Summary: Building a Machine Learning Based Text Understanding System
Stephen G. Soderland
Department of Radiology, UCLA
Los Angeles, CA 90024
Text understanding systems are approaching the
point of being a practical technology as long as the
system is trained for a narrowly defined domain.
Machine learning and statistical approaches can
minimize the effort involved in adapting a text
understanding system to a new domain.
This paper presents a system whose goal is deep
understanding, limited only by the necessity of
designing a formal representation of the target
concepts relevant to the domain. This system is an
advance over previous machine learning based
systems because of its richer output representation,
and an advance over equally expressive text
understanding systems because of its more
extensive use of machine learning.