Modeling of learning curves with applications to POS tagging
详细信息 查看全文
- 作者：Manuel Vilares Ferro ; vilares@uvigo.es" class="auth_mail" title="E-mail the corresponding author ; Ví ; ctor Manuel Darriba Bilbao darriba@uvigo.es" class="auth_mail" title="E-mail the corresponding author ; Francisco José ; Ribadas Pena ribadas@uvigo.es" class="auth_mail" title="E-mail the corresponding author
- 关键词：Correctness ; Functional sequences ; Learning curves ; POS tagging ; Proximity criterion ; Robustness
- 刊名：Computer Speech & Language
- 出版年：2017
- 出版时间：January 2017
- 年：2017
- 卷：41
- 期：Complete
- 页码：1-28
- 全文大小：3486 K
文摘
- class="listitem" id="list_ulist0010">
- class="label">•
We estimate the learning curve for accuracy from a portion of the corpus.
- class="label">•
Technique iterative, with a termination criterion based on an error threshold.
- class="label">•
We show to what extent errors impact decision-making on accuracy-based criteria.
- class="label">•
Tests performed on 10 taggers and 3 corpora (English and Spanish).
- class="label">•
Results, not tied to a given implementation, confirm feasibility and robustness.