CN104281599A - Method and device for recommending information to user in social network - Google Patents
Method and device for recommending information to user in social network Download PDFInfo
- Publication number
- CN104281599A CN104281599A CN201310281786.4A CN201310281786A CN104281599A CN 104281599 A CN104281599 A CN 104281599A CN 201310281786 A CN201310281786 A CN 201310281786A CN 104281599 A CN104281599 A CN 104281599A
- Authority
- CN
- China
- Prior art keywords
- user
- good friend
- information
- level
- friend
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Abstract
The invention discloses a method and device for recommending information to a user in a social network. The method comprises the steps that a plurality of friend users of a user in the social network are classified into two or more first-level friend user groups; for each of the two or more first-level friend user groups, the tendency of the user for each first-level friend user group is calculated, so that two or more corresponding tendency values are obtained; according to the two or more corresponding tendency values, related information is recommended to the user on the basis of the friend information, in the social network, of all elements in the two or more first-level friend user groups.
Description
Technical field
The present invention relates to the information recommendation in social networks.
Background technology
Social networks (SNS) is frequently used by people in recent years.Social networks is that people carry out social platform with the individual or group such as with same background or hobby.On social networks, user has the personal profiles and his network linking that represent personal identification usually.Social networks allow user such as plusing good friend, put up personal information (such as data, state, photograph), such as, exchange with good friend by photos and sending messages, the mode such as to play games, or share the information of news, comment or video and so on to good friend.The form of social networks can be website, software or application etc.User can pass through computing machine, and the mobile device of the electronic equipment or such as intelligent mobile phone, panel computer and so on that comprise such as desktop PC, laptop computer and so on visits social networks.Usually depend on comprise this type of subscriber equipment, server etc. social networking system to realize the function of social networks.Be wherein a critical function of social networks to user's recommendation information, it has very great help to raising user's liveness and reservation user.
Summary of the invention
The invention provides a kind of method to the user's recommendation information in social networks, comprising: the multiple good friend users of user in social networks are classified as two or more one-level good friend user set; For each in two or more one-level good friend user set, calculate the tendentiousness that user gathers each one-level good friend user, thus obtain two or more corresponding propensity value; According to two or more corresponding propensity value, based on the respective friend information of each element in social networks in two or more one-level good friend user set, recommend the information of being correlated with to user.
Present invention also offers a kind of for the device to the user's recommendation information in social networks, comprising: for the multiple good friend users of user in social networks being classified as the device of two or more one-level good friend user set; For for each in two or more one-level good friend user set, calculate the tendentiousness that user gathers each one-level good friend user, thus obtain the device of two or more corresponding propensity value; And for according to two or more corresponding propensity value, based on the respective friend information of each element in social networks in two or more one-level good friend user set, recommend the device of the information of being correlated with to user.
Accompanying drawing explanation
Fig. 1 is the schematic diagram of a part for the user's friend relation illustrated in social networks;
Fig. 2 is the process flow diagram that method is according to an embodiment of the invention shown;
Fig. 3 is the schematic diagram of the step that the information that classifying step is relevant with recommendation is according to an embodiment of the invention shown;
Fig. 4 is the process flow diagram of some sub-step of the step that the information of recommending according to an embodiment of the invention to be correlated with is shown; And
Fig. 5 illustrates according to an embodiment of the invention to the example of the display of user's recommendation information.
Embodiment
There is the social networks of numerous species, such as system of real name social network sites, micro-blog, instantaneous communication system, location-based social application, video sharing, social gaming etc.Although the type of social networks is different, one of its basic function is recommendation information.The information of recommending comprises the information etc. of other users in user's possibility interested news information, video information or this social networks.User can select various operation according to the information of recommending, such as, browse the news information of recommendation, watch the video of recommendation, access the page of recommended user or added as a friend.Wherein, the function of recommending the information (in social networks be commonly called commending friends) relevant with user is very favourable, this is because carry out plusing good friend by the mode of user search there is the low and not high defect of the degree of correlation of efficiency, and recommend suitable good friend that user can be made to have more and relevant good friend to user, be extremely conducive to any active ues and the reservation user that increase social networks.
For commending friends information, the friend recommendation method that social network winding thread uses comprises that secondary good friend (FOF) recommends, good friend bunch recommends, data is recommended and address list recommendation etc., wherein secondary friend recommendation have Candidate Set extensively, belong to basic data collection, can illustrate to there is advantages such as contacting between two people by the common good friend of user and secondary good friend.Along with the development of social networks, social circle detects and more and more receives publicity, and social circle is such as the set comprising the good friend with some common trait or common background.Compared with promoting overall good friend's number of user, the good friend's number promoting the social circle more relevant to user can make friend recommendation more can play adding users liveness and retain the effect of user.
For friend recommendation, there is very large decay in current friend recommendation algorithm, namely recommendation results cannot just lose the degree of correlation with user the prolonged time.This is because the degree of correlation of social circle and user changes in time and secondary good friend itself choose decay.Therefore the information recommendation of such as friend recommendation needs the information recommendation because usually carrying out such as friend recommendation and so on such as degree of correlation considering social circle's moment.
Fig. 1 diagrammatically illustrates a part for the user's friend relation in social networks.As shown in Figure 1, in social networks, user A can have the multiple good friend users such as comprising F01, F02, F03.Friend relation (representing with the line between user) is generally two-way, each other two users of good friend can have such as send mutually instant messages, check data, the authority such as access photograph album.Making user have more good friend can make this user more enliven on social network sites and more have less possibility to leave, and this is favourable for social network sites.
As shown in Figure 1, good friend user (or one-level good friend user) F01, F02, F03 of user A can have the good friend users such as FOF101, FOF102, FOF103, FOF201, FOF202, FOF301.These not yet become good friend with A, but are called as the secondary good friend of user A with the user that the good friend user of user A is good friend.The secondary good friend FOF103 of visual user A has the friend relation with F01 and F02 in FIG, then itself and A have at least F01 and F02 two common good friends.
User can carry out plusing good friend by sending good friend's application to another user, through the other side, good friend's application agrees to that after good friend applies for, both become good friend.Recommend to allow to provide more to user with the information of other users in this social networks to user, there is with it good friend of the larger degree of correlation.Set based on the good friend user of user comes then to make recommendation results be not easy to lose the degree of correlation with user to user's recommendation information targetedly.
Fig. 5 is the example at the interface to user's recommendation information, and wherein FOF201, FOF103, FOF301, FOF101 are the secondary good friends as recommendation results.User can to recommendation results carry out application add the other side for good friend (502), select no longer recommend this person (503) and access the operations such as its private page (501).
Fig. 2 is the process flow diagram that method is according to an embodiment of the invention shown.
As shown in Figure 2, method comprises according to an embodiment of the invention: the multiple good friend users of user in social networks are classified as two or more one-level good friend user set (S202); For each in two or more one-level good friend user set, calculate the tendentiousness that user gathers each one-level good friend user, thus obtain two or more corresponding propensity value (S203); And according to two or more corresponding propensity value, based on the respective friend information of each element in social networks in two or more one-level good friend user set, recommend the information (S204) of being correlated with to user.Then can return above-mentioned steps, above-mentioned steps was run repeatedly by the predetermined cycle, to guarantee that correlated results is updated (S205) at this week after date.
In Fig. 2, the classifying step (S202) of embodiment comprises and the good friend user of each user is classified as multiple one-level good friend user set (community).In this step, produce have multiple good friend user mutually between the data list of weight, and data list is substituted in community's detection algorithm to sort out described multiple good friend user.Realize the information that it needs whether to have between the good friend user of the interactive information between good friend's user list of acquisition user, the existing good friend's grouping information of user, two good friend users and user friend relation.First (1) set up represent user two good friends between weight (weight) <friend-id1 of degree of relationship, friend-id2, weight>, wherein friend-id1, friend-id2 represent two good friend users of user, and the value of weight can be 0 to 1.Then (2) can revise weight value based on the existing good friend's grouping of user, this can realize in the following way: if two good friend users are classified as one group (such as user is operated by the buddy list shown social network sites) by user, then weight is improved 0.1, until be added to maximal value 1.(3) can revise weight value based on the interactive information of the user of two in this social networks subsequently, account form such as:
Wherein n be comprise mutual behavior of writing number of times, m is the mutual behavior number of times reading private page, and t behavior of the reading number of times that to be other mutual.Above-mentioned data can substitute in community's detection algorithm as input and sort out the good friend user of this user and the good friend user through sorting out added corresponding one-level good friend user set, this community's detection algorithm such as walktrap algorithm by (4) afterwards.
Classifying step (S202) comprises further and sorting to the element (i.e. the good friend user of user) in each one-level good friend user set of user.Can based on the level of interaction Pref (hostid of this good friend's user to user, friendid) the active degree active (id) of (wherein hoistid representative of consumer) and this good friend user calculates this good friend user relative to the one-level good friend weight of user and according to this one-level good friend weight sequencing, this weight
The account form of score (hosted, friendid) is such as:
score(hostid，friendid)＝Pref(hostid，frendid)×0.618+acitve(friendid)×0.382(2)
The account form of level of interaction is such as:
Pref
n+1＝0.76*Pref
n+ΔPref
n+1 (3)
Wherein n is the large period number from setting up friend relation, and large period can be predetermined (such as 1 day).And wherein weight
kwithin shorter time cycle (such as 1 hour), user is to the weight calculation of the highest behavior of the one-level good friend weight of this good friend user, the score value of behavior is defined as follows: m is in large period (such as 1 day), has how many minor cycles (such as 1 hour).Weight
kdetermination mode be such as:
The active degree (active (id)) of good friend user is also depended in the calculating of one-level good friend user weight, and the determination mode of active degree is such as:
The head office in log in total duration * 0.5+ in the past 7 cycles in active (id)=7 cycles of past be number of times * 0.5 (6) then by active degree normalization, make result value between 0 ~ 1.
The example of categorization results is as shown in 301 in Fig. 3 and 302, and wherein 301 and 302 is be an one-level good friend user set respectively.F01 and F02 belong to same one-level good friend user set, F01 be ordered in F02 before F03 then belong to another one-level good friend user set.
As shown in Figure 2, method in an embodiment to user's recommendation information also comprises for each in two or more one-level good friend user set, calculate the tendentiousness that user gathers each one-level good friend user, thus obtain two or more corresponding propensity value (S203).In this step, based on the historical activity information of received user in social networks, calculate two or more one-level good friend user set (community_i) corresponding two or more corresponding propensity value Pref (hosted, community
i).Wherein historical activity information comprise following in one or more: the number of times add_count of the operation that user adds as a friend for the application of once recommended information; User no longer recommends the number of operations forbit_count of the information relevant with this good friend for the selection of once recommended information; The number of times show_count of the recommended display of the information of once recommending, and user enters the number of times go_to_prf_count of the page be associated to relevant information.Can receive in the interface such as shown in Figure 5 such as by clicking the information of the aforesaid operations that respective block is carried out.The account form of propensity value is such as follows:
The wherein secondary good friend number (will hereafter mention) that extends for this one-level good friend user gathers of n, in addition wherein:
Then to Γ ref (hostid, community
i) be normalized:
Wherein m is the number of one-level good friend user set.
In alternative embodiments, in the step calculating propensity value, the level of interaction of each element in each in can gathering based on user and two or more one-level good friend user, calculates two or more corresponding propensity value.On average propensity value can be calculated to the level of interaction that Top N number of one-level good friend user gathers interior good friend:
Pref(hostid，community
i)＝∑pref(hostid，community
i)/N (10)
Pref(hostid，community
i)＝pref(hostid，community
i)/∑pref(hostid，community
i)(11)
In another alternative embodiment, in the step calculating propensity value, two or more corresponding propensity value can be calculated by the mode of the model pre-estimating of such as PCA and so on.
The mode of the calculating propensity value of foregoing description can use according to particular case.
According to embodiment, recommend to user the step of the information of being correlated with (S204) to comprise the respective good friend of each element in each one-level good friend user set of selection, the result met the demands is added the step (S402) of the extension set of this one-level good friend user set; The step of (S403) is filtered to each element extended in set, and for the element still retained after filtration, calculate it about the correlation of user and the step (S404) of sequence.
In an embodiment, the respective good friend of each element in first selecting each one-level good friend user to gather, adds the extension set (S402) of this one-level good friend user set by the result met the demands.As shown in Figure 3, FOF101, FOF103, FOF201 are the secondary good friend extended from primary user set 301.There is friend relation to be called to be an in-degree (degree) of this secondary good friend and set from certain one-level good friend user element (good friend user) of gathering in secondary good friend and this set that community_i extends by one, according to the quantity (size (community_i)) of the element in each in different two or more one-level good friend users set, then secondary good friend adds and extends set and need meet the following conditions:
(1) when one-level good friend user set comprises the element of less than 20, namely during size (community_i) < 20, the degree > size (community_i) * 0.8 of secondary good friend.This be due to set interior element number less time, in-degree need meet higher threshold value, at this moment can ensure the degree of correlation of candidate source;
(2) when 32 > size (community_i) >=20, degree >=16 of secondary good friend;
(3) when size (community_i) >=32, the degree > size (community_i) * 0.5 of secondary good friend.Namely when set interior element is more, the threshold value of in-degree can suitably be relaxed, and makes candidate source quantity more, to keep data integrity when selecting.
Then the element (i.e. secondary good friend) in each extension is filtered (S403).Wherein judge the secondary good friend whether Zeng Zuowei recommendation results display of potential recommendation, when the information relevant to secondary good friend once the recommended number of times shown then filtered after exceeding certain threshold value (such as: 10 times); Such as, if a user has been filtered higher duration (such as: 7 large periods, 7 days), then the recommended number of times of Information Level of this secondary good friend has been set to 0, again judges whether to filter this user according to aforementioned principle.
Then the correlation (S404) of secondary good friend about user of the recommendation still retained after filtering is calculated.Calculating can use machine learning model to calculate weight and according to weight sequencing.Wherein aspect of model set such as: the i) matching degree of static aspect data, comprising: school, company, local, hobby; Ii) liveness factor, comprising: log in frequency, good friend's number iii) user and secondary good friend Dynamic link library each other, comprising: common good friend's number, whether have visit record etc.Whether the positive sample that wherein model label marks is such as have application to add as a friend; Negative sample is such as defined as and shows more than 5 times as recommendation results, but the sample do not added as a friend yet.Wherein machine learning model can select Logistic Regression in addition.The result of last correlation is score (hostid, fofid), and sorts according to this correlation.
According to embodiment, recommend to user the step of the information of being correlated with (S204) also to comprise and based on the two or more propensity value calculated, the secondary good friend extended is selected and sorts and show as recommendation results.Wherein selected and sorted comprises at least following two kinds of modes:
(1) with corresponding propensity value for proportional distribution described relevant information to be recommended.Such as pre-determine and recommend 80 good friends altogether, if Pref is (hostid, community
1)=0.4 is the one-level good friend user set that propensity value is the highest, then get 80*0.4 community
1extension set in the information that preceding secondary good friend is correlated with that sorts be placed on, by that analogy.
(2) calculate the ranking value of each good friend user of each element, and entirely sort together according to the respective good friend of ranking value to each element.Ranking value can be expressed as:
rank(hoted，fofid)＝Pref(hostid，community
i)×score(hostid，fofid)×score(hostid，friendid)(12)
Do full sequence according to rank (hosted, fofid), get the relevant information of the secondary good friend of front predetermined number (such as 80) as last recommendation results.
In addition because the secondary good friend in the extension set of each one-level good friend user set may repeat, therefore duplicate removal can be carried out after the above step.
To user recommend information display as shown in Figure 5.It should be noted that Fig. 5 is only for the object of example, this display can comprise more or less recommendation information, other user interactions projects or different layouts.
In addition, the step of the above-mentioned method to user's recommendation information can such as according to predetermined cycle (such as 1 day) circular flow (S205), the recommendation of information is made to be real-time, to guarantee that recommendation results can upgrade, and do not decay with the tendentiousness reduction of one-level good friend user set.
According to embodiments of the invention, additionally provide a kind of for the device to the user's recommendation information in social networks, comprising: for the multiple good friend users of user in social networks being classified as the device of two or more one-level good friend user set; For for each in two or more one-level good friend user set, calculate the tendentiousness that user gathers each one-level good friend user, thus obtain the device of two or more corresponding propensity value; And for according to two or more corresponding propensity value, based on the respective friend information of each element in social networks in two or more one-level good friend user set, recommend the device of the information of being correlated with to user.
Device for sorting out comprises: for generation of have multiple good friend user mutually between the data list of weight, and data list is substituted in community's detection algorithm the device coming to sort out multiple good friend user.
Device for calculating propensity value is configured to: based on the historical activity information of received user in social networks, calculates two or more corresponding propensity value.
Wherein historical activity information comprise following in one or more: the number of times of the operation that user adds as a friend for the application of once recommended information; User no longer recommends the number of operations of the information relevant with this good friend for the selection of once recommended information; The recommended number of times of the information of once recommending; And user enters the number of times of the page be associated to relevant information.
Device for calculating propensity value is also configured to: the level of interaction of each element in each in gathering based on user and two or more one-level good friend user, calculates two or more corresponding propensity value.
Device for calculating propensity value is also configured to: calculate two or more corresponding propensity value by the mode of model pre-estimating.
The respective friend information of each element in social networks comprise following in one or more: the quantity of the element in each in two or more one-level good friend user set; The quantity of the friend relation of all elements during the respective good friend of element and the one-level good friend user at its place gather; The information relevant to respective good friend is recommended number of times once; Limit the information recommended duration relevant to respective good friend; And the respective good friend calculated is about the correlation of user.
Comprise for recommending the step of the information of being correlated with: for the device of corresponding propensity value for proportional distribution relevant information to be recommended.
Device for the step of recommending the information of being correlated with also comprises: for calculating the ranking value of each good friend of each element, and the respective good friend of each element is carried out together to the device of full sequence according to ranking value.
Although the step shown for purposes of illustration above and device, the present invention is not limited to these steps or device, and wherein some step can combine or change order.In force, some step or device even can omit in the appropriate case.With reference to instructions of the present invention and accompanying drawing, those skilled in the art are easy to make various change of the present invention and do not exceed scope of the present invention.
Claims (18)
1., to a method for the user's recommendation information in social networks, comprising:
The multiple good friend users of described user in described social networks are classified as two or more one-level good friend user set;
For each in described two or more one-level good friend user set, calculate the tendentiousness of described user to each described one-level good friend user set, thus obtain two or more corresponding propensity value; And
According to described two or more corresponding propensity value, based on the respective friend information of each element in described social networks in described two or more one-level good friend user set, recommend the information of being correlated with to described user.
2. method according to claim 1, the step of wherein said classification comprises: produce have described multiple good friend user mutually between the data list of weight, and described data list is substituted in community's detection algorithm to sort out described multiple good friend user.
3. method according to claim 1, wherein in the step of the described propensity value of described calculating, based on the received historical activity information of described user in described social networks, calculates described two or more corresponding propensity value.
4. method according to claim 3, wherein said historical activity information comprise following in one or more:
The number of times of the operation that described user adds as a friend for the application of once recommended information;
Described user no longer recommends the number of operations of described good friend for the selection of described once recommended information;
The recommended number of times of described information of once recommending; And
Described user enters the number of times of the page be associated to described relevant information.
5. method according to claim 1, wherein in the step of the described propensity value of described calculating, described in each in gathering based on described user and described two or more described one-level good friend user, the level of interaction of each element, calculates described two or more corresponding propensity value.
6. method according to claim 1, wherein in the step of the described propensity value of described calculating, calculates described two or more corresponding propensity value by the mode of model pre-estimating.
7. method according to claim 1, wherein said respective friend information comprise following in one or more:
The quantity of the described element in each in described two or more one-level good friend user set;
The quantity of the friend relation of all described element during the respective good friend of described element and the described one-level good friend user at its place gather;
The information relevant to described respective good friend once recommended number of times;
Limit the information recommended duration relevant to described respective good friend; And
The described respective good friend calculated is about the correlation of described user.
8. method according to claim 1, the step of the information that wherein said recommendation is correlated with comprises: with corresponding propensity value for proportional distribution described relevant information to be recommended.
9. method according to claim 1, the step of the information that wherein said recommendation is correlated with also comprises: the ranking value calculating each good friend user described of each element described, and entirely sorts together according to the described respective good friend of described ranking value to each element described.
10., for the device to the user's recommendation information in social networks, comprising:
For the multiple good friend users of described user in described social networks being classified as the device of two or more one-level good friend user set;
For for each in described two or more one-level good friend user set, calculate the tendentiousness of described user to each described one-level good friend user set, thus obtain the device of two or more corresponding propensity value; And
For according to described two or more corresponding propensity value, based on the respective friend information of each element in described social networks in described two or more one-level good friend user set, recommend the device of the information of being correlated with to described user.
11. devices according to claim 10, the wherein said device for sorting out comprises: for generation of have described multiple good friend user mutually between the data list of weight, and described data list is substituted in community's detection algorithm the device sorting out described multiple good friend user.
12. devices according to claim 10, the wherein said device for calculating described propensity value is configured to: based on the received historical activity information of described user in described social networks, calculates described two or more corresponding propensity value.
13. devices according to claim 12, wherein said historical activity information comprise following in one or more:
The number of times of the operation that described user adds as a friend for the application of once recommended information;
Described user no longer recommends the number of operations of described good friend for the selection of described once recommended information;
The recommended number of times of described information of once recommending; And
Described user enters the number of times of the page be associated to described relevant information.
14. devices according to claim 10, wherein be configured at the described device for calculating described propensity value: the level of interaction of each element described in each in gathering based on described user and described two or more described one-level good friend user, calculates described two or more corresponding propensity value.
15. devices according to claim 10, the wherein said device for calculating described propensity value is configured to: calculate described two or more corresponding propensity value by the mode of model pre-estimating.
16. devices according to claim 10, wherein said respective friend information comprise following in one or more:
The quantity of the described element in each in described two or more one-level good friend user set;
The quantity of the friend relation of all described element during the respective good friend of described element and the described one-level good friend user at its place gather;
The information relevant to described respective good friend once recommended number of times;
Limit the information recommended duration relevant to described respective good friend; And
The described respective good friend calculated is about the correlation of described user.
17. devices according to claim 10, wherein said for recommending the step of the information of being correlated with to comprise: for the device of corresponding propensity value for proportional distribution described relevant information to be recommended.
18. devices according to claim 10, wherein said for recommending the device of the information of being correlated with also to comprise: for calculating the ranking value of each good friend described in each element described, and the described respective good friend of each element described to be carried out together to the device of full sequence according to described ranking value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310281786.4A CN104281599A (en) | 2013-07-02 | 2013-07-02 | Method and device for recommending information to user in social network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310281786.4A CN104281599A (en) | 2013-07-02 | 2013-07-02 | Method and device for recommending information to user in social network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104281599A true CN104281599A (en) | 2015-01-14 |
Family
ID=52256477
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310281786.4A Pending CN104281599A (en) | 2013-07-02 | 2013-07-02 | Method and device for recommending information to user in social network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104281599A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106550004A (en) * | 2015-09-23 | 2017-03-29 | 阿里巴巴集团控股有限公司 | The recommendation method and device of business object |
| CN109154945A (en) * | 2017-04-24 | 2019-01-04 | 微软技术许可有限责任公司 | New connection based on data attribute is recommended |
| US10628677B2 (en) | 2016-03-14 | 2020-04-21 | Tencent Technology (Shenzhen) Company Limited | Partner matching method in costarring video, terminal, and computer readable storage medium |
| CN113343112A (en) * | 2021-06-30 | 2021-09-03 | 北京百易数字技术有限公司 | Customer relationship management method and system based on social media |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102637183A (en) * | 2011-02-12 | 2012-08-15 | 北京千橡网景科技发展有限公司 | Method and device for recommending friends to user in social network |
| CN102664828A (en) * | 2012-04-26 | 2012-09-12 | 复旦大学 | System and method for friend recommendation in social network service (SNS) network |
| US20130031016A1 (en) * | 2011-07-25 | 2013-01-31 | Isaac Rubinstein | System for posting profiles of anonymous individuals and for enabling potential employers to discover and to communicate with the anonymous individuals |
| CN102915307A (en) * | 2011-08-02 | 2013-02-06 | 腾讯科技（深圳）有限公司 | Device and method for recommending personalized information and information processing system |
| CN103138954A (en) * | 2011-12-02 | 2013-06-05 | 中国移动通信集团公司 | Recommending method, recommending system and recommending server of referenced item |
| CN103166918A (en) * | 2011-12-12 | 2013-06-19 | 阿里巴巴集团控股有限公司 | Data recommendation method and device |
-
2013
- 2013-07-02 CN CN201310281786.4A patent/CN104281599A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102637183A (en) * | 2011-02-12 | 2012-08-15 | 北京千橡网景科技发展有限公司 | Method and device for recommending friends to user in social network |
| US20130031016A1 (en) * | 2011-07-25 | 2013-01-31 | Isaac Rubinstein | System for posting profiles of anonymous individuals and for enabling potential employers to discover and to communicate with the anonymous individuals |
| CN102915307A (en) * | 2011-08-02 | 2013-02-06 | 腾讯科技（深圳）有限公司 | Device and method for recommending personalized information and information processing system |
| CN103138954A (en) * | 2011-12-02 | 2013-06-05 | 中国移动通信集团公司 | Recommending method, recommending system and recommending server of referenced item |
| CN103166918A (en) * | 2011-12-12 | 2013-06-19 | 阿里巴巴集团控股有限公司 | Data recommendation method and device |
| CN102664828A (en) * | 2012-04-26 | 2012-09-12 | 复旦大学 | System and method for friend recommendation in social network service (SNS) network |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106550004A (en) * | 2015-09-23 | 2017-03-29 | 阿里巴巴集团控股有限公司 | The recommendation method and device of business object |
| CN106550004B (en) * | 2015-09-23 | 2019-12-10 | 阿里巴巴集团控股有限公司 | Service object recommendation method and device |
| US10628677B2 (en) | 2016-03-14 | 2020-04-21 | Tencent Technology (Shenzhen) Company Limited | Partner matching method in costarring video, terminal, and computer readable storage medium |
| CN109154945A (en) * | 2017-04-24 | 2019-01-04 | 微软技术许可有限责任公司 | New connection based on data attribute is recommended |
| CN113343112A (en) * | 2021-06-30 | 2021-09-03 | 北京百易数字技术有限公司 | Customer relationship management method and system based on social media |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhu et al. | Predicting user activity level in social networks | |
| US20190104197A1 (en) | Discovering signature of electronic social networks | |
| Epstein et al. | Taming data complexity in lifelogs: exploring visual cuts of personal informatics data | |
| CN104021233B (en) | Social network friend recommendation method based on community discovery | |
| US9152969B2 (en) | Recommendation ranking system with distrust | |
| US10178197B2 (en) | Metadata prediction of objects in a social networking system using crowd sourcing | |
| US8972275B2 (en) | Optimization of social media engagement | |
| Saavedra et al. | Asymmetric disassembly and robustness in declining networks | |
| JP5560367B2 (en) | Method, system and server for managing dynamic information of friends in a network | |
| US20160055541A1 (en) | Personalized recommendation system and methods using automatic identification of user preferences | |
| CN106940705A (en) | A kind of method and apparatus for being used to build user's portrait | |
| EP2693669A2 (en) | Content feed for facilitating topic discovery in social networking environments | |
| US20090287687A1 (en) | System and method for recommending venues and events of interest to a user | |
| CN106022800A (en) | User feature data processing method and device | |
| CN108416616A (en) | The sort method and device of complaints and denunciation classification | |
| CN107548500A (en) | Event anomalies based on user's routine model | |
| CN111723292B (en) | Recommendation method, system, electronic equipment and storage medium based on graph neural network | |
| CN105446970A (en) | Item recommendation method and device | |
| EP2581869A1 (en) | Content quality and user engagement in social platforms | |
| Xiong et al. | Understanding user-to-User interaction on government microblogs: An exponential random graph model with the homophily and emotional effect | |
| JP2016502709A (en) | Method and system for determining PYMK usage and content based on a value model | |
| CN110637317A (en) | Distributed node cluster for establishing digital contact points across multiple devices on a digital communications network | |
| CN110322281A (en) | The method for digging and device of similar users | |
| CN104281599A (en) | Method and device for recommending information to user in social network | |
| De Salve et al. | Predicting influential users in online social network groups |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150114 |
|
| WD01 | Invention patent application deemed withdrawn after publication |