• State Verified Answer
  • +1 person also asked this people also asked this
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 181 subscribers
  • Views 2514 views
  • Users 0 members are here
Options
Related

Is it possible to do fuzzy match check for term translation in Xbench?

Lingfeng Ding

Hi,

On an account, we need to check term translation compliance in Xbench. For some languages, like Croatian, due to declensions or suffixes, a lot of false positives will be reported by Xbench. This prevents linguists from spotting real errors. I want to check if it's possible for Xbench to do fuzzy match check or stem check for term translation? Please understand that the account has their own online translation system. So, Terminology Verifier in Trados Studio is not an option here. 

Thank you

Lingfeng Ding
Rate translation
×
Suggest better translation
×
Parents
  • +1

    The Key Terms check in Xbench uses a very simple heuristic that does not have stemming and lemmatization features.

    It is not designed to be a full fledge terminology check but rather a check of the few key terms that, given its importance for the project in question, can make bearable a few false positives.

    A typical use case for key terms are the few dozens of new terms coined during the project with several people working and that are more prone for inconsistencies.

    In Xbench Search grammar, which is a combination of Regex and PowerSearch, the current key term check is basically the following search:

    • Source: "<SourceTerm>"
    • Target Teram: -"<TargetTerm"
    • Search Mode: Regex
    • PowerSeach: Enabled

    where the angle brackets mean "word boundary", and the PowerSearch minus sign means "and does not have".

    You can use Xbench checklists to add fuzzyness and stemming. For example:

    • Source: "<SourceT.*>"
    • Target Teram: -"<TargetT.*>"
    • Search Mode: Regex
    • PowerSeach: Enabled

    ...could be a terminology check with stemming support if SourceT and TargetT were the stems.

    These 5-minute videos are an introduction to checklists:

    This 5-minute video explains PowerSearch:

    To avoid hijacking this forum with Xbench questions, if you have any questions on Xbench checklists, I suggest to place your questions in the Xbench forum: forum.xbench.net
Reply
  • +1

    The Key Terms check in Xbench uses a very simple heuristic that does not have stemming and lemmatization features.

    It is not designed to be a full fledge terminology check but rather a check of the few key terms that, given its importance for the project in question, can make bearable a few false positives.

    A typical use case for key terms are the few dozens of new terms coined during the project with several people working and that are more prone for inconsistencies.

    In Xbench Search grammar, which is a combination of Regex and PowerSearch, the current key term check is basically the following search:

    • Source: "<SourceTerm>"
    • Target Teram: -"<TargetTerm"
    • Search Mode: Regex
    • PowerSeach: Enabled

    where the angle brackets mean "word boundary", and the PowerSearch minus sign means "and does not have".

    You can use Xbench checklists to add fuzzyness and stemming. For example:

    • Source: "<SourceT.*>"
    • Target Teram: -"<TargetT.*>"
    • Search Mode: Regex
    • PowerSeach: Enabled

    ...could be a terminology check with stemming support if SourceT and TargetT were the stems.

    These 5-minute videos are an introduction to checklists:

    This 5-minute video explains PowerSearch:

    To avoid hijacking this forum with Xbench questions, if you have any questions on Xbench checklists, I suggest to place your questions in the Xbench forum: forum.xbench.net
Children