ing causality and replication. By contrast, experimental studies offer high internal validity, discovery
of causal relationships, and ease of replication, but
the rigid control over the settings and interactions of
an experiment can limit generalizability. New methods work toward bridging these two research contexts, either by bringing the lab to the field to recruit
more diverse participants in a natural setting or by
simulating natural settings and interactions of the
field in the lab. This workshop showcased more naturalistic experimental paradigms, innovative tools
and methods, and challenges in conducting research
to optimize both internal and ecological validity.
The workshop was organized by Dominic DiFran-zo, Natalie Bazarova, Shyam Sundar, and Jeff Hancock. No report was submitted to AI Magazine.
Emoji Understanding and
Applications in Social Media
The Emoji Understanding and Applications in Social
Media workshop brought computer and social sci-
ence researchers and industry practitioners together
to discuss and exchange ideas on understanding the
social, cultural, communicative, and linguistic roles
of emoji, while leading discussions on building nov-
el computational methods to understand them.
Emoji have become a popular way to enhance electronic communication. They play distinct social and
communicative roles compared to other forms of
written language. The ability to automatically
process, derive meaning, and interpret text infused
with emoji is essential as society embraces emoji as a
standard form of online communication. Yet a broad
variety of reasons make this a challenging task for
traditional natural language processing techniques.
Key reasons include the pictorial nature of emoji; the
fact that (the same) emoji can be used in different
contexts to express different meanings; the broad
variety of contexts in which a sender may choose to
use an emoji (such as the sociocultural aspects and
mutual relationships between the sender and the
receiver); and the fact that emoji are used and interpreted differently across different cultures and communities around the world. The goal of this workshop was to stimulate research on understanding
these challenges in emoji use and on developing
novel approaches to analyze, interpret, and understand them.
Tyler Schnoebelen (principal product manager,
Integrate.ai) gave the opening keynote speech, “Emo-
ji Are Great and/or They Will Destroy the World.” He
discussed the connections of emoji to particular
styles and ideologies. Specifically, he compared the
way that celebrities, journalists (from right, left, cen-
ter), and different genders use emoji, while bridging
social theories with computational linguistics. San-
jaya Wijeratne, Amit Sheth, and Horacio Saggion pre-
sented the tutorial “Improving Emoji Understanding
Tasks Using EmojiNet,” where Wijeratne discussed
the way in which knowledge bases of emoji mean-
ings can be used to solve problems of emoji similari-
ty and emoji sense disambiguation.
The eight research papers presented at the workshop covered a wide variety of topics ranging from
building and improving emoji lexicons to receiver
interpretation of emoji. Several papers discussed
emoji research problems that are of interest to the
computer science researchers, such as applying temporal variability of emoji usage patterns for improving emoji prediction, learning emoji embedding
models via emoji co-occurrence networks, and building emoji lexicons. Other papers were focused on linguistics and social science research problems, such as
examining the resemblance of repetitions in emoji
use to beat gestures, how emoji are used to express
solidarity in social media messages during crisis
events, and how receiver interpretation of emoji
changes across different genders.
A highly interdisciplinary panel, “The Challenges
in Emoji Understanding,” provided an animated and
engaging forum to the attendees to discuss the open
emoji research problems with leading researchers and
practitioners. The panel consisted of Jennifer Lee
(vice chair, Unicode Emoji Subcommittee), Keith
Winstein (assistant professor, Stanford University),
Eric Goldman (professor, Santa Clara University),
Rachael Tatman (data scientist, Kaggle), and
Francesco Barbieri (postdoctoral researcher, Universi-tat Pompeu Fabra, Spain). The panel discussion topics included representing emoji using Unicode code-points, submitting new emoji proposals, challenges
in using emoji as a new language, implications of
emoji interpretation in the law, and building new
computer algorithms to understand emoji.
Sanjaya Wijeratne, Emre Kiciman, Horacio Saggion, and Amit Sheth served as cochairs of the workshop. The accepted papers are published under CEUR
Workshop Proceedings Volume 2130. The workshop
program is covered in a WIRED article by Arielle
Pardes.
The workshop was organized by Sanjaya Wijeratne, Emre Kiciman, Horacio Saggion, and Amit
Sheth. This report was written by Sanjaya Wijeratne
and Amit Sheth.
Event Analytics Using
Social Media Data
Social media channels enjoy many advantages over
traditional media channels, such as ubiquity, mobil-
ity, immediacy, and the seamless communication in
reporting, covering and sharing real-world events
such as the Boston bombings, the NBA finals, and the
US presidential elections. Given these advantages,
social media posts on Twitter, Facebook, Instagram,
WhatsApp and the like can typically reflect events as
they happen, in real time. Despite these benefits,
