JAYCOR Work in Support of the Navy Center for Applied Research in Artificial Intelligence
Abstract
The purpose of the TERSE project has been to develop a natural language system that can perform the function of highlighting important information in Navy CASREPs. This task involves research in many different areas, all dependent on one another. The most central issue in designing ths system is knowledge representation. This issue is central because the design of other portions of the system will be effected by the choice of knowledge representation. For example, the system which analyzes the raw natural language text must be designed in such a way that it can transform its input into this knowledge representation. The application component is also tightly coupled with the representation because it looks for information contained in it and makes inferences from that information. A knowledge representation can have a great effect on the ease with which the system can be defined and maintained. A knowledge representation is not just the set of data structures designed to hold knowledge for an application. A knowledge representation is a combination of data structures and interpretive procedures that, if used in the right way in a program, will lead to knowledgeable behavior. It is important that the representation exhibit the ability to capture this heterogeneous information in such a way that makes it easy to handle and understand for the human designers because in large representation systems, things can quickly get unmanageable. It is not our goal to represent everything that a human expert who would read these messages might understand. It has been our effort to only capture the meaning of the text at a level that facilitates our particular application. Keywords: Text processing; Natural language.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 31, 1987
- Accession Number
- ADA196108
Entities
People
- Kathryn Di Benigno