate the matching degree between the second user and described first user described in each, described Matching Model is to set up according to having history mutual information amount between the essential information of mutual corresponding two the historical match user of historical information and corresponding two historical match user in social networks in advance.

In the present embodiment, related " model " can represent, set up, represent the computational tool of corresponding relation between input variable and output variable according to the known historical variable value with corresponding relation between input variable and output variable, can be used to the variate-value of any input variable to calculate the variate-value of corresponding output variable, predict thereby can realize according to historical friend-making or match condition the possibility that the good friend of current recommendation is accepted by user.

For example, Matching Model herein can represent the corresponding relation between the matching degree between essential information and this two users of two users, also be, in this Matching Model, input variable is two users' essential information, output variable is two users' matching degree, wherein, matching degree can be for representing that the information interaction between these two users gos deep into degree, calculated by the information interaction amount between these two users, for example, matching degree can be one [0, 1] numerical value in, it more shows that close to 1 two information interchanges between user are more, more deep.In addition, as the input variable of Matching Model, two users' of mating essential information can form separately respectively an orderly proper vector, dimension in each user's orderly proper vector for example can comprise sex, head image information, inhabitation province, inhabitation city, year of birth, birth month, go out the birthday, wages scope, level of education, job overall, marital status, industry, housing conditions, purchase car situation, birth province, City of birth, love state, whether login first, user's head portrait quality, height, body weight, constellation, blood group, the essential informations such as microblogging state and/or individual monologue state.

As previously mentioned, Matching Model can be according in social networks before this once the historical information between corresponding two historical users of mutual mistake mutual (for example comment on, put praise, collect, reply, instant communication function (as private chat function), do not like, draw black) situation and setting up.Particularly, two users that once carried out the information interaction in a certain amount of front can be considered to corresponding two historical match user, the historical matching degree of these corresponding two historical match user can be calculated by historical information interactive quantity between the two, on this basis, the essential information of these corresponding two historical match user matching degree historical with it can be considered to have known corresponding relation.Therefore, owing to there being corresponding two historical match user of once carrying out in a large number information interaction in social networks, historical variable value that can be using the essential information of these corresponding two historical match user as Matching Model input variable, historical variable value using the historical matching degree of these corresponding two historical match user as Matching Model output variable, trains the Matching Model that represents corresponding relation between corresponding two users' essential information and corresponding two users' matching degree.Wherein, go deep into the variation of corresponding relation between degree in order to make Matching Model can constantly adapt to user basic information and information interaction in social networks, can regularly adopt de novo information interaction to upgrade the Matching Model of social networks, for example, can upgrade weekly one time Matching Model.

In the time being first user commending friends, substantially meet the second user of first user friend-making condition for each, can using each the second user's essential information respectively with the variate-value as input variable together with the essential information of first user, calculate the variate-value of output variable by the Matching Model of having trained, as the matching degree of each second user and first user, for predicting that the information interaction that each second user and first user may occur gos deep into degree.

Be understandable that, in the present embodiment, " model " can be for example decision tree, random forest, support vector machine, Naive Bayes Classification, artificial neural network etc.For example, the present embodiment can adopt decision Tree algorithms to realize, and decision tree adopts the beta pruning of cost complicacy, can avoid like this training data over-fitting, and the leaf node transforming is compared other beta pruning algorithms and had higher accuracy rate and supporting rate.In addition, decision Tree algorithms is the prediction of the sample of supporting location attribute value set also, and training result supports serializing and persistence, can be by quick renewal training sample and model to meet online efficient Real time request.But it should be noted that, decision Tree algorithms is only a kind of computational tool example that each " model " in the present embodiment can adopt, and adopts respectively which kind of computational tool to realize for each " model " in the present embodiment, and the present embodiment does not limit this.

What in addition, in the present embodiment, related term " history " represented is event or the event correlative factor that current recommendation process occurred before.For example, " historical match user " can be illustrated in corresponding two users that information interaction occurred before current recommendation process to each other, correspondingly, " historical information is mutual " can represent corresponding two information interactions that historical match user was carried out before current recommendation process.

Step 203, according to the matching degree between the second user described in each and described first user, from the second user described in each, filter out at least one the 3rd user and recommend to described first user.

Screen the 3rd user from the second user time, the matching degree between each second user and first user can be used as a foundation of screening.Like this, pass through Matching Model, meet each second user of first user friend-making condition for essential information, the information interaction that can estimate between each second user and first user is goed deep into degree, and from the second user, filter out the 3rd user that more may go deep into first user information interaction with this, more meet the preference of first user to make to recommend the 3rd user of first user.Therefore, on the one hand, because the good friend user of first user is recommended in screening in the user of the similar user institute preference without similar to first user in essential information, not only can make the 3rd user who recommends first user more meet the preference of first user, and in social networks, not there is the similar user of first user or the good friend user of similar user preference realizes to first user commending friends; On the other hand, due to estimate matching degree between each second user and first user only need first user essential information and without the history friend-making behavior of first user, even for historical friend-making behavior or the very few first users of information interaction such as new registration users, also can be implemented as its commending friends, is the cold start-up problem that the historical friend-making behaviors such as new registration user or the very few user of information interaction realize commending friends thereby solve.

In addition, conventionally each second user also has friend-making condition separately, in order to make to recommend two parties to be afterwards all interested in to carry out friend-making behavior with the other side, can also ensure that the essential information of described first user all meets the friend-making condition of the 3rd user described in each, can make first user and the 3rd user's essential information all meet the other side's friend-making condition like this when the 3rd user in screening.

Be understandable that, in the present embodiment, screen the 3rd user from the second user time, can be only according to this foundation of matching degree between each second user and first user, or also can consider the multiple foundations including matching degree.

In the possible embodiment of the first that adopt multiple foundation screening the 3rd user, because the matching degree between each second user and first user is to predict taking the mutual situation of the historical information between all users in social networks as foundation, its reflection be the friend-making situation of user in whole social networks, but can not reflect the personalization preferences of first user, therefore, for the 3rd user who makes to filter out is more partial to the preference of first user personalization, can also be on the basis of matching degree, increase again the foundation of a preference, screen the 3rd user to consider matching degree and preference, wherein, preference can be taking the history friend-making behavior of first user as according to doping, so just can make the 3rd user's screening more meet the personalization preferences of first user.

Particularly, screen in the 3rd user's embodiment according to matching degree and preference at the same time, on the method step basis shown in Fig. 2, for example can also be according to the essential information of the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, and, filter out at least one the 3rd user to described first user recommendation from the second user described in each time, the also preference to the second user described in each according to described first user.Wherein, described friend-making preference pattern can be for setting up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance.

It should be noted that, friend-making preference pattern is herein that first user is set up separately for first user, it can represent as the user's of first user friend-making object of action essential information and first user the corresponding relation between this object user's preference, also be, in this friend-making preference pattern, input variable is the essential information as the user of first user friend-making object of action, output variable is the preference of first user to this object user, wherein, preference can be for the preference degree that represents that first user embodies this object user's friend-making behavior, by first user, this object user's friend-making behavior situation is calculated.For example, preference can be the numerical value in [0,1], and it more shows that close to 1 first user is to more preference of this object user, and it more shows that close to 0 first user is to more not preference of this object user.Wherein, can be divided into positive feedback behavior (even behavior that preference increases), negative feedback behavior (even behavior that preference reduces) and not have remarkable feedback behavior (i.e. the behavior without impact on preference) for calculating the friend-making behavior of preference.Wherein, positive feedback for example can comprise the behaviors such as comment, reply, instant messaging (as private chat), point are praised, collection, negative feedback for example can comprise draws black, the behavior such as lose interest in, and does not have remarkable feedback behavior for example can comprise to click the behavior such as check.In addition, as the input variable of friend-making preference pattern, for example can comprise appearance scoring, occupation, age, income, educational background, area, height, house situation, purchase car situation, praised quantity, be collected the dimensions such as quantity as the user's of first user friend-making object of action essential information.

Be understandable that, friend-making preference pattern can be that the history friend-making behavior situation of once carrying out before this according to first user in social networks is set up.Particularly, once carried out the historical object user of historical friend-making behavior for first user, first user can be calculated all historical friend-making behavior of this history object by first user the historical preference of this history object, on this basis, this history object user's essential information and first user can be considered to have known corresponding relation to its historical preference.Therefore, owing to existing many first users once to carry out the historical object user of historical friend-making behavior in social networks, historical variable value that can be using these historical object users' essential information as friend-making preference pattern input variable, using first user to these historical object users' historical preference the historical variable value as friend-making preference pattern output variable, train the object user's who represents first user friend-making behavior essential information and the friend-making preference pattern of first user to corresponding relation between this object user's preference.Wherein, for the preference pattern that makes to make friends can constantly adapt to the variation of first user personalization preferences in social networks, can regularly adopt the de novo friend-making behavior of first user to upgrade the friend-making preference pattern of first user.

In the time being first user commending friends, meet the second user of first user friend-making condition for each essential information, can be using each the second user's essential information respectively as the variate-value of input variable, calculate the variate-value of output variable by the friend-making preference pattern of having trained, preference as first user to each the second user, for the preference degree of predicting that friend-making behavior that first user may occur each second user can embody.

In the possible embodiment of the second that adopt multiple foundation screening the 3rd user, owing to often existing some seldom to use in each second user or can not the degree of depth using the user of social networks, these users often can not carry out making friends preferably interactive with first user and cause recommending successful not after recommending first user, therefore, for fear of will the second bad user of social networks service condition being recommended to first user, can also be on the basis of matching degree, increase again a foundation that uses sign degree, to consider matching degree and screen the 3rd user with sign degree, wherein, use sign degree can represent the service condition of each the second user to social networks, so just can make to avoid those to be filtered into the 3rd user to the second bad user of social networks service condition.

Particularly, at the same time according to matching degree with screen with sign degree in the 3rd user's embodiment, on the method step basis shown in Fig. 2, for example can also be according to essential information and/or the historical behavior of the second user described in each, for calculating, the second user described in each represents the use sign degree of the second user to social networks service condition described in each, and, filter out at least one the 3rd user to described first user recommendation from the second user described in each time, can be also according to the use sign degree of the second user described in each.

It should be noted that, screening can be for example each the second user's liveness, sincerity degree, popular degree as each the second user's of foundation use sign degree when the 3rd user and exchange any one or more in openness.

Described liveness can represent that the second user described in each triggers the frequent degree of historical behavior in described social networks, can be calculated as dimension by each the second user's multiple historical behavior.Wherein, for the historical behavior that calculates liveness for example can comprise login, start state, put praise, the dimension such as comment, each dimension for example can adopt the diminishing utility function of certain parameter to be converted to indicated value, then, by the indicated value linear weighted function of each dimension, obtains liveness.By each the second user's liveness, can embody the active degree of each the second user in social networks, therefore, screen the 3rd user according to each the second user's liveness, can be implemented as first user and recommend the second comparatively active user, and avoid recommending those sluggish the second users.

Described sincerity degree can represent by the second user described in each that quantity that waits stage, essential information integrated degree, essential information really degree and object that historical behavior produces in described social networks embodies each described in the second user use the sincerity degree of described social networks.Wherein, for the dimension of calculating sincerity degree for example can comprise data authenticity, user gradation, draw black number, send Information Number, data completeness, recently login time, dynamically issue number, authentication grade score and image and must grade according to grade, each dimension for example can adopt the diminishing utility function of certain parameter to be converted to indicated value, then by the indicated value linear weighted function of each dimension, obtain sincerity degree.By each the second user's sincerity degree, can embody the sincerity situation that each second user uses social networks, its sincerity degree is higher shows that it uses social networks more energetically, its sincerity degree is lower shows that it may be more the user that malicious user or inactive participation are used social networks, therefore, screen the 3rd user according to each the second user's sincerity degree, can be implemented as first user and recommend comparatively actively to use the second user of social networks, and avoid recommending those to belong to the second user of participation use social networks malicious user or inactive.

Described popular degree can represent that the second user described in each is carried out the frequent degree of historical behavior in described social networks by other users, can be calculated as dimension by the multiple historical behavior using each second user as object of action, wherein, for example can comprise the accessed number of every day, the dimension such as be commented on, praised for the historical behavior that calculates popular degree, each dimension for example can adopt the diminishing utility function of certain parameter to be converted to indicated value, then by the indicated value linear weighted function of each dimension, obtain the popular degree on the same day.Further, each day popular degree in short-term nearest the second user (as nearest a week) can also be calculated to recent popular degree according to time decay weighted accumulation, even can also further each day in nearest the second user long-term (as nearest 30 days) popular degree accumulative total be obtained to total popular degree.By each the second user's popular degree, can embody each the second user's pouplarity, therefore, screen the 3rd user according to each the second user's popular degree, can be implemented as first user and recommend comparatively welcome the second user.

Described interchange openness can represent the feedback degree of the second user to the history friend-making behavior that in described social networks, other users initiatively trigger described in each, can receive the feedback information quantity that history friend-making behavior quantity that other users initiatively trigger and the second user send by each the second user and calculate.For example, the second user's interchange openness can be feedback quantity and the ratio that initiatively exchanges quantity, wherein, initiatively exchanging quantity is the number of users initiatively exchanging with the second user, and feedback quantity is that the second user is to initiatively exchanging the number of users that has carried out feedback.By each the second user's interchange openness, can embody each second user and the active of first user be exchanged to the possibility giving a response, therefore, screen the 3rd user according to each the second user's interchange openness, not only can be implemented as first user recommends to respond the second user more frequently to initiatively exchanging, can also make on the other hand to receive in social networks the less user priority of exchange of information initiatively recommended and reduce isolated user, can also avoid on the one hand again recommending receiving in social networks initiatively the more popular user of exchange of information and avoid popular user by the too much situation of information harassing and wrecking.

Be understandable that,, can select liveness, sincerity degree, popular degree and exchange any one or more use sign degree in openness as the foundation of screening the 3rd user according to actual demand according to matching degree with screen the 3rd user's embodiment with sign degree for simultaneously.In addition, also it should be noted that, in screening when the 3rd user, except above-mentioned embodiment, can also the while realize according to matching degree, preference with sign degree.

In some embodiments of the present embodiment, in the time adopting multiple foundation screening the 3rd user, for the ease of screening, can first recommend acceptance for each second user utilizes its each foundation to calculate one, and then screen the 3rd user according to each the second user's recommendation acceptance.Furthermore, in order to make the good friend who recommends to first user further meet the preference of first user, after can utilizing the historical recommendation process once occurring in social networks, history is recommended user to predict that to the recommended user's of history acceptance level each second user is in the recommended possibility that first user is accepted afterwards, as each the second user's recommendation acceptance.

For example, for simultaneously according to matching degree, preference and use signs degree as screening in some embodiments of foundation, with reference to Fig. 3, step 203 specifically can comprise:

Step 301, according to the matching degree of the second user and described first user described in each, the preference of described first user to the second user described in each and the second user's use sign degree described in each, utilize and recommend forecast model, calculate the recommendation acceptance of described first user to the second user described in each, described recommendation forecast model is in advance according to historical matching degree of being recommended user and historical recommended user in described social networks, history is recommended the preference of user to the recommended user of history, historical recommended user's use sign degree, and, history is recommended user to set up the represented recommendation acceptance of the recommended user's of history history friend-making behavior.

In a recommendation process, using the user that accepts friend recommendation as being recommended user, recommended user's user as recommended user using recommended to this, for the matching degree of being recommended between user and recommended user, recommended the preference of user to recommended user, recommended user's use sign degree and recommended the recommendation acceptance of user to recommended user, recommendation forecast model herein can represent this matching degree, preference, use the corresponding relation between sign degree and this recommendation acceptance, also be, in this recommendation forecast model, input variable can at least comprise the matching degree of being recommended between user and recommended user, recommended the preference of user to recommended user, recommended user's use sign degree, output variable is recommended the recommendation acceptance of user to recommended user, the acceptance level of recommending acceptance to be recommended user to embody recommended user's friend-making behavior for expression, by being recommended user to calculate recommended user's friend-making behavior situation.Wherein, can be divided into positive feedback behavior (even behavior of recommending acceptance to increase) and negative feedback behavior (even behavior of recommending acceptance to reduce) for the friend-making behavior of calculated recommendation acceptance, positive feedback for example can comprise a little praise, the behavior such as collection, negative feedback for example can comprise draw black, the behavior such as do not like.In addition, except aforementioned matching degree, preference and use sign degree, recommend the input variable of forecast model for example can also comprise recommended user be whether new registration user, recommended user whether in the near future (as nearest one week interior) logined the dimensions such as social networks.In addition,, except the aforementioned example of enumerating for " model " of the present embodiment, recommending forecast model can also be for example that logistic returns mode.

It should be noted that, recommend forecast model to be recommended user to set up the recommended user's of history history friend-making behavior situation according to historical in the historical recommendation process once carrying out before this in social networks.Particularly, for a historical recommendation process, its history is recommended user to recommend acceptance to be recommended user to be calculated by user's history friend-making behavior history by historical after recommending to the recommended user's of history history, and history is recommended the historical matching degree between user and historical recommended user, history is recommended the historical preference of user to the recommended user of history and historical recommended user's use sign degree to calculate by the aforesaid corresponding embodiment of the present embodiment, on this basis, the historical matching degree that this history recommendation process is corresponding, historical preference and the historical use sign degree history corresponding with this history recommendation process recommend acceptance can be considered to have known corresponding relation.Therefore, owing to there being the historical recommendation process once occurring in a large number in social networks, can be using historical matching degree corresponding to these historical recommendation process, historical preference and the historical sign degree that uses as the historical variable value of recommending forecast model input variable, recommend acceptance as the historical variable value of recommending forecast model output variable using history corresponding to these historical recommendation process, train the recommendation forecast model that represents corresponding relation between matching degree, preference and use sign degree that recommendation process is corresponding and recommendation acceptance corresponding to recommendation process.

In the time being first user commending friends, meet the second user of first user friend-making condition for each essential information, can be using the aforementioned matching degree calculating as each second user, the preference variate-value as input variable together with use sign degree, calculate the variate-value of output variable by the recommendation forecast model of having trained, recommendation acceptance as first user to each the second user, for predicting in the acceptance level of recommending each second user friend-making behavior that first user may occur afterwards to embody.

Wherein, for the social networks of some scene, recommending forecast model can be to set up respectively for the dissimilar user that recommended, and adopts the recommendation forecast model identical with first user type to calculate the second user's recommendation acceptance while being dissimilar first user commending friends.For example, in marriage and making friend's social networks, can set up respectively one for men's family and lady's family and recommend forecast model, and can adopt the recommendation forecast model at men's family while being men's family commending friends, during for lady's family commending friends, can adopt the recommendation forecast model at lady's family.

Step 302, recommendation acceptance according to described first user to the second user described in each filter out at least one the 3rd user and recommend to described first user from the second user described in each.

When according to recommendation acceptance screening the 3rd user, in some embodiments, for example can to each the second user's recommendation acceptance order from high to low, each second user be formed to an initial stage recommendation list according to first user, then recommend to first user as the 3rd user according to the second user of the select progressively some of initial stage recommendation list.

Be understandable that, only, according to recommending order from high to low of acceptance to choose the 3rd user, can make the good friend who recommends to user is all the popular user who comparatively enlivens, and new registration user and the more not popular recommended possibility of user are lower.For fear of this " Matthew effect ", make all types of user there is comparatively balanced recommended probability, in other embodiments, can login situation according to difference the second user is classified, and from different classes of, extract respectively some second users and recommend first user as the 3rd user.Specific to screening in the 3rd user's embodiment according to recommendation acceptance, step 302 for example can comprise: the second user described in each is divided into multiple candidate user classifications according to the login situation in social networks; Respectively for each candidate user classification, according to described first user in described candidate user classification described in each recommendation acceptance of the second user by high order on earth, from the second user described in each of described candidate user classification, extract three user corresponding with the ratio of choosing of described candidate user classification; Gathering the 3rd user who extracts out in candidate user classification described in each recommends to described first user.Particularly, in the time screening respectively the 3rd user from the second user of each candidate user classification, can be on the basis of aforementioned initial stage recommendation list, put in order according to initial stage recommendation list is vertical, be respectively each candidate user classification extract its classification the second user of corresponding ratio quantity as the 3rd user, thereby form a mixing recommendation list with all the 3rd users that extract.Wherein, user candidate classification for example can comprise new registration user, online user, in the recent period (as nearest one week interior) login user, login user etc. not in the recent period.In all the 3rd users, the ratio that the second user in each user candidate classification occupies can be to determine according to the ratio of all types of user in same day social networks.Be understandable that, logining situation with respect to user in the basis of aforementioned liveness, sincerity degree and recommendation acceptance is only one of them difference as multiple dimensions, in the division foundation of candidate user classification, user's login situation is unique dimension, so just can make all types of user that popular degree is different have comparatively balanced recommended probability.

It should be noted that, according to recommend acceptance according to login situation category filter the 3rd user's embodiment in, can also require the essential information of first user to meet the 3rd user that the filters out condition of making friends, all be interested in to carry out friend-making behavior with the other side with two parties after making to recommend.Particularly, in some embodiments, for example can be on the basis of aforementioned mixing recommendation list, the 3rd user's order in mixing list is reset, both sides' essential information is met to make friends the 3rd user of condition of the other side and rearrange recommendation list front portion, and residue the 3rd user is aligned to recommendation list rear portion, reset recommendation list thereby form, to recommend to first user according to the order of resetting recommendation list.

Then return to Fig. 2.

After step 203 is complete, if the 3rd number of users filtering out is very few, can also delete the wherein insensitive condition of some first user to its friend-making condition history being set according to first user, to the 3rd number of users filtering out is supplemented.

By the technical scheme of the present embodiment, on the one hand, because the good friend user of first user is recommended in screening in the user of the similar user institute preference without similar to first user in essential information, not only can make the 3rd user who recommends first user more meet the preference of first user, and in social networks, not there is the similar user of first user or the good friend user of similar user preference realizes to first user commending friends; On the other hand, due to estimate matching degree between each second user and first user only need first user essential information and without the history friend-making behavior of first user, even for historical friend-making behavior or the very few first users of information interaction such as new registration users, also can be implemented as its commending friends, is the cold start-up problem that the historical friend-making behaviors such as new registration user or the very few user of information interaction realize commending friends thereby solve.

Referring to Fig. 4, show the process flow diagram of another embodiment of method of commending friends in the present invention.In the present embodiment, for example specifically can comprise the steps:

Step 401, obtain the friend-making condition of first user, and search essential information and meet each second user of described friend-making condition.

Step 402, according to the essential information of the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance.

Wherein, friend-making preference pattern can be that first user is set up separately for first user, it can represent as the user's of first user friend-making object of action essential information and first user the corresponding relation between this object user's preference, also be, in this friend-making preference pattern, input variable is the essential information as the user of first user friend-making object of action, output variable is the preference of first user to this object user, wherein, preference can be for the preference degree that represents that first user embodies this object user's friend-making behavior, by first user, this object user's friend-making behavior situation is calculated.For example, preference can be the numerical value in [0,1], and it more shows that close to 1 first user is to more preference of this object user, and it more shows that close to 0 first user is to more not preference of this object user.Wherein, can be divided into positive feedback behavior (even behavior that preference increases), negative feedback behavior (even behavior that preference reduces) and not have remarkable feedback behavior (i.e. the behavior without impact on preference) for calculating the friend-making behavior of preference.Wherein, positive feedback for example can comprise the behaviors such as comment, reply, instant messaging (as private chat), point are praised, collection, negative feedback for example can comprise draws black, the behavior such as lose interest in, and does not have remarkable feedback behavior for example can comprise to click the behavior such as check.In addition, as the input variable of friend-making preference pattern, for example can comprise appearance scoring, occupation, age, income, educational background, location, height, house situation, purchase car situation, praised quantity, be collected the dimensions such as quantity as the user's of first user friend-making object of action essential information.

Be understandable that, friend-making preference pattern can be that the history friend-making behavior situation of once carrying out before this according to first user in social networks is set up.Particularly, once carried out the historical object user of historical friend-making behavior for first user, first user can be calculated all historical friend-making behavior of this history object by first user the historical preference of this history object, on this basis, this history object user's essential information and first user can be considered to have known corresponding relation to its historical preference.Therefore, owing to existing many first users once it to be carried out the historical object user of historical friend-making behavior in social networks, historical variable value that can be using these historical object users' essential information as friend-making preference pattern input variable, using first user to these historical object users' historical preference the historical variable value as friend-making preference pattern output variable, train the object user's who represents first user friend-making behavior essential information and the friend-making preference pattern of first user to corresponding relation between this object user's preference.Wherein, for the preference pattern that makes to make friends can constantly adapt to the variation of first user personalization preferences in social networks, can regularly adopt the de novo friend-making behavior of first user to upgrade the friend-making preference pattern of first user.

In the time being first user commending friends, substantially meet the second user of first user friend-making condition for each, can be using each the second user's essential information respectively as the variate-value of input variable, calculate the variate-value of output variable by the friend-making preference pattern of having trained, preference as first user to each second user and first user, for the preference degree of predicting that friend-making behavior that first user may occur each second user can embody.

Step 403, preference according to described first user to the second user described in each filter out at least one the 3rd user and recommend to described first user from the second user described in each.

By friend-making preference pattern, meet each second user of first user friend-making condition for essential information, can realize according to the history friend-making behavior of first user and estimate the preference degree of first user to each the second user, and from the second user, filter out the 3rd user that more may go deep into first user information interaction with this, more meet the preference of first user to make to recommend the 3rd user of first user.

Therefore, the technical scheme of the present embodiment, on the one hand, because the good friend user of first user is recommended in screening in the user of the similar user institute preference without similar to first user in essential information, not only can make the 3rd user who recommends first user more meet the preference of first user, and in social networks, not there is the similar user of first user or the good friend user of similar user preference realizes to first user commending friends; On the other hand, because first user is that history friend-making behavior based on first user is estimated to each the second user's preference, can avoid the 3rd similar user of essential information who recommends to first user to be tending towards identical, this is not only conducive to user and gets to know good friend in scope widely, but also can make the 3rd user who filters out more be partial to the personalization preferences of first user.

example devices

After having introduced the method for exemplary embodiment of the invention, next, with reference to figure 5～10 describe exemplary embodiment of the invention, for the equipment of commending friends.

Referring to Fig. 5, show the structural drawing of equipment one embodiment of commending friends in the present invention.In the present embodiment, described equipment for example specifically can comprise:

Friend-making Condition Matching module 501, for obtaining the friend-making condition of first user, and searches essential information and meets each second user of described friend-making condition;

Matching degree computing module 502, for according to the essential information of described first user and the second user's essential information described in each, utilize Matching Model, calculate the matching degree between the second user and described first user described in each, described Matching Model is to set up according to having history mutual information amount between the essential information of mutual corresponding two the historical match user of historical information and corresponding two historical match user in social networks in advance;

The first recommending module 503 for according to the matching degree between the second user described in each and described first user, filters out at least one the 3rd user and recommends to described first user from the second user described in each.

Referring to Fig. 6, show the structural drawing of another embodiment of equipment of commending friends in the present invention.In the present embodiment, except all structures shown in earlier figures 5, described equipment for example can also comprise:

Preference computing module 601, for the essential information according to the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance, and

Wherein, when described the first recommending module 503 filters out at least one the 3rd user to described first user recommendation from the second user described in each, the also preference to the second user described in each according to described first user.

Referring to Fig. 7, show the structural drawing of the another embodiment of equipment of commending friends in the present invention.In the present embodiment, except all structures shown in earlier figures 6, described equipment for example can also comprise:

Use signs degree computing module 701, for essential information and/or historical behavior according to the second user described in each, for the second user described in each calculates expression the second user use sign degree to social networks service condition described in each, and

Wherein, when described the first recommending module 503 filters out at least one the 3rd user to described first user recommendation from the second user described in each, also according to the use sign degree of the second user described in each.

Furthermore, optional, described use sign degree for example specifically can comprise liveness, sincerity degree, popular degree and exchange any one or more in openness;

Described liveness can represent that the second user described in each triggers the frequent degree of historical behavior in described social networks;

Described sincerity degree can represent by the second user described in each that quantity that waits stage, essential information integrated degree, essential information really degree and object that historical behavior produces in described social networks embodies each described in the second user use the sincerity degree of described social networks;

Described popular degree can represent that the second user described in each is carried out the frequent degree of historical behavior in described social networks by other users;

Described interchange openness can represent the feedback degree of the second user history friend-making behavior that other users initiatively trigger in to described social networks described in each.

Referring to Fig. 8, show the structural drawing of the first recommending module 503 1 embodiments in the embodiment of the present invention.In the present embodiment, described the first recommending module 503 for example specifically can comprise:

Recommend acceptance calculating sub module 801, for the matching degree according to the second user and described first user described in each, the preference of described first user to the second user described in each and the second user's use sign degree described in each, utilize and recommend forecast model, calculate the recommendation acceptance of described first user to the second user described in each, described recommendation forecast model is in advance according to historical matching degree of being recommended user and historical recommended user in described social networks, history is recommended the preference of user to the recommended user of history, historical recommended user's use sign degree, and, history is recommended user to set up the represented recommendation acceptance of the recommended user's of history history friend-making behavior,

User recommends submodule 802, for the recommendation acceptance to the second user described in each according to described first user, filters out at least one the 3rd user and recommend to described first user from the second user described in each.

Referring to Fig. 9, show user in the embodiment of the present invention and recommend the structural drawing of submodule 802 1 embodiments.In the present embodiment, described user recommends submodule 802 for example specifically can comprise:

Candidate user classification submodule 901, for being divided into multiple candidate user classifications by the second user described in each according to the login situation at social networks;

Recommend user to extract submodule 902, be used for respectively for each candidate user classification, according to described first user in described candidate user classification described in each recommendation acceptance of the second user by high order on earth, from the second user described in each of described candidate user classification, extract three user corresponding with the ratio of choosing of described candidate user classification;

Gather user and recommend submodule 903, recommend to described first user for gathering the 3rd user that candidate user classification is extracted out described in each.

Wherein, alternatively, in some embodiments of the embodiment of the present invention, the essential information of described first user can all meet the friend-making condition of the 3rd user described in each.

Referring to Figure 10, show the equipment structural drawing of an embodiment again of commending friends in the present invention.In the present embodiment, described equipment for example specifically can comprise:

Preference computing module 601, for the essential information according to the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance;

The second recommending module 1001 for the preference to the second user described in each according to described first user, filters out at least one the 3rd user and recommends to described first user from the second user described in each.

By apparatus embodiments provided by the invention, not only can make the good friend who recommends to this user more meet this user's preference, and without screen commending friends from the good friend user of this user's similar user institute preference, also can realize to this user's commending friends when very few thereby do not there is this user's similar user or the good friend user of similar user preference in social networks, reduced cannot commending friends situation occur.

Although it should be noted that some devices or the sub-device of the equipment of having mentioned commending friends in above-detailed, this division is only exemplary not enforceable.In fact, according to the embodiment of the present invention, the feature of above-described two or more devices and function can be specialized in a device.Otherwise, the feature of an above-described device and function can Further Division for to be specialized by multiple devices.

In addition, although described in the accompanying drawings the operation of the inventive method with particular order,, this not requires or hint must be carried out these operations according to this particular order, or the operation shown in must carrying out all could realize the result of expecting.Additionally or alternatively, can omit some step, multiple steps be merged into a step and carry out, and/or a step is decomposed into multiple steps carries out.

Although described spirit of the present invention and principle with reference to some embodiments, but should be appreciated that, the present invention is not limited to disclosed embodiment, the division of each side is not meant that to the feature in these aspects can not combine to be benefited yet, and this division is only the convenience in order to explain.The present invention is intended to contain interior included various amendments and the equivalent arrangements of spirit and scope of claims.

For example, coming with preference in the present embodiment of commending friends, also can realize in conjunction with the various possible embodiment in the aforementioned embodiment that carrys out commending friends with matching degree.For example, in some embodiments of the present embodiment, can also screen the 3rd user according to matching degree and preference simultaneously.And for example, in other embodiments of the present embodiment, can also be simultaneously screen the 3rd user according to matching degree, preference and at least one with sign degree.For another example, in the other embodiment of the present embodiment, can be according to screening the 3rd user by matching degree, preference, the recommendation acceptance that calculates with sign degree.Again for another example, in some embodiments again of the present embodiment, can first the second user be divided in different candidate user classifications, then screen respectively the 3rd user from each candidate user classification.Again for another example, at again again in some embodiments of the present embodiment, can require the 3rd user's who filters out essential information to meet the friend-making condition of first user.

Brief description of the drawings

Read detailed description below by reference to accompanying drawing, above-mentioned and other objects of exemplary embodiment of the invention, feature and advantage will become easy to understand.In the accompanying drawings, show some embodiments of the present invention in exemplary and nonrestrictive mode, wherein:

Fig. 1 schematically shows the framework schematic diagram of an exemplary application scene of embodiments of the present invention;

Fig. 2 schematically shows the process flow diagram of method one embodiment of commending friends in the present invention;

Fig. 3 schematically shows the process flow diagram that screens the 3rd user's one embodiment in the embodiment of the present invention;

Fig. 4 schematically shows the process flow diagram of another embodiment of method of commending friends in the present invention;

Fig. 5 schematically shows the structural drawing of equipment one embodiment of commending friends in the present invention;

Fig. 6 schematically shows the structural drawing of another embodiment of equipment of commending friends in the present invention;

Fig. 7 schematically shows the structural drawing of the another embodiment of equipment of commending friends in the present invention;

Fig. 8 schematically shows the structural drawing of the first recommending module 503 1 embodiments in the embodiment of the present invention;

Fig. 9 schematically shows user in the embodiment of the present invention recommends the structural drawing of submodule 802 1 embodiments;

Figure 10 schematically shows the equipment structural drawing of an embodiment again of commending friends in the present invention;

In the accompanying drawings, identical or corresponding label represents identical or corresponding part.

Embodiment

Below with reference to some illustrative embodiments, principle of the present invention and spirit are described.Should be appreciated that providing these embodiments is only used to make those skilled in the art can understand better and then realize the present invention, and not limit the scope of the invention by any way.On the contrary, it is in order to make the disclosure more thorough and complete that these embodiments are provided, and the scope of the present disclosure intactly can be conveyed to those skilled in the art.

One skilled in the art will appreciate that embodiments of the present invention can be implemented as a kind of system, device, equipment, method or computer program.Therefore, the disclosure can specific implementation be following form, that is: hardware, software (comprising firmware, resident software, microcode etc.), or the form of hardware and software combination completely completely.

A kind of method and apparatus of commending friends has been proposed according to the embodiment of the present invention.

In this article, it will be appreciated that, that related term " first user ", " the second user ", " the 3rd user " represents respectively is the good friend user that the user who accepts recommendation in same commending friends process, satisfied acceptance recommend user to make friends the user of condition and finally recommend out.Be understandable that, in different recommendation process, first user can be any user in social networks, and for example, certain second user in a recommendation process or the 3rd user can be also the first users of another recommendation process.For " first user " accepting in recommendation process to recommend, the present invention does not limit.In addition, any number of elements in accompanying drawing is all unrestricted for example, and any name is all only for distinguishing, and does not have any limitation.

Below with reference to some representative embodiments of the present invention, explain in detail principle of the present invention and spirit.

Claims

1. a method, comprising:

Obtain the friend-making condition of first user, and search essential information and meet each second user of described friend-making condition;

According to the essential information of described first user and the second user's essential information described in each, utilize Matching Model, calculate the matching degree between the second user and described first user described in each, described Matching Model is to set up according to having history mutual information amount between the essential information of mutual corresponding two the historical match user of historical information and corresponding two historical match user in social networks in advance;

According to the matching degree between the second user described in each and described first user, from the second user described in each, filter out at least one the 3rd user and recommend to described first user.

2. method according to claim 1, also comprises:

According to the essential information of the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance, and

Wherein, filter out at least one the 3rd user to described first user recommendation from the second user described in each time, the also preference to the second user described in each according to described first user.

3. method according to claim 2, also comprises:

According to essential information and/or the historical behavior of the second user described in each, for calculating, the second user described in each represents the use sign degree of the second user to social networks service condition described in each, and

Wherein, filter out at least one the 3rd user to described first user recommendation from the second user described in each time, also according to the use sign degree of the second user described in each.

4. method according to claim 3, wherein, described use sign degree comprises liveness, sincerity degree, popular degree and exchanges any one or more in openness;

Described liveness represents that the second user described in each triggers the frequent degree of historical behavior in described social networks;

Described sincerity degree represent by the second user described in each that quantity that waits stage, essential information integrated degree, essential information really degree and object that historical behavior produces in described social networks embodies each described in the second user use the sincerity degree of described social networks;

Described popular degree represents that the second user described in each is carried out the frequent degree of historical behavior in described social networks by other users;

Described interchange openness represents the feedback degree of the second user history friend-making behavior that other users initiatively trigger in to described social networks described in each.

5. method according to claim 3 wherein, filters out at least one the 3rd user and recommends to described first user from the second user described in each, comprising:

According to the matching degree of the second user and described first user described in each, the preference of described first user to the second user described in each and the second user's use sign degree described in each, utilize and recommend forecast model, calculate the recommendation acceptance of described first user to the second user described in each, described recommendation forecast model is in advance according to historical matching degree of being recommended user and historical recommended user in described social networks, history is recommended the preference of user to the recommended user of history, historical recommended user's use sign degree, and, history is recommended user to set up the represented recommendation acceptance of the recommended user's of history history friend-making behavior,

Recommendation acceptance according to described first user to the second user described in each filters out at least one the 3rd user and recommends to described first user from the second user described in each.

6. method according to claim 5, described according to described first user the recommendation acceptance to the second user described in each, from the second user described in each, filter out at least one the 3rd user and recommend to described first user, comprising:

The second user described in each is divided into multiple candidate user classifications according to the login situation in social networks;

Respectively for each candidate user classification, according to described first user in described candidate user classification described in each recommendation acceptance of the second user by high order on earth, from the second user described in each of described candidate user classification, extract three user corresponding with the ratio of choosing of described candidate user classification;

Gathering the 3rd user who extracts out in candidate user classification described in each recommends to described first user.

7. according to the method described in claim 1 or 6, wherein, the essential information of described first user all meets the friend-making condition of the 3rd user described in each.

8. a method, comprising:

According to the essential information of the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance;

Preference according to described first user to the second user described in each filters out at least one the 3rd user and recommends to described first user from the second user described in each.

9. an equipment, comprising:

Friend-making Condition Matching module, for obtaining the friend-making condition of first user, and searches essential information and meets each second user of described friend-making condition;

Matching degree computing module, for according to the essential information of described first user and the second user's essential information described in each, utilize Matching Model, calculate the matching degree between the second user and described first user described in each, described Matching Model is to set up according to having history mutual information amount between the essential information of mutual corresponding two the historical match user of historical information and corresponding two historical match user in social networks in advance;

The first recommending module for according to the matching degree between the second user described in each and described first user, filters out at least one the 3rd user and recommends to described first user from the second user described in each.

10. equipment according to claim 9, also comprises:

Preference computing module, for the essential information according to the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance, and

Wherein, when described the first recommending module filters out at least one the 3rd user to described first user recommendation from the second user described in each, the also preference to the second user described in each according to described first user.

11. equipment according to claim 10, also comprise:

Use signs degree computing module, for essential information and/or historical behavior according to the second user described in each, for the second user described in each calculates expression the second user use sign degree to social networks service condition described in each, and

Wherein, when described the first recommending module filters out at least one the 3rd user to described first user recommendation from the second user described in each, also according to the use sign degree of the second user described in each.

12. equipment according to claim 11, wherein, described use sign degree comprises liveness, sincerity degree, popular degree and exchanges any one or more in openness;

13. equipment according to claim 11, wherein, described the first recommending module comprises:

Recommend acceptance calculating sub module, for the matching degree according to the second user and described first user described in each, the preference of described first user to the second user described in each and the second user's use sign degree described in each, utilize and recommend forecast model, calculate the recommendation acceptance of described first user to the second user described in each, described recommendation forecast model is in advance according to historical matching degree of being recommended user and historical recommended user in described social networks, history is recommended the preference of user to the recommended user of history, historical recommended user's use sign degree, and, history is recommended user to set up the represented recommendation acceptance of the recommended user's of history history friend-making behavior,

User recommends submodule, for the recommendation acceptance to the second user described in each according to described first user, filters out at least one the 3rd user and recommend to described first user from the second user described in each.

14. equipment according to claim 13, described user recommends submodule to comprise:

Candidate user classification submodule, for being divided into multiple candidate user classifications by the second user described in each according to the login situation at social networks;

Recommend user to extract submodule, be used for respectively for each candidate user classification, according to described first user in described candidate user classification described in each recommendation acceptance of the second user by high order on earth, from the second user described in each of described candidate user classification, extract three user corresponding with the ratio of choosing of described candidate user classification;

Gather user and recommend submodule, recommend to described first user for gathering the 3rd user that candidate user classification is extracted out described in each.

15. according to the equipment described in claim 9 or 14, and wherein, the essential information of described first user all meets the friend-making condition of the 3rd user described in each.

16. 1 kinds of equipment, comprising:

Preference computing module, for the essential information according to the second user described in each, utilize the friend-making preference pattern of described first user, calculate the preference of described first user to the second user described in each, described friend-making preference pattern is to set up according to the history friend-making behavior of described first user and as the user's of described historical friend-making object of action essential information in advance;

The second recommending module for the preference to the second user described in each according to described first user, filters out at least one the 3rd user and recommends to described first user from the second user described in each.