by focusing on the applications and implications of
web and social media data for demographic research.
The workshop brought together researchers from
different disciplines studying digital data for different purposes. Workshop participants interacted with
each other through a combination of presentations
and interactive activities.
This workshop was organized by the Max Planck
Institute for Demographic Research in collaboration
with the scientific panel on digital demography of
the International Union for the Scientific Study of
Population. It was the fourth workshop held at the
International Conference on Web and Social Media
during the last 4 years revolving around the theme
of demography and social media. The continuous
presence of demographers at the International Conference on Web and Social Media is a sign of the importance of social media studies for the demographic
community as well as the commitment of demographers to advance research on web and social media
and their societal implications.
Presentations were either short ( 10 minutes for
presenting plus 5 minutes of discussion) or long ( 15
minutes for presenting plus 5 minutes for discussion).
This maximized the exposure to a range of topics in
digital trace research. It also increases the exchange
of ideas between different research fields. Keynote
speeches were delivered by Yelena Mejova (research
leader at the Institute for Scientific Interchange in
Turin, Italy) and Maria Sironi (associate professor in
the Department of Social Science at University College London).
The workshop was also interactive and fun, thanks
to a series of group activities. These included speed
dating for ideas (an ice-breaker); online data access, privacy, and use (a series of minipanels); name-dropping
(paper titles for potential collaborations); and working with your academic match (a closing activity).
An online evaluation conducted after the workshop showed that participants found the event enjoyable and helpful for networking. In particular,
they appreciated the interactive activities but would
have preferred fewer presentations. These recommendations will be considered for a future edition of the
Sofia Gil-Clavel, Diego Alburez-Gutierrez, and Emilio
Zagheni were the co-chairs of the workshop.
Driven Complex Networks
The growing availability of multifaceted social media
data gives rise to unprecedented opportunities for
unveiling complex real-world online behaviors. This
also supports the proliferation of complex network
models where the expressive power of the graph-based
relational structure is enhanced through exposing
several types of features that are peculiar to the social
The aim of the Modeling and Mining Social-Media-Driven Complex Networks workshop was to get
an insight on current trends regarding both modeling and mining aspects concerning any type of data-driven network that can be inferred from social media
data contexts, such as heterogeneous, multilayer, temporal, location-aware, and probabilistic networks.
The workshop consisted of four sessions spanning
a full day, and included nine presentations of accepted papers and an invited talk. During this time,
researchers and practitioners discussed the relation
between social media and complex networks from
several points of view, including both data-driven
modeling (for example, modeling of location-based,
mobility-based, and topic-based networks) and tasks
focusing on the analysis and manipulation of network structures (for example, network anonymization,
link prediction, influence, and information spread).
Several discussions brought up the possibility of
inferring the topology of a network from features
other than the presence of explicit relations between
entities, for example, geolocation of tweets, mobility of users, and topical information. This indicated
that while in most cases it may be intuitive to represent social media data by the use of features that natively define a topology among entities (for example,
friendships, interactions, ratings, and so on), in several contexts it may be more fruitful to remodel information network data on data-driven features that
characterize latent relations between entities tailored
to a given problem.
Another recurring discussion was over how to conveniently exploit textual content issued by social
media. It has been shown how textual content can
be exploited to define topic-driven interactions and
communities that can be used to characterize a specific phenomenon (for example, the political scenario
of a country), but also how graph structures contained in text flows (such as chat logs) can be leveraged to characterize the text itself (for example, by
identifying abusive language).
Misinformation spread, a major research topic in
recent years, has continued to be an interesting research avenue in social media analysis. This is not
surprising, given the impact that this phenomenon
has had in major events like political elections.
In recognition of this topic, the workshop hosted
an invited talk by Robert West (Ecole Polytechnique
Fédérale de Lausanne) on Message Distortion in Information Cascades. During his talk West discussed
how information diffusion processes based on iterative summarization can strongly alter the original
content of a message (for example, a scientific article),
even in the absence of a malicious intent.
The workshop showed how such tools are extremely
flexible and key-enabling for modeling and analyzing
phenomena in different domains. At the same time,
the discussions brought to light how coping with
social-media-driven complex networks can be challenging, due to the high heterogeneity of the data
and the absence of a standard way to process them.