Issues in the TERSE Project
Abstract
This work is part of the TERSE project--an effort to understand and make use of information in free text portions of Navy CASREP reports. Naval ships are required to send out a CASREP report on every piece of equipment which experiences a failure that can not be corrected within 24 hours. The Navy is faced with several problems concerning these messages: 1) The shear number of messages coming in; 2) Sorting out time critical information from non-time critical information; 3) Forwarding relevant information to different organizations. Because CASREP reports also contain formatted sections, some information detection and dissemination can be done fairly easily with computers. The task of processing the information contained in the narrative portions of the messages is more difficult. Our goal has been to parse the narrative text and map the information transmitted in that section into a computer-based representation. After this mapping is accomplished, specialized application programs can access the knowledge base containing the information and perform their specific tasks. This paper deals with knowledge representation as it related to the problems of representing information communicated in natural language text. It discusses transitioning this project to the real world and potential problems arising from the past approach to knowledge representation.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 12, 1988
- Accession Number
- ADA196145
Entities
People
- M. K. Di Benigno