Author: Miquel Camprodon <br />
Title: Analysis of Community Structure in a Young Internet Domain <br />
Time: Friday september 18th, 10am <br />
Place: Sala d'actes de la FIB, edifici B6, Campus Nord <br />
Advisors: Jordi Delgado and Ricard Gavaldà <br />
Program: Master in Computing, LSI Departament, UPC<br />
<br />
Abstract: <br />
<br />
In this work we analyse a young top level Internet domain, .cat, which opened <br />
in February 2007. We use available data from seven months, since May 2007 to <br />
March 2009, performing an study of different properties. <br />
The work starts with basic graph properties to characterise it, like <br />
number of vertices, edges, connected components, and in and out-degree <br />
distributions. We obtained that it is similar to other national web domains. <br />
Then we compare different community detection methods in order to select <br />
the bests in terms of modularity and similarity, and we study of robustness <br />
of the network. Our results showed that .cat presents some community structure. <br />
Finally we perform an in-depth study of the communities detected at several <br />
points in time, depending on words from sites. We proposed a new measure, <br />
significance, allowing us to consider only the most relevant words for each community.<br />
<br />
This thesis has been possible by a cooperation agreement with the PuntCat foundation, <br />
which administers the .cat domain. <br />
<br />
A full version of the thesis will be available soon from UPC Commons.