kiran garimella Gianmarco De Francisci Morales Aristides Gionis Michael Mathioudakis 1 Controversies are everywhere Twitter 3 Garimella et al WSDM 2016 Blogs 4 Instagram 5 ID: 529764
Download Presentation The PPT/PDF document "Reducing controversy by connecting oppos..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Reducing controversy by connecting opposing views
kiran garimellaGianmarco De Francisci MoralesAristides GionisMichael Mathioudakis
1Slide2
Controversies are everywhere!Slide3
Twitter
3
Garimella et al. WSDM 2016Slide4
Blogs
4Slide5
Instagram
5Slide6
US Senate VOTes
6Slide7
How can we bridge the divide?
This Paper7Slide8
Reducing controversy by connecting opposing views
Connect the two sidesTwitterModel interactions as a graphRetweet graph Nodes: users, Edges: retweetsMeasure of Polarization
How well does information flow between the two sides?Based on random walks
How Can we bridge the divide?
8Slide9
Problem: Given a graph and two sides, find the
k best edges that maximize the reduction in polarization9Slide10
Reducing controversy by connecting opposing views
Side 1
Side 2
Reducing Controversy by Connecting Opposing Views
10Slide11
Reducing controversy by connecting opposing views
GreedyLook for all pairs of nodesFind the k pairs that give the highest reduction in the polarization scoreO(n2), n: number of nodes
Algorithms
11Slide12
Reducing controversy by connecting opposing views
Side 1
Side 2
Our algorithm
The best edges are
between the
highest
degree
nodes
12Slide13
Reducing controversy by connecting opposing views
Side 1
Side 2
Our algorithm
The
best edges are
between the
highest degree nodes
13
O
(p
2
), p << nSlide14
Reducing controversy by connecting opposing views
High degree users Highly retweeted usersWe can not recommend @realDonaldTrump to retweet @BarackObamaNot practical, not likely to materialize
But it Might not materialize
14Slide15
Reducing controversy by connecting opposing views
Take into account the probability of the user accepting the recommendation Learn probabilities from the dataAcceptance probability
15Slide16
Details about the polarization score
Acceptance probability modelHow we combine polarization minimization and acceptance probabilityExperiments and findings
Visit the
posteR
…
16