Although considerable amount of work has been conducted recently of how to predict links between users in online social media, studies exploiting different kinds of knowledge sources for the link pre- diction problem are rare. In this paper latest results of a project are presented that studies the extent to which interactions – in our case di- rected and bi-directed message communication – between users in online social networks can be predicted by looking at features obtained from social network and position data. To that end, we conducted two experi- ments in the virtual world of Second Life. As our results reveal, position data features are a great source to predict interacts between users in on- line social networks and outperform social network features significantly. However, if we try to predict reciprocal message communication between users, social network features seem to be superior.
More than Just Lines on a Map: Best Practices for U.S Bike Routes
Predicting Interactions In Online Social Networks: An Experiment in Second Life; MSM Workshop; Hypertext 2013, Paris
1. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions in !
Online Social Networks!
!
An Experiment in Second Life!
Michael Steurer and Christoph Trattner!
Graz University of Technology!
Austria!
2. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Motivation!
! Predicting Communication!
! Interactions between users !
! Reciprocity of interactions!
! No Research for Combined Dataset!
!
! Used Datasets!
! Online Social Network!
! Location-based Social Network!
3. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Second Life!
http://notizen.typepad.com/aus_der_provinz/sl061018_001_1.jpg!
4. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Event Data!
5. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Position Data!
http://notizen.typepad.com/aus_der_provinz/sl061018_001_1.jpg!
6. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Online Social Data!
7. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Collecting Data!
! Event Data!
! 10 months starting in March 2012!
! 218,245 unique events !
! Location-based Social Data!
! 13 Million data samples!
! 190,160 unique users !
! Online Social Data!
! 135,181 users with interactions on their “Wall” !
8. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Networks!
! Created Networks!
! Online Social Network!
! Location-based Social Network!
! Topological Features!
! Common Neighbors, Adamic Adar, …!
! Homophilic Features!
! Groups, Regions, Distance, …!
9. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Experimental Setup!
! Merged Networks + Features!
! Binary Classification Problem!
!
! Predict Interactions!
! Co-Occurrence!
! Interaction – No Interaction!
! Predict Reciprocity!
! Co-Occurrence!
! Reciprocal – Not Reciprocal!
10. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Feature Differences!
! Predict Interactions!
! Significant differences!
! Common Groups – 1.9 vs. 0.4!
! Distance – 10.3m vs. 38.1m!
! Predict Reciprocity!
! Topological features significantly different!
! Common Groups – 2.0 vs. 1.81!
! Distance – 9.35m vs. 11.19m!
11. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
!
! Verified Stability – SVM, Random Forest!
Prediction Model!LogisticRegression!
Feature Set! Interactions! Reciprocity!
Online Social Network! 0.863 AUC! 0.679 AUC!
Location-based Social Network! 0.919 AUC! 0.551 AUC!
All Features! 0.953 AUC! 0.709 AUC!
12. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions In Online Social Networks!
Conclusion!
! Predicting Interactions or Reciprocity!
! Location-based social network features!
! Online social network features!
! Best Features!
! Average distance between users, days seen!
! Common Neighbors!
! Future Work!
! Predicting Partnership!
! Use Time Information!
13. Information Systems and Computer Media - Graz University of Technology!
Predicting Interactions in !
Online Social Networks!
!
An Experiment in Second Life!
Michael Steurer and Christoph Trattner!
Graz University of Technology!
Austria!