XBench is an amazing tool created by Spanish translation company ApSIC, primarily as a way to view and check terminology in glossaries or exported translation memories. XBench can be used by translators to ensure consistency between glossaries and TMs. Also, it contains a powerful QA function which will QA an 'ongoing translation' (i.e., edited files) against a TM and glossary and identify any mismatches. It also identifies any numerical mismatches.
At ForeignExchange Translations, XBench plays an important role in our Metriq quality system. We rely on it to:
- support linguists during translation (instead of checking an Excel glossary or having MultiTerm running in the background, linguists use XBench - they can even combine lookups with Internet searches for a given term!);
- analyze changes made edit, proofread, and client in-country review;
- act as a QA check tool (we check for adherence to terminology, untranslated segments, inconsistencies in target/source segments, making sure that products names are not changed, tag and numeric mismatches);
- check quality of TMs (e.g., checking for duplicates, untranslated segments, inconsistencies) and matching terminology between TMs and glossaries.
Amazingly enough, XBench is a free download from the ApSIC web site. A good overview of its functionality and use are available at Openoffice.org.
If you haven't yet tried XBench, check it out. Highly recommended stuff!
Did you like this post? Subscribe to Medical Translation Insight via email or RSS!
Categories: quality, terminology management, tools
4 Comments:
Subscribe to:
Post Comments (Atom)





Xbench is not only great free tool, but it has an excellent support, which helped me within a day to resolve an issue I had with the program.
The custom checklists feature is very handy too. The only problem I have is - it doesn't really find non-translated segments. Maybe one have to pre-segment file first?
Wasaty, you are true. Trados files need to be presegmented in order to detect the untranslated segments. With SDLX files this does not happen. Maybe this could be a suggestion to send to ApSIC in order to improve it.