Boolean Semantics for Natural Language

Abstract

This report aims to identify the areas where NLP (natural language processing) is expected to be successful, and to suggest how people and organisations can make use of the opportunities it will create, as users, suppliers or investors.(p.4) Tim Johnson has written a very useful guide to the commercial side of NLP applications.It is a large, comprehensive report that runs 459 double-spaced pages.Johnson is a graduate of imperial College and has previously written a report on Expert Systems (July 1984), so he is familiar with work in other areas of machine intelligence.This report is better technically than one might expect from a layman.At $395 it is probably most appropriate for managers or researchers who are actively involved in the marketplace and for institutional or library copies.The major sections of the report are Management Summary, Markets, Technology, Applications, and Company Profiles.The first two sections are of primary interest either to those who are unfamiliar with the major application areas of NLP (described as mainframe and micro database interfaces, dialogue interfaces, content scanning, text editing, machine translation, and talkwriter) or with the business side of projected markets.For example, total US market projections are given as $15M in 1985, $420M in 1990, and $1500M in 1995.These projections are also broken down by application area.For active researchers in NLP, the technology section will not contain any major surprises.A very brief tour of approaches in syntactic and semantic parsing is followed by a review of current systems in both the mainframe and micro markets.These include Intellect, Ramis II English, Plume, Themis, Easytalk, Clout, NaturalLink, Savvy, Microdata Natural Language, Safeguard Cash-Management System, Logos's Intelligent Translator, ALPS Computer Translation System, Smart Translator, and Weidner MicroCAT.For each of these systems, there is a brief discussion of the product and a summary table that lists pertinent data such as availability, price, computer requirements, implementation language, opportunities for customizing, the software interface, the underlying NL technology, input requirements (menu-driven, ill-formed input, spelling correction), vocabulary size, and dialog management facilities (e.g., ellipsis and anaphora).