Researching Trust and Distrust

Allison Druin, Associate Dean for Research, iSchool

This past week, Assistant Professor and HCIL Director, Jen Golbeck, received a best paper award from the 2011 IEEEInternational Conference on Social Computing (#socialcom).  Her award-winning paper was co-authored with ThomasDuBois, University of Maryland computer science alumni and now a post-doc at  Virginia Tech, and Aravind Srinivasan, a Professor in Computer Science and UMIACS.  Their work focused on some important questions surrounding trust and distrust by examining social networks.

 Assistant Professor and HCIL Director, Jen Golbeck,

Recently, I sat down with Jen for a “virtual interview” to better understand this important research area:

[Allison] When you study “trust” are you looking at how people trust each other on social networks or if they trust the social network?

[Jen] I'm looking at trust between people. There is work on people trusting systems (sometimes called trust in automation), but I'm more interested in finding ways to compute how much one person trusts another person.

[Allison] How can you make predictions about people from how much they trust each other?

[Jen] There are a lot of ways. Some of my earlier work looked at paths that connected people through the social network and the trust that their intermediate friends had for one another. More recently, I have been working on analyzing traits of each individual in the relationship and using that, along with structural social network features, to predict trust relationships. It's more exciting because it's more realistic - we don't often know how much all the intermediate friends trust each other, so working from more commonly available data is important.

[Allison] What’s the most surprising result you’ve found from your most recent work?

[Jen] Lately we have been trying to predict people's personality traits by analyzing their Facebook and Twitter profiles. It turns out we can do that quite accurately, even with very limited information from the users. We hope this is something we can eventually use to help understand people's relationships.

[Allison] Should people be concerned with what they put out on the web?

[Jen] Of course, always. A lot of that is independent of the work I'm doing. There is some very sophisticated data mining taking place on the web. That can be used to improve users' online experience, which is good, but we don't get to control who uses it and for what. If the idea of people analyzing you is bothersome, it is best to really turn up privacy settings. However, that doesn't necessarily protect your data. For example, on Facebook, if your friends install an app, that app can access some of your data, even if you don't consent to it. Thus, it's safe to assume that anything you share online is accessible to companies with any variety of intentions. If you don't want them to know something, it's better to keep it off the web.

[Allison] What’s next for you in this area of research?

[Jen] I'm going to be pushing in the direction of understanding users. The personality research has been very fruitful, but we are in the very early stages of that work. I hope to find more individual features to profile and new techniques to predict them.

[Allison] If doctoral students or new faculty are trying to get into this area of research what advice do you have for them?

[Jen] The most important thing is to try to find a problem in a space that is untouched. There is so much that can be done in studying social networks, analyzing users, and improving systems with that information. If you do the next steps on existing research, all of your results will be incremental and boring. Find a problem that has hardly been touched but where you see great potential. That sets you up to be the thought leader on an important problem, and it means you will have an exciting time discovering new things.

[Allison] Thanks, Jen.