The broad goal of my research is to enable the
development of inexpensive natural language applications that can be
customized by normal users (i.e., not language engineers) to their
domain of interest.
Natural language applications include both
applications that interact with users using human language, and
applications that help users manage information encoded in human
language. Examples include information retrieval and extraction,
question answering, and help systems.
In pursuit of this goal, I use machine learning
methodology as an alternative to hand-construction. This methodology
has benefits similar to those of high-level programming languages --
namely, concise specification, higher level of abstraction and greater