COLING'94: ABSTRACT of the Tutorials

Session 1 9:30-12:30, August 2 (Tuesday)

Theme:
Parsing Technologies: the Right Tool for the Right Problem.
Lecturer:
Yves Schabes
Mitsubishi Electric Research Laboratories, Cambridge, USA

Abstract

This tutorial will give an overview of the most recent parsing technologies and their applications. The tutorial describes the main algorithms and their applications for various classes of formalisms such as finite state transducers for morphological analysis, finite state grammars for part of speech disambiguation, local grammars for noun-phrase spotting, and more traditional grammars such as context-free grammars and lexicalized context-free grammars for higher level applications. We will also illustrate how parsing technologies have become more robust and accurate by accommodating both linguistic description and statistical data.

Discussion of the design and implementation of a sample of useful software will be featured.

Session 2 14:00-17:00, August 2 (Tuesday)

Title:
Natural language generation - at the Crossroads of Cognitive Science and Engineering
Lecturer:
Michael Zock
Scientique Laboratoire d'Informatique pour la Mecanique

Abstract

Natural language generation is a dynamic field that lies at the crossroads of various disciplines: linguistics, psychology, rhetorics, computer science (to name just those). What is striking is the fact that, despite its inherent interdisciplinarity, most scholars largely ignore the work of the neighbouring disciplines.

The goal of this tutorial is twofold:

The following issues will be addressed : The tutorial will be organized in a way similar to most generators: we will start from deep generation (determination of content, planning of text structure) and then move on to surface generation (lexical choice, syntactic structure).

Special emphasis will be given to the fact that the problem of natural language generation can only be solved in the realm of cognitive science, that is, withing a framework where linguists, psychologists and computer scientists meet and agree to work in concert.

Session 3 9:30-12:30, August 3 (Wendsday)

Theme:
Future Directions of MT -- Language, Meanings and Translation
Lecturer:
Jun'ichi Tsujii
Professor of UMIST, UK

Abstract

The tutorial gives a summary of recent developments of various paradigms of MT, including Linguistics-based MT, Knowledge-based MT, Example-based MT and Statistics-based MT. The comparison of these paradigms reveal different conceptions of the relationship between translation and understanding which these paradigms maintain. The turtorial also gives a thorough classification of difficulties actual MT systems encounter and suggests possible ways of resolving them. It also discusses possible MT architectures in which various MT paradigms can properly be integrated.

Session 4 14:00-17:00, August 3 (Wendsday)

Theme:
Corpus Based Natural Languge Processing
Lecturer:
Mitchell Marcus
Professor of University of Pennsylvania, USA

Abstract

A dramatic shift of research paradigm has taken place within NLP during the past several years. Probabilistic and information-theoretic techniques are increasingly being combined with traditional methods of symbolic computing, automatically utilizing the implicit and explicit lingustic information in newly-developed on-line corpora of both text and transcribed speech. Many of these data bases add rich linguistic annotation to corpora of natural language text. Already, these methods have led to a dramatic improvement in the performance of a variety of NLP systems. This tutorial will focus on these trends, surveying the application of empirical corpus-based techniques to a range of problems, including such areas as part of speech determination, syntactic parsing, lexical semantics, and discourse structure. It will also discuss the development of annotated corpora themselves.

Session 5 9:30-12:30, August 4 (Thursday)

Theme:
Towards Intelligent Multimedia Interfaces
Lecturer:
Mark Maybury
MITRE, USA

Abstract

Multimedia communication is ubiquitous in daily life. When we converse with one another, we utilize a wide array of media to interact, including spoken language, gestures, and drawings. We exploit multiple sensory systems or modes of communication including vision, audition, and taction. Although humans have a natural facility for managing and exploiting multiple input and output media, computers do not. Consequently, providing machines with the ability to interpret multimedia input and generate coordinated multimedia output would be a valuable facility for a number of key application such as information retrieval and analysis, training, and decision support. This tutorial focuses specifically on those intelligent interfaces that exploit multiple media and modes to facilitate human-computer communication. The purpose of this tutorial is to introduce the emerging literature and set of techniques for building multimedia and multimodal interfaces, i.e., those interfaces that interpret and generate multiple media, e.g., spoken and written natural language, graphics, non-speech audio, maps, animation.

Session 6 14:00-17:00, August 4 (Thursday)

Theme:
Current Topics in Pragmatics
Lecturer:
Yorick Wilks
Professor of University of Sheffield, UK

Abstract

The tutorial will survey and discuss some recent developments in pragmatics for natural language processing: in particular belief representationa nd manipulation systems for modeling the comprehension and generation of dialogue. We shall also discuss developments in the segmentation and cohererence of discourse, and some attempts to model phenomena such as lexical disambiguation and pronoun resolution within such frameworks.

________________________________________________________________________

shimazu@atom.ntt.jp (Akira Shimazu)

[Back to COLING'94 Home Page]