Changes between Version 13 and Version 14 of JINO-2

Timestamp:

Jan 10, 2013, 4:20:38 PM (5 years ago)

Author:

nicolas.dugue@…

Comment:

--

JINO-2

@@ -31 +31 @@
 === Data Analysis of Cloud and P2P Applications ===
 
-* 16h15-17h00 Raphaël Fournier-S'niehotta (LIP6, Paris) '''Detection and analysis of pedophile activity in P2P networks'''
+* 16h15-17h00 Raphaël Fournier-S'niehotta (LIP6, Paris) '''Detection and analysis of paedophile activity in P2P networks'''[[BR]]After collecting very large datasets of search engines queries on P2P networks, we study paedophile activity. We first design and validate a paedophile query detection tool. We then use it to estimate the fraction of paedophile queries and users. We study the evolution of paedophile activity, which significantly improved between 2009 and 2012. We also shed light on the social integration of paedophile users by observing the privileged moments to issue such queries.
 
 * 17h00-17h45 Nicolas Dugué (LIFO, Université d'Orléans), Anthony Perez (LIFO, Université d'Orléans) '''Detecting social capitalists on Twitter using similarity measures'''[[BR]]Social networks such as Twitter or Facebook are part of the phenomenon called Big Data, a term used to describe very large and complex data sets. To represent these networks, the connections between users can be easily represented using (directed) graphs. In this presentation, we are mainly focused on two different aspects of social network analysis. First, our goal is to find an efficient and high-level way to store and process a social network graph, using reasonable computing resources (processor and memory). We believe that this is an important research interest, since it provides a more democratic method to deal with large graphs. Next, we turn our attention to the study of social capitalists, a specific kind of users on Twitter. Roughly speaking, such users try to gain visibility by following other users regardless of their contents. Using two similarity measures called overlap index and ratio, we show that such users may be detected and classified very efficiently.