Using comment information available from Digg we define
a co-participation network between users. We focus on
the analysis of this implicit network, and study the behavioral
characteristics of users. Using an entropy measure,
we infer that users at Digg are not highly focused and
participate across a wide range of topics. We also use the
comment data and social network derived features to predict
the popularity of online content linked at Digg using a
classification and regression framework. We show promising
results for predicting the popularity scores even after limiting
our feature extraction to the first few hours of comment
activity that follows a Digg submission.
I am grateful to my advisor, Dr. Huzefa Rangwala, who pushed me real hard and stayed with me to get it done!
I am trying to move to wordpress and restructure my blog, but till then, I don't have any section to upload my paper. Anyways, it's available through George Mason University's Technical Reports Series for 2009, which can be located here: http://cs.gmu.edu/~tr-admin/papers/GMU-CS-TR-2009-7.pdf. Also, the paper will soon be available through IEEE.