Improving conversational spoken language machine translation via pronoun recovery

Yanlin Hu*, Heyan Huang, Ping Jian, Yuhang Guo

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Machine translation for social communication is necessary in daily life. However, spoken language translation faces many challenges especially in the translation of zero pronouns which is absent in the source language but appear in the target language. Dropping of pronouns severely affects the machine translation from pronoun dropped language such as Chinese to other languages. This phenomenon occurs more frequently in the conversational spoken language. In order to solve this problem, we insert the position of missing pronouns into the source side, then we use the word alignment method to filter the pronouns in order to pick up the pronouns which are really helpful for the machine translation. We achieve improvement on the translation of chat, message and telephone conversational corpus.

Original languageEnglish
Title of host publicationSocial Media Processing - 4th National Conference, SMP 2015, Proceedings
EditorsMaosong Sun, Xichun Zhang, Zhenyu Wang, Xuanjing Huang
PublisherSpringer Verlag
Pages209-216
Number of pages8
ISBN (Print)9789811000799
DOIs
Publication statusPublished - 2015
Event4th National Conference on Social Media Processing, SMP 2015 - Guangzhou, China
Duration: 16 Nov 201517 Nov 2015

Publication series

NameCommunications in Computer and Information Science
Volume568
ISSN (Print)1865-0929

Conference

Conference4th National Conference on Social Media Processing, SMP 2015
Country/TerritoryChina
CityGuangzhou
Period16/11/1517/11/15

Keywords

  • Conversational spoken language
  • Machine translation
  • Pronouns recovery
  • Word alignment

Fingerprint

Dive into the research topics of 'Improving conversational spoken language machine translation via pronoun recovery'. Together they form a unique fingerprint.

Cite this