Project Name: Personalized User Interfaces in Real World Contexts

Project Leader

Michael Terry

Project Co-Leader

Joanna McGrenere

Researchers

• Charles Clarke
• Michael Terry
• Temy Tidafi
• Robert Woodbury
• Joanna McGrenere (UBC)
• Andrea Bunt (Manitoba)

Theme Distribution

• Theme 1 : 20%
• Theme 2 : 10%
• Theme 4 : 10%
• Theme 5 : 60%

Project Description

Modern new media applications offer a wide range of sophisticated capabilities, but at the cost of an overwhelming number of options and commands. As one example, Adobe Photoshop boasts over 500 commands and options in its application menus. Such sophistication is a natural byproduct of supporting a wide range of tasks and users, but it can easily overwhelm both novices ''and'' experienced users, especially when learning to perform new tasks.

Given the ever-growing complexity of applications, new interface paradigms are needed to enable users to effectively and efficiently manage this complexity. Adaptable (user controlled customization), adaptive (system controlled), and mixed-initiative (shared user and system control) have all been proposed to address this overarching problem, and all have known strengths and weaknesses. However, to date, few of these approaches have had much impact on the market place. In part, this lack of impact can be attributed to research and design methods largely informed by relatively brief evaluations in controlled laboratory settings, which, by definition, lack the rich context, constraints, and pressures of day-to-day work. More widespread, long-term, ecologically valid studies of these approaches are thus needed to better inform interface designs in this space (Grossman, et al, 2009; McGrenere et al, 2002).

This project will investigate ''interface personalization'' techniques to address problems of complexity and learnability in modern interfaces. These personalization techniques will draw upon recent advances in machine learning and statistical inferencing techniques to enable the system to better understand users' tasks, intentions, and goals, thereby allowing the system to play a role in adapting the interface and teaching the user how to best use the application. To address the need for long-term, ecologically valid evaluations, we will contextualize this work in mature open source applications. Historically, the open source community has been greatly underutilized in interactive systems research; this research will more thoroughly investigate how this enthusiastic, energetic community can be best utilized to conduct large-scale, longitudinal research. Partnering with this community will not only catalyze design and evaluations of these interaction paradigms, it will also assist in creating corpora of application usage data, vital for evolving models of user behaviors, tasks, and goals.

Excellence of the Research

Development of Highly Qualified Personnel (HQP)

This project will be contextualized within mature open source content-creation applications. As such, students involved in this project will be working with sophisticated, real-world applications that will be deployed and used by hundreds or thousands of users. This is a unique opportunity for students to make both theoretical and practical contributions whose impact can easily be felt by thousands of people.

Networking and Partnerships

To date, interactive systems research has not leveraged the possibilities of the open source community. This work will not only partner researchers in academia and industry (Autodesk), it will also create new ties to the much larger open source community. Closer ties between research, industry, and open source communities enables radical new forms of research. For example, while commercial software companies must carefully control their brand and image in the marketplace (which, in turn, influences the design of the software produced and the risks that can be taken with those designs), the open source community can more freely experiment, without fear of economic side-effects. One of the things this research project will explore is the notion of performing large-scale experimentation in the open source software community, with the lessons learned transferring to commercial software products once the value has been demonstrated.

Knowledge and Technology Exchange and Exploitation

See above.

References to the Literature

Grossman, T., Fitzmaurice, G., and Attar, R. 2009. A survey of software learnability: metrics, methodologies and guidelines. In Proceedings of the 27th international Conference on Human Factors in Computing Systems (Boston, MA, USA, April 04 - 09, 2009). CHI '09. ACM, New York, NY, 649-658.

McGrenere, J., R.M. Baecker, and K.S. Booth. An evaluation of a multiple interface design solution for bloated software. In Proceedings of the ACM Conference on Human Factors in Computing Systems (CHI 2002), pages 163170, 2002.

Publications

Presentations

• 2 Minute Madness 2010