|
CS Home
Dept Info/Contacts
People
Research
Events
Courses
Undergrad Programs:
Computer Science
Informatics
Graduate Program
Prospective Students
Faculty Hiring
Employment
Resources
Help Lab Hours
Student Groups
Support the Department: Weeg Professorship
|
|
Social Influence and Information Diffusion in Large Online Networks
Friday, Nov 6, 2009
4:00-5:00pm, 140 SH (Schaeffer Hall)
Abstract
The widespread adoption of online social networks and social media services has
had a profound impact on the way that information is created and disseminated.
These communities provide fertile ground for research that investigates how
local interactions between individuals give rise to complex emergent phenomena.
In this talk, I will present a number of empirical findings on information
diffusion from two large-scale studies of two online communities: Second Life
and Twitter. Second Life is a massively multiplayer virtual world in which all
content is user-generated. Using 130 days of time resolved social network and
transfer data, we will examine the role of social networks in the adoption of
content. We propose a simple model based on maximum likelihood estimation that
demonstrates the effect of an individual's neighbors (friends) on the rate of
adoption of content. Adoption rates quicken as the number of friends adopting
increases, and this effect varies with the connectivity of a particular user.
In addition, we examine the role of individuals in distributing content,
showing that a few users account for a disproportionate number of adoption
events in Second Life. In our second study, which focuses on Twitter, I will
show a number of results that shed light on the role of individuals in the
reposting of URLs. We find that the behavior of few, highly connected individuals
accounts for a large proportion of subsequent URL posts. Furthermore, we show
how the rate of reposting content decreases as a function of social distance.
Finally, I will present preliminary results on the relationship between
popularity on Twitter and content type using a large set of human-generated
annotations from Amazon Mechanical Turk.
|
