One aims to streamline big data, making it faster and easier for companies to share. The other wants to guard folks‚Äô personal information online. Both are doctoral students at UC Irvine‚Äôs Donald Bren School of Information & Computer Sciences and recipients of 2013 Google Ph.D. Fellowships.
Yingyi Bu and Bart Knijnenburg will have all their University of California fees paid and receive $33,000 each for research for two years running. They have also been assigned mentors at the Internet giant, which says on its website that ‚Äúnurturing and maintaining strong relations with the academic community is a top priority at Google. ‚Ä¶ This year we awarded 15 unique fellowships to some amazing students in the U.S. and Canada.‚ÄĚ
Competition for the prestigious grants is tough, and ICS Dean Hal Stern is thrilled that both of the school‚Äôs candidates were selected by a company that has traditionally named winners from such well-known institutions as UC Berkeley and Stanford University.
‚ÄúThis was not at all expected. I was obviously hoping at least one would win, but I didn‚Äôt expect two,‚ÄĚ Stern says. ‚ÄúI‚Äôm very excited because we believe that we‚Äôre one of the top computing programs in the country, and this is a nice reinforcement of that.‚ÄĚ
He notes that the funding will also help the school expand its graduate support in tough budgetary times by freeing up public money for other students. ‚ÄúThese awards are powerful in terms of both their recognition and their financial impact,‚ÄĚ Stern says.
Bu and Knijnenburg share his joy.
‚ÄúOh, I feel excited, because this means that UCI is doing world-class things,‚ÄĚ says Bu, 31, a native of China who has been studying computer science on campus for three years. His fellowship project involves designing more efficient computing structures.
‚ÄúIt‚Äôs a good opportunity to have a close connection with a Google researcher,‚ÄĚ he says. ‚ÄúIt‚Äôs also an opportunity for me to make the system we‚Äôre building have more impact in the real world.‚ÄĚ
Knijnenburg, 29, says of the fellowship: ‚ÄúMy doctoral adviser told me not to expect to get it because they always give it to the ‚Äėbig‚Äô schools. When I got the email saying I‚Äôd won, I was really surprised.‚ÄĚ
Asked by Google not to publicly share the news for a month, he did go out and celebrate with a few friends.
Keeping such interactions private is at the heart of Knijnenburg‚Äôs work, which focuses on informatics, or the relation of humans to computers. Specifically, he wants users‚Äô data automatically kept more private if their online activity seems to indicate that‚Äôs what they would prefer.
‚ÄúMy topic for Google is to help people make privacy decisions in varying ways,‚ÄĚ Knijnenburg says. ‚ÄúSome people are very private; some people are not. The obvious solution is to give people more control.‚ÄĚ
That‚Äôs easier said than done, he admits, with the proliferation of online cookies that track usage for companies and even government agencies. Knijnenburg believes that rather than individual consumers trying to figure out how to opt out of complicated data gathering systems, it could be done for them if their habits show that‚Äôs what they‚Äôd want.
‚ÄúI, better than anyone, know I should protect my bank account information online, and I know how to do it too,‚ÄĚ he says. ‚ÄúBut it‚Äôs just too complex. And it shouldn‚Äôt be. I also at times don‚Äôt feel like deciding whether to try to encrypt something. So I choose immediate gratification over the risk of giving away personal information. I‚Äôm only human.‚ÄĚ
Knijnenburg, who‚Äôs also been at UC Irvine for three years, has already patented a method to determine how people make privacy decisions, and he thinks it could be put to good use automating the often overwhelming controls of social networks.
On the surface, Bu‚Äôs focus appears to be just the opposite: He‚Äôs building new architecture by which the estimated 8 billion Web pages now in existence could be checked against each other more rapidly for ingesting, storing, managing, indexing, querying, and analyzing trends and other information.
As computers have grown increasingly adept at automatically compiling data, it has become unwieldy and time-consuming to sort all the terabytes. Bu is researching ways to better ‚Äústack‚ÄĚ huge piles of software so that programs can operate together with greater efficiency.
‚ÄúMy primary area of interest is large-scale data management systems, especially data-intensive computing systems,‚ÄĚ he says. One current project is Pregelix, an open-source implementation of a ‚Äúbulk-synchronous‚ÄĚ programming model for extensive graph analytics.
Bu points out that Knijnenburg‚Äôs work could be aided by his own efforts.
‚ÄúHis privacy protection techniques could probably be much faster with our system,‚ÄĚ he says. ‚ÄúHe also needs to deal with mountains of data. He could let these algorithms run on ‚Äėshare-nothing‚Äô machines.‚ÄĚ