Transcript Slides

CMU SCS
Mobility and Social Networks
in the cloud
Christos Faloutsos
CMU
CMU SCS
Interactions
(1) More data
Social net’s
mobility
(3) tools
More data
NSF'12
cloud
C. Faloutsos (CMU)
(2) Insights;
Research problems
2
CMU SCS
Interactions
(1) Mobility -> social networks
– More info/context (who is where, and when)
– Easier to interact with friends
NSF'12
C. Faloutsos (CMU)
3
CMU SCS
Interactions
(1) More data
Social net’s
mobility
(3) tools
More data
NSF'12
cloud
C. Faloutsos (CMU)
(2) Insights;
Research problems
4
CMU SCS
Interactions
(2) Social networks -> cloud
Properties of social networks have high
impact on algo design
– ‘six degrees’
– Power-law degree distributions
– ‘No good cuts’
NSF'12
C. Faloutsos (CMU)
5
CMU SCS
Interactions
(1) More data
Social net’s
mobility
(3) tools
More data
NSF'12
cloud
C. Faloutsos (CMU)
(2) Insights;
Research problems
6
CMU SCS
Interactions
(3) Social networks <- cloud
Conversely, mass of data needs cloud / mapreduce etc techniques to be
–
–
–
–
NSF'12
Stored (sharded / replicated etc)
Searched (show me quickly my friends)
Analyzed (link prediction)
Monetized (which ad to show to whom)
C. Faloutsos (CMU)
7
CMU SCS
Research challenges for mining
• Understand how networks form; how
humans interact
– -> sociology
– -> marketing
– -> …
• How rumors/ideas/memes spread
• Mobility patterns of humans (bio virus
prop.)
NSF'12
C. Faloutsos (CMU)
8
CMU SCS
Research challenges for mining
• Spot anomalies, i.e.
– account hijacking;
– spammers;
– behavior change (disgruntled employees)
NSF'12
C. Faloutsos (CMU)
9