Dan Cosley glamour shot

Dan Cosley (aka DanCo PhD, or drc44 aat cornell ddot edu)

Affiliations:
[Cornell University] [Cornell Information Science] [Reimagination Lab] [CeRI (Cornell eRulemaking Initiative)]
[Interaction Design Lab] [GroupLens Research] [CommunityLab]

Quick Links:
[The Pensieve website (try it!)] [Teaching info] [My publications] [My full CV (pdf)]

News and overview

Welcome to my professional site. As of fall 2008, I'm on the tenure track as an assistant professor in Cornell University's information science department. Prior to that I spent two years as a visiting assistant professor working with Geri Gay's Interaction Design Lab and teaching HCI classes. It was and is good.

A PhD from the University of Minnesota's computer science department helped make this all possible; John Riedl and Loren Terveen are my advisors, and the thesis is Helping Hands: Design for Member-Maintained Online Communities. It's a pretty good read, for a dissertation. If you need to do a background check, here's my CV; below, I hit a few highlights.

Projects

My main interest for a long time has been helping people make sense of and manage information, both individually and as groups. More recently this has grown to include leveraging people's current behaviors online, along with social science theory, to produce individual and social goods that otherwise would not have been created. More details on what this means below, and even more details on my own projects at the Reimagination Lab website.

Pensieve: technology supporting reminiscence

The project I'm currently most excited about is code-named Pensieve until we get sued by J. K. Rowling. This project has an intense, personal inspiration: I rarely remember the past, and mostly I remember bad things. And, although I've kept a blog for 11 years specifically to remind me of the past, I rarely look at it. Then I realized I can write programs to remind me to look at it. Now I get text messages from my past and I like it. Pensieve is about generalizing this to other people, other media, and other contexts and adding support for social reminiscing, which my interviews with potential participants has convinced me is really important.

You can try it yourself:

SuggestBot, intelligent task routing, and modeling community

Recommender systems are another example of leveraging behavior and helping people manage information. Using consumption and preference information that people already provide or are willing to provide cheaply, they can help people find new information to explore, as with MovieLens. This raises a number of issues, starting with making accurate predictions for individuals, but quickly moving into a number of interesting HCI issues such as recommending for groups, helping new users enter the system, evaluating the effectiveness of recommendations, and understanding how those recommendations bias users, all of which I have worked on in the past.

We can use recommendations to do more than help individuals manage their own information overload problems. Karau and Williams' collective effort model predicts that people will be more motivated to contribute to a group good if you reduce their cost of doing so. This leads to the idea of intelligent task routing--asking people to help a community by recommending specific tasks they're likely to already know how to do or likely to enjoy doing. I developed this work as my dissertation, working with Dan Frankowski, John Riedl, and Loren Terveen, culminating (so far) in the SuggestBot tool I wrote for Wikipedia, reported in an IUI 2007 paper [PDF] [ACM DL].

There is still much to do under this umbrella: understanding how to use other theories of how people become attached to groups to make more sophisticated and varied recommendations (of people, of projects, and so on). I'd still like to find students who want to work on this. It stalled a little while I was working with the Interaction Design Lab but I want to make it happen. If you want to help, send me an email.

Being Heard

This project is new enough that it doesn't even quite have a coherent description, but the root is this: one time, I spent quite a long time pouring my heart out to a friend about the girl I thought I was going to marry. After what can only be described as extended angst, his reply: "I learned a new rollerblade trick. Wanna see?"

Recently, I've realized that I don't actually know very much about my patterns of communication with most of the people in my life. What do they look like? What do we talk about? When do they happen? And, can I combine cool vizualizations of conversational behavior over time with clever theories about the ways people make sense of converstaions and relationships to figure it out?

More generally, based on what I saw at CHI 2009, this project is part of what is a growing trend in HCI around developing interfaces that help people become self-aware. Whether it's activity detection and ambient awareness, deriving interesting info from financial records, tracking and visualizing personal data in general, or plain old persuasive computing, this seems like a growth area in HCI and one that has massive potential research and real-world impact.

Other Cornell Collaborations

Geri and her IDL helped me get involved in a number of other interesting projects during my two years of postdocery there before I became tenure track:

And, more generally, I have found a number of fellow travelers through the excellent interdisciplinary culture at Cornell who were working on projects that my knowledge of Wikipedia, social science theory, and recommendation systems have allowed me to contribute to, including:

I also did and helped with a whole bunch of work around recommendation systems with my advisors and labmates back at GroupLens, who I miss. This included group recommendations, recommending research papers, metrics for recommendation effectiveness, the influence of recommender interfaces on ratings, tagging in recommender systems, and learning to recommend for new users.

Finally, there are a number of miscellaneous projects, including developing something resembling a social news aggregator in 1998, building a tagging system with Jeremy Goecks in 2001, helping Google learn to play Who Wants To Be a Millionaire with Tony Lam (et al.), looking at re-identification through preferences and discussions with Dan Frankowski (et al.), and so on. Someday I'll do something far enough ahead of the curve to get rich or famous. Maybe. That part's not actually that important to me.

You can download related publications here or see the list in my curriculum vita. I also have a research statement as of 2011 that tells the story in more detail. For people who are interested in seeing how these things evolve, here's a version from 2008.

Teaching

Fall 2012: INFO/COMM 3450 + INFO 4940, Human Computer Interaction. [Website].

I care deeply about teaching, and most students and observing professors regard me as a solid teacher; for evidence, and more philosophy, see my teaching statement.

Prior semesters:

In general, I believe students learn the most when they're thinking and doing. To that end, I assign lots of work and try to structure classes, especially advanced ones, so that most of the class time involves student activities rather than lectures. I'm a pretty good speaker, but speaking's not the goal, learning is. I managed to get through the whole course in fall 2008 with my longest stretch of lecture under 20 minutes, and though it's awkward to have less "control" of the situation, I think it's mostly a better situation.

I've been teaching for a while; in addition to the courses above at Cornell, I have taught at the University of Minnesota (2005) and James Madison University (1998-2000).

Education

Other professional activities and awards

I have reviewed for a number of conferences and journals. I have served on the Minnesota CS department's faculty recruiting committee and the JMU CS department's curriculum committee. I recently helped write two successful 3-year NSF grants and a Hatch grant. I've received Graduate School (2000) and Guidant Fellowships (2005) at Minnesota. I've had a visiting appointment at Cornell and, now that I've jumped onto the track (or been tied to it, depending on my mood that day), I can't even make the "but I have not yet been a tenure-track faculty member" joke and point to that silly college essay. But I still like it enough to want to point to it.

Personal

Apart from the last bit, this is somewhat more formal than is normal for me. You might get a better read on Dan the person from my personal page.

Older Updates

I think it might be useful to keep the older updates and links around, so here they are.