Skip to main content

Translation Quality for Professionals

Objective

The situation of professional translation is characterised by increasing demands (integration into localisation workflow) and increasing time constraints (simultaneous shipment). The main objective of TQPro is to improve the quality of professional translation, by developing software support for each phase of the workflow. TQPro tools support different types of workflow, and provide support for preparation, translation and revision phases, both for conventional and machine translation. Tools to be developed will be: verification of terminology, extraction of lexicon entries from corpora for fast resource build-up, pattern matchers and intelligent search and replace, and others. TQPro will also increase the quality of the translation tools themselves (translation memories, machine translation). Partners expect a significant increase in productivity without quality loss

Objectives:
The main objective of TQPro is to improve the quality of professional translation.
Sub-goals are:
1. Provide quality control tools, to verify the quality of input and output texts and terminology used.
2. Improve the quality of machine translation systems to make it a quality tool for professionals. TQPro will provide better linguistic capabilities and customising tools for vertical application MT, i.e. MT within specific, well-defined domains.
3. Support quality in the workflow, by providing models for better integration, and by providing productivity tools.
4. Speed up the production of resources for human and machine translation: Verify the quality of memories, by indicating misaligned segments; and maintain terminology in multi-language workflows.
5. Develop standards for text and terminology interchange. This is to support multi-vendor environments.

Work description:
1. Preparation.
Define the workflow, collect linguistic resources for development and test, and write the tool specifications.
2. Standards.
Develop standards for terminology exchange (OLIF), for text handling (OTEXT), and APIs for calling translation components from external applications, based on existing activities.
3. Resources.
Build tools which extract full bilingual MT and terminology entries from corpora to shorten ramp-up time for new translation tasks in new vertical domains.
4. Preparation tools.
These tools are: verification of terminology (is it known and legal?), intelligent pattern matching for text preparation, and MT translatability check (is the text suitable for MT?).
5. Translation Quality.
This is checking translation memories for terminological errors and misalignment, and increasing the quality of MT by integrating information extraction and probabilistic techniques, and by customisation and tuning tools.
6. Revision tools.
Tools check the completeness of translation (missing and un-translated segments); they do term verification for human translation (has the intended terminology really been used?) and MT diagnostics for MT output to identify segments of critical quality, and provide intelligent search and replace for fast revision.
7. Integration.
Tools will be integrated into a client-server environment, with a demonstrator Web interface. They will support the different workflows described in WP1.
8. Test and Evaluation, to verify the achievement of initial objectives: Functionality, performance, and productivity gains.

Milestones:
1. New software tools (version 1 due project month 12, version 2 due project month 21):

term verification for source and target texts,
lexicon build-up for human and machine lexicons,
translatability check,
intelligent text processing (pattern matching, search & replace),
completeness check,
MT diagnostics.
2. Improved translation components (MT systems, memories) in example applications.
3. Improved standards for text handling, lexicon exchange, and API calls for translation functions.

Call for proposal

Data not available

Coordinator

SAP AKTIENGESELLSCHAFT SYSTEME, ANWENDUNGEN, PRODUKTE IN DER DATENVERARBEITUNG
Address
Neurottstrasse 16
69190 Walldorf
Germany
 

Participants (3)

CAT TECHNOLOGIES SRL
Italy
Address
Vilale Resegone 9
20020 Arese
 
CENTER FOR SPROGTEKNOLOGI
Denmark
Address
Njalsgade 80
2300 Copenhagen
 
LOGOSCRIPT SL
Spain
Address
Calle Tarragona 149-157, 7O2a
08015 Barcelona