Rule-based vs. probabilistic surface realisation of definite description

Authors:
Francis Marques V. dos Santos;Daniel Bastos Pereira;Ivandré Paraboni
Affiliations:
University of São Paulo (EACH), São Paulo, Brazil;University of São Paulo (EACH), São Paulo, Brazil;University of São Paulo (EACH), São Paulo, Brazil
Venue:
Companion Proceedings of the XIV Brazilian Symposium on Multimedia and the Web
Year:
2008

Citing 4
Cited 0

Building natural language generation systems

Building natural language generation systems
Generation that exploits corpus-based statistical knowledge

COLING '98 Proceedings of the 17th international conference on Computational linguistics - Volume 1
Evaluating algorithms for the generation of referring expressions using a balanced corpus

ENLG '07 Proceedings of the Eleventh European Workshop on Natural Language Generation
Building a semantically transparent corpus for the generation of referring expressions

INLG '06 Proceedings of the Fourth International Natural Language Generation Conference

Quantified Score

Hi-index	0.00

Visualization

Abstract

We describe the evaluation work of two standard approaches to the surface realisation of definite descriptions as Portuguese text. Taking as an input a non-linguistic representation of the description to be generated, a rule-based approach makes use of grammar constraints to compute the appropriate surface string, whereas a competing probabilistic model applies n-gram statistics to the same task. The results of both systems are compared against a corpus of human-produced descriptions and their advantages are discussed.