REAL ENGLISH: A TRANSLATOR TO ENABLE NATURAL LANGUAGE MAN-MACHINE CONVERSATION
Abstract
This dissertation presents a pragmatic interpreter/translator called Real English to serve as a natural language man-machine communication interface in a multi-mode on-line information retrieval system. This multi-mode feature affords the user a library-like searching tool by giving him access to a dictionary, lexicon, thesaurus, synonym table, and classification tables expressing binary relations as well as the document file representing the field of discourse. The user is thereby allowed a greater freedom in search strategy. Real English will (1) syntactically analyze the user's message by means of a strong analysis grammar to produce a tree representing the interrelationships of the grammatical entities comprising the message, (2) use this tree together with a pragmatic grammar to establish the set of commands necessary to fulfill the request, and (3) form the proper syntax for each command. The strong linguistic foundation of the syntax analyzer endows the system with the power of flexibility. As experience shows that certain new structures occur and should therefore be a part of the system, they may be incorporated into the system by the grammarian without a major overhaul of the procedures to date. The user is permitted to phrase his requests in any convenient form (i.e. declarative, imperative, interrogative, or fragmented sentence referred to as conversationally dependent sentences). Thus, instead of placing the user in the difficult position of learning a new language, the system is given the responsibility of responding in and to the user's language, i.e. the man-machine conversation is carried out in a natural language.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1969
- Accession Number
- AD0689280
Entities
People
- Harvey Gautin
Organizations
- Moore School of Electrical Engineering