logo SBA

ETD

Archivio digitale delle tesi discusse presso l’Università di Pisa

Tesi etd-05232012-181321


Tipo di tesi
Tesi di dottorato di ricerca
Autore
COSCIA, MICHELE
URN
etd-05232012-181321
Titolo
Multidimensional Network analysis
Settore scientifico disciplinare
INF/01
Corso di studi
INFORMATICA
Relatori
tutor Prof. Pedreschi, Dino
correlatore Giannotti, Fosca
Parole chiave
  • community discovery
  • complex networks
  • data mining
  • multidimensional analysis
Data inizio appello
25/06/2012
Consultabilità
Completa
Riassunto
This thesis is focused on the study of multidimensional networks. A multidimensional network is a network in which among the nodes there may be multiple different qualitative and quantitative relations. Traditionally, complex network analysis has focused on networks with only one kind of relation. Even with this constraint, monodimensional networks posed many analytic challenges, being representations of ubiquitous complex systems in nature. However, it is a matter of common experience that the constraint of considering only one single relation at a time limits the set of real world phenomena that can be represented with complex networks. When multiple different relations act at the same time, traditional complex network analysis cannot provide suitable analytic tools. To provide the suitable tools for this scenario is exactly the aim of this thesis: the creation and study of a Multidimensional Network Analysis, to extend the toolbox of complex network analysis and grasp the complexity of real world phenomena. The urgency and need for a multidimensional network analysis is here presented, along with an empirical proof of the ubiquity of this multifaceted reality in different complex networks, and some related works that in the last two years were proposed in this novel setting, yet to be systematically defined. Then, we tackle the foundations of the multidimensional setting at different levels, both by looking at the basic extensions of the known model and by developing novel algorithms and frameworks for well-understood and useful problems, such as community discovery (our main case study), temporal analysis, link prediction and more. We conclude this thesis with two real world scenarios: a monodimensional study of international trade, that may be improved with our proposed multidimensional analysis; and the analysis of literature and bibliography in the field of classical archaeology, used to show how natural and useful the choice of a multidimensional network analysis strategy is in a problem traditionally tackled with different techniques.
File