research activities

Usable user modeling

Towards man-machine symbiosis

J.C.R. Licklider predicted that computers could eventually become humans' intellectual, social and creative partners. One of the critical steps towards this vision is to enable machines to better understand humans. Recent advancements in machine learning and user modeling, along with advances in sensing technologies and information extraction, are finally starting to enable computers a glimpse at the lives of their users.

But why do we want this, and how will computers relate to humans?

A simple answer is that one of the first applications of user modeling is to help people manage their everyday tasks and responsibilities. Automation will continue to enter our lives to save us from the details of how things work. Simple examples are smart thermostats and automatic transmission systems in cars. But anyone who has ever used a smart thermostat (or switched from using a manual to automatic tranmission) will tell you, there is a cost to automation. A feeling of a loss of control, frustration, and unpredictability can plague automation, and often leads to mistrust of the system when things go wrong repeatedly. This is why I am interested in researching the usability of user modeling systems, to investigate ways that people can form a productive relationship with systems that automate their lives -- (given that this automation is never going to be perfect!)

Are you making a new Clippy?

No, and yes. Systems that watch users actions and take proactive action, such as Office Assistant could potentially save people time and effort in dealing with complex computer systems.

As the number of features in software applications continues to grow exponbentially, it will be come increasingly difficult for people to know the best way to accomplish their tasks. Outside desktop applications, people's ever-increasing social roles and mobile work practices, fueled by instant anywhere communications networks, mean that proactive assitance will likely also find a role in managing the complexity of people's complex social and professional lives.

Thus, proactive systems that infer people's tasks and needs, to identify ways to help will become increasingly important. However, it will likely look, feel, and act very little like beloved Clippy...

For more information...

Please see my Ph.D. proposal.

Social ubiquitous computing

Reducing the effects of distance

The layout of our physical environment has been found to have a profound effect on our social networks. In particular how our workplaces are arranged shapes who we meet, work with, and even, to some extent, our professional success. Yet, we seldom have control over the layouts of our workspaces; space allocation is frequently arbitrary and predetermined.

The structure of our social networks has been found to be important at multiple levels -- beyond determining who we hang out with in our spare time. According to the Granovetter theory of weak ties, a large portion of our new ideas and opportunities tend to come from the outer reaches of our social networks. Simultaneously, our close ties have been shown to provide us the greatest amount of personal feedback and emotional support that keep us happy.

If people could have greater control over whom they meet, people could build arbitrary large social networks as they please, and more easily find friends with whom they want to associate most closely. I have been exploring ways by which ubiquitous computing technology could empower people to have greater control over they social networks, by designing new technologies that facilitate lightweight social interactions in situations where such interactions occur seldomly. Project ok-net surrounded connecting geographically separate social spaces within the workplace to allow people whose offices are farther apart to become socially aware of one another. Serendipity for kiosks, with Nathan Eagle, was a project using cell phones that sought to "break the ice" and reduce social barriers to conversation among strangers in public spaces. Peerworks, a recent collaboration with Scotty Ostler, works to help people in shared physical public spaces learn about one another and share information and data.

Other stuff...