At the Nov 2008 TikiFest in Paris, Marc Laporte, Youcef Bey and Alain Désilets discussed what could be done in TikiWiki, to incorporate the kinds of basic Computer Assisted Translation (CAT) tools that professional translators can't survive without nowadays. The result was the roadmap below.

Who will use it, and what for?

Types of organisation

• International NGOs with distributed offices/volunteers
  • Goal: Support volunteer-bassed crowsourcing of translation.
• Small translation offices
  • Goal: Use internally to do Just in Time, Agile translation
• Software Localisers
  • Crowdsource localisation to their users (note: Already done by Pootle, so don't focus too much on that).

Use cases

• I am translating and can't find how to translate a term, or phrase. Give me some suggestions found in some resource (ex: dict, termbase, translation memory).
• I looked for a translation and couldn't find it in any of the available resources. Now I found the answer and want to contribute it back to some resource.
• I am getting a bunch of suggestions. I need to know which one is better.
• I have some data in TMX or TBX format, or other. I want to import it into TikiWiki.
• The content I need to translate is in MS-Word. How to do this?
• I am not translating, but need to access a dictionary (Note: not core Use Case, but might come for free).

What tools?

Make a list of the kinds of CAT tools we might want to support.

Architecture

Three possible approaches:

• Build a "poor-man's" version from scratch directly into TikiWiki.
• Integrate an existing Open Source software into TikiWiki (ex: OmegaT).
• Interface with existing web services.

Existing building-blocks

Standards

• XLIFF
• TBX
• TMX

Open Source software

• OmegaT
• PhpDictionary: http://sourceforge.net/projects/phpdictionary/
• GlossWord: http://sourceforge.net/projects/glossword/
• bitext2tmx: http://sourceforge.net/projects/bitext2tmx/
  • Java based aligner.
• Tracell
  • https://sourceforge.net/projects/tracell/
  • Tracell is a Web-based Computer Assisted Translation(WCAT) tool.
  • Project in planning stage.
• PHP implementation of Dijkstra's algorithm
  • http://en.giswiki.net/wiki/Dijkstra%27s_algorithm#PHP
  • Maybe useful if we need to implement automatic text alignment algorithm inside TikiWiki.
• etc...

Commercial Software

• TRADOS
• etc...

Free web services

• GDT
• etc...

Commercial web services

• TERMIUM
• etc..

Iteration Plan

• Search box at bottom of Translation dialog (or in module or popup). You type (later, a copy-paste hotkey would be nice) a word or phrase, and it searches in Wikipedia and Wiktionary and OmegaWiki.
• Implement wrappers for other free online dictionary (ex: GDT).
• Import existing dictionary (could be TBX, CSV, etc....).
  • Character substitutions in page names
  • Quick Edit module should detect and warn about special characters in page names
• Site has its own local terminology database. User can enter entries. Note: Talk to Marc and Louis-Philippe, cause they already have a glossary feature that's almost ready to go out.
  • http://etmis.org/ in January 2009
• Implement automatic alignment editor to deal with pages that already exist, which were not created using a segment by segment editor. AD is taking the lead on this.
• Change translation editor so it shows each segment in its own edit box. Note: Ask Alain about what he and Seb Paquet did when they thought about this kind of interface. There is a wiki page somewhere on the site, can't remember where.
• When user saves a translation, aligned segments go into a TM. The TM is an extra table. The save format of the page itself stays the same (i.e. a string containing the whole markup for the page).
  • Saved info: both segments, source page, date, etc.
• Search field on translation editor searches not only in dictionaries and TBs but also in TM.
  • Search could use a few enhancements

Todos