Senior Microsoft Corp. Researcher Susan Dumais predicts that in 10 years, we will look back on today’s search interfaces and recognize them as a simple and limited way to interact with information. After all, she explains, a 5-inch-long rectangle with a long list of text results beneath it doesn’t do much to help people make sense of the billions upon billions of unorganized bits of data in the world.
Dumais sees plenty of room for improvement, both in how people specify their information needs and how the results of queries are presented to them. If she has her way, a decade from now people will be able to easily locate the information they need and use the results in context without even realizing they’re searching.
If anyone can make the notion of a search function disappear, Dumais is a likely candidate. Her years of research work recently earned her a coveted spot in the CHI Academy, an honorary group of leaders in the field of human-computer interaction (HCI). The award was presented by the Association for Computing Machinery (ACM) SIGCHI, the ACM’s Special Interest Group on Computer-Human Interaction, which brings together people working on the design, evaluation, implementation and study of interactive computing systems for human use. Dumais was recognized for her cumulative contributions to the study of HCI, her influence on the work of others and her development of new research directions.
Microsoft research podcast
What’s Your Story: Lex Story
Model maker and fabricator Lex Story helps bring research to life through prototyping. He discusses his take on failure; the encouragement and advice that has supported his pursuit of art and science; and the sabbatical that might inspire his next career move.
“It’s a great honor to be recognized by your peers, not only for sustained contributions but also for shaping some of the thinking in the field,” said Dumais, a senior researcher in the Adaptive Systems & Interaction (ASI) group at Microsoft Research since 1997.
Eric Horvitz, senior researcher and group manager of Microsoft Research’s ASI group, said, “It’s been an absolute honor to have the opportunity to work closely with Sue. I’ve enjoyed brainstorming and collaborating with her on far-out possibilities that she manages to bring to life. Few people know that Sue was a high-speed luge competitor. While she doesn’t hurtle down icy slides these days, I see her daily continuing her high-speed twists and turns as she rapidly innovates in the conceptual realms of search and interaction.”
Dumais is the second Microsoft researcher to be honored by the CHI Academy. Her colleague Senior Researcher Jonathan Grudin received the award last year for his contributions to the HCI field.
Dumais works with obvious passion at the intersection of HCI and information retrieval. Her research focuses on information access and management, an area that covers everything from searching and browsing to simply finding information on one’s desktop when creating a document. Her quest: Make it easy for people for find, use and make sense of information.
The topic may seem narrow, but Dumais’ research covers a wide territory. Her Stuff I’ve Seen project came about when she became interested in the problem of information reuse several years ago. A good deal of knowledge work involves finding and reusing information that has previously been seen (rather than discovering information in the first place), yet few information management tools support this. Stuff I’ve Seen is a tool that makes it easy for users to find information they’ve seen before, whether it was seen as e-mail, attachments, files, Web pages, appointments, tablet journal entries or other formats. This is done by providing unified access to different sources of information and providing a fast and flexible interface with quick sorting, filtering, previews and thumbnails.
More recently, Dumais has been going in other new research directions. In one project, she and her colleagues are working to personalize search capabilities, using knowledge of a person’s previous interactions with information to produce results that are more relevant to the individual. In another project, she’s striving to contextualize search functionality.
“The idea is to understand when people search, why people search and whether we can make it easier for them to get results without leaving the application they’re in,” Dumais said. “It’s recognizing the fact that search is not the end goal. We want to show people results in context and help them integrate those results into whatever they’re doing.”
Dumais is quick to point out that the research she engages in isn’t just theoretical. For example, she works closely with Microsoft’s MSN Search group to keep improving the core MSN, Web search engine and related capabilities. Much of that research entails trying to understand when and why people are satisfied with their results. Some of her early work in text classification also has been implemented in Microsoft, SharePoint, Portal Server, Microsoft’s enterprise search engine, and in spam filters. In addition, much of the core philosophy from the Stuff I’ve Seen project ultimately found its way into the MSN Toolbar and desktop search engine.
That’s part of what makes Microsoft Research an amazing place, according to Dumais. “I have the freedom to analyze interesting and important problems and do solid research,” she said, “but we’re not just doing lab studies and thinking great ideas. We have the resources to develop research prototypes that apply the learning and help solve those problems.”
Microsoft Research also has given Dumais the opportunity to publish numerous papers in academic journals and at professional conferences, including CHI, the leading conference in the HCI arena; the Special Interest Group on Information Retrieval (SIGIR), the pre-eminent information retrieval conference; and the Conference on Uncertainty in Artificial Intelligence (UAI), which focuses on intelligent systems.
Dumais has shaped the HCI community by supervising student interns every year for the past 25 years, collaborating on projects with academic and industry partners, serving on Ph.D. committees, and teaching at various universities. Currently Dumais is an adjunct faculty member in the Information School at the University of Washington.
Her own educational background is a perfect fit for blending algorithms and interfaces. Dumais concedes that she didn’t set out to be a research scientist. As a child, she envisioned herself as a firefighter and, later, perhaps a doctor or lawyer. She entered Bates College in Lewiston, Maine, with a plan to major in math and go on to law school. But in her sophomore year of college, a course called Mathematical Models of Human Memory and Perception changed her mind.
“It was really exciting to see the abstract infrastructure of math being used in ways that could help us understand how people interact with the world, and also to design systems that enable them to do that better,” she explained. “I’ve been fascinated by that combination ever since.”
Armed with a bachelor’s degree in math and psychology, Dumais headed to graduate school. She came away from Indiana University with a Ph.D. in cognitive and mathematical psychology, a discipline that seeks to understand how people perceive, organize and retrieve information. Upon completing her doctorate, Dumais went to work as a researcher at Bell Labs in New Jersey, in one of the first research groups in the industry dedicated to HCI. There, she began research on a statistical method for concept-based information retrieval called latent semantic indexing.
That early work has proved influential in the HCI, information retrieval, psychology and education communities. Latent semantic indexing is a statistical technique for extracting and representing the similarities of words by analyzing large collections of text. The resulting representation is more abstract than individual words, and addresses information access-problems that stem from the nature of human vocabulary and word usage. Twenty-five years later, it continues to be relevant in solving fundamental problems in information retrieval.
Algorithms 
Search and information retrieval