User behavior modeling of content consumption and generation in online social networks
Online Social Networks (OSNs) are based over computer systems that should perform under user generated load. It is therefore important to understand how the user behaves in OSNs to evaluate the generated load. However there are various kinds of user behaviors in such services and we are lack of comprehensive studies of various user behaviors.
In this thesis we study the user behaviors in OSNs from three point of views, (a) the online contents consumption behavior, (b) the favoring behavior, and (c) the online content production behavior.
First we study the online contents consumption behavior in respect to the popularity of online contents. We model the popularity of online contents with explanatory factors by using the Cox proportional hazard regression model and predict their popularity. Second we investigate the user favoring behaviors, i.e., how users favor other users' contents and how they reciprocally favor their contents. Third, we look into how online users generate their contents in OSNs and specifically we scrutinize their content generation behaviors and model them with the stretched exponential distribution.
Defence : 03/25/2011 - 14h - Site Jussieu 25-26/105 Jury members : Chadi Barakat, INRIA, [Raporteur]
Sue Moon, KAIST, [Raporteur]
Mohamed Ali Kaafar, INRIA
Krishna Gummadi, MPI
Mr. Serge Fdida, UPMC
Mr. Kavé Salamatian, Université de Savoie