Computational evaluation of a noun-driven syntax representation
Abstract
This work explores a possible solution to an observed mismatch between a verb-centered syntactic analysis paradigm and noun-centered information sources. The latter may form not only the larger information stores online, access to which is improved by progress in natural language understanding systems but may also extend to most texts in general. Some basic categorization of texts along these lines provides testing classes by which to evaluate the performance of a noun-driven syntax representation. A syntax parser was developed and instrumented to provide analysis. The noun-driven representation matches or outperforms more traditional representations; results are discussed in conjunction with findings about the parser's performance and the adequacy of the text categories.
Degree
M.A.
Advisors
Raskin, Purdue University.
Subject Area
Linguistics|Information science|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.