Download Research Tools
In conducting research, we often look to the past for answers. Today, at Stanford University, I had the opportunity to look to the future. This morning I watched four excellent presentations delivered by the teams of students enrolled in CS210, Project-Based Computer Science Innovation & Development; and this afternoon, I attended the class' fair, modeled after a trade show, where I was able to delve more deeply into each of the projects. If the inquisitiveness, passion and determination of the students I met today are any indication, the future of our profession is in very good hands.
The goal of CS210 is to provide computer science students with an opportunity to collaborate on a real-world project provided by a corporate partner. The challenge of the project Microsoft External Research handed over to the students was to make satellite data more accessible to environmental scientists. Specifically, Team Nimbus was tasked with reducing the costs, time and complexity associated with managing satellite images while at the same time improving the reliability of those images, which are often difficult to manipulate on a desktop.
The result of the team's work is CloudLab, which utilizes the Windows Azure platform to remove the heaviest work from the desktop and put it in the cloud, where there is far more computing power and accessibility. During the development of CloudLab, scientists at the Lawrence Berkeley National Laboratory served as the team's customers. For the students in the class, which was taught by Jay Borenstein, the benefits go far beyond a passing grade. Throughout the class, students gained practical insight into many applied aspects of computer science, such as source control and agile programming methodologies. By working on a real project with the potential to have an impact on industry, the students became better informed about what they may wish to pursue professionally.
Beyond the experience gained by the students, Stanford will use its up-close view of what's important throughout the industry to continue refining its academic offerings. For me, this collaboration effort provided the chance to get to know people whose names I'm confident will one day be familiar to us all. Finally, and most importantly, the experience is a compelling reminder, for all of us throughout the global research community, of how important it is to look at our work and all of its challenges through the perspectives of others as often as possible.
Dan Fay, director, Earth, Energy, and Environment, Microsoft External Research
"Software Engineering!? What do I know of computers and software?" So said Archbishop Tutu as he welcomed 700 computer scientists to the 32nd annual International Conference on Software Engineering (ICSE), held the first week of May in Cape Town, South Africa. The venerable icon, Nobel prize winner and champion of human rights went on to add "I am not so old or disconnected from the modern world that I don't realise that computers and software permeate every aspect of our modern lives and ... that technology can be enabling and can play a crucial role in raising educational standards, in improving the quality of life and in helping commerce and industry." It is these three aspects - education, life and industry - that the Software Engineering Innovation Foundation (SEIF) Awards were set up to address. And it was indeed fitting that they were presented at the ICSE conference where several of the 12 recipients were present.
Watch the introduction at ICSE from Archbishop Tutu
The SEIF Awards are a joint initiative between the Computer Science team of Microsoft External Research, the Research in Software Engineering Group (RiSE) and the Microsoft Visual Studio 2010 team. Microsoft aimed to partner with the worldwide academic community to advance the tools and technologies that are used in and with Visual Studio 2010. A total of 85 submissions were received and after a rigorous evaluation process, we chose 12 projects to support for a year. In May 2011, the community will get together at ICSE 2011 in Hawaii for a workshop to share the results of their work. The submitted projects came from all over the world. Of the winning projects, four were from North America, three each from Europe and South America and one each from China and India.
A strong focus of the work from RiSE is detecting, correcting and preventing bugs in code. Sunghun Kim of the Hong Kong University of Science and Technology proposes to first integrate a state-of-the-art bug prediction algorithm, called Change Classification, into Visual Studio 2010, so that developers will be quickly notified (using colored underlines) about code areas that are predicted to be buggy. Kim describes his method as seeming like a guardian angel looking over your shoulder, pointing out bugs and suggesting plausible fixes.
From left: Jane Prey (Director of Gender Diversity, MSR), Yuriy Brun and Reid Holmes (representing David Notkin), Alessandro Orso, Nachi Nagappan (MSR), Gail Murphy, Stefano Tonetta, Wolfram Schulte (Research Area Manager, RiSE Group), Karin Breitman, Guido de Caso (with Uchitel), Sebastien Uchitel
Closer to home, at the University of Washington, David Notkin and his students plan to explore the role speculation can play in software development. Their interest in speculation has been piqued both by the potential availability of using "cycles for quality" (e.g., from multi-core) and also by the need for breakthroughs in how environments augment the abilities of developers. Their goal is to warn a developer, as early as possible, that changes in code will conﬂict with the work of another developer on their team. By limiting or avoiding these conﬂicts, they hope to decrease the amount of time spent ﬁxing "broken" builds, enabling developers to focus more directly on their tasks rather than source-control management (SCM) problems. Implementing the approach will involve combining a simpliﬁed abstract model of SCM systems to derive a common interface that can manage the kinds of conﬂicts they would like to detect, and building speciﬁc SCM connectors (e.g., for Microsoft's Team Foundation Server, or CodePlex's Mercurial source control repository) that can interact with the variety of SCM systems used by real projects. Notkin and his students aim to build a user interface specifically for Visual Studio 2010, that can be used to alert developers when their actions conﬂict with other development trees.
In the education space, Pankaj Jalote at IIIT-D in Delhi is of the opinion that an introductory course in software engineering in a computer science program remains one of the hardest subjects to teach. While focusing on concepts and techniques is essential, there is often not sufficient time for the tools to be given the appropriate amount of attention. Real software engineering is now very tool intensive and a large set of tools is needed to cover the different aspects of the software development lifecycle - including requirements modeling, requirements documentation, project planning, design, coding related, code management and those related to testing. Generally a host of different tools might be used in a project, coming from a variety of sources. This project will develop a prototype course done almost fully within Visual Studio 2010, with its existing rich tool set, while also integrating other tools such as spreadsheets. Jalote maintains that using Visual Studio 2010 can have a major impact on the teaching: besides mastering the concepts and techniques, students will also learn through the use of a proper environment and tools how software is really developed, and will gain skills that are highly desired by the industry.
These are just some of the projects for 2010/11. For a full list of projects, as well as announcements for SEIF II stay tuned to our website and blog.
Judith Bishop, director, Computer Science, Microsoft External Research
Last week, at the Microsoft Research sixth annual Latin American Faculty Summit in Guaruja, Brazil, Rob Fatland, program manager with Microsoft Research, and Humberto da Rocha, professor in the Department of Atmospheric Sciences at the Universidade de São Paulo, led an intriguing presentation about their Atlantic Rainforest Micrometeorology Sensor Network Pilot Study. It’s a study that took place a mere 130 miles from where the Faculty Summit was held—a local project that could have broad environmental impact worldwide.
It all began about a year and a half ago when representatives from the Sao Paulo Research Foundation (FAPESP) and the National Institute for Space Research (INPE) asked Microsoft Research for assistance in developing a senor network with capabilities of operating under rainforest conditions. The inquiry led us to reach out to our colleagues at Johns Hopkins University, where a wireless sensor network program called Life Under Your Feet was previously developed, in collaboration with Microsoft Research, for soil-ecology research. So in October 2008, Microsoft Research brought the necessary experts and organizations together and this rainforest study was born.
The project was guided by three broad goals: first, to create a scientifically valid data set; second, to successfully engineer the project so that the technology could function fully in settings as challenging as a rainforest; and third, to develop a system that can serve as a model and be replicated in other research.
As a result of this close collaboration between researchers and scientists at Microsoft Research, Johns Hopkins University, and Universidad de São Paulo, the sensor network was deployed for four weeks in the rainforest, where it captured data every 30 seconds on the temperature, humidity and light throughout the canopy. In addition, a weather station recorded the staples of meteorology: rainfall, wind conditions and barometric pressure.
The result was that the team gathered an incredibly large and accurate data set that scientists are now analyzing to help them understand rainforest ecosystems. Additionally, the technology we’ve used in this study could be applied to a variety of situations across the globe, such as monitoring receding glaciers in the Arctic or measuring seismic activity for better earthquake predictions.
As researchers, we understand that our planet and our climate are undergoing change. Our challenge is that there’s still so much science can’t tell us—so many details that aren’t understood due to lack of data. This Atlantic rainforest project is one example of science and technology working together to understand a complex ecosystem.
Regardless of where we live, this type of research is a benefit for us all. For more information and to see a video of the Atlantic rainforest research site, you can visit our Atlantic Rainforest project page.