In my (short?) stint as a researcher, I’ve had the great pleasure of attending a wide variety of conferences. Everywhere I go, there seems to be one of two recurring themes. Either a community has trouble understanding what it does, or it has difficulty doing more than one thing.

The panel discussion at ACM Ubicomp 2011 seems to hint that this conference falls into the former category. Being ubiquitous quite literally means “appearing everywhere;” as technology pervades every facet of society, it is difficult to think of research that would not fall under this banner. Well, perhaps theoretical algorithm work is outside of this scope. But given that the conference saw work ranging from the large (city-level) to the small (individual-level), and the application scenarios were as varied as ever, it seems that ubiquitous research is… well… pervasive?

In doing so, Ubicomp joins the rest of the “who are we?” gang, including the AI community (what is artificial intelligence?), the data mining community (what should being a data miner mean?), and the recommender community (I remember a panel at one of the past conferences: “RecSys isn’t anybody’s home conference. We all belong elsewhere”).

The problem with this is that, in theory, doing everything dilutes the value of any one thing that you do: understanding the goal of the community really affects the quality and impact of the work that can be done. But is this really the case? Personally, I am not sure if I agree. In particular, I remembered a post that I wrote about SIGIR 2010 where I quoted another researcher when I wrote:

the research has fallen into a methodological rut; the conference is “trapped by a very successful paradigm […where] people can do complex work, the quality of that work can be measured, and progress made.”

In this case, the conference seemed to have become good at doing one thing at the expense of being interesting. So, it seems that lacking identity as a community is a quality that matters; or, put another way, too much identity = yawn.

What is really important about any conference? Does a community really matter? Two things immediately come to mind.

Attendee Mixing/Diversity

The conferences I’ve been to that felt the strangest were the least diverse ones, in terms of the audience. KDD was very interesting because there are so many walks of life that are somehow involved in the “big-data” bandwagon. While the Ubicomp work certainly scores highly in diversity, I did notice that the ‘community’ is full of faces that are very familiar to each other (e.g., session chairs calling out to audience members by name). Does that make bigger = better?

Building bridges

I’ve drawn a lot of value from conferences that put people with diverging opinions and backgrounds in the same room. The researcher beside the industry; the technically-minded beside the humanist.

Ultimately, though, none of this stuff matters (unless you are trying to sell registrations to your conference). What excites me the most about conferences is seeing people work on important Problems. Seeing research that matters, that addresses huge problems in the world, whether by chipping away at it one piece at a time or slamming a huge dent into it; and, particularly, stuff that could not be easily solved by any one company (should they suddenly find a business model in it).

I would re-post the question I ended with there: why do we have presentation-based conferences? I’m still not sure. Maybe it’s as flawed as peer review (being a consequence of it).

If you are curious, here’s a list of the conferences I’ve attended in the last few years, sorted by theme.

• Pervasive and ubiquitous computing (Pervasive 2011 and ACM UbiComp 2011),
• Data mining(IEEE ICDM 2010, ACM KDD 2011, ECML/PKDD 2010),
• Artificial intelligence (IJCAI 2009),
• Information retrieval (ACM SIGIR 2009 and 2010),
• Recommender systems (ACM RecSys 2007, 2008, 2010, and soon I will head to 2011),
• Generic applied computing (ACM SAC 2008)
• and Computational trust (IFIPTM 2008).

One thing I’ve learned from making this list (other than that there are some great conferences out there that I would love to go to and have not had the chance to yet) is that the conference I’ve attended the most is ACM RecSys. So much for it not being anybody’s home. This year, I’ll be hosting one of the workshops on personalization in mobile applications - PeMA 2011.