Probabilistic Refinement Algorithms for the Generation of Referring Expressions

R. Altamirano, C. Areces, and L. Benotti. Probabilistic Refinement Algorithms for the Generation of Referring Expressions. In Proceedings of the 24th International Conference on Computational Linguistics (COLING 2012), pp. 53–62, Mumbai, India, 2012.

Download

[pdf] 

Abstract

We propose an algorithm for the generation of referring expressions that adapts the approach of Areces et al. (2008, 2011) to include overspecification and probabilities learned from corpora. After introducing the algorithm, we discuss how probabilities required as input can be computed for any given domain for which a suitable corpus of REs is available, and how the probabilities can be adjusted for new scenes in the domain using a machine learning approach. We exemplify how to compute probabilities over the GRE3D7 corpus of Viethen (2011). The resulting algorithm is able to generate different referring expressions for the same target with a frequency similar to that observed in corpora. We empirically evaluate the new algorithm over the GRE3D7 corpus, and show that the probability distribution of the generated referring expressions match the one found in the corpus with high accuracy.

BibTeX

@InProceedings{Altamirano2012,
  author =       "R. Altamirano and C. Areces and L. Benotti",
  booktitle =    "Proceedings of the 24th International Conference on
                 Computational Linguistics (COLING 2012)",
  title =        "Probabilistic Refinement Algorithms for the Generation
                 of Referring Expressions",
  year =         "2012",
  address =      "Mumbai, India",
  pages =        "53--62",
  abstract =     "We propose an algorithm for the generation of
                 referring expressions that adapts the approach of
                 Areces et al. (2008, 2011) to include overspecification
                 and probabilities learned from corpora. After
                 introducing the algorithm, we discuss how probabilities
                 required as input can be computed for any given domain
                 for which a suitable corpus of REs is available, and
                 how the probabilities can be adjusted for new scenes in
                 the domain using a machine learning approach. We
                 exemplify how to compute probabilities over the GRE3D7
                 corpus of Viethen (2011). The resulting algorithm is
                 able to generate different referring expressions for
                 the same target with a frequency similar to that
                 observed in corpora. We empirically evaluate the new
                 algorithm over the GRE3D7 corpus, and show that the
                 probability distribution of the generated referring
                 expressions match the one found in the corpus with high
                 accuracy.",
  owner =        "areces",
  timestamp =    "2012.10.30",
  URL =          "http://www.aclweb.org/anthology/C12-2006",
}