At ICDE 2018, VMware will present the 2017 early career Systems Research Award to Tim Kraska, Assistant Professor of Computer Science at Brown University, who will be joining MIT in the spring as an Associate Professor. Professor Kraska’s university will receive a gift of US $100,000 in support of his research on data management systems.
Tim Kraska, Assistant Professor of Computer Science at Brown University and VMware’s 2017 Systems Research Award Recipient
Celebrating holistic systems approach
Professor Kraska has been widely recognized for his early work on hybrid human-machine data management. “Tim was amongst the first in the database community to recognize and realize the potential of crowdsourcing platforms for data processing tasks”, says Ugur Cetintemel, computer science department chair at Brown. A key underpinning of this research is an observation that humans are often needed to resolve queries that database systems and search engines are unable to address in isolation. Tim and his colleagues have contributed substantially to progress in human-machine data management with implementations, theory, and applications. On the systems side, his work includes a pioneering reference architecture (CrowdDB) for hybrid crowdsourced queries. In terms of theory, Tim developed semantics and statistical methods for analyzing crowdsourced queries in distinguished papers on “Crowdsourced Enumeration Queries” and “Estimating the Impact of Unknown Unknowns”. Tim has also explored multiple applications of these ideas, for example developing crowd-sourced solutions to challenges in “big data” integration and cleaning. One of the purposes of the VMware Systems Research Award is to celebrate the style of holistic systems treatment that Tim and his collaborators have role-modeled so effectively with that body of early research.
A powerful new vision for the future of database systems
Over the past four years at Brown, Tim has continued to role-model this holistic treatment by formulating and tackling research problems that together represent a powerful new vision for the future of database systems. The theme of human-computer interaction, which is likely to be so pivotal in the era of pervasive machine learning and artificial intelligence, has been an important element of Professor Kraska’s work. For example, Vizdom supports interactive visualization and exploration of datasets through pen and touch. Tim’s recent project “Quantifying Uncertainty in Data Exploration” (QUDE) recognizes the human traps and pitfalls inherent in data exploration and recommender systems, and he is formulating new ways to quantify and counteract the risk of misleading “false positive” data correlations (sometimes called “p-hacking”). Based on the new requirements of human-in-the-loop data management and processing, Tim and his colleagues are also putting the underlying systems on new foundations.
One compelling example is Tupleware, an end-to-end data processing system that can achieve up to three orders of magnitude performance improvement over alternative systems like Spark or Hadoop. Tupleware was amongst the first systems to exploit modern “just-in-time” compilation techniques to efficiently implement user-defined queries; the approach has been influential in both industry and academia. Another example is “Network Attached Memory” (NAM) which recognizes that traditional database architectures are unable to fully exploit emerging technologies like RDMA. RDMA technology overturns the assumption that networking is the bottleneck in query processing; the NAM architecture separates storage from processing to achieve nearly two orders of magnitude performance improvement. Says Professor Sam Madden of MIT, “I expect that most high-performance database systems in the future will adopt a similar architecture”.
VMware’s System Research Award highlights early career faculty who have made exceptional research contributions to one or more emerging areas of Computer Science, and who have achieved notable impact with their work. This year, the selection committee was chaired by Professor Mike Stonebraker (MIT) and included Professor Greg Ganger (CMU), Professor James Larus (EPFL), Chris Ramming (VMware), Professor Jennifer Rexford (Princeton), Pratap Subrahmanyam (VMware) and Raj Yavatkar (VMware emeritus). VMware sees the award as one way to support and give back to the academic research community, which plays a crucial role in exploring new technology.
–Chris Ramming, Senior Director, VMware Research & Innovation