Document-Level Machine Translation Evaluation with Gist Consistency and Text Cohesion

Zhengxian Gong; Min Zhang; Guodong Zhou

Document-Level Machine Translation Evaluation with Gist Consistency and Text Cohesion

Zhengxian Gong
, Min Zhang
, Guodong Zhou^*

^*Corresponding author for this work

Soochow University

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

13 Scopus citations

Abstract

Current Statistical Machine Translation (SMT) is significantly affected by Machine Translation (MT) evaluation metric. Nowadays the emergence of document-level MT research increases the demand for corresponding evaluation metric. This paper proposes two superior yet low-cost quantitative objective methods to enhance traditional MT metric by modeling document-level phenomena from the perspectives of gist consistency and text cohesion. The experimental results show the proposed metrics can obtain better correlation with human judgments than traditional metrics on evaluating document-level translation quality.

Original language	English
Pages	33-40
Number of pages	8
State	Published - 2015
Externally published	Yes
Event	2nd Workshop on Discourse in Machine Translation, DiscoMT 2015 - Lisbon, Portugal Duration: 17 Sep 2015 → …

Conference

Conference	2nd Workshop on Discourse in Machine Translation, DiscoMT 2015
Country/Territory	Portugal
City	Lisbon
Period	17/09/15 → …

Cite this

@conference{3292032923b54d168dd5e423a70f1055,

title = "Document-Level Machine Translation Evaluation with Gist Consistency and Text Cohesion",

abstract = "Current Statistical Machine Translation (SMT) is significantly affected by Machine Translation (MT) evaluation metric. Nowadays the emergence of document-level MT research increases the demand for corresponding evaluation metric. This paper proposes two superior yet low-cost quantitative objective methods to enhance traditional MT metric by modeling document-level phenomena from the perspectives of gist consistency and text cohesion. The experimental results show the proposed metrics can obtain better correlation with human judgments than traditional metrics on evaluating document-level translation quality.",

author = "Zhengxian Gong and Min Zhang and Guodong Zhou",

note = "Publisher Copyright: {\textcopyright}2015 Association for Computational Linguistics.; 2nd Workshop on Discourse in Machine Translation, DiscoMT 2015 ; Conference date: 17-09-2015",

year = "2015",

language = "英语",

pages = "33--40",

}

TY - CONF

T1 - Document-Level Machine Translation Evaluation with Gist Consistency and Text Cohesion

AU - Gong, Zhengxian

AU - Zhang, Min

AU - Zhou, Guodong

PY - 2015

Y1 - 2015

N2 - Current Statistical Machine Translation (SMT) is significantly affected by Machine Translation (MT) evaluation metric. Nowadays the emergence of document-level MT research increases the demand for corresponding evaluation metric. This paper proposes two superior yet low-cost quantitative objective methods to enhance traditional MT metric by modeling document-level phenomena from the perspectives of gist consistency and text cohesion. The experimental results show the proposed metrics can obtain better correlation with human judgments than traditional metrics on evaluating document-level translation quality.

AB - Current Statistical Machine Translation (SMT) is significantly affected by Machine Translation (MT) evaluation metric. Nowadays the emergence of document-level MT research increases the demand for corresponding evaluation metric. This paper proposes two superior yet low-cost quantitative objective methods to enhance traditional MT metric by modeling document-level phenomena from the perspectives of gist consistency and text cohesion. The experimental results show the proposed metrics can obtain better correlation with human judgments than traditional metrics on evaluating document-level translation quality.

UR - https://www.scopus.com/pages/publications/85123443030

M3 - 论文

AN - SCOPUS:85123443030

SP - 33

EP - 40

T2 - 2nd Workshop on Discourse in Machine Translation, DiscoMT 2015

Y2 - 17 September 2015

ER -

Document-Level Machine Translation Evaluation with Gist Consistency and Text Cohesion

Abstract

Conference

Other files and links

Fingerprint

Cite this