DATE: Thursday, August 19, 2004
TIME: Noon - 1 pm
PLACE: Wean Hall 8220

Eyal de Lara
University of Toronto

Community-Driven Adaptation

Mobile devices, such as cell phones and network-enabled personal digital assistants (PDAs) are fast becoming the platform of choice for accessing Web content, and are expected to play a leading role in the pervasive or ubiquitous computing environments of the future. Unfortunately, most Web content is intended for consumption on powerful desktop computers, and therefore has to be adapted to meet the limited resources of mobile devices. In recent years, systems that automatically adapt content to specific devices have surfaced as a promising approach for coping with the resource constraints and high degree of heterogeneity in pervasive-computing environments. The main challenge in automatic content adaptation is designing adaptation policies that make optimal use of resources and maximize user satisfaction.

In this talk, I will describe Community-Driven Adaptation (CDA), a novel technique for automatic content adaptation that adapts content based on feedback provided by users. CDA groups users into communities based on common characteristics (e.g., device type or preferences), and assumes that members of the same community share requirements for adaptation, so that adapted content that is suitable for some members of the community is likely to be acceptable to other members as well. When responding to a mobile user request, CDA first makes an initial prediction on how to adapt the content. Next, CDA monitors the user as she modifies the adapted content to make it more appropriate to her tasks (e.g., make content more readable by removing an irrelevant toolbar, convert text to speech, or increase or reduce a video’s frame rate). The user modifications constitute an implicit feedback mechanism that CDA uses to improve its adaptation predictions for future accesses by other members of the user’s community. Experiments conducted with a prototype that adapts image-rich web pages for browsing over bandwidth-limited links show that compared to existing approaches to automatic adaptation, CDA reduces wastage of network bandwidth by up to 90% and requires less user interaction to correct bad adaptation decisions.

Eyal de Lara received his Ph.D. and M.Sc. from Rice University in 2002 and 1999, and a B.Sc. from the Instituto Tecnologico de Monterrey in 1995. In 2002, he joined the Department of Computer Science at the University of Toronto as an Assistant Professor. His research interests include distributed systems, networking, and mobile and pervasive computing.

For Further Seminar Info Contact:
or visit