Biology Reference
In-Depth Information
frequently incomplete and/or limited to a specific context.
A multitude of heuristic models for population structure
and mobility (e.g., gravity laws of various types) have been
endorsed, yet they frequently generate conflicting results
and all suffer from the shortcomings of being connected to
the arbitrariness of administrative boundaries, lacking
microscopic knowledge relating the individual dynamics to
population interactions, and lacking network and spatial
proximity correlations. Put simply, despite the large effort
and relevant advances in the study of human transport, no
general framework of human interactions and mobility
based on microdynamic principles and able to bridge all
spatial scales exists. The new era of the social web and the
data deluge is, however, raising the limits scientists have
long been struggling with. From the development of new
pervasive technologies to the use of proxy data, such as the
digital traces that individuals leave with their mobile
devices, it is nowadays possible to characterize the network
of human interactions from the scale of the single indi-
vidual to the level of entire populations.
The first steps towards a cohesive framework for the
collection of detailed data on human behavior and mobility
started by passively tracking human interactions through
the use of cell phones
[24]
. Each cell phone was equipped
with custom software that would regularly check which cell
phone towers and Bluetooth devices were within range.
Since Bluetooth has a range that is limited to just a few
meters, they were able to track the location of each user
with a hitherto unprecedented resolution up to room level.
This pioneering study introduced the concept of 'reality
mining', the idea that technology had finally reached
a point where it was not only feasible but also easy to
unobtrusively track individual behavior and social inter-
actions in real time. What followed was an avalanche of
works extending this approach in various directions, both
by using pre-existing data in innovative ways and by
creating new tools and techniques to obtain data that had
never been available before. Similar experiments have been
used recently to generate predictive power on influenza
spreading in communities by mapping the social pattern of
individuals along with their health status
[25]
.
Analogous experiments are also performed with
progressively smaller, lightweight, active Radio-
Frequency Identification (RFID) tags
[26]
. Each tag emits
a low-power directed radio signal with a range of up to 1
m. Whenever two tags are able to exchange signals for an
extended period of time, this is a clear sign that the people
wearing them are facing each other and in close prox-
imity, providing a good indication that a face-to-face
conversation is taking place. The authors distributed such
tags to hundreds of volunteers in various settings, and for
the first time were able to track the way in which face-to-
face interactions occur in real-world settings ranging from
school and hospitals to conferences and workplaces.
In a pioneering work, Dirk Brockmann and co-workers
have shown that popular sites for currency tracking (euro-
billtracker, wheresgeorge) can be used to gather a massive
number of records on money dispersal, and to use those as
a proxy on humans to collect a wealth of novel and
unprecedented data. This work opens a novel path to the
general exploitation of proxy data for human interaction
and mobility based on the evidence that humans leave
abundant traces of their interactions and mobility patterns
within various types of data-driven websites. The pervasive
use of mobile and Wi-Fi technologies in our daily life is
also changing the way we can measure human mobility.
Modern mobile phones and PDAs combine sophisticate
technologies such as Bluetooth, GPS, and WiFi, constantly
producing detailed traces of our daily activities. The recent
study of Gonzalez and co-workers on human mobility
based on mobile phone data to track the movements of
100 000 people over 6 months is just the most explosive
example of how these kinds of data are going to shatter our
methodology in the field and critically revise our knowl-
edge of social dynamics.
One of the main challenges offered by these networks
lies in their complexity and multi-scale nature. As a large
body of work spurred by first paper on complex networks
has shown, most real-world networks present dynamical
self-organization and the lack of characteristic scales, main
hallmarks for complex systems. The various statistical
distributions characterizing these networks are generally
heavy-tailed, skewed and varying over several orders of
magnitude. This is not just true for the degree distribution
P(k) characterizing the probability that each node in the
system is connected to k neighboring nodes, it is observed
for the intensity carried by the connecting links, transport
flows and other basic quantities.
Analogously, similar heterogeneities are found at much
larger resolution scale. At the urban scale of a single city, an
impressive characterization of the human interactions flows
is represented by the TRANSIM study
[17]
. This study
focused on the network of locations in the city of Portland,
Oregon, including homes, offices, shops and recreational
areas. The temporal links between locations represent the
flow of individuals going from one place to another at
a given time. The resulting network is characterized by
broad distributions of the degrees and of the flows of
individuals traveling on a given connection
[17]
. Strong
heterogeneities are thus present not only at the topological
level, but also at the level of the traffic on the network:
a simultaneous characterization of the system in terms both
of topology and weights associated to connections is
needed to integrate the different levels of complexity in
a unifying picture
[27]
.
Similar results have been found in commuting patterns
among cities and counties within a given geographical
region/country. In this case,
the nodes of the network
