By Janie Chang
August 22, 2012 9:00 AM PT
Microsoft Research maintains strong ties to the academic world through a variety of programs, including its popular hosting of summer interns. All Microsoft labs support this program, and it would be hard to find a researcher who did not consider it a rewarding effort in academic outreach.
While interns benefit from working with some of the best minds in computer science, researchers benefit just as much from the fresh perspectives their young colleagues bring. In fact, some Microsoft Research interns bring expertise from disciplines that have little to do with computer science—or so it might seem.
Patrick Meegan is pursuing a Master of Fine Arts degree at the University of Southern California’s School of Cinematic Arts. This made him an ideal candidate to work on a Rich Interactive Narratives (RIN) project with Microsoft Research India.
“The beauty of RIN,” Meegan says, “is that the technology hides in the shadows and lets the content shine, so my focus has been on the nature of digital storytelling itself. My film-editing experience has proven invaluable, since the RIN authoring tool is similar to non-linear video-editing software.”
Meegan’s RIN production, an original interactive story called We Bring Fire, incorporates live-action video, super-slow-motion video, panoramas, documents, original music, and drawings. Like his mentor, Joseph Joy, Meegan believes this new medium requires a different approach to deliver interactivity that feels natural and inherent to the story.
“To this end, Joseph pushes all of us on the content side to develop our unique visions,” Meegan says, “which, in turn, pushes the team’s developers to figure out technical solutions that transcend a specific creative project.”
For Meegan, a highlight of his internship has been the opportunity to work with so many talented individuals.
“Curtis Wong, who has grappled with questions about the deeper nature of storytelling ever since the earliest days of interactive digital media, shared some great insights,” Meegan says. “I also got help from Sing Bing Kang of the Interactive Visual Media Group and his intern, Eduardo Gastal. Then there was Research Media Works, our in-house video-production group, whose generosity with tools and support allowed me to devote my attention to the story I’m telling.
“RIN is a medium where you’re not bound to a script or a single linear plot,” Meegan reflects. “You’re not sure how the story is going to end until you finish. It reminds me of what our managing director, P. Anandan said, that in research, you can’t set out knowing exactly what you want to do. Then it wouldn’t be research.”
Iris Howley is studying ways to incorporate social factors into educational technologies. At Microsoft Research’s FUSE Labs, Howley, a Ph.D. candidate in Human-Computer Interaction at Carnegie Mellon University, is working on a project complementary to her interests, one that makes use of So.cl, an interest- and knowledge-sharing social-networking website.
“My research goal is to help users overcome obstacles to effective participation in the community,” Howley says. “My mentor, Todd Newman, and I wanted a project that would fit my overarching research goal, but also provide valuable knowledge to the So.cl team and external research communities. The project examines how people share knowledge, who interacts with what knowledge, and ways to connect users to the content they're most likely to respond to. It’s fascinating!”
Howley’s past research experience provided her with a tool set for examining language and behavior, which she found extremely useful for better understanding So.cl community participation and knowledge sharing.
She finds it interesting to be on a team that maintains a social-networking website for public use.
“There are pressures on and structures within the team that do not exist in smaller, academic research teams,” Howley explains. “Having questions answered, acquiring data, and seeking feedback all require adjusting to a different process.”
Outside the lab, Howley considers life in Seattle a highlight of her time as an intern.
“There's nothing quite like watching seaplanes take off from the park down the street or picking blackberries from the side of the road as you bike to the local beach.”
What does it really mean for online content to “go viral”? Can this intuitive concept be made mathematically precise? These are the questions that keep Ashton Anderson busy during his internship at Microsoft Research New York City. A Ph.D. candidate in Computer Science at Stanford University, Anderson is working with mentor Sharad Goel, as well as Jake Hofman and Duncan Watts, to examine the “greatest hits” of online content—the most popular news stories, images, videos, and more.
“People have this idea that content diffuses through social ties, much like a virus.” Anderson says. “But in previous work, Sharad and Duncan found that for most content, this almost never happens. It's much more likely that most people get exposed to it from a major hub, like a news website.”
It's still possible, though, that some popular pieces of content—the “greatest hits”—spread virally. Anderson’s project explores this possibility by analyzing the structure of the spreading patterns of the most popular stories, videos, and images. After first nailing down a precise definition of virality and identifying examples of truly viral content, the next step will be trying to find which features of content are more likely to lead to viral spread.
“I love basically everything about my internship,” Anderson says, when asked about highlights. “My research group is absolutely top-notch, and they're also very fun to be around! There's a wonderful atmosphere that supports both stimulating conversations and hilarious banter. And I'm really loving spending my summer in New York City. What could be nicer?”
Bei "Penny" Pan studies traffic. A graduate student from the University of Southern California, she was working on her Ph.D. researching Los Angeles traffic when she met her mentor, Yu Zheng, during a conference and discovered that he was studying traffic in her hometown, Beijing. Both saw an opportunity at Microsoft Research Asia for Pan to leverage her experience with Los Angeles traffic to study the situation in Beijing.
“I’m working on the detection of anomalous traffic events in a city using GPS trajectories generated by a large number of taxicabs,” Pan says. “The event could be anything unusual, such as accidents, broken traffic lights, or even disasters like floods or road collapses. The insight is that GPS-equipped taxis can be regarded as mobile sensors, constantly probing the traffic rhythm of a city. This work is different from other studies because it concerns events that lead a large number of drivers to alter their regular behaviors, rather than those that alter the behaviors of only a few drivers.”
Armed with domain knowledge of how a traffic accident evolves, it has been easy for Pan to interpret irregular behaviors when an event happens and to derive more interesting insights from the data.
“Ultimately, we want to create an online event-detection system,” Pan explains, “so that when anomalous events happen, the system can report events and send out real-time alerts and suggest alternative routes for drivers. This system would benefit not only individual drivers by providing them more efficient driving experiences, but also governments and transportation agencies, who need to monitor urban traffic and develop more effective policies and plans.”
What has been the highlight of Pan’s internship?
“The work itself is a great motivation for me,” she says. “In graduate school, we are always aiming at publishing papers, but here, our goal is far more than that. We want our achievements to enhance people’s lives.”
Vasilis Syrgkanis, a Cornell University Ph.D. student in Computer Science, is fascinated by online markets such as auction marketplaces, crowdsourcing markets, and ad markets. He formulates theoretical models of such markets to study their efficiency—and, potentially, to develop ideas on how to make small functional modifications to improve efficiency.
He couldn’t be happier with his time at Microsoft Research New England. With the help of manager Jennifer Chayes, Christian Borgs, other researchers, and fellow interns, Syrgkanis has been working on a number of projects focused at the intersection of economics and computer science.
Syrgkanis admits that the lab’s vibrant collaborative environment, interesting topics, and availability of talented researchers and interns made it tough keeping to just one research project. His main project has been studying how to design incentive mechanisms in a way that resists manipulation.
“Consider a company that hopes to do viral marketing through free product offers on social media,” Syrgkanis explains. “They want to make the offer to users whose behaviors indicate they are good potential brand advocates. The challenge is to identify the most valuable users via the users’ posts, discussions, status updates, and chats. However, just knowing such an offer exists could lead a user to alter her or his normal behavior to win the freebie. Therefore, we need to examine how one should create offers in a way that doesn’t distort the user’s authentic, normal behavior.”
Syrgkanis feels positive about the contribution he has made, thanks to his experience in analyzing the efficiency of systems that cope with “selfish” individuals who try to maximize personal gain at the expense of an overall social goal.
“My knowledge in this area proved helpful in many of my projects at Microsoft Research,” he says, “and I also managed to use some of the techniques I had developed in my thesis research.”
Jessica Lingel is working towards a Ph.D. in Library and Information Science at Rutgers University. This domain expertise has been put to good use at Microsoft Research Cambridge, where Lingel has been guided by mentor Tim Regan.
“Tim was primarily interested in how the tools used to build software shape the actual products that get developed,” Lingel says, “as well as how notions of tool use and craft are, or aren’t, apparent in current paradigms of education in computer science. This research has been really interesting in turning up connections between something we generally associate as very old—craft—and something we associate as very new—software. Putting software development, as a learned set of skills that relies on a set of tools, into the context of craft is helping us tackle some core assumptions about technology, ownership, and collaboration.”
Lingel interviewed software engineers about tools they have used for a long time and asked them to bring in actual examples of code from different parts of their histories as software developers. Her experience at Rutgers, which focuses on information practices and technological appropriations at the level of community, proved especially valuable.
“I used similar approaches of layering different interviewing strategies that range from focus groups to cognitive mapping to interviews,” Lingel explains, “to unpack the different phenomena related to information, technology, and everyday life.”
Lingel enjoys the welcoming, supportive environment at Microsoft Research and the freedom to take ownership of her project.
“That was great, and I felt like I was contributing a different set of methodological experiences and theoretical frameworks to this project, as well as sense of connection to craft as a practice.”