Automatic Processing of Navy Message Narrative.
NAVAL RESEARCH LAB WASHINGTON DC
Pagination or Media Count:
This report describes an experimental system that employs techniques of computational linguistics and artificial intelligence to extract information from Navy messages. The long-term goal of this work is to develop capabilities that will enable systems to handle a broad spectrum of messages, from highly formatted messages with little English description to messages consisting entirely of English narrative. In this report, an experimental system that takes the first step towards automated understanding of Navy messages was constructed and implemented. The system extracts informational content from reports about shipboard equipment failure and uses the content to assign a distribution list to each message and to generate a summary of the equipment failure. The system can be extended to other Navy message types and used for other military applications. This approach, called information formatting, uses an explicit grammar of English and a classification of the semantic relationships within the domain to derive a tabular representation of the information in a message narrative. Prototype knowledge bases for distribution and summarization were implemented in OPS5 and use the information format to automatically construct a distribution list and summary. A comparison of computer-generated summaries with those obtained manually showed good agreement, indicating that it is possible to automatically process message narrative and generate appropriate, and ultimately useful, summaries. Keywords Natural language understanding Knowledge representation Production rule systems.