UGURU: a natural language UNIX consultant

Show full item record

Title: UGURU: a natural language UNIX consultant
Author: Hanson, John
Abstract: UGURU is a natural language conversation program, implemented in Prolog, which can manage a wide knowledge base of facts about Unix. The range and wording of questions that it understands are based on surveys taken of students, mostly Unix beginners. UGURU is also designed to accept statements in English that can be added as facts to the knowledge base. Each fact is represented as a "binding set:" a verb-oriented semantic net with the characteristics of directed acyclic graphs. The main actions taken by UGURU are divided between two primary modules, a parser and a retriever. To produce a binding set from an input, the parser incorporates a new kind of object-oriented grammar of several levels, parallel tracing of distinct parse trees by independent units called recognizers, the concurrent use of both syntactic and semantic knowledge, and a "pragmatic criterion" that requires the system to mimic the sequence of human parsing. The retriever, invoked to answer input questions, seeks to match the binding set representing the question to a fact in the knowledge base by performing semantic transformations on the two sets.
Record URI: http://hdl.handle.net/1850/10133
Date: 1987

Files in this item

Files Size Format View
JHansonThesis02-22-1987.pdf 2.747Mb PDF View/Open

The following license files are associated with this item:

This item appears in the following Collection(s)

Show full item record

Search RIT DML


Advanced Search

Browse