Improving the generation of personalised descriptions

Thiago Castro Ferreira, Ivandré Paraboni


Abstract
Referring expression generation (REG) models that use speaker-dependent information require a considerable amount of training data produced by every individual speaker, or may otherwise perform poorly. In this work we propose a simple personalised method for this task, in which speakers are grouped into profiles according to their referential behaviour. Intrinsic evaluation shows that the use of speaker’s profiles generally outperforms the personalised method found in previous work.
Anthology ID:
W17-3536
Volume:
Proceedings of the 10th International Conference on Natural Language Generation
Month:
September
Year:
2017
Address:
Santiago de Compostela, Spain
Editors:
Jose M. Alonso, Alberto Bugarín, Ehud Reiter
Venue:
INLG
SIG:
SIGGEN
Publisher:
Association for Computational Linguistics
Note:
Pages:
233–237
Language:
URL:
https://aclanthology.org/W17-3536
DOI:
10.18653/v1/W17-3536
Bibkey:
Cite (ACL):
Thiago Castro Ferreira and Ivandré Paraboni. 2017. Improving the generation of personalised descriptions. In Proceedings of the 10th International Conference on Natural Language Generation, pages 233–237, Santiago de Compostela, Spain. Association for Computational Linguistics.
Cite (Informal):
Improving the generation of personalised descriptions (Castro Ferreira & Paraboni, INLG 2017)
Copy Citation:
PDF:
https://aclanthology.org/W17-3536.pdf