CN107633260A - A kind of social network opinion leader method for digging based on cluster - Google Patents

A kind of social network opinion leader method for digging based on cluster Download PDF

Info

Publication number
CN107633260A
CN107633260A CN201710729792.XA CN201710729792A CN107633260A CN 107633260 A CN107633260 A CN 107633260A CN 201710729792 A CN201710729792 A CN 201710729792A CN 107633260 A CN107633260 A CN 107633260A
Authority
CN
China
Prior art keywords
mrow
user
opinion
leader
information
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.)
Granted
Application number
CN201710729792.XA
Other languages
Chinese (zh)
Other versions
CN107633260B (en
Inventor
张波
张倩
李美子
潘建国
赵勤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Normal University
University of Shanghai for Science and Technology
Original Assignee
Shanghai Normal University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shanghai Normal University filed Critical Shanghai Normal University
Priority to CN201710729792.XA priority Critical patent/CN107633260B/en
Publication of CN107633260A publication Critical patent/CN107633260A/en
Application granted granted Critical
Publication of CN107633260B publication Critical patent/CN107633260B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present invention relates to a kind of social network opinion leader method for digging based on cluster, to obtain the leader of opinion in social network user, comprise the following steps：1) social networks model is established, and obtain the in-degree of each user in social networks model, Betweenness Centrality and in-degree group gathers coefficient；2) user is clustered using K means clustering algorithms according to the coefficient that gathers of the in-degree of user, Betweenness Centrality and in-degree group, leader of opinion's candidate user collection L is obtained in cluster result；3) user activity and user force of user in leader of opinion's candidate user collection L is calculated, and user's leading capacity is calculated according to user activity and user force；4) leader of opinion is obtained in leader of opinion's candidate user collection L according to user's leading capacity.Compared with prior art, the present invention has the advantages that to consider that comprehensive, assessment is accurate, it is accurate to calculate.

Description

A kind of social network opinion leader method for digging based on cluster
Technical field
The present invention relates to social networks technical field, is dug more particularly, to a kind of social network opinion leader based on cluster Pick method.
Background technology
Social network opinion leader has significant impact in terms of thought, impression and action to people.And due to social networks Opening, they are more powerful than domestic consumer in terms of information propagation.Certainly, the research to leader of opinion is society One of most important research of network user's analysis field is handed over, is widely used in the analysis prediction of information propagation, Public-opinion directing and prison Superintend and direct and the business development of community network.
Leader of opinion's mining process of processing in to(for) big data is a challenge all the time.Most leaders of opinion, which excavates, to be calculated Method to the influence power of the whole network user assess without differentiation, and user is more in social networks, and the time complexity of calculating process is got over It is high.Cha M et al. use user's number of degrees, and user is mentioned number, deliver content and the numerical value such as are forwarded or quote and carry out analysis mining and anticipate See leader；Bright et al. the Sentiment orientations that user is added on the basis of LeaderRank of Xu Jun and user activity carry out opinion neck Sleeve excavates；The related cum rights microblogging graph model of Wu's Xian brightness et al. structure topic, graph model is found using the thought of random walk Central point, the leader of opinion in microblogging is excavated with this；New et al. the first identifications of Cao Jiu obtain theme community, then respectively from knot Three structure, behavior and emotion dimensions are measured to user force, propose that MFP algorithms excavate leader of opinion；Chen Yuan et al. is pressed Structural hole position in community network, centrad position and marginal position carry out leader of opinion's identification；Song Qian is pretty et al. to be considered User activity and user force calculate user's leading capacity, and the user excavated in conjunction with user-centricity in social networks leads Person；Wu Chao et al. replys relation structure under specific topics, according to model and replys graph of a relation, considers money order receipt to be signed and returned to the sender people's emotion and inclines Leader of opinion is identified to, money order receipt to be signed and returned to the sender Path Clustering and money order receipt to be signed and returned to the sender text similarity.Above-mentioned technology not in social networks to Family is classified, and filters out the user for being likely to become leader of opinion, but to excavate social network opinion leader from cluster angle Provide the foundation.
The content of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide one kind considers comprehensive, assessment Accurately, the accurately social network opinion leader method for digging based on cluster is calculated.
The purpose of the present invention can be achieved through the following technical solutions：
A kind of social network opinion leader method for digging based on cluster, to obtain the neck of the opinion in social network user Sleeve, comprises the following steps：
1) establish social networks model, and obtain the in-degree of each user in social networks model, Betweenness Centrality and In-degree group gathers coefficient；
2) K-means clustering algorithms pair are used according to the coefficient that gathers of the in-degree of user, Betweenness Centrality and in-degree group User is clustered, and leader of opinion's candidate user collection L is obtained in cluster result；
3) user activity and user force of user in leader of opinion's candidate user collection L are calculated, and is lived according to user Jerk and user force calculate user's leading capacity；
4) leader of opinion is obtained in leader of opinion's candidate user collection L according to user's leading capacity.
In described step 1), the calculating formula of the in-degree of user is：
Wherein, DI(u) in-degree for being user u, δvuIt is defined as when user v is user u follower, then having δvu=1, when When user v is not user u follower, then there is δvu=0, V are user's set in social networks.
In described step 1), the calculating formula for gathering coefficient of in-degree group is：
Wherein, CI(u) coefficient is gathered for user u in-degree group, in-degree group user sum that n is user u, P is User u in-degree group set, M (v) are the directed edge sum for having with user v necessary being between the user of direct frontier juncture system, N (v) it is to have the total number of users of direct frontier juncture system with user v.
In described step 1), the calculating formula of the Betweenness Centrality of user is：
Wherein, BI(u) Betweenness Centrality for being user u, σmn(u) pass through in the shortest path between user m and user n User u shortest path number, σmnThe sum of shortest path between user m and user n.
In described step 2), in the cluster that cluster result obtains selection meet simultaneously the element in-degree at cluster center it is maximum, User in the cluster for gathering coefficient maximum and maximum three conditions of Betweenness Centrality of in-degree group adds leader of opinion candidate User collects L.
When in the absence of the element in-degree for meeting cluster center simultaneously it is maximum, in-degree group gather that coefficient is maximum and intermediary in During the cluster of maximum three conditions of disposition, then selection meets the cluster of any two conditions simultaneously, and user therein is added into opinion Leader's candidate user collection L.
In described step 3), the calculating formula of user activity is：
UA (u)=α1FP'(u)+α2FF'(u)+α3FE'(u)
α1+α2+α3=1
ΔTp=Tnow-Tfirstpublish
ΔTf=Tnow-Tfirstforward
ΔTe=Tnow-Tfirstevaluate
Wherein, UA (u) is user u user activity, and FP (u) is the frequency that user u releases news, and FF (u) is user u The frequency of forwarding information, FE (u) be user's u comment informations frequency, FP'(u), FF'(u), FE'(u) be respectively FP (u), FF (u), the value after FE (u) min-max standardizations, Δ TpThe time T of data is being obtained for user unowWith releasing news earliest Time TfirstpublishBetween interval,It is user u in Δ TpTime in the sum that releases news, Δ TfFor User u is obtaining the time T of datanowWith the time T of earliest forwarding informationfirstforwardBetween interval,For User u is in Δ TfTime in forwarding information sum, Δ TeThe time T of data is being obtained for user unowWith earliest comment information Time TfirstevaluateBetween interval,It is user u in Δ TeTime in forwarding information sum, α1、α2、 α3For the weights after being distributed by step analysis.
In described step 3), the calculating formula of user force is：
Wherein, UI (u) be user u user force, UI (v) be user v user force, ADvuFor user v to Family u attention rate, the follower that fans (u) is user u gather, | V | it is total number of users in social networks, D (v) is user v's Degree sum, CR (u) are user u information coverage, and CRi (u) is information i information coverage, and pub (u) ∪ for (u) are to use Family u is issued or the set of forwarding information, | for (u) | it is the sum of user's u forwarding informations, | pub (u) | released news for user u Sum, δviIt is defined as when user v forwards or commented on user u information i, information i covers user v, then has δvi=1, when User v does not forward or commented on user u information i, and information i is not covered with user v, then has δvi=0 is, | for (v) | to use The sum of other people information of family v forwarding, | eva (v) | the sum of other people information is commented on for user v, | for (v.source=u) | For user v forwarding user's u information sum, | eva (v.source=u) | for user v comment user's u information sum, | For (k) | the sum for forwarding other people information for being user k, | pub (k) | the sum to be released news for user k, focus (v) they are use Family v follower's set,.
In described step 3), the calculating formula of user's leading capacity is：
ULD (u)=UI (u) × UA (u)
Wherein, ULD (u) is user u user's leading capacity.
Described step 4) specifically includes following steps：
All users in leader of opinion's candidate user collection L are sorted from big to small according to user's leading capacity, K before selection As leader of opinion.
Compared with prior art, the present invention has advantages below：
First, consider comprehensive：Leader of opinion's candidate user collection is filtered out in terms of topological attribute, is determined in terms of user property The technological frame of leader of opinion, considers topological attribute and user property, avoids analyzing knot caused by only using part attribute Fruit one-sidedness.
2nd, it is accurate to assess：User's in-degree group gathers definition and the computational methods of coefficient.Consider there is direct frontier juncture with user In-degree group of the follower of system and indirect frontier juncture system as user, the coefficient that gathers of user's in-degree group more accurately have evaluated Close relation between the follower of user's in-degree group member, relation is closer, is more possible to spread the letter that user propagates Breath.User's in-degree group gathers element of the coefficient as cluster analysis, can effectively improve the level of cluster analysis, so as to get meaning See that leader's candidate user set member has sufficient leader of opinion's feature.
3rd, it is accurate to calculate：The computational methods of user's leading capacity include：The calculating of user activity and the meter of user force Calculate.The leader of opinion that user activity effectively ensures to excavate is active in social networks.Consider user's shadow comprehensively simultaneously Ring the miscellaneous sources of power：Itself (information coverage) and the influence power of follower are contributed so that the calculating of user force is more To be accurate.
Brief description of the drawings
Fig. 1 is flow chart of the method for the present invention.
Fig. 2 is an example network node diagram of the invention.
Embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.
Embodiment
The present invention is defined to social networks first：
Social networks model：Social networks form turns to figure G=(V, E, R), and V represents user's set in social networks, E The set of relation between user is described, R is the matrix for representing N × N of relation between user.
Topological attribute：Topological attribute is digraph G midpoints and point, the collection of relation function between Bian Yubian, and point and side Close.
User property：User property is quantitative relation of the user in social networks between various actions.
User's leading capacity：Leader of opinion is the people for having in social networks diffusion information capability, and user's leading capacity is to this The quantization of ability.User's leading capacity depends primarily on user activity and user force.
Follower and follower：Follower and follower occur in pairs, in social networks user u and user v it Between directed edge be present, and directed edge source point is u, terminal v, then user u is user v follower, and user v is user u pass Note person.
Leader of opinion：Leader of opinion is the K user that user's leading capacity is maximum in social networks, is represented with O, while social Non-opinion leader user in network is domestic consumer, is represented with C.
Leader of opinion's candidate user collection：Leader of opinion's candidate user collection L is before accurately user's leading capacity of user is calculated It is considered as that the most possible user as leader of opinion gathers.
As shown in figure 1, provide the technology that the present invention excavates social network opinion leader from cluster angle
Concretely comprising the following steps for social network opinion leader technology is excavated from cluster angle：
A. the social networks concept related to leader of opinion's excavation and modeling are defined.
B. (1) calculates the in-degree and Betweenness Centrality of user according to the relation of directed edge between user；Statistics can pass through one The user of bar side or two frontier juncture note users form the in-degree group of user, calculate user's in-degree group and gather coefficient；(2) combine In-degree, Betweenness Centrality and the in-degree group of above-mentioned user gathers coefficient, the K-means clusters after being improved using K-means++ Algorithm clusters to user；(3) in obtained multiple clusters, there will be a larger in-degree, larger Betweenness Centrality, and compared with The user that big in-degree group gathers in the cluster of coefficient adds leader of opinion's candidate user and concentrated.
C. the number of the issue of (1) difference counting user, forwarding and comment information, and information is released news and obtained at first Time interval, at first forwarding information and obtain information time interval, at first comment information and obtain information time interval, Calculating the Subscriber Unit time averagely issues, forwards and comment information number, and user activity is calculated；(2) counting user institute Information sum and the information of forwarding and the comment sum of reception, and forwarding number and comment of the user to the information of each follower Number, calculate the attention rate for each user that user is paid close attention to it；(3) forwarding or comment user issue and forwarding information are calculated Number the information coverage of user is calculated；(4) each follower of user is considered as into user to the attention rate of user to chase after With contribution weight of the person to user force, the influence power that user obtains at follower is calculated, combining information coverage rate calculates User force.
D. the user activity and user force that basis calculates, the leading capacity of user is calculated；The user that will be calculated Leading capacity sorts in descending order, and it is leader of opinion to take K maximum user of family leading capacity.
(1) selection of leader of opinion's candidate user collection
When user is the leader of opinion in social networks, has numerous users and select to pay close attention to the user, follow the user, Show on figure G that the terminal for being there are multiple summits is the user, in-degree of the user on social networks G is expressed as：
Wherein, the δ when user v is user u followervu=1, the δ when user v is not user u followervu=0. Obviously, the in-degree of user reflects whether user can turn into leader of opinion, the high use of user's in-degree from the quantity of user follower The family user lower than user in-degree more likely turns into leader of opinion.In-degree group gathers coefficient from relation between user follower Consider that user turns into the possibility of leader of opinion in compactness, when there is a directed edge to point to u from v, user v is user u In-degree group member, when there is another directed edge to point to v from w, user w is also user u in-degree group member, in-degree group Gather coefficient to be expressed as：
Wherein, P is user u in-degree group set, and n is user u in-degree group user sum, and N (v) is and user v There is the total number of users of direct frontier juncture system., M (v) is and user v has the directed edge of necessary being between the user of direct frontier juncture system total Number.In addition, Betweenness Centrality is the embodiment of user's carrying information ability in whole social networks G, it is expressed as：
Wherein σmnIt is the sum of shortest path between user m and user n, σmn(u) it is most short between user m and user n Pass through user u shortest path number, B in pathI(u) shortest path that can weigh any two user in social networks passes through User u probability.
Cluster analysis is carried out to user using K-means++ improved k-means algorithms；Selected in the cluster that cluster obtains Meet the cluster of condition：The element in-degree at cluster center is maximum, and in-degree group gathers coefficient maximum, and Betweenness Centrality is maximum, and (three is maximum Condition), if not meeting the cluster of three maximal conditions, the cluster for meeting two maximal conditions may be selected, user in cluster is added into opinion neck In sleeve candidate user collection L.
(2) calculating of user activity
User is not the receive information in social networks always, is issued, forwarding, is commented in the unit interval by calculating user The Information Number of opinion can calculate liveness of the user on social networks.The frequency that user u releases news is defined as：
ΔTp=Tnow-Tfirstpublish
Wherein, TnowBe obtain data time, TfirstpublishIt is that user u releases news earliest in the data of acquisition Time,It is user u in Δ TpTime in the set that releases news,It is user u in Δ TpWhen The interior sum to release news.FP (u) values are bigger, and it is more active to illustrate that user u releases news.The frequency definition of user's forwarding information For：
ΔTf=Tnow-Tfirstforward
Wherein, TfirstforwardIt is the time of the earliest forwarding informations of user u in the data of acquisition,It is user u In Δ TfTime in forwarding information set,It is user u in Δ TfTime in forwarding information sum.FF (u) value is bigger, illustrates that user's u forwarding informations are more active.The frequency of user comment information is defined as：
Wherein, TfirstevaluateIt is the time of the earliest comment informations of user u in the data of acquisition,It is user U is in Δ TeTime in comment information set,It is user u in Δ TeTime in comment information sum.FE (u) value is bigger, and it is more active to illustrate that user u comments on other people information.
Release news frequency to user, forwarding information frequency, and comment information frequency carries out min-max standardizations：
Wherein,WithIt is leader of opinion's candidate user collection L respectively The minimum frequency that releases news of middle user, minimum forwarding information frequency and minimum comment information frequency.WithIt is that user is maximum in leader of opinion's candidate user collection L respectively Release news frequency, maximum forwarding information frequency and maximum comment information frequency.
Released news frequency, forwarding information frequency and comment information frequency according to user, comments leader's candidate user The user activity of user is concentrated, is defined as follows：
UA (u)=α1FP'(u)+α2FF'(u)+α3FE'(u)
Wherein α1+α2+α3=1, α1, α2And α3Value using step analysis (Analytic Hierarchy Process, AHP) it is allocated.
(3) calculating of user force
1) calculating of the user to follower's degree of concern
User may have multiple followers, but user might not forward or comment on the information of each follower, use Family v is defined as follows to user u attention rate：
Wherein focus (v) is user v follower's set, | for (u) | it is the sum of user's u forwarding informations, | pub (u) | it is the sum that user u releases news, | eva (v) | it is the sum of other people information of user comment, | for (v.source=u) | be The sum of user v forwarding users u information, | eva (v.source=u) | it is the sum of user v comment users u information.ADvu Value it is bigger, illustrate that concerns of the user v to user u is more.
2) calculating of user profile coverage rate
For each user in social networks, the information coverage of user illustrates the influence of user to a certain extent Power, the information coverage of user are defined as follows：
Wherein, δviEqual to the information i that 1 explanation user v forwarded or commented on user u, information i covers user v, δvi Equal to the information i that 0 explanation user did not forward or commented on user u, information i is not covered with user v.| V | it is social networks Middle total number of users, pub (u) ∪ for (u) are the information aggregates of user u issues or forwarding, and the information that user u is issued or forwarded is all Belong to user u information.
3) calculating of user force
The information coverage of synthetic user, the follower of user is to the attention rate of user and the influence power of user follower The influence power of user is obtained, the influence power of user is defined as follows：
Wherein fans (u) is follower's set of user person, and when user is one in leader of opinion user's Candidate Set During member, the initial effects power of user is 1, and otherwise the initial effects power of user is total number of users in the degree and social networks of user Ratio.
4) selection of leader of opinion
The liveness of synthetic user and the influence power of user obtain the leading capacity of user, and user's leading capacity is defined as follows：
ULD (u)=UI (u) × UA (u)
User's leading capacity of user in leader of opinion's candidate user collection L is sorted according to order from big to small, selection row The forward preceding K user of sequence is leader of opinion.
Embodiment：
Following examples are provided to illustrate to excavate the technology of leader of opinion in social networks (such as the institute of accompanying drawing 2 from cluster angle Show).
1st, the selection of leader of opinion's candidate user collection
1) in-degree calculates
User V1 in-degrees D in accompanying drawing 2I=1, (V1) user V2 in-degrees DI(V2)=2, user V3 in-degrees DI(V3)=2, use Family V4 in-degrees DI(V4)=2, user V5 in-degrees DI(V5)=2.
2) in-degree group gathers coefficient calculating
For user V1, it be V2 to have side to point to V1 user, and the user for having side sensing V2 is V1 and V3, therefore user V1 In-degree group is (V1, V2, V3),
For user V2, the user for having side to point to V2 has side to point to V3, therefore user V2 in-degree for V1, V3, while V5 Group is (V1, V2, V3, V5),
For user V3, the user for having side to point to V3 has side to point to V1 for V1, V5, while V2, and V4 has side to point to V5, therefore V3 in-degree group is (V1, V2, V3, V4, V5),
For user V4, the user for having side sensing V4 is that V3, V5, while V1 have side to point to V3, and V2 points to V5, therefore user V4 in-degree group is (V1, V2, V3, V4, V5),
For user V5, the user for having side sensing V5 is that V2, V4, while V3 have side to point to V4, and V1 has side to point to V2, therefore User V5 in-degree group is (V1, V2, V3, V4, V5),
3) Betweenness Centrality calculates
Shortest path present in accompanying drawing is：
There is 1 V2 → V1 → V3 by V1 shortest path in shortest path, while V2 to V3 shortest path sum is 2 Bar, therefore V1 Betweenness Centrality is：
There are 5 by V2 shortest path in shortest path, V1 → V2 → V5, V3 → V2 → V1, V3 → V2 → V5, V4 → V5 → V3 → V2 → V1, V5 → V3 → V2 → V1, while V3 to V5 shortest path sum is 2, therefore V2 intermediary center Property is：
There are 5 by V3 shortest path in shortest path, V1 → V3 → V4, V4 → V5 → V3 → V2 → V1, V4 → V5 → V3 → V2, V5 → V3 → V2 → V1, V5 → V3 → V2, therefore V3 Betweenness Centrality is：BI(V3)=1+1+1+1+1=5
There is 1, V3 → V4 → V5 by V4 shortest path in shortest path, while V3 to V5 shortest path sum is 2, therefore V4 Betweenness Centrality is：
There are 5 by V5 shortest path in shortest path, V2 → V5 → V3, V2 → V5 → V4, V4 → V5 → V3 → V2 → V1, V4 → V5 → V3 → V2, V4 → V5 → V3, while V2 to V3 shortest path sum is 2, therefore V5 intermediary center Property is：
Value to each user u topological attribute group is DI(u), CI(u),BI(u) min-max normalizeds are carried out, are made Clustered with the improved K-means algorithms of K-means++ algorithms, in this example, select user being divided into two clusters. To cluster 1 (V1, V4), cluster 2 (V2, V3, V5), and because the center of cluster 2 for clustering to obtain in in-degree and Betweenness Centrality is all big In the center of cluster 1, therefore it is leader of opinion's candidate user collection to select cluster 2 (V2, V3, V5), L=(V2, V3, V5).
2. user's leading capacity calculates
1) calculating of user activity
A) issue, forward, comment information frequency calculates
It can be obtained according to the content of table 1, for user V2, Tnow=2017-4-22, release news number 20, Tfirstpublish= 2017-4-1, calculate to obtain Δ Tp=21,Forwarding information 6, Tfirstforward=2017-4-2, Calculate to obtain Δ Tf=20,Comment information number 10, Tfirstevaluate=2017-4-3, calculate to obtain Δ Te=19,
For user V3, Tnow=2017-4-22, release news number 15, Tfirstpublish=2017-3-20, is calculated ΔTp=33,Forwarding information 10, Tfirstforward=2017-4-5, calculate to obtain Δ Tf=17,Comment information 5, Tfirstevaluate=2017-4-15, calculate to obtain Δ Te=7,
For user V5, Tnow=2017-4-22, release news number 5, Tfirstpublish=2017-3-1, calculate to obtain Δ Tp=52,Forwarding information 15, Tfirstforward=2017-4-10, calculate to obtain Δ Tf=12,Comment information 5, Tfirstevaluate=2017-4-1, calculate to obtain Δ Te=21,
The user related information table of table 1.
User V1 V2 V3 V4 V5
Release news number 1 20 15 3 5
Release news at first the time 2017-3-1 2017-4-1 2017-3-20 2017-2-18 2017-3-1
Forwarding information number 3 6 10 5 15
The forwarding information time at first 2017-3-9 2017-4-2 2017-4-5 2017-4-7 2017-4-10
Comment information number 3 10 5 4 5
The comment information time at first 2017-3-3 2017-4-3 2017-4-15 2017-1-1 2017-4-1
Obtain information time 2017-4-22 2017-4-22 2017-4-22 2017-4-22 2017-4-22
B) issue, forward, comment information frequency standard
To the frequency that releases news, forwarding information frequency, the progress min-max standardizations of frequency information frequency obtain following Table
C) user activity calculates
Issue, forwarding, comment information frequency shared weight in user activity are determined using analytic hierarchy process (AHP).Structure is sentenced Disconnected matrix
Obtain weight value α1=0.636985, α2=0.2582850, α3=0.1047294 calculates：
UA (V2)=0.636985 × 1+0.2582850 × 0+0.1047294 × 0.62=0.70
UA (V3)=0.636985 × 0.41+0.2582850 × 0.31+0.1047294 × 1=0.45
UA (V5)=0.636985 × 0+0.2582850 × 1+0.1047294 × 0=0.26
2. the calculating of user force
Obtained according to the content analysis of table 2：
The user of table 2. forwards and comment information source table
User Forwarding information source user Comment information source user
V1 V2:1, V3:2 V2:1, V3：2
V2 V5：3, V4：2, V1：1 V5：7, V1：2, V3：1
V3 V2：4, V4：5, V5：1 V2：2, V4：3
V4 V5：3, V3：2 V5：4
V5 V3：10, V4：5 V4：5
Note：V：N indicates that the source of n bar information is user V
A) user's attention rate
For user V2, follower has user V1 and user V3:
Wherein V1 follower is user V2 and user V3, therefore user V1 is to user V2 attention rate：
Wherein V3 follower is user V2 and user V4, therefore user V3 is to user V2 attention rate：
For user V3, follower has user V1 and user V5:
Wherein V1 follower is user V2 and user V3, therefore user V1 is to user V3 attention rate：
Wherein V5 follower is user V3 and user V4, therefore user V5 is to user V3 attention rate：
For user V5, follower has user V2 and V4:
Wherein V2 follower is user V1 and user V5, therefore user V2 is to user V5 attention rate:
Wherein V4 follower is user V5, therefore user V4 is to user V5 attention rate:
B) calculating of user profile coverage rate
According to table 3, table 4, the content of table 5 is analyzed.
The user V2 of table 3. information can be caused to cover situation table
User V2 information encoding Covering number
2 3
4 2
15 4
20 2
26 1
Note：Other information covering number is 0；
The user V3 of table 4. information can be caused to cover situation table
User V3 information encoding Covering number
6 3
9 1
25 4
Note：Other information covering number is 0；
The user V5 of table 5. information can be caused to cover situation table
User V5 information encoding Covering number
1 2
8 2
13 4
19 3
20 1
Note：Other information covering number is 0；
User V2 information coverage calculates：
User V3 information coverage calculates：
User V5 information coverage calculates：
C) the calculating of user force
In summary user's attention rate and user profile coverage rate, are calculated user force.
In user collects V,Therefore it is initialV2, V3, V5 ∈ L, initial UI (V2)=UI (V3)=UI (V5)=1.
User V2 follower is V1 and V3, therefore UI (V2)=CR (V2)+AD(V1)(V2)×UI(V1)+AD(V3)(V2)× UI (V3)=0.09+0.17 × 0.6+0.31 × 1=0.502 user V3 follower is V1 and V5, therefore UI (V3)=CR (V3)+AD(V1)(V3)×UI(V1)+AD(V5)(V3)× UI (V5)=0.06+0.33 × 0.6+0.38 × 1=0.638 user V5's Follower is V2 and V4, therefore
UI (V5)=CR (V5)+AD(V2)(V5)×UI(V2)+AD(V4)(V5)× UI (V4)=0.12+0.52 × 1+0.78 × The calculating of 0.6=1.1083. user's leading capacity
Synthetic user liveness and user force, user's leading capacity is calculated：
ULD (V2)=UI (V2) × UA (V2)=0.70 × 0.502=0.3514
ULD (V3)=UI (V3) × UA (V3)=0.45 × 0.638=0.2871
ULD (V5)=UI (V5) × UA (V5)=0.26 × 1.108=0.28808
Descending sort is carried out to the leading capacity of user in L,
ULD (V2) ＞ ULD (V5) ＞ ULD (V3)
If necessary to excavate a leader of opinion, then user V2 is leader of opinion；If necessary to excavate two leaders of opinion, Then user V2 and user V5 is leader of opinion.
In summary, the topological attribute of the invention based on social networks, makes full use of the user of user in social networks to belong to Property, propose to excavate the technology of social network opinion leader from the angle of cluster.Due to conventional leader of opinion's digging technology, seldom In view of only having certain customers to have the condition as leader of opinion in social networks, for this problem, the present invention utilizes society Hand over the topological attribute of user in network to be clustered, filter out with leader of opinion's candidate user as leader of opinion's condition Collection, then concentrate the leading capacity of user to analyze leader of opinion's candidate user, excavate not only with liveness but also with influence The leader of opinion of power.

Claims (10)

  1. A kind of 1. social network opinion leader method for digging based on cluster, to obtain the neck of the opinion in social network user Sleeve, it is characterised in that comprise the following steps：
    1) social networks model is established, and obtains the in-degree of each user, Betweenness Centrality and in-degree in social networks model Group gathers coefficient；
    2) coefficient is gathered using K-means clustering algorithms to user according to the in-degree of user, Betweenness Centrality and in-degree group Clustered, leader of opinion's candidate user collection L is obtained in cluster result；
    3) user activity and user force of user in leader of opinion's candidate user collection L is calculated, and according to user activity User's leading capacity is calculated with user force；
    4) leader of opinion is obtained in leader of opinion's candidate user collection L according to user's leading capacity.
  2. A kind of 2. social network opinion leader method for digging based on cluster according to claim 1, it is characterised in that institute In the step 1) stated, the calculating formula of the in-degree of user is：
    <mrow> <msup> <mi>D</mi> <mi>I</mi> </msup> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>v</mi> <mo>&amp;Element;</mo> <mi>V</mi> <mo>,</mo> <mi>v</mi> <mo>&amp;NotEqual;</mo> <mi>u</mi> </mrow> </munder> <msub> <mi>&amp;delta;</mi> <mrow> <mi>v</mi> <mi>u</mi> </mrow> </msub> </mrow>
    Wherein, DI(u) in-degree for being user u, δvuIt is defined as when user v is user u follower, then having δvu=1, work as user When v is not user u follower, then there is δvu=0, V are user's set in social networks.
  3. A kind of 3. social network opinion leader method for digging based on cluster according to claim 1, it is characterised in that institute In the step 1) stated, the calculating formula for gathering coefficient of in-degree group is：
    <mrow> <msup> <mi>C</mi> <mi>I</mi> </msup> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mi>n</mi> </mfrac> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>v</mi> <mo>&amp;Element;</mo> <mi>P</mi> </mrow> </munder> <mfrac> <mrow> <mi>M</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> </mrow> <mrow> <mi>N</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mi>N</mi> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> </mfrac> </mrow>
    Wherein, CI(u) coefficient is gathered for user u in-degree group, the in-degree group user sum that n is user u, P is user u The set of in-degree group, M (v) is the directed edge sum for having with user v necessary being between the user of direct frontier juncture system, N (v) be with User v has the total number of users of direct frontier juncture system.
  4. A kind of 4. social network opinion leader method for digging based on cluster according to claim 1, it is characterised in that institute In the step 1) stated, the calculating formula of the Betweenness Centrality of user is：
    <mrow> <msup> <mi>B</mi> <mi>I</mi> </msup> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>m</mi> <mo>&amp;Element;</mo> <mi>V</mi> </mrow> </munder> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>n</mi> <mo>&amp;NotEqual;</mo> <mi>m</mi> <mo>&amp;Element;</mo> <mi>V</mi> </mrow> </munder> <mfrac> <mrow> <msub> <mi>&amp;sigma;</mi> <mrow> <mi>m</mi> <mi>n</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> </mrow> <msub> <mi>&amp;sigma;</mi> <mrow> <mi>m</mi> <mi>n</mi> </mrow> </msub> </mfrac> </mrow>
    Wherein, BI(u) Betweenness Centrality for being user u, σmn(u) user is passed through in the shortest path between user m and user n U shortest path number, σmnThe sum of shortest path between user m and user n.
  5. A kind of 5. social network opinion leader method for digging based on cluster according to claim 1, it is characterised in that institute In the step 2) stated, select to meet simultaneously that the element in-degree at cluster center is maximum, in-degree group in the cluster that cluster result obtains The user gathered in the cluster of coefficient maximum and maximum three conditions of Betweenness Centrality adds leader of opinion's candidate user collection L.
  6. A kind of 6. social network opinion leader method for digging based on cluster according to claim 5, it is characterised in that when In the absence of the element in-degree that meets cluster center simultaneously it is maximum, in-degree group gather that coefficient is maximum and Betweenness Centrality maximum three During the cluster of item condition, then selection meets the cluster of any two conditions simultaneously, and user therein is added into leader of opinion candidate and used Family collection L.
  7. A kind of 7. social network opinion leader method for digging based on cluster according to claim 1, it is characterised in that institute In the step 3) stated, the calculating formula of user activity is：
    UA (u)=α1FP'(u)+α2FF'(u)+α3FE'(u)
    α1+α2+α3=1
    <mrow> <mi>F</mi> <mi>P</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <mo>|</mo> <mi>p</mi> <mi>u</mi> <mi>b</mi> <msup> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mrow> <msub> <mi>&amp;Delta;T</mi> <mi>p</mi> </msub> </mrow> </msup> <mo>|</mo> </mrow> <mrow> <msub> <mi>&amp;Delta;T</mi> <mi>p</mi> </msub> </mrow> </mfrac> </mrow>
    ΔTp=Tnow-Tfirstpublish
    <mrow> <mi>F</mi> <mi>F</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <mo>|</mo> <mi>f</mi> <mi>o</mi> <mi>r</mi> <msup> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mrow> <msub> <mi>&amp;Delta;T</mi> <mi>f</mi> </msub> </mrow> </msup> <mo>|</mo> </mrow> <mrow> <msub> <mi>&amp;Delta;T</mi> <mi>f</mi> </msub> </mrow> </mfrac> </mrow>
    ΔTf=Tnow-Tfirstforward
    <mrow> <mi>F</mi> <mi>E</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <mo>|</mo> <mi>e</mi> <mi>v</mi> <mi>a</mi> <msup> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mrow> <msub> <mi>&amp;Delta;T</mi> <mi>e</mi> </msub> </mrow> </msup> <mo>|</mo> </mrow> <mrow> <msub> <mi>&amp;Delta;T</mi> <mi>e</mi> </msub> </mrow> </mfrac> </mrow>
    ΔTe=Tnow-Tfirstevaluate
    Wherein, UA (u) is user u user activity, and FP (u) is the frequency that user u releases news, and FF (u) forwards for user u The frequency of information, FE (u) be user's u comment informations frequency, FP'(u), FF'(u), FE'(u) be respectively FP (u), FF (u), Value after FE (u) min-max standardizations, Δ TpThe time T of data is being obtained for user unowWith release news earliest when Between TfirstpublishBetween interval,It is user u in Δ TpTime in the sum that releases news, Δ TfFor with Family u is obtaining the time T of datanowWith the time T of earliest forwarding informationfirstforwardBetween interval,For with Family u is in Δ TfTime in forwarding information sum, Δ TeThe time T of data is being obtained for user unowWith earliest comment information Time TfirstevaluateBetween interval,It is user u in Δ TeTime in forwarding information sum, α1、α2、α3 For the weights after being distributed by step analysis.
  8. A kind of 8. social network opinion leader method for digging based on cluster according to claim 7, it is characterised in that institute In the step 3) stated, the calculating formula of user force is：
    <mrow> <mi>U</mi> <mi>I</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>C</mi> <mi>R</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>+</mo> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>v</mi> <mo>&amp;Element;</mo> <mi>f</mi> <mi>a</mi> <mi>n</mi> <mi>s</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> </mrow> </munder> <msub> <mi>AD</mi> <mrow> <mi>v</mi> <mi>u</mi> </mrow> </msub> <mo>&amp;CenterDot;</mo> <mi>U</mi> <mi>I</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> </mrow>
    <mrow> <mi>U</mi> <mi>I</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mn>1</mn> </mtd> <mtd> <mrow> <mi>v</mi> <mo>&amp;Element;</mo> <mi>L</mi> <mo>,</mo> <mi>u</mi> <mo>&amp;Element;</mo> <mi>L</mi> <mo>,</mo> <mi>v</mi> <mo>&amp;Element;</mo> <mi>f</mi> <mi>a</mi> <mi>n</mi> <mi>s</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mfrac> <mrow> <mi>D</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> </mrow> <mrow> <mo>|</mo> <mi>V</mi> <mo>|</mo> </mrow> </mfrac> </mtd> <mtd> <mrow> <mi>v</mi> <mo>&amp;NotElement;</mo> <mi>L</mi> <mo>,</mo> <mi>u</mi> <mo>&amp;Element;</mo> <mi>L</mi> <mo>,</mo> <mi>v</mi> <mo>&amp;Element;</mo> <mi>f</mi> <mi>a</mi> <mi>n</mi> <mi>s</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow>
    <mrow> <mi>C</mi> <mi>R</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mi>&amp;epsiv;</mi> <mi>p</mi> <mi>u</mi> <mi>b</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>&amp;cup;</mo> <mi>f</mi> <mi>o</mi> <mi>r</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> </mrow> </munder> <mi>C</mi> <mi>R</mi> <mi>i</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> </mrow> <mrow> <mo>|</mo> <mi>p</mi> <mi>u</mi> <mi>b</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>|</mo> <mo>+</mo> <mo>|</mo> <mi>f</mi> <mi>o</mi> <mi>r</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>|</mo> </mrow> </mfrac> </mrow>
    <mrow> <mi>C</mi> <mi>R</mi> <mi>i</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>v</mi> <mi>&amp;epsiv;</mi> <mi>V</mi> </mrow> </munder> <msub> <mi>&amp;delta;</mi> <mrow> <mi>v</mi> <mi>i</mi> </mrow> </msub> </mrow> <mrow> <mo>|</mo> <mi>V</mi> <mo>|</mo> </mrow> </mfrac> </mrow>
    <mrow> <msub> <mi>AD</mi> <mrow> <mi>v</mi> <mi>u</mi> </mrow> </msub> <mo>=</mo> <mfrac> <mrow> <mo>|</mo> <mi>f</mi> <mi>o</mi> <mi>r</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>|</mo> <mo>+</mo> <mo>|</mo> <mi>p</mi> <mi>u</mi> <mi>b</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>|</mo> </mrow> <mrow> <munder> <mo>&amp;Sigma;</mo> <mrow> <mi>k</mi> <mo>&amp;Element;</mo> <mi>f</mi> <mi>o</mi> <mi>c</mi> <mi>u</mi> <mi>s</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> </mrow> </munder> <mrow> <mo>(</mo> <mo>|</mo> <mi>f</mi> <mi>o</mi> <mi>r</mi> <mo>(</mo> <mi>k</mi> <mo>)</mo> <mo>|</mo> <mo>+</mo> <mo>|</mo> <mi>p</mi> <mi>u</mi> <mi>b</mi> <mo>(</mo> <mi>k</mi> <mo>)</mo> <mo>|</mo> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>&amp;times;</mo> <mfrac> <mrow> <mo>|</mo> <mi>f</mi> <mi>o</mi> <mi>r</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>.</mo> <mi>s</mi> <mi>o</mi> <mi>u</mi> <mi>r</mi> <mi>c</mi> <mi>e</mi> <mo>=</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>|</mo> <mo>+</mo> <mo>|</mo> <mi>e</mi> <mi>v</mi> <mi>a</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>.</mo> <mi>s</mi> <mi>o</mi> <mi>u</mi> <mi>r</mi> <mi>c</mi> <mi>e</mi> <mo>=</mo> <mi>u</mi> <mo>)</mo> </mrow> <mo>|</mo> </mrow> <mrow> <mo>|</mo> <mi>f</mi> <mi>o</mi> <mi>r</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> <mo>|</mo> <mo>+</mo> <mo>|</mo> <mi>e</mi> <mi>v</mi> <mi>a</mi> <mrow> <mo>(</mo> <mi>v</mi> <mo>)</mo> </mrow> <mo>|</mo> </mrow> </mfrac> </mrow>
    Wherein, UI (u) be user u user force, UI (v) be user v user force, ADvuIt is user v to user u Attention rate, fans (u) be user u follower gather, | V | be social networks in total number of users, D (v) be user v degree it is total Number, CR (u) are user u information coverage, and CRi (u) is information i information coverage, and pub (u) ∪ for (u) send out for user u The set of cloth or forwarding information, | for (u) | it is the sum of user's u forwarding informations, | pub (u) | released news for user u total Number, δviIt is defined as when user v forwards or commented on user u information i, information i covers user v, then has δvi=1, work as user V does not forward or commented on user u information i, and information i is not covered with user v, then has δvi=0 is, | for (v) | it is user v Forwarding other people information sum, | eva (v) | comment on the sums of other people information for user v, | for (v.source=u) | to use The sum of family v forwarding user's u information, | eva (v.source=u) | the sum for the comment user's u information for being user v, | for (k) | the sum for forwarding other people information for being user k, | pub (k) | the sum to be released news for user k, focus (v) is user v Follower set,.
  9. A kind of 9. social network opinion leader method for digging based on cluster according to claim 8, it is characterised in that institute In the step 3) stated, the calculating formula of user's leading capacity is：
    ULD (u)=UI (u) × UA (u)
    Wherein, ULD (u) is user u user's leading capacity.
  10. A kind of 10. social network opinion leader method for digging based on cluster according to claim 6, it is characterised in that Described step 4) specifically includes following steps：
    All users in leader of opinion's candidate user collection L are sorted from big to small according to user's leading capacity, K conduct before selection Leader of opinion.
CN201710729792.XA 2017-08-23 2017-08-23 Social network opinion leader mining method based on clustering Active CN107633260B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710729792.XA CN107633260B (en) 2017-08-23 2017-08-23 Social network opinion leader mining method based on clustering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710729792.XA CN107633260B (en) 2017-08-23 2017-08-23 Social network opinion leader mining method based on clustering

Publications (2)

Publication Number Publication Date
CN107633260A true CN107633260A (en) 2018-01-26
CN107633260B CN107633260B (en) 2020-10-16

Family

ID=61099724

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710729792.XA Active CN107633260B (en) 2017-08-23 2017-08-23 Social network opinion leader mining method based on clustering

Country Status (1)

Country Link
CN (1) CN107633260B (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108647247A (en) * 2018-04-16 2018-10-12 国家计算机网络与信息安全管理中心 Key node recognition methods is propagated based on the micro-blog information for improving PageRank algorithms
CN109146700A (en) * 2018-08-14 2019-01-04 西华大学 A kind of influence power feature extracting method for social networks leader
CN109657105A (en) * 2018-12-25 2019-04-19 杭州铭智云教育科技有限公司 A method of obtaining target user
CN109919794A (en) * 2019-03-14 2019-06-21 哈尔滨工程大学 A kind of microblog users method for evaluating trust based on belief propagation
CN110110084A (en) * 2019-04-23 2019-08-09 北京科技大学 The recognition methods of high quality user-generated content
CN110489658A (en) * 2019-07-12 2019-11-22 北京邮电大学 Online social network opinion leader method for digging based on digraph model
CN110737804A (en) * 2019-09-20 2020-01-31 华中科技大学 graph processing memory access optimization method and system based on activity level layout
CN111681120A (en) * 2020-05-20 2020-09-18 卓尔智联(武汉)研究院有限公司 Core user determination method, device and storage medium
CN112667876A (en) * 2020-12-24 2021-04-16 湖北第二师范学院 Opinion leader group identification method based on PSOTVCF-Kmeans algorithm
CN112785156A (en) * 2021-01-23 2021-05-11 罗家德 Industrial leader identification method based on clustering and comprehensive evaluation
CN113158082A (en) * 2021-05-13 2021-07-23 聂佼颖 Artificial intelligence-based media content reality degree analysis method

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102890702A (en) * 2012-07-19 2013-01-23 中国人民解放军国防科学技术大学 Internet forum-oriented opinion leader mining method
CN103024017A (en) * 2012-12-04 2013-04-03 武汉大学 Method for distinguishing important goals and community groups of social network
CN103136331A (en) * 2013-01-18 2013-06-05 西北工业大学 Micro blog network opinion leader identification method
CN104035987A (en) * 2014-05-30 2014-09-10 南京邮电大学 Method for ranking microblog network user influence
CN104615717A (en) * 2015-02-05 2015-05-13 北京航空航天大学 Multi-dimension assessment method for social network emergency
CN105260474A (en) * 2015-10-29 2016-01-20 俞定国 Microblog user influence computing method based on information interaction network

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102890702A (en) * 2012-07-19 2013-01-23 中国人民解放军国防科学技术大学 Internet forum-oriented opinion leader mining method
CN103024017A (en) * 2012-12-04 2013-04-03 武汉大学 Method for distinguishing important goals and community groups of social network
CN103136331A (en) * 2013-01-18 2013-06-05 西北工业大学 Micro blog network opinion leader identification method
CN104035987A (en) * 2014-05-30 2014-09-10 南京邮电大学 Method for ranking microblog network user influence
CN104615717A (en) * 2015-02-05 2015-05-13 北京航空航天大学 Multi-dimension assessment method for social network emergency
CN105260474A (en) * 2015-10-29 2016-01-20 俞定国 Microblog user influence computing method based on information interaction network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
TORE OPSAHL等: "Node Centrality in Weighted Networks: Generalizing Degree and Shortest Paths", 《SOCIAL NETWORKS》 *
李纲，陈璟浩: "网络舆情突发事件参与者群体构成及行为的实证研究", 《信息资源管理学报》 *
胡勇等: "网络舆论形成过程中意见领袖形成模型研究", 《四川大学学报(自然科学版)》 *

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108647247A (en) * 2018-04-16 2018-10-12 国家计算机网络与信息安全管理中心 Key node recognition methods is propagated based on the micro-blog information for improving PageRank algorithms
CN109146700A (en) * 2018-08-14 2019-01-04 西华大学 A kind of influence power feature extracting method for social networks leader
CN109146700B (en) * 2018-08-14 2021-11-02 西华大学 Method for extracting influence characteristics of social network leader
CN109657105B (en) * 2018-12-25 2021-10-22 杭州灿八科技有限公司 Method for acquiring target user
CN109657105A (en) * 2018-12-25 2019-04-19 杭州铭智云教育科技有限公司 A method of obtaining target user
CN109919794A (en) * 2019-03-14 2019-06-21 哈尔滨工程大学 A kind of microblog users method for evaluating trust based on belief propagation
CN109919794B (en) * 2019-03-14 2022-07-29 哈尔滨工程大学 Microblog user trust evaluation method based on trust propagation
CN110110084A (en) * 2019-04-23 2019-08-09 北京科技大学 The recognition methods of high quality user-generated content
CN110489658A (en) * 2019-07-12 2019-11-22 北京邮电大学 Online social network opinion leader method for digging based on digraph model
CN110737804A (en) * 2019-09-20 2020-01-31 华中科技大学 graph processing memory access optimization method and system based on activity level layout
CN110737804B (en) * 2019-09-20 2022-04-22 华中科技大学 Graph processing access optimization method and system based on activity degree layout
CN111681120A (en) * 2020-05-20 2020-09-18 卓尔智联(武汉)研究院有限公司 Core user determination method, device and storage medium
CN112667876A (en) * 2020-12-24 2021-04-16 湖北第二师范学院 Opinion leader group identification method based on PSOTVCF-Kmeans algorithm
CN112667876B (en) * 2020-12-24 2024-04-09 湖北第二师范学院 Opinion leader group identification method based on PSOTVCF-Kmeans algorithm
CN112785156A (en) * 2021-01-23 2021-05-11 罗家德 Industrial leader identification method based on clustering and comprehensive evaluation
CN112785156B (en) * 2021-01-23 2024-04-30 罗家德 Industrial collar and sleeve identification method based on clustering and comprehensive evaluation
CN113158082A (en) * 2021-05-13 2021-07-23 聂佼颖 Artificial intelligence-based media content reality degree analysis method
CN113158082B (en) * 2021-05-13 2023-01-17 和鸿广科技(上海)有限公司 Artificial intelligence-based media content reality degree analysis method

Also Published As

Publication number Publication date
CN107633260B (en) 2020-10-16

Similar Documents

Publication Publication Date Title
CN107633260A (en) A kind of social network opinion leader method for digging based on cluster
Adali et al. Measuring behavioral trust in social networks
CN103605791B (en) Information transmission system and information-pushing method
CN106202211B (en) Integrated microblog rumor identification method based on microblog types
JP2021524949A (en) Target recommendation method and device, storage medium and terminal device
CN101431485B (en) Method and system for automatically recommending internet information
CN100479572C (en) Method and system for monitoring suspicious user of rubbish SMS
CN103914491B (en) To the data digging method and system of high-quality user-generated content
CN106953887A (en) A kind of personalized Organisation recommendations method of fine granularity radio station audio content
Nasim et al. Real-time detection of content polluters in partially observable Twitter networks
CN104077723A (en) Social network recommending system and social network recommending method
CN108632252A (en) A kind of private network agreement iteration conversed analysis method, apparatus and server
CN104954360B (en) Sharing contents screen method and device
CN106951446A (en) Financial Information method for pushing and device
CN108874641A (en) A kind of analysis method that burying point data and device
Lane et al. Connecting personal-scale sensing and networked community behavior to infer human activities
Velayutham et al. Bot identification: Helping analysts for right data in twitter
Shafiq et al. Effective packet number for 5G IM wechat application at early stage traffic classification
CN109522460A (en) A kind of public sentiment monitoring method and system based on social networks community division
CN107015993A (en) A kind of user type recognition methods and device
CN105740323A (en) Resource characteristic and user preference-based resource matching recommendation method and system
Tang et al. Dynamic community detection with temporal Dirichlet process
Boffa et al. Context-aware advertisment recommendation on twitter through rough sets
CN106570174B (en) Big data analysis method and system running on CMMB super hotspot service platform
CN104268214A (en) Micro-blog user relationship based user gender identification method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CB03 Change of inventor or designer information
CB03 Change of inventor or designer information

Inventor after: Pan Jianguo

Inventor after: Zhang Bo

Inventor after: Zhang Qian

Inventor after: Li Meizi

Inventor after: Zhao Qin

Inventor before: Zhang Bo

Inventor before: Zhang Qian

Inventor before: Li Meizi

Inventor before: Pan Jianguo

Inventor before: Zhao Qin