Graph-based generation of referring expressions
Computational Linguistics
A Reference Architecture for Natural Language Generation Systems
Natural Language Engineering
Intrinsic vs. extrinsic evaluation measures for referring expression generation
HLT-Short '08 Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics on Human Language Technologies: Short Papers
Trainable speaker-based referring expression generation
CoNLL '08 Proceedings of the Twelfth Conference on Computational Natural Language Learning
Evaluating algorithms for the generation of referring expressions using a balanced corpus
ENLG '07 Proceedings of the Eleventh European Workshop on Natural Language Generation
The TUNA-REG Challenge 2009: overview and evaluation results
ENLG '09 Proceedings of the 12th European Workshop on Natural Language Generation
Learning content selection rules for generating object descriptions in dialogue
Journal of Artificial Intelligence Research
Building a semantically transparent corpus for the generation of referring expressions
INLG '06 Proceedings of the Fourth International Natural Language Generation Conference
Using spatial reference frames to generate grounded textual summaries of georeferenced data
INLG '08 Proceedings of the Fifth International Natural Language Generation Conference
NIL-UCM: most-frequent-value-first attribute selection and best-scoring-choice realization
INLG '08 Proceedings of the Fifth International Natural Language Generation Conference
Does size matter: how much data is required to train a REG algorithm?
HLT '11 Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies: short papers - Volume 2
| Hi-index | 0.00 |
One important subtask of Referring Expression Generation (REG) algorithms is to select the attributes in a definite description for a given object. In this paper, we study how much training data is required for algorithms to do this properly. We compare two REG algorithms in terms of their performance: the classic Incremental Algorithm and the more recent Graph algorithm. Both rely on a notion of preferred attributes that can be learned from human descriptions. In our experiments, preferences are learned from training sets that vary in size, in two domains and languages. The results show that depending on the algorithm and the complexity of the domain, training on a handful of descriptions can already lead to a performance that is not significantly different from training on a much larger data set.