When: Thursday, May 08, 3:00 p.m.- 4:00 p.m.

Where: 1305Newell-Simon Hall

Jimmy Lin, University of Maryland

LTI Seminar

Abstract:
Over the past couple of decades, the field of computational linguistics, and more broadly, text processing, has seen the emergence and later dominance of empirical techniques and data-driven research. An impediment to research progress today is the need for scalable algorithms to cope with the vast quantities of available data. Recently, MapReduce has emerged as an attractive alternative to traditional parallel programming models for developing distributed applications. It provides a simple functional abstraction that hides many system-level issues, allowing the researcher to focus on actually solving the problem. At least for text processing, MapReduce has the potential to provide solutions that are fast (in terms of running time), easy (in terms of implementation), and cheap (in terms of hardware costs). In this talk, I will overview several "cloud computing" efforts at the University of Maryland, one of the pilot institutions in the Academic Cloud Computing Initiative sponsored by Google and IBM. Ongoing projects include a course that brings undergraduates and graduates together to work on open research problems. Project topics include statistical machine translation, email analysis, image processing, and biological sequence alignment.

<< Back