Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Generating Definite Descriptions NonIncrementality, Inference, and Data
 

Summary: Generating Definite Descriptions
Non­Incrementality, Inference, and Data
Claire Gardent, Hélène Manuélian,
Kristina Striegnitz, Marilisa Amoia
1. Introduction
The generation of referring expressions is a central task for systems that
automatically generate natural language texts. Indeed, the correct use of
natural language referential devices is crucial for generating successful
utterances, i.e., utterances that are easily and correctly understood by the
hearer, because referring expressions play an important role in linking an
utterance to the previous discourse, the non­linguistic situation the utterance
is produced in, and the knowledge of speaker and hearer.
One algorithm that has been particularly influential in this field is the
incremental algorithm for generating referring expressions presented in (Dale
and Reiter 1995).
In this paper, we both extend this basic algorithm to deal with more
complex, inference based, definite descriptions and propose an alternative,
non­incremental algorithm which circumvents the shortcomings arising from
incrementality. Moreover, we present the results of several corpus studies on
definite descriptions in French which suggest some research directions that

  

Source: Amoia, Marilisa - Equipe Talaris, INRIA-Lorraine

 

Collections: Computer Technologies and Information Sciences