LINGUIST List 8.738

Fri May 16 1997

Confs: Natural Language Processing

Editor for this issue: T. Daniel Seely <seelylinguistlist.org>

Directory

1. Gosse Bouma, Topics in Constraint-Based Natural Language Processing

			WORKSHOP ANNOUNCEMENT


	Topics in Constraint-Based Natural Language Processing 


SPEAKERS

	Suresh Manandhar, University of York
	Gosse Bouma, Groningen University
	Dale Gerdemann,	 Tuebingen University 
 	Thilo Goetz,		"
 	Gerald Penn,		" 
 	Guido Minnen,		" 
 	Shuly Wintner,		" 
 	Andreas Zahnert,	" (tentative)

COORDINATION

	Gosse Bouma, Groningen University
	Dale Gerdemann, Tuebingen University

HOST

	BCN summerschool 97 (June 30 - July 11, Groningen University) 

COURSE DESCRIPTION

Linguistic theories have increasingly been expressed in a declarative or
constraint-based style, which is highly appropriate for describing the
competence of native speakers. The grammar becomes a description of the
facts about a language, which is neutral with respect to any particular 
mode of processing. It is the burden of the computational linguist to find
particular strategies for using this declarative knowledge so that it can 
be used for performance oriented tasks such as recognizing, parsing or 
generation. 

In this course, we will examine processing techniques for such
declarative, constraint-based grammars. We will begin with parsing and 
generation techniques appropriate for simple unification-based grammars 
such as PATR-II and Definite Clause Grammars. We then will look at the 
motivation for extending these systems into typed feature structure-based 
systems such as the Troll system used extensively at the University 
of Tuebingen and the widely used ALE system. Various notions of typing will
be examined and compared from linguistic and processing points of view. 
Typed feature structures will lead us then into a discussion of more 
general constraint based grammars (since constraints are often attached to 
types). The constraints in such a system are simply descriptions which must 
hold true of every well-formed linguistic object. Unification in these 
grammars no longer plays any role, though it may still be used in an
implementation as a way of doing constraint solving. 

In the second part of the course, we will look at some newer compilation
techniques, which have been applied to typed feature structure grammars. In
particular, we will look at how Prolog-style abstract machines can be 
adapted for this domain. We will see that having types on feature 
structures is a great advantage, since it allows memory to be allocated 
only for appropriate features. A variety of optimizations of such machines 
will be explored. Finally, we will make a case study of one particular 
kind of constraint, namely lexical rules, where it has been argued that 
special purpose compilation is called for. We will see, for example, how 
finite state automata can be applied to model the possible feeding orders 
between such rules.

COURSE READING

 (the course will cover topics from the following)

 * Hassan Ait-Kaci, (1991), "Warren's Abstract Machine : A Tutorial
 Reconstruction", MIT Press
 * Bob Carpenter, (1992), "The Logic of Typed Feature Structures",
 Cambridge UP.
 * Bob Carpenter and Yan Qu, (1995), "An Abstract Machine for Attribute
 Value Logics. 4th IWPT.
 (http://macduff.andrew.cmu.edu/ftp/iwpt4.ps.Z )
 * Dale Gerdemann and Paul King, (1994), "The Correct and Efficient
 Implementation of Appropriateness Specifications for Typed Feature
 Structures", COLING.
 (http://www.sfs.nphil.uni-tuebingen.de/~dg/)
 * Thilo Goetz and Detmar Meurers, (1996), "The importance of being lazy -
 Using lazy evaluation to process queries to HPSG grammars", TALN 96.
 (http://www.sfs.nphil.uni-tuebingen.de/~dm/)
 * Thilo Goetz and Detmar Meurers, (1995) Compiling HPSG type constraints
 into definite clause program", ACL.
 (http://www.sfs.nphil.uni-tuebingen.de/~dm/)
 * Detmar Meurers and Guido Minnen, (1995) A Computational Treatment of
 HPSG Lexical Rules as Covariation in Lexical Entries", 5th NLULP.
 (http://www.sfs.nphil.uni-tuebingen.de/~dm/)
 * Shalom Wintner, (1997), An Abstract Machine for Unification Grammars,
 PhD Thesis, Univ. of Haifa
 (http://www.cs.technion.ac.il/~shuly/)


BACKGROUND READING

 (participants are expected to be familiar with the following)

 * Bob Carpenter and Gerald Penn, (1994), ALE 2.0 Users Guide.
 (http://macduff.andrew.cmu.edu/ale/)
 * Carl Pollard and Ivan Sag, (1994), "Head-Driven Phrase Structure
 Grammar", Chicago UP, CSLI series.

 Any book on NLP in Prolog. Examples are:

 * Gerald Gazdar and Chris Mellish, (1989), "Natural Language Processing
 in Prolog", Addison-Wesley.
 * Fernando Pereira and Stuart Shieber, (1987), "Prolog and Natural
 Language Analysis", Chicago UP, CSLI series.
 * Michael Covington, (1994), "Natural Language Processing for Prolog
 Programmers", Prentice Hall.

FURTHER INFORMATION

	The BCN Summerschool Web site 

	http://www.bcn.rug.nl/bcn/events/index.html

	gives an overview of other activities during the summer school, 
	and contains an electronic registration form (registration 
	deadline is June, 18).


This course is part of a cooperation program between Graduate School
BCN/Centre for Language and Cognition Groningen at the University of
Groningen and the Seminar fuer Sprachwissenshaft at the
Eberhard-Karls-Universitaet Tuebingen. The program is funded by NWO,
foundation for Language, Speech and Logic.


- 
Gosse Bouma, Alfa-informatica, RUG, Postbus 716, 9700 AS Groningen
gosselet.rug.nl tel. +31-50-3635937 fax +31-50-3636855