WHO:	Big Picture Seminar: Dan Gildea
TOPIC:	Syntax for Statistical Machine Translation
ABSTRACT:	I will present the work done at this year's Johns Hopkins summer research workshop on machine translation. In recent evaluations of machine translation systems, statistical systems based on probabilistic models have outperformed classical approaches base d on interpretation, transfer, and generation. Nonetheless, the output of statistical systems often contains obvious grammatical errors. Our group aimed to integrate syntactic structure into statistical models to address this problem. I will discuss th e types of syntactic features developed and the rescoring framework with which they were integrated into the baseline translation system.››à›
WHEN:	9/26/2003 12:30:00 PM
WHERE:	Computer Science Building 703

Events Homepage