The Measure of a Model

Rebecca Bruce; Janyce Wiebe; Ted Pedersen

The Measure of a Model

Rebecca Bruce
, Janyce Wiebe
, Ted Pedersen

Research output: Contribution to conference › Paper › peer-review

Abstract

This paper describes measures for evaluating the three determinants of how well a probabilistic classifier performs on a given test set. These determinants are the appropriateness, for the test set, of the results of (1) feature selection, (2) formulation of the parametric form of the model, and (3) parameter estimation. These are part of any model formulation procedure, even if not broken out as separate steps, so the tradeoffs explored in this paper are relevant to a wide variety of methods. The measures are demonstrated in a large experiment, in which they are used to analyze the results of roughly 300 classifiers that perform word-sense disambiguation.

Original language	English (US)
Pages	101-112
Number of pages	12
State	Published - 1996
Externally published	Yes
Event	1st Conference on Empirical Methods in Natural Language Processing, EMNLP 1996 - Philadelphia, United States Duration: May 17 1996 → May 18 1996

Conference

Conference	1st Conference on Empirical Methods in Natural Language Processing, EMNLP 1996
Country/Territory	United States
City	Philadelphia
Period	5/17/96 → 5/18/96

Bibliographical note

Publisher Copyright:
© 1996 Proceedings of the Conference on Empirical Methods in Natural Language Processing, EMNLP 1996. All rights reserved.

Find It

Find It at U of M Libraries

Cite this

@conference{8a3ae0d7d17d4c7b923d0bf43e03ac55,

title = "The Measure of a Model",

abstract = "This paper describes measures for evaluating the three determinants of how well a probabilistic classifier performs on a given test set. These determinants are the appropriateness, for the test set, of the results of (1) feature selection, (2) formulation of the parametric form of the model, and (3) parameter estimation. These are part of any model formulation procedure, even if not broken out as separate steps, so the tradeoffs explored in this paper are relevant to a wide variety of methods. The measures are demonstrated in a large experiment, in which they are used to analyze the results of roughly 300 classifiers that perform word-sense disambiguation.",

author = "Rebecca Bruce and Janyce Wiebe and Ted Pedersen",

note = "Publisher Copyright: {\textcopyright} 1996 Proceedings of the Conference on Empirical Methods in Natural Language Processing, EMNLP 1996. All rights reserved.; 1st Conference on Empirical Methods in Natural Language Processing, EMNLP 1996 ; Conference date: 17-05-1996 Through 18-05-1996",

year = "1996",

language = "English (US)",

pages = "101--112",

}

TY - CONF

T1 - The Measure of a Model

AU - Bruce, Rebecca

AU - Wiebe, Janyce

AU - Pedersen, Ted

PY - 1996

Y1 - 1996

N2 - This paper describes measures for evaluating the three determinants of how well a probabilistic classifier performs on a given test set. These determinants are the appropriateness, for the test set, of the results of (1) feature selection, (2) formulation of the parametric form of the model, and (3) parameter estimation. These are part of any model formulation procedure, even if not broken out as separate steps, so the tradeoffs explored in this paper are relevant to a wide variety of methods. The measures are demonstrated in a large experiment, in which they are used to analyze the results of roughly 300 classifiers that perform word-sense disambiguation.

AB - This paper describes measures for evaluating the three determinants of how well a probabilistic classifier performs on a given test set. These determinants are the appropriateness, for the test set, of the results of (1) feature selection, (2) formulation of the parametric form of the model, and (3) parameter estimation. These are part of any model formulation procedure, even if not broken out as separate steps, so the tradeoffs explored in this paper are relevant to a wide variety of methods. The measures are demonstrated in a large experiment, in which they are used to analyze the results of roughly 300 classifiers that perform word-sense disambiguation.

UR - https://www.scopus.com/pages/publications/0040310140

UR - https://www.scopus.com/pages/publications/0040310140#tab=citedBy

M3 - Paper

AN - SCOPUS:0040310140

SP - 101

EP - 112

T2 - 1st Conference on Empirical Methods in Natural Language Processing, EMNLP 1996

Y2 - 17 May 1996 through 18 May 1996

ER -

The Measure of a Model

Abstract

Conference

Bibliographical note

Find It

Other files and links

Fingerprint

Cite this