User Guide

SYSTRAN Enterprise Server 7 – Online Tools User Guide 10
Special Terms Used in this Guide
The following table defines a few of the special terms used in this Guide.
Term
Definition
Bilingual
Corpus
In Training Server, this is an aligned parallel corpus, meaning side-by-side
display of the source and target translation of the text data.
Corpus
(plural Corpora). A corpus is a structured set of texts in a single language
(monolingual corpus) or text data in multiple languages (multilingual
corpus). An example of a multilingual corpus is the aligned parallel corpus
which displays side-by- side source text and multiple translations of same
text data.
DNT
Do Not Translate: identifies a text sequence that should not be translated
Language Pair
The source and target language
NFW
Not Found Word: a term or expression not found in the translation
resources
RBMT
Rule-Based Machine Translation
SDM
SYSTRAN Dictionary Manager
SMT
Statistical Machine Translation
SPE
Statistical Post Editing
STPM
SYSTRAN Translation Project Manager
Test Corpus
A bilingual corpus used to calculate automatic scores by comparing the
machine translated source language segments in the test corpus with the
target language segments in the test corpus.
Training
Corpus
A bilingual corpus used in a training task (Hybrid Training and Statistical
Training) that generates the first version of the statistical resources used by
SYSTRAN Hybrid MT Engine.
Tuning Corpus
A bilingual corpus used in a training task (Hybrid Training and Statistical
Training) to fine-tune the statistical resources used by SYSTRAN Hybrid
MT Engine to improve translation quality.
TU/TUs
Translation Unit(s). A TU consists of a source language segment and one
or several target language segments.
UD
User Dictionary
About Language Translation Software
Language Translation Software, also known as Machine Translation, provides a set of tools
by which electronic text is automatically translated from one language (like English) into
another (like Spanish).
The software quickly yields a gisting-level translation (the ability to understand the general
idea of what is written) and even better translation quality when users take the time to train
the software to learn their company or subject terminology.
SYSTRAN offers robust tools to help you improve the quality of your translations and
streamline your post-editing and quality assurance workflows.
While SYSTRAN strives to provide the highest possible automatic translation quality, it
is important to note that the quality of the source text has a significant effect on