Yoav Artzi Amit Levy CSE 510 HCI Spring 2010 Project final presentation Instance Based Network Representation Community detection Representing instances detection Graph representation This may or may not really happened ID: 319348
Download Presentation The PPT/PDF document "Instance Based Social Network Representa..." 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
Instance Based Social Network Representation
Yoav ArtziAmit Levy
CSE 510: HCI
Spring 2010
Project final presentationSlide2
Instance Based Network Representation
Community detection
Representing instances detection
Graph representationSlide3
*This may or may not really happenedSlide4
My CTO? Oh, maybe I shouldn’t post it yet.
*This may or may not really happenedSlide5
Challenges
How to answer the central question?Too many dimensionsRespect people’s
privacy
Only a few chances to get it rightSlide6
Approach
Ask user to associate individuals with clustersAnswering a higher level question
How do users intuitively
perceive their
network?
Evaluate if algorithm captures user’s perception of their social networkSlide7
Our Experiment
3 question types
Composed using the user’s Facebook network
User gets 10 questions of a single typeSlide8
Our ExperimentSlide9
Our ExperimentSlide10
Our ExperimentSlide11
Results
11
10
10
10
10
10
10
10
10
7
7
8Slide12
Results
43
42
43Slide13
Limitations
Selection biasCould not cover all variables
Representation algorithms left out
Binary use of affiliation and location
Limited data for analysis
For example, no friend count, only mutual friends, interactions,
etcSlide14
What We Learned
Quiz-like possibilities for privacy configurationPrivacy matters and forgotten
“… scary, it brought up photos of friends that are accidently in my Facebook… Is that the goal? To show that half of them are not really connected to you?”
“… it was really fun for me, this little game… “ Slide15
What We Learned
People are worried about privacy on FacebookThey need to see who has access to their dataAlgorithmic approaches might help communicate privacySlide16
Future Work
More of the same (gather more data)Explore more personal variables
Relating success to network properties
Size, path lengths, clustering co-efficient
Use
our
approach to communicate privacy in FacebookEvaluate “in the wild”
Explore other uses of social network clusteringHelp create groups for privacy settings