Attending conferences is one of the most rewarding experiences.
Conferences provide an excellent opportunity to meet researchers, exchange
ideas, experience good food and explore best of the cities in the world.
I had one such experience recently, when I visited Cologne to present our
paper at ICWSM.
Conference Introduction
International Conference on Web and Social Media (ICWSM) is a premier venue
that attracts computer scientists, social scientists, psychologists and
economists from around the globe. These researchers gather to discuss ideas,
share knowledge and exchange information with a common theme of online social
media.
Like every other year, this year too had an intriguing lineup of keynotes,
papers, posters and workshops.
Keynotes
Each morning of the conference began with a keynote. Each keynote speaker
talked about their recent research ventures for about an hour, and these
addresses were clearly the highlight of the conference.
1.
Dr. Suresh Venkatasubramanian talked about Algorithmic Fairness.
He argued that we humans are susceptible to bad judgements. For instance,
it was noted that judges often convict more right before lunch, and pardon
more right after lunch! Algorithms are mathematical constructs, and they
are expected to be fair, but they often pick up the biases present in the
data, and end up being unfair. For example, a Google search on 'beauty',
ends up with a bunch of white women. He talked about the need and possible
ways of detecting such unfairness in algorithms, and how we can formulate
algorithms that are fair.
2.
Dr. Amir Goldberg talked about decoding culture from digital traces.
He presented insights from emails of a typical organization, where he
measured the cultural fitness from the email text, and successfully
predicted promotion, dismissal and voluntary exits of employees.
3.
Dr. Lise Getoor talked about collective reasoning for richly structured socio-behavioral data. She described in detail a
highly scalable open
source probabilistic programming language which could help in tasks
like group classification, link prediction and entity resolution.
Workshops
ICWSM hosted many workshops that were held a day before the main conference.
Of the various tracks, I attended the
Wiki workshop that dealt with the
research carried out on Wikipedia.
In particular, I liked the invited addresses of
Dr. Claudia Wagner and
Dr. Fabian Suchanek. Claudia talked about gender inequalities in Wikipedia.
She highlighted that it was far more difficult for women to enter
Wikipedia, as of all the men and women celebrities on Wikipedia, women were
more notable. Fabian gave a very interesting overview - 'A Hitchhiker's
Guide to Ontology'. He is the lead researcher behind the well known
Knowledge Base
YAGO. We too, as a lab, work on problems related to Knowledge
Bases, like
NELL. Hence, hearing about the journey of YAGO from its founders
was one of the most capturing moments of the conference.
Interesting Papers
ICWSM had a bunch of very interesting
papers, following are my favourites:
1.
Predictability of Popularity: Gaps between Prediction and Understanding
Authors : Benjamin Shulman, Amit Sharma, Dan Cosley
The paper looked at the problem of estimating the popularity of a
post/tweet/song, both at the time of posting, and a few minutes after posting.
Coherent to previous work, they found it hard to predict right at the time of
posting, whereas, if they allowed the post to stay for a short duration, they
could quite successfully predict the popularity. However, the initial popularity
(i.e the number of likes/retweets/upvotes received in the initial few minutes)
dominates the entire feature set. This effect is observed across many different
platforms independently. They underscored the fact the predicting is not same
as understanding, as one could very successfully predict the popularity after
observing the initial few trends, but, this gives us no insight about why a
post would become popular.
2.
Who Did What: Editor Role Identification in Wikipedia [Honorable Mention]
Authors : Diyi Yang, Aaron Halfaker, Robert Kraut, Eduard Hovy
The paper deciphers the hidden roles that editors play in Wikipedia,
some contributors might be fact checkers, whereas others might be good at
referencing relevant Wikipedia pages (Wikipedians!). Another few might take up
the task of appropriate markup usage, and of course, some would be significant
content writers. Clearly, a wide variety of roles are played, and identifying
such roles not just help us better understand Wikipedia contributions, but they
also allow for better laser targeted recommendations. (for instance, page on
Indian PM Narendra Modi needs a Wikipedian).
3.
"Blissfully Happy" or "Ready to Fight": Varying Interpretations of Emoji
Authors : Hannah Miller, Jacob Thebault-Spieker, Shuo Chang, Isaac Johnson,
Loren Terveen, Brent Hecht
We use emoticons in everyday life, and they are a quick way to express feelings
and emotions. Even though they are widely used across different platforms, they
face a lot of issues, because, a given Emoji is rendered differently in
different platforms and devices. This provides room for misinterpretation.
The paper is full of interesting results about how different Emojis are
misconstrued as noted from a user study.
Our Paper
Fitting the ICWSM theme, our work revolved around understanding what (and why)
questions remain unanswered on social question answering websites.
We inquired factors that can prevent a question from getting answered,
or conversely, the factors that make any given questions a great question.
For more details, refer to the
paper, or a
blog about the same.
Other Fun Moments
As it was my first international trip, I was super excited about the
conference. I roamed around Cologne quite a bit, and during the adjoining
weekend, I went on to Paris, with a couple of my school friends
who work in Europe. It was a great time catching up with them, and
witnessing the beautiful city of Paris.
Coming from India, Europe was different in more ways than one.
I was amazed multiple times, on various accounts. Following are, in no
particular order, some of highlights of my non-conference adventures.
- Cologne is sparsely populated! And it adds to the peace and general
beauty of the city
- To my surprise, tap water in Europe, is pure enough and one can consume
it directly
- The World's second largest Cathedral is located in Cologne
- The opening reception of the conference was hosted in a Chocolate
Museum!
- Crepes are very popular in Paris, they are delicious - a must eat!
- Eiffel Tower sparkles every hour during night hours, making it all the
more spectacular
- Museums are a commonplace in Europe, I went to Ludwig Museum in Cologne
and Louvre in Paris. Yes, Louvre has the celebrated painting of Mona Lisa.
- Emirates has wonderful air-shows, that lets a passenger see through front
and rear cameras of the aircraft. I was always curious about how the runway
appeared to the pilot, right before landing
Acknowledgments
It was a fun filled trip, and it wouldn't have been possible without the
help of many individuals and organizations. Firstly, I would like to thank
Dr. Partha Talukdar, for his time and advice throughout the project and
the trip. This was joint work with Yogesh Dahiya, another member from MALL
lab, whose ideas and contributions were indispensable. Lastly, many
thanks to Google Research, Accenture and AAAI organization for sponsoring
the research and the visit.
