My CMS

Matteo Magnani, Institute of Information Science and Engineering (ISTI), CNR, Italy
Luca Rossi, Dept. of Communication Studies and Humanities, University of Urbino, Italy

Tutorial summary
The last two decades have witnessed the proliferation of several Social Network Sites (SNSs).
While it is not clear whether only one or few big SNSs will survive in the near future, or multiple
specialized services will still exist separately, we can claim that a model based on a single layer
of social connections will never be able to accurately describe our complex and layered online
social experience: while Facebook connections can explain a lot about a user’s social life, his/her
professional network may require an analysis of LinkedIn connections and his/her information
consumption practices might be better explained by looking at his/her Twitter network.
Decades before the advent of SNSs a similar layered scenario had already been described by
sociologists like Goffman for which individuals perform on multiple stages, creating a sort of
fragmented public personality whose different components relate to different audiences (and thus
networks) and anthropologists like Gluckman who observed human relationships characterized
by their multiplexity. While this view has been widely used by early digital­culture researchers, it
has not been regularly applied together with Social Network Analysis (SNA) methods to study
online SNSs.
However recent works have re­defined the foundations of multi­layer network models highlighting
the opportunity to apply SNA approaches to a wide range of complex social relationships as well
as study the mutual influences between different co­existing networks.
This tutorial will review the main theoretical models, data gathering methods and analytical tools
to deal with multiple networks and to understand how a multi­layer network perspective may
change our knowledge of user behaviours. Multiple online network analysis is a recent and
growing field, with long­standing theoretical bases rooted in classical sociological analysis and
multiplex social network analysis methods. As such, it presents numerous research opportunities both for experienced researchers and young academics looking for a field of
specialization.
Prerequisites and outcomes
Participants will benefit from a general knowledge of the basic concepts in Social Network
Analysis, in particular centrality measures. However, the required concepts will be briefly recalled
as needed during the tutorial.
The main intended learning outcomes are the following:
­ Know the historical roots of multiple network analysis.
­ Recognize the specific aspects emphasized by different models for multiple networks.
­ Theorize the main socio­computational challenges of multiple network analysis.
­ Perform multiple network analysis tasks on real data.
­ Identify the most promising research directions in the area.
Detailed course description
The tutorial is divided into five main sections. We indicate the general topic of each section with a
few selected suggested readings ­ a more exhaustive list will be distributed to the participants
during the tutorial.
1) Historical foundations of Multiple Network Analysis
While the topic of multiple network analysis has recently seen a rise in general interest, largely
consequent to the new wave of interest that has been addressed to single layer networks from
many different research fields, it can be rooted into a long­standing research tradition. In order to
introduce the topic we will examine the early literature that in many different research areas (from
anthropology to sociology) considered the multiplex nature of human beings. These studies,
spanning several kinds of communication, have introduced the idea that it may not be
methodologically correct to analyze a partial network by isolating just a specific kind of tie.
Starting from these premises we will show how social sciences have often considered
multiplexity even out of the context of social network analysis.
Suggested readings:
Skvoretz J and Agneessens F (2007) Reciprocity, multiplexity, and exchange: Measures. Quality
& quantity, Springer, 41(3).
Minor MJ (1983) New directions in multiplexity analysis. Applied network analysis.
2) Models & measures for Multiple Networks
In this section we will introduce the main models and measures. We will briefly review models
allowing multiple node types (also called heterogeneous or multi­type networks), models allowing
multiple relationship types (also called multi­dimensional networks), multi­slice models and
models explicitly representing the co­existence of multiple networks (also called multi­layer(ed)
or multi­stratum networks). Then we will focus on the main measures. Here we will introduce two
different approaches, respectively reducing multiple networks to a single traditional network and
keeping the layers separate. We will define and exemplify degree and neighborhood centrality,
dimension relevance, multi­layer distance.
Suggested readings:
Berlingerio M, Coscia M and Giannotti F (2011) Finding and Characterizing Communities in
Multidimensional Networks. In: 2011 International Conference on Advances in Social Networks
Analysis and Mining, IEEE computer Society.
Magnani M and Rossi L (2011) The ML­model for multi­layer social networks. In: The 2011
International Conference on Advances in Social Network Analysis and Mining, Los Alamitos, CA,
USA, IEEE computer Society.
3) Formation & Evolution of Multiple Network
Network formation models are among the most important tools in Network Science and Social
Network Analysis. A typical application of artificially generated networks is to provide null models
that can be used to test new measures and make comparisons with real networks, so that
significant patterns can be highlighted in the real data. In addition, these models are useful to test
hypotheses on the dynamics underlying network evolution. However, most existing generative
models have been developed to describe the evolution of single networks. In this section we will
review some very recent works modelling the co­evolution of multiple networks.
Suggested readings:
B. Podobnik, D. Horvatić, M. Dickison, and H. E. Stanley (2012) Preferential Attachment in the
Interaction between Dynamically Generated Interdependent Networks, Europhys. Lett. (EPL) 100,
50004
Magnani M and Rossi L (2013) Formation of multiple networks. In: Social Computing,
Behavioral­Cultural Modeling and Prediction, Springer.
4) Clustering & Community detection in Multiple Networks
Although several community detection algorithms for single social networks exist, the discovery
of communities spanning multiple networks is still a largely unexplored topic. At the same time,
some recent works have identified new computational approaches to tackle this complex
problem. In this section we will present a selection of community detection methods for multiple
networks, highlighting the research context where they emerged and showing applications to real
data.
Suggested readings:
Mucha PJ, Richardson T, Macon K, et al. (2010) Community Structure in Time­Dependent,
Multiscale, and Multiplex Networks. Science, American Association for the Advancement of
Science, 328(5980), 876–878, Available from: http://dx.doi.org/10.1126/science.1184819.
Brigitte Boden, Stephan Günnemann, Holger Hoffmann, Thomas Seidl (2012) Mining coherent
subgraphs in multi­layer graphs with edge labels. KDD.
5) Multiple Network Data: Retrieval and Ethical issues
The collection of well structured multiple network data can be a difficult task. Within this last part
of the tutorial we are going through some of the related problems. We will also present some
available multiple network datasets. At the same time we will present some thoughts on the
practical and ethical aspects involved in multiple network data collection.
 
Biographies
Matteo Magnani graduated in Computer Science at the University of Bologna in 2002 (110/110
with mention). He studied at the University of Marne la Vallée (undergraduate level) and the
Imperial College London (postgraduate research level). In 2006 he obtained a PhD in Computer
Science (Bologna) where in 2011 he also graduated in Violin (110/110 with mention). He has
received a Rotary Prize for the best student of the Science Faculty (UniBO), and his mother is
very proud of him (or at least this is what she officially says). Until May 2012 he was an assistant
professor (RTD) at the Dept. of Computer Science, University of Bologna and he has held a
position at research assistant professor level at the Data Intensive Systems group, Dept. of
Computer Science, Aarhus University, Denmark. He is currently at KDD Lab, ISTI, CNR, Pisa
(Italy), and since August 2013 he will be Associate Professor at the Department of Computing
Science, Uppsala University, Sweden.
His main research interests span Database and Information Management systems, specifically
uncertain information management and multidimensional database queries, Network Science
and Social Computing. He has written around 1.5 Kg of papers on these topics (when printed on
heavy A4 size sheets). He has several years of teaching experience and has obtained the
Pedagogical Training Certificate at Aarhus University.
 
Luca Rossi is Assistant Professor of Media Analysis at the Department of Communication
Studies and Humanities, University of Urbino Carlo Bo, Italy. He works on SNA techniques
applied to Social Media data and to the analysis of audience practices. He presented his work in
many international conferences, among others: IR, SBP, ASONAM, SunBelt, ICWSM. He has
teaching experience both at undergraduate level where he teaches Sociology of New Media and
Media Analysis and at the graduate level where he teaches Social Network Analysis techniques
as a compulsory class of the PhD program in Sociology of Communication at the University of
Urbino. Since August 2013 he will be at IT University in Copenhagen, Denmark.
Matteo and Luca have a growing experiences in the field. In 2011 they won the Best Paper
Award at the ASONAM conference for their seminal paper “The ML­model for multi­layer social
network analysis” where they defined concepts and methods to study multi­layer online social
networks. In 2012 they organized the first International Workshop on Complex Social Network
Analysis, they are organizing the Symposium on Multiple Network Analysis and Mining (satellite
event of NetSci 2013) and they have authored the entry on Data Structures and methods for
mining multiple social networks for the upcoming Encyclopedia of Social Network Analysis and
Mining (Springer). Putting together two different backgrounds (respectively, computational and
sociological) they will also be able to provide insights on opportunities and challenges of doing
interdisciplinary research on these topics. Together, they have successfully attracted funding
from Working Capital (Telecom Italia), PRIN and FIRB (MIUR ­ Italian Ministry for education,
University and Research) schemes.
Contact information
Matteo Magnani: ISTI, CNR, Via Moruzzi 1, Pisa, IT. email matteo.magnani@isti.cnr.it, phone +39
333 3833579.
Luca Rossi: Dept. of Communication Studies and Humanities, Via Saffi 15, 61029 Urbino, IT.
email luca.rossi@uniurb.it phone +39 0722 305726