GRAPH: The Costs of Redundancy in Referring Expressions

Emiel Krahmer; Mariet Theune; Jette Viethen; Iris Hendrickx

GRAPH: The Costs of Redundancy in Referring Expressions

Emiel Krahmer, Mariet Theune, Jette Viethen, Iris Hendrickx

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic

10 Citations (Scopus)

48 Downloads (Pure)

Abstract

We describe a graph-based generation system that participated in the TUNA attribute selection and realisation task of the REG 2008 Challenge. Using a stochastic cost function (with certain properties for free), and trying attributes from cheapest to more expensive, the system achieves overall .76 DICE and .54 MASI scores for attribute selection on the development set. For realisation, it turns out that in some cases higher attribute selection accuracy leads to larger differences between system-generated and human descriptions.

Original language	Undefined
Title of host publication	Proceedings of the 5th International Natural Language Generation Conference (INLG 2008)
Place of Publication	USA
Publisher	Association for Computational Linguistics (ACL)
Pages	227-229
Number of pages	3
ISBN (Print)	not assigned
Publication status	Published - Jun 2008
Event	5th International Natural Language Generation Conference, INLG 2008 - Salt Fork, Ohio, USA Duration: 12 Jun 2008 → 14 Jun 2008

Publication series

Name
Publisher	The Association for Computational Linguistics
Number	DTR08-9

Conference

Conference	5th International Natural Language Generation Conference, INLG 2008
Period	12/06/08 → 14/06/08
Other	12-14 June 2008

Keywords

EWI-13092
IR-64885
METIS-251092

Access to Document

Cite this

@inproceedings{5d43f41eef5045a1b1f6bdeeac057c66,

title = "GRAPH: The Costs of Redundancy in Referring Expressions",

abstract = "We describe a graph-based generation system that participated in the TUNA attribute selection and realisation task of the REG 2008 Challenge. Using a stochastic cost function (with certain properties for free), and trying attributes from cheapest to more expensive, the system achieves overall .76 DICE and .54 MASI scores for attribute selection on the development set. For realisation, it turns out that in some cases higher attribute selection accuracy leads to larger differences between system-generated and human descriptions.",

keywords = "EWI-13092, IR-64885, METIS-251092",

author = "Emiel Krahmer and Mariet Theune and Jette Viethen and Iris Hendrickx",

year = "2008",

month = jun,

language = "Undefined",

isbn = "not assigned",

publisher = "Association for Computational Linguistics (ACL)",

number = "DTR08-9",

pages = "227--229",

booktitle = "Proceedings of the 5th International Natural Language Generation Conference (INLG 2008)",

address = "United States",

note = "5th International Natural Language Generation Conference, INLG 2008 ; Conference date: 12-06-2008 Through 14-06-2008",

}

GRAPH: The Costs of Redundancy in Referring Expressions. / Krahmer, Emiel; Theune, Mariet; Viethen, Jette et al.
Proceedings of the 5th International Natural Language Generation Conference (INLG 2008). USA: Association for Computational Linguistics (ACL), 2008. p. 227-229.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic

TY - GEN

T1 - GRAPH: The Costs of Redundancy in Referring Expressions

AU - Krahmer, Emiel

AU - Theune, Mariet

AU - Viethen, Jette

AU - Hendrickx, Iris

PY - 2008/6

Y1 - 2008/6

N2 - We describe a graph-based generation system that participated in the TUNA attribute selection and realisation task of the REG 2008 Challenge. Using a stochastic cost function (with certain properties for free), and trying attributes from cheapest to more expensive, the system achieves overall .76 DICE and .54 MASI scores for attribute selection on the development set. For realisation, it turns out that in some cases higher attribute selection accuracy leads to larger differences between system-generated and human descriptions.

AB - We describe a graph-based generation system that participated in the TUNA attribute selection and realisation task of the REG 2008 Challenge. Using a stochastic cost function (with certain properties for free), and trying attributes from cheapest to more expensive, the system achieves overall .76 DICE and .54 MASI scores for attribute selection on the development set. For realisation, it turns out that in some cases higher attribute selection accuracy leads to larger differences between system-generated and human descriptions.

KW - EWI-13092

KW - IR-64885

KW - METIS-251092

M3 - Conference contribution

SN - not assigned

SP - 227

EP - 229

BT - Proceedings of the 5th International Natural Language Generation Conference (INLG 2008)

PB - Association for Computational Linguistics (ACL)

CY - USA

T2 - 5th International Natural Language Generation Conference, INLG 2008

Y2 - 12 June 2008 through 14 June 2008

ER -