cross-community influence in discussion fora
Post on 21-Oct-2014
358 views
DESCRIPTION
TRANSCRIPT
© Copyright 2010 Digital Enterprise Research Institute. All rights reserved.
Digital Enterprise Research Institute www.deri.ie
Cross-Community Influence in Discussion Fora
Václav Belák, Samantha Lam, Conor Hayes
[email protected] http://www.StefanDecker.org/
Digital Enterprise Research Institute www.deri.ie
Motivation
1
2
3
4
5
6
7
forum A forum B
• Online social communities represent an important cultural and business asset in context of many services on the Web
• Management and exploitation of these communities has thus become important and one way to do it is to focus on influential actors
• Social influence has been intensively studied in SNA, but can we extend the notion of influence to the level of communities?
Digital Enterprise Research Institute www.deri.ie
Research Questions
• How can we identify communities persistently affecting other communities?
• Given a specific community, which communities does it influence? Which communities are dependent on the activity of others?
• Over time, how can we identify that a community is being increasingly influenced or even overtaken by another community?
Digital Enterprise Research Institute www.deri.ie
Methods: Definition of Impact
• We propose to take two factors into account: 1. degree of community membership of the users 2. centrality of the users within each community
• we used in-degree (# replies of a user) • For general case of n users and k communities define:
• n × k membership matrix M • n × k centrality matrix C
• Cross-community k × k impact matrix J can then be obtained as a product of the two matrices:
• Communities have usually different sizes, we therefore work with normalised impact matrix:
Ji, j =Ji, jMl,il=1
n!
M=10.20
00.81
!
"
###
$
%
&&&,C =
2100
0105
!
"
###
$
%
&&&
J =MTC = 4 28 13
!
"#
$
%&
Digital Enterprise Research Institute www.deri.ie
Methods: Impact-based Measures
• Diagonal elements of J contain independence values (self-impact) • Total impact a community has on others is its importance • Total impact other communities have on a community is the community’s dependence • Level of dispersion (heterogeneity) of importance/dependence of
community i can be measured as an entropy of a an i-th row/column of the impact matrix
• Is a community broadly influential or does it influence only few other communities?
J = 4 28 13
!
"#
$
%&
Digital Enterprise Research Institute www.deri.ie
Evaluation Data-Set
• 10 years of data of the largest Irish discussion board system • Segmented using 1 week sliding window
• 1 week window represents approx. 84% of cross-fora posting activity
• 448 snapshots in total • 636 communities, 73k users, 8M posts
Digital Enterprise Research Institute www.deri.ie
Clustering Fora By I. and D.
Aggregate impact matrices from the individual snapshots and cluster the communities (by k-means) embedded in the row and column spaces of the aggregate matrix.
0.3 0.4 0.5 0.6 0.7
0.0
0.5
1.0
1.5
2.0
2.5
row entropy
log(
impo
rtanc
e)
●
●
●
●
●
82133
●
●
●
●●
● ●●
●
2 4 6 8 10
0.4 0.5 0.6 0.7 0.8
01
23
4
column entropy
log(
depe
nden
ce)
●7
●
●
●●
● ●●
● ●
3 5 7 9
J1 = 1 23 3
!
"#
$
%&, J2 = 5 2
3 5
!
"#
$
%&
Jagg = J1 + J2
2= 3 2
3 4
!
"#
$
%&
Digital Enterprise Research Institute www.deri.ie
Overall I/D over Time
Take the communities with the highest importance and dependence at each week and plot them over time.
Wee
k 1
Wee
k 25
W
eek
50
Wee
k 75
W
eek
100
Wee
k 12
5 W
eek
150
Wee
k 17
5 W
eek
200
Wee
k 22
5 W
eek
250
Wee
k 27
5 W
eek
300
Wee
k 32
5 W
eek
350
Wee
k 37
5 W
eek
400
Wee
k 42
5
Soccer
Politics
IrelandOffline
Help Desk
Counter−Strike
Humour
Humanities
Computers & Tech.
Half−Life
Quake
After Hours
0 0.2 0.4 0.6 0.8Value
Color Key
Wee
k 1
Wee
k 25
W
eek
50
Wee
k 75
W
eek
100
Wee
k 12
5 W
eek
150
Wee
k 17
5 W
eek
200
Wee
k 22
5 W
eek
250
Wee
k 27
5 W
eek
300
Wee
k 32
5 W
eek
350
Wee
k 37
5 W
eek
400
Wee
k 42
5
PBANKnights of the R.T.Spell CzechsEventsThe Cuckoo's NestLubnipAsk Doctor DementoHoLLThe ThunderdomeTipp InstFNWAIThe IlluminatiDigital Art & DesignHistory & HeritagePearTree HouseLord of the RingsComeonbanusModeratorsFreemasonsNewbies & FAQFeedbackScienceHelp DeskReaverTelevisionHumourRecycle BinHumanitiesWebgamesWork & JobsLiteratureAfter HoursSportsComputers & Tech.GamesFilmsRole Playing
0 0.05 0.1 0.15 0.2 0.25 0.3Value
Color Key
Digital Enterprise Research Institute www.deri.ie
Cross-Community Infl. over Time
Count cases when community i’s impact on j was higher than j’s independence and plot the pairs with the highest counts.
Count From (i) To (j)
29 Moderators Reported Posts
22 FNWAI Poker
17 The Thunderdome After Hours
14 PI Mods Personal Issues
Digital Enterprise Research Institute www.deri.ie
Moderation of Pers. Issues
150 200 250 300 350 400 450
01
23
45
67
week
impa
ct
●
●
●
●
●
●
●
●
●
●
●
●
●
●
PI ModsModeratorsindependence
Digital Enterprise Research Institute www.deri.ie
Conclusion
• The evaluation demonstrated that the framework • is able to identify highly influential and dependent communities • can be used for efficient monitoring of the cross-community
activity, perhaps even for early alerts • can identify which communities to stimulate (e.g. by posting a
message) s.t. the stimulus spreads efficiently • We aim to extend it with content analysis
• E.g. What are the most influential communities with respect to a particular topic?
• We will also investigate empirically-observed topic cascades and modify our models accordingly if needed
• Finally, our goal is to propose a method for measuring significance of cross-community impact
• Belák V., Lam S., Hayes C. Cross-Community Influence in Discussion Fora. ICWSM 2012.
• Belák V., Lam S., Hayes C. Targeting Communities to Maximise Information Diffusion. MSND/WWW 2012.
Digital Enterprise Research Institute www.deri.ie
Fold, No, Wait, All In!
240 260 280 300 320 340
05
1015
week
impa
ct
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●●
FNWAI to PokerPoker to FNWAIPoker's indep.