Jeffrey P. Bigham

Contact Information

University of Rochester
Department of Computer Science
Rochester, NY

jbigham@cs.rochester.edu

Travel Schedule

UID Group@MIT CSAIL until Jan 1.

Recent News

Words to Click

CS Faculty Jobs   CSC 210   DUB   VizWiz  TurKit   Twitter   WebAnywhere   WebAnywhere@GoogleCode   WebProgrammingCourse  

Tweet Tweet:

 "
    fetching my latest tweet...
"

Research Summary

My research seeks to help people more effectively access information.

I'm now at the University of Rochester. If you think you might be interested in working with me, please contact me. If you're interested in pursuing your Ph.D. here, your first step is to apply to the computer science department.

A Brief History of My Research:

At the University of Washington, I started the WebInSight Group with my advisor Richard E. Ladner. I invented and released the WebAnywhere non-visual web-browsing web application, which has recently taken on a life of its own. I also collaborated with the AIM Research Group headed by Jacob O. Wobbrock of the Information School, and actively participated in DUB, our inter-departmental HCI group.

The WebInSight group started with a system for automatically adding alternative text to web images. We're currently pursuing three complementary projects: WebinSitu seeks a better understanding of the browsing behavior of blind web users, WebAnywhere enables web access for blind people from any computer that happens to be available, and Accessmonkey will allow users to improve accessibility without relying on web developers. I worked with the CoScripter team at IBM Almaden to develop TrailBlazer, an intelligent, non-visual guide through web pages. Unlike many web macro recorders, TrailBlazer can generalize the knowledge it records to new tasks.

I've also collaborated with the web accessibility team at IBM Japan to crowdsource accessibility. This began with my work on Accessmonkey, and has since been extended to a really great system called Social Accessibility, which helps connect blind web users experiencing accessibility problems with sighted users who can help fix them.

I've helped to make iPhones (visual touchscreens) accessible to blind users with Slide Rule and helped to improve the density of sign language available in advanced STEM fields with the ASL-STEM Forum to help deaf students succeed.

I've recently been exploring audio CAPTCHAs in hopes that we might finally make the frustrating tests of humanhood usable by everyone. As a first step, we redesigned the interface to audio CAPTCHAs and improved the success rate of blind web users by 59%, illustrating the challenge and potential for impact of creaing more intuitive non-visual interfaces. Read my CHI paper on audio CAPTCHAs.

In addition to research, I'm an active advocate for accessible design and for the richness and challenge of developing intuitive non-visual interfaces. Be sure to check out my Web Accesssibility Research Blog.

Mining Web Mash-Ups from Web Interactions

The web provides unrivaled access to information, but we're often stuck with the interfaces someone out there thought to build for us. By leveraging ideas from information extraction, inference over user behavior, and end user programming, Transcendence enables users to explore the web hidden behind forms in ways not currently possible. Here's a nice video of the first version of this system Transcendence in action, and here are the slides from my IUI Presentation about Transcendence.

As a follow-up, I collaborated with Jeffrey Nichols and Ryan Kaminsky to explore how we might automatically connect different forms together by observing the interactions of the people using them. Check out our TX2 UIST paper.


Videos!


My 90 second elevator pitch for WebAnywhere at EmTech 2009.

A video description of the WebAnywhere non-visual web browser, available anywhere.

A widely-distributed outreach video in which I (and others) explain why computer science is cool.

Getting around the restrictions of web forms by demonstration - Transcendence Project Video.

Web Accessibility Research

Why work on web accessibility? The potential social impact is huge - the web presents great opportunity for blind users because of the enormous amount of information made available in a format that can be converted to speech or braille. Techniques developed for non-visual access directly impact other types of interfaces, such as small-screen mobile devices, web access over the phone in the developing world, and even automatic understanding of web content by automatic agents (who also cannot see). Developing good non-visual interfaces is a rich, interactive and challenging design space, and an important area of HCI research that has not been adequately explored.

Non-visual access to the web remains difficult for blind users for a number of reasons, including:

  1. Accessibility
    1. The content in images, animations and movies aren't accessible and alternatives are often not provided.
    2. Many sites restrict access using an inaccessible visual CAPTCHA, or a difficult audio CAPTCHA.
    3. Dynamic Javascript menus and Flash often require specific input devices, such as a mouse.
  2. Usability
    1. Complex sites are difficult and inefficient to navigate using the serial interface exposed by screen readers.
    2. Audio CAPTCHAs are not designed for non-visual use, leading to unacceptable success rates in solving them.
  3. Availability
    1. Many people cannot afford expensive assistive technology.
    2. Assistive technology is not installed on most computers.

Web accessibility research is an incredible field. This area provides the opportunity to explore and improve interactions in a relatively unexplored space (not just incremental improvements here!) with the chance to make positive social impact in the short term. My WebAnywhere system is currently being used by 700 teachers, students, and other ordinary folks each a week, and it only recently progressed beyond a research project. That's a short turn-around time from innovation to impact!

I recently received a grant from the Mellon Foundation ($50k) to further the development of WebAnywhere. At various points, I've had the NCTI, Osbergs, NSF, Boeing, Microsoft, TRACE and NASA to thank for my funding.

Jeffrey P. Bigham
jbigham@cs.rochester.edu