00100	
00200	
00300	
00400			TABLE OF CONTENTS
00500	
00600	INTRODUCTION					     ?
00700		The problem of recognizing natural language
00800		dialogue expressions in real time. Previous
00900		approaches. The problem of simulating paranoid
01000		linguistic behavior in a psychiatric interview.
01100		Summary of a method for transforming natural
01200		language input expressions until a pattern
01300		is obtained which completely or fuzzily
01400		matches a more abstract stored pattern.
01500	
01600	
01700	OVERVIEW					 ?
01800		The paranoid model (PARRY2) consists of a
01900		RECOGNIZE module which performs the task of
02000		recognizing the input and a RESPOND module
02100		which decides how to respond. The RECOGNIZE
02200		module functions independently of the RESPOND
02300		module except in the case of anaphoric
02400		references which it provides on request from
02500		the language recognizer.
02600	
02700	
02800	PREPROCESSING					  ?
02900		Dictionary lookup and translations. How
03000		misspellings and typing errors are handled.
03100	
03200	
03300	SEGMENTING					  ?
03400		Bracketing the pattern into shorter segments.
03500		A "simple" pattern contains no delimiters; a
03600		"compound" pattern is made up of two or more
03700		simple patterns.
03800	
03900	
04000	MATCHING INDIVIDUAL SEGMENTS                      ?
04100		Negations and anaphora. Matching the pattern
04200		with stored patterns having pointers to
04300		response functions in memory. If a complete
04400		match is not found, a fuzzy match is attempted
04500		by deleting elements from the pattern one at
04600		time. If no match is found, the RESPOND module
04700		must decide what to do.
04800	
04900	
05000	COMPOUND-PATTERN MATCH                                ?
05100		Complete and fuzzy matching when the pattern
05200		contains two or more segments.
05300	
05400	
05500	ADVANTAGES AND LIMITATIONS                         ?
05600		The advantages of ignoring as irrelevant some
05700		of what is recognized and what is not recognized
05800		at all. The complete language recognition process
05900		of the algorithm requires less than one second
06000		of real time. How the data base "learns". The
06100		measurement of linguistic improvement.
06200	
06300	
06400	REFERENCES                                         ?
06500	
06600	
06700	APPENDIX 1                                          ?
06800		A diagnostic psychiatric interview which
06900		illustrates some of the model's linguistic
07000		capabilities.
07100	
07200	
07300	APPENDIX 2                                           ?
07400		A listing of the dictionary illustrating 
07500		the algorithm's recognizable input words and 
07600		the word class names they are translated into.
07700	
07800	
07900	APPENDIX 3                                           ?
08000		A listing of the simple patterns.