The average broadband and cellular users now spend more time in online social networks than in face-to-face interactions with people other than their immediate families.
Social networks tend to create tight-knit groups characterized by a high density of connections, and these connections are often good predictors of users’ tastes and future connections. But, finding these communities is both computationally and statistically challenging.
The Wireless Networking and Communications Group at The University of Texas at Austin’s Cockrell School of Engineering developed a new graph-clustering technique that helps with community detection, user profiling, link prediction and collaborative filtering. Learn how researchers were able to both reach globally optimal solutions with better statistical properties and provide an algorithm that easily scales.
Presented by The University of Texas
View
official listing on sxsw.com
