[Forum SIS] Seminario: Samuele Stivanello, Ecology of human activities
Antonio Pievatolo
antonio.pievatolo a mi.imati.cnr.it
Mar 11 Giu 2019 19:02:08 CEST
Lunedi' 17 giugno alle 14:45, presso il CNR-IMATI, in via A. Corti, 12,
Milano, Aula Pentagonale edificio Bassini II piano, si terrą il seguente
seminario:
Samuele Stivanello, Dipartimento di Matematica, Universita' di Padova
Title: Ecology of human activities
Abstract: One of the most studied macro-ecological pattern is the
Relative Species Abundance (RSA), describing the distribution of the
population among species. Various natural ecosystems share a common
shape of the empirical RSA and through years ecologists have used
several recipes - some of them are rule of thumb prescriptions - to
determine global RSA features from local information. Recently our group
has developed a rigorous statistical framework able to predict global
scale biodiversity from scattered local plots ([2], [3]). Such a
framework is derived from a simple mathematical model informed by basic
biological assumptions. In this talk we illustrate how to
adapt/generalize such an approach to human activities. In particular we
consider four datasets of man-made systems and/or network mediated human
activities: e-mail communication, Twitter posts, Wikipedia articles and
Gutenberg books. Once we set the proper correspondence to what we
consider species and individuals of a species, our approach reveals RSA
is scale-free in each mentioned dataset with a power law form maintained
- with roughly the same exponent - through the different human
activities considered. In turn, RSA scale-invariance allows us to use
activity information on local scale to predict hidden features of the
human dynamics at the global scale. E-mail communication is particular
relevant to illustrate our method as well as our results and possible
applications. We consider the senders’ activity network where each node
is a sender and a directed link from node A to node B represents an
email issued from user A to user B. We set the identity of a sender to
identify the species and the number of sent emails to be the individuals
pertaining to a species. Thus, for instance, if user A has sent n emails
we say that the species A has n individuals. Now, suppose an observer
has access to a small sample of sent emails, or equivalently to partial
information on links and nodes of the email communication network. Our
method is capable to infer the number of nodes (i.e. the number of
users) and the degree distribution (i.e. the topology) of the whole
network, thus revealing features of the dynamics unknown to the
observer. Moreover, our framework predicts how the number of users grows
with the number of links recorded, which represents another well known
pattern in ecology called the Species-Accumulation Curve (SAC). These
findings may have applications in resource management, optimal network
design and information diffusion on graphs. This is a work in
collaboration with M. Formentin, A. Tovo and S. Favaro [1].
[1] M. Formentin, A. Tovo, S. Stivanello, S. Favaro. Ecology of human
activities. In preparation, 2019+.
[2] A. Tovo, S. Suweis, M. Formentin, M. Favretti, I. Volkov, J. R.
Banavar, S. Azaele, A. Maritan. Upscaling species richness and
abundances in tropical forests. Science Advances 3 (10) (2017) e1701438.
[3] A. Tovo, M. Formentin, S. Suweis, S. Stivanello, S. Azaele. A.
Maritan. Inferring macro-ecological patterns from local species’
occurrences. Methods in Ecology and Evolution, submitted, 2019.
--
Dr Antonio Pievatolo
IMATI-CNR
http://www.imati.cnr.it/joomla/index.php/people?layout=edit&id=101
ph. +39 02 23699 520
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.stat.unipg.it/pipermail/sis/attachments/20190611/3d26a2df/attachment-0001.html>
Maggiori informazioni sulla lista
Sis