Translation Memory (TM) - PowerPoint PPT Presentation

1 / 10
About This Presentation
Title:

Translation Memory (TM)

Description:

Translation Memory (TM) Ashish What is Translation Memory A multilingual text archive/database containing multilingual texts, allowing storage and retrieval of ... – PowerPoint PPT presentation

Number of Views:209
Avg rating:3.0/5.0
Slides: 11
Provided by: Ashi77
Category:

less

Transcript and Presenter's Notes

Title: Translation Memory (TM)


1
Translation Memory (TM)
  • Ashish

2
What is Translation Memory
  • A multilingual text archive/database containing
    multilingual texts, allowing storage and
    retrieval of aligned multilingual text segments
  • not MT machine translation system
  • Database that is used to aid human translators.
  • Database consists of text segments in a source
    language and their translations into target
    language.

3
CAT VS TM
  • CAT computer aided translation system
  • TM translation memory tool
  • CAT is more general software which usually
    includes TM
  • CAT also includes glossaries, dictionaries,
    collocation databases, alignment tool etc.

4
Segment
  • Segments
  • Usually sentences,
  • Blocks,
  • Paragraphs,
  • Sentence phrases.
  • Typical segments
  • Sentences, messages, titles
  • Word level translation do not come under domain
    of TM.
  • Word level translation implemented as glossaries
    in machine aided translation system.

5
Significance
  • Translator has to translate only new sentences.
  • It work best on texts which are highly repetitive
  • Legal, Technical, administration domain
  • Translation of Technical manual for new
    version/model
  • Such repetition is present in
  • User manuals,
  • Annual reports,
  • Legal contract papers,
  • Patents
  • For team work
  • Provide consistency
  • Helps restrict vocabulary

6
Concept
  • Source text is fed to the system
  • System generates possible translation matches
  • User select the best match
  • Matching is exact or fuzzy

7
Exact Vs fuzzy
  • Exact
  • Accurate
  • Easy to implement
  • Adequate in applications where new documents
    differ a little from old one
  • Fuzzy
  • Gives non exact matches
  • -Exact-Similar-Less similar-

8
Issues
  • Automatic alignment of phrases within sentence of
    bilingual data.
  • Segment size
  • Small (word/phrase level)
  • More matches
  • Less accuracy (ambiguity)
  • Large (paragraph/sentence)
  • Few matches
  • More accuracy

9
CAT-TM tools
  • free
  • OmegaT
  • Implemented in Java (open source)
  • Single user
  • TM, fuzzy
  • Sentence level
  • Transolution
  • Implemented in Python (open source)
  • TM, fuzzy
  • Single user
  • Commercial
  • Trados
  • TM, fuzzy
  • supports terminology dictionary,
  • Network support ?
  • AidTransStudio
  • Single user
  • TM
  • Support dictionary lookup

10
References
  • http//en.wikipedia.org/wiki/Translation_memory
  • OmegaT user manual
  • nlp.fi.muni.cz/nlp/nlp-prace/referaty/nepil/TM-eva
    luation.ps
  • http//www.webbsnet.com/translation/thesis.html
  • www.lingobit.com/solutions/misc/translation-memory
    .htm
  • Automatic Translation Memory Fuzzy Match
    Post-Editing A Step beyond Traditional TM/MT
    Integration, LREC-2004
Write a Comment
User Comments (0)
About PowerShow.com