Database Reference
In-Depth Information
1.0
10
0
C 0
C 2
C 1
10
-1
0.8
10
-2
0.6
10
-3
t
= 10
0
h
t
= 10
1
h
t
= 10
2
h
t
= 10
3
h
0.4
10
-4
0.2
10
-5
r
g
=1
.5
5
0.0
012345
10
-6
678910
1
10
100
1,000
Figure 15.8 The GPS trajectories of tens of thousands of cars observed for one week in
the city of Milan, Italy, and the power-law distribution of users' radius of gyration and
travel length (left); the work-home commuting patterns mined from the previous data set
by trajectory clustering and the normal distribution of travel length within each discovered
pattern (right).
subpopulations the global diversity vanishes and similar behavior emerges. The
dual scenario of global diversity (whose manifestation is the emergence of scale-
free distributions) and local regularity (within clusters, or behavioral profiles)
is perceived today as the signature of social phenomena, and seems to repre-
sent a foundational tenet of computational social sciences. Although network
science and data mining emerged from different scientific communities using
largely different tools, we need to reconcile the
macro/global
approach of the
first with the
micro/local
approach of the second within a unifying theoretical
framework, because each can benefit from the other and together they have the
potential to support realistic and accurate models for simulation and what-if
reasoning of social phenomena. This vision of convergence among computer
science, complexity science, and the social sciences is shared today by large
research initiatives, such as the FuturICT program.
2
15.4 Bibliographic Notes
Erdos and Renyi
(
1959
) is the seminal paper that introduced random graphs.
The famous small-world model was presented in
Watts and Strogatz
(
1998
), and
the first argumentations on the small-world phenomenon and the cliquishness
2
http://www.futurict.eu
