Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Rapid Incremental Parsing with Repair Steven P. Abney
 

Summary: Rapid Incremental Parsing with Repair
Steven P. Abney
1990
Abstract
This work describes a method of achieving rapid, reliable parsing
of natural text through the application of three techniques: (1) resolv-
ing small questions sequentially, (2) repairing errors directly, instead of
searching through a non-deterministic space, and (3) recognizing major
constituents before analyzing the details of their internal structure. The
resulting parser, which I call CASS, is fast and accurate. It parses a mil-
lion words in 5-6 hours; that is as fast as the fastest parsers reported in
the literature. Its accuracy at recognizing chunks (mid-level constituents)
and at identifying subjects and predicates is 95% or better.
Introduction
Much of the information contained in online texts will remain inaccessible until
we have natural language parsers that are faster and more reliable than the
current state of the art. In the work described here I explore the possibility
of achieving rapid, reliable parsing of natural text by application of some sim-
ple techniques: (1) resolving small questions sequentially, (2) repairing errors
from earlier processing as they become apparent downstream, and (3) recog-

  

Source: Abney, Steven P. - School of Information, University of Michigan

 

Collections: Computer Technologies and Information Sciences