IC2S2 Wrapup.

Ian Mulvany
Jul 20, 2017 · 5 min read

So, last week I attended the third international conference on computational social science. This is an area of research that is now at the heart of the role that I am doing at SAGE — building tools to support this emerging field. The conference was in Köln in Germany, and there were probably between 300 and 400 attendees.

I wrote a bunch of posts about this conference, and you can look over them here:

The Keynotes were all videoed, and the 2018 conference has already been announced!.

I wanted to sketch out my impressions coming away from the meeting.

For a small venue, there were a lot of talks. There were about five or six parallel sessions, and I only managed to get to a small number of the parallel sessions. The talks were given 15 minutes each, and the organisation and timekeeping was excellent, but the rooms were scattered somewhat around the venue, so popping in from one parallel session to another didn’t feel like something that was going to be easy to do. As a result I had a huge feeling of missing out. I really enjoyed most of the talks that I listened to, and just imagine that I would have gotten a lot more were I to have been able to see all the talks. The solution has to be to figure out a way to video all of the tracks in the future, and not just the keynotes.

This was a wonderfully interdisciplinary conference, I had conversations with people with the following kinds of backgrounds: AI, computer science, physics, sociology, demographics, geography, bibliometrics, education, economics. I think this is a discipline in that is making interdisciplinary research actually work.

That said, myself and my colleagues noticed a trend for some papers that were presented to be naive in their methods, especially in some cases where people with a CS background were tackling topics from areas where there have been decades of work on methodology (linguistics spring to mind). That speaks to a need for further and deeper dialog between these different co-mingling backgrounds.

Many of the speakers were relying on traditional methods (surveys) to flesh out their insights, in other cases researchers were taking the survey to new scales by deploying onto large platforms like Wikipedia and Facebook. This tells me that the field is as much about transformation as it is about revolution, in methodologies.

The wide array of disciplines present were reflected well in the keynotes, with some talks looking at real systems that had been observed, others looking at foundational aspects of the mathematics of the tools being used, others looking at how to use data at scale and new computational methods to test underlying theory. I felt that the quality of all of the keynotes was high. A few of them were outstanding.

One interesting question I had was with a researcher who asked whether we had created any new foundational social knowledge through the interface with computational methods and data. He felt that the field has been making the claim that a kind of new way of doing theory of social science would emerge and he has not seen that happen. He agreed that these tools provide more power to social researchers. I think this is a very interesting question. Internally at SAGE we have liked the analogy of data at scale acting as akin to the creation of a new instrument of investigation for the social sciences. The telescope and microscope by themselves didn’t create new theories, but did lead to their discovery in the physical sciences. The core theories in the social sciences have solidly emerged over the last century. Initially perhaps we may find that this new discipline will help us to make stronger knowledge claims which depend on merging both theory and grounding that with evidence and data. My favourite talks in the conference had that combination of data and theory. Will that lead to new types of theory? I just don’t know, but there is clearly a lot of work that can be tackled in ways that would have been previously impossible to do.

We have been thinking about what we can do to help social scientists get equipped with technical skills. I head a number of people at the conference asking how, as computer scientists, they could get a grounding in social theories, to help them communicate better with their new colleagues, and also to help them start to understand their own results better. This interface between theory and practice is so exciting.

So my favourite talks were the following:

Dashun Wang talked about producing success in scientific careers and in product launches. The work on scientific careers was interesting as it shows that there is a neat way to predict future citations to papers. As I’ve worked in this space for so long this was naturally interesting to me. They key message from this talk is that there is hope, your best days in your career may still be in front of you!

Kathleen Carley gave a terrifying talk on how bot networks manipulate twitter to end up changing human behaviour. We all know about fake news, but seeing some of the fine grained mechanisms laid bare left me worried. It was a great talk, but had a touch of the dystopian about it.

Jeff Hancock gave a beautiful talk on how we create models of these digital systems that we use, and how those models lead to our emotional reaction to those systems. I was astonished that most people who use Facebook don’t know that their newsfeed is customised by Facebook. There are so many issues that this research raises, from algorithmic accountability, to ethics, through to the future of our own relationships with AI and robot driven systems.

My favourite keynote was by Emily Falk who used fMRI to investigate the brain processes involved in the decision to share information. This was mapped to what we know about which kinds of things become widely shared. I loved the breadth of this research, connecting sharing behaviour, content models, brain behaviour, the physiology of communication.

My standout talk of the entire conference was from Ridhi Kashyap on looking at the issue of missing women and sex selective abortion. It was just a ten minute talk, but she showed how using google search volume can give access to a critical social issue for which there are no other reliable data sources. The issue at hand is people using ultrasounds to determine the sex of their child before birth, and then choosing to abort female foetuses. By looking at google search term volume for “ultrasounds” you can actually begin to get a handle on how big an issue this is in areas of India for which data was previously not available.

In terms of the conference overall I would have liked to see some scope for open debate around some of the fundamental topics of the field, particularly around ethics and data management. I’d like to try to arrange something along these lines for the conference next year, I’d love to hear your thoughts!

)
Ian Mulvany

Written by

Head of Product Innovation for SAGE Publications.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade