If you want to avoid the discursive portions and jump directly to a list of papers with full citations, go ahead.
My big hobby horse is making sense of the Web on a personal and small-group level. Right now the primary way I'm looking at doing this with is something called collaborative filtering, or more generally, recommender systems. I'm more interested in reasons that users should believe and trust recommendations than raw algorithmic power. See the REFEREE paper below, section 4, for my most cogent argument why.
Another set of interests is around a grant we have with people at Carnegie Mellon and University of Michigan that aims to implement social science theories as computations you can use to help shape a community. For example, similar people like each other. Compute similarity, and use it to introduce people or form discussion groups. Or, people contribute to groups when they know their contributions matter. Unique contributions matter -- so data mine your users' profiles to learn special things that a given use might be able to contribute, and tell them. One particular angle, member maintained communities, is particularly interesting. I did some related work, in class-organized class resources, during my Masters' program, and have picked the general idea back up for the dissertation.
My other current interests are UI design, particularly "intelligent user interfaces" (a misnomer, I think, as the goal is probably intelligent programs that make the user's life easier, interface-visible or not). I'm also looking at sociable computers: computers that can act as hosts, that can help people find each other, etc. If electronic communication takes away from human interaction, as is claimed, then perhaps we should find a way to help put it back. One way to do this would be to use similarity information (such as that gleaned by recommender systems). Such a natural idea, yet so hard to study well.
I'm also interested in computer uses in education, including distance learning (a necessary but interesting evil), and tools for enhancing classrooms, virtual or physical. However, this part of the package is currently quiescent.
How do similarity of interests and demographic similarity affect the way people interact when performing multi-person tasks online? None, and some, at least for the interests, demographics, and task we used in our first experiment, a paper on which was recently accepted to GROUP 2003. I did this work with Pam Ludford and Loren Terveen, and we think we didn't use a specific enough notion of interests and the tasks were too general. Pam led the charge on a sequel accepted at CHI 2004: better tasks, a more compelling notion of similarity, and telling people special contributions they might make to a discussion topic lead to more group participation and results that online community builders can put into practice. For CHI 2005, we are back and better than ever, telling people about how editorial oversight increases people's motivation to contribute to a member-maintained community. Peers are as effective as experts in giving oversight, which means use everyone, not just editors. We also have a paper at the AAAI Spring Symposium paper that reports on the above work as well as throwing out some ideas to be molded by the fire of scholarly critique.
I'm working with the GroupLens research group here at U of M, at least as long as I don't really screw up. The main project the group has going right now is MovieLens, a practical example of collaborative filtering suitable for running experiments with. It will help you find movies you like, if you tell it how you feel about a few movies you've already seen. The way it does this is by matching you up (anonymously) with people who have similar tastes. Check it out.
I've done several projects related to MovieLens. The first was implementing small improvements to the group prediction feature that Mark O'Connor built. This allowed members to invite non-users to join groups, and we studied how users reacted to the group recommendations. It turns out that despite being pretty simplistic, people use them and like them a lot. ECSCW 2001 was thrilled to hear about them, although Joe had to give the talk since he was already in Europe and the conference was Sept. 15, 2001, meaning I had no chance of flying over because of the events of Sept. 11, 2001.
I also helped Mamun Rashid write a paper about algorithms for learning new users' preferences (rating the few movies you've already seen mentioned above). We found good algorithms, and told people about them at IUI 2002.
Finally, Tony Lam and I collaborated on a study of how the prediction a recommender system displays can influence the way people give ratings. The answer is: more than you'd think (or want!) This research was published at CHI 2003 and written up in the New York Times, although you have to pay for the privilege of NYT. Read the research paper, it's free.
TechLens has the goal of recommending research papers. We haven't done a lot with it (except discuss how to do it), although Sean McNee led an effort to try out a number of algorithms, some CF-based, some not. There's a tradeoff between novelty and perceived relevance to a particular paper, and different algorithms let you make different tradeoffs. I was a minor contributor. CSCW 2002 ate it up.
A project which I completed as an intern at NEC Research Institute in summer of 2001 with Steve Lawrence and David Pennock, two extremely prolific researchers. A next step in collaborative filtering is integrating content and collaborative filtering. REFEREE is a platform for testing recommenders (with or without integration of content) using ResearchIndex. Our belief is that testing recommendations offline is not the right way to test them: recommendation is not classification. Instead, you should measure user satisfaction and the impact that recommendations have on users. See the VLDB 2002 paper for more info. Note: pretty much everyone involved has moved on from NEC, and it is, as far as I can tell, not actively supported any more.
I was a minor player on a paper headed by Tony Lam (he was at NEC with me in 2001) which used Google as a knowledge base for a program to play Who Wants to Be a Millionaire. Alas, computers are banned by the official rules, plus nobody remembers the show anyways, but combining search engines with some heuristics and a basic decision-making module results in a player which is, as the paper puts it, "six questions to human" -- and it's the easy six questions. Steve and Dave were also in on this paper, which you can check out at UAI 2003 or just download below.
Smartshopper's goal was to study everyday computing and UI design in the context of a Palm application that you can use to help budget and manage your shopping. There's still a page for it, but it stopped working with Palm OS 4.0 and we never went back in and fixed it.
Actually, you can skip the send us data part, as no active research is ongoing. We had well over 40,000 downloads from zdnet and download.com, and a lot of people liked it, so bully for us. If only we had a nickel for every user... Alas, now they require people to pay $$$ to list their software, so it's gone.At any rate, we got a poster accepted at the SIGCSE 2001 ACM student research competition. I've been told that this would be worth studying more, but I have not found the time.
My Masters thesis is called "MaSH: Making Serendipity Happen". It helps a group (say, a club or a class) to build its own website a la the Open Directory project, but without category editors and with the ability to comment on and rank sites added to the directory. It was of some success in a beginning computer class.
I've also turned it into a conference paper that was published at WebNet 2001. Hooray for me. The software itself is not available at this time, because I'm embarrassed by it now. It may rise again someday, but for now, one must be satisfied by the papers.
I also built a little online review system, Poser, that turned out to be popular with students even though it's not very good. It'd be useful to rebuild faster, stronger, and better. This work was done with Mark Lattanzi and Ben McDowell, and we got a small education journal publication out of it. It was used in several classes in the James Madison University CS department. I don't claim it's a great paper, but you gotta start somewhere.
My first real CS project was Agware; I was the junior officer to one Matt Labarge. It started with wide goals and evolved into a moderately useful mailing list facility with a handy interface. It's dead now -- completely subsumed by better things -- but I'll always remember it fondly.
Cosley, D., Frankowski, D., Kiesler, S., Terveen, L., & Riedl, J. (2005). How Oversight Improves Member-Maintained Communities. In Proceedings of CHI 2005, Portland, OR, pp. 11-20. [PDF] [PS] [ACM DL]
Cosley, D. (2005). Mining Social Theory to Build Member Maintained Communities. In Proceedings of KCVC 2005, Palo Alto, CA. [PDF] [PS]
Ling, K., Beenen, G., Ludford, P., Wang, X., Chang, K., Li, X., Cosley, D., Frankowski, D., Terveen, L., Rashid, A. M., Resnick, P., & Kraut, R. (2005). Using social psychology to motivate contributions to online communities. Jour nal of Computer-Mediated Communication, 10(4). [JCMC]
Ludford, P.J., Cosley, D., Frankowski, D., & Terveen, L. (2004). Think Different: Increasing Online Community Participation Using Uniqueness and Group Dissimilarity. In Proceedings of CHI 2004, Vienna, Austria, pp. 631-638. [PDF] [PS] [ACM DL]
Cosley, D., Ludford, P., & Terveen, L. (2003). Studying the Effect of Similarity in Online Task-Focused Interactions. In Proceedings of Group 2003 Conference (GROUP 2003), Sanibel Island, FL, pp. 321-329. [PDF] [PS] [ACM DL]
Lam, S.K., Pennock, D.M., Cosley, D., & Lawrence, S. (2003). 1 Billion Pages = 1 Million Dollars? Mining the Web to Play "Who Wants to be a Millionaire?". In Proceedings of the 19th Conference on Uncertainty in Artificial Intelligence (UAI2003), Acapulco, Mexico, pp. 337-345. [PDF] [PS]
Cosley, D., Lam, S.K., Albert, I., Konstan, J., & Riedl, J. (2003). Is Seeing Believing? How Recommender Systems Influence Users' Opinions. In Proceedings of CHI 2003, Fort Lauderdale, FL, pp. 585-592. [PDF] [PS] [ACM DL]
McNee, S., Albert, I., Cosley, D., Gopalkrishnan, P., Lam, S.K., Rashid, A.M., Konstan, J.A., & Riedl, J. (2002). On the Recommending of Citations for Research Papers. In Proceedings of ACM 2002 Conference on Computer Supported Cooperative Work (CSCW 2002), New Orleans, LA, pp. 116-125. [PDF] [PS] [ACM DL]
Cosley, D., Lawrence, S., & Pennock, D.M. (2002). REFEREE: an open framework for practical testing of recommender systems using ResearchIndex. In Proceedings of the 28th International Conference on Very Large Data Bases (VLDB 2002), Hong Kong, China, pp. 35-46. [PDF] [PS]
Rashid, A.M., Albert, I., Cosley, D., Lam, S.K., McNee, S., Konstan, J.A., & Riedl, J. (2002). Getting to Know You: Learning New User Preferences in Recommender Systems. In Proceedings of the 2002 International Conference on Intelligent User Interfaces (IUI 2002), San Francisco, CA, pp. 127-134. [PDF] [PS] [ACM DL]
Goecks, J., & Cosley, D. (2002). NuggetMine: Intelligent Groupware for Opportunistically Sharing Information Nuggets. In Proceedings of the 2002 International Conference on Intelligent User Interfaces (IUI 2002), San Francisco, CA, pp. 106-113. [PDF] [PS] [ACM DL]
Cosley, D. (2001). Using Online Tools to Enhance Classrooms: A Case Study with MaSH (Making Serendipity Happen). In Proceedings of WebNet 2001, Orlando, Florida, pp. 233-238. [PDF] [PS]
O'Connor, M., Cosley, D., Konstan, J. A., & Riedl, J. (2001). PolyLens: A Recommender System for Groups of Users. In Proceedings of ECSCW 2001, Bonn, Germany, pp. 199-218. [PDF] [No .ps -- big!]
Goecks, J., Cosley, D., Razieli, Z., Good, N., & Pham, P. (2001). Discovering Design Principles of Everyday Computing Devices via a Case Study. ACM Student Research Competition at SIGCSE 2001, Charlotte, NC, p. 449. (note: they left everyone out of the proceedings except Jeremy.) [No electronic format]
Cosley, D. (1999). MaSH: Making Serendipity Happen. Master's Thesis, James Madison University. [PDF] [No .ps -- big!] [MS Word]
Lattanzi, M., & Cosley, D. (1998). Poser, an Online Review Tool in Java. Computer Science Education, 8(3), pp. 251-264. [PDF] [PS]
[mail Dan] [Dan home] [top of page] [top of topic] [up one level] [UMN CS Dept.] [MovieLens]
The views and opinions expressed in this page are strictly those of the page author.
The contents of this page have not been reviewed or approved by the University of Minnesota.