Using MultiTerm with a concept-oriented approach and the option "Check for possible non-use of target terms" produces lots of false positives - what are we supposed to do?

I am in the process of re-organizing my organization's glossary. In several sources, including SDL sources like the White Paper on Terminology Management, the concept-oriented approach to term management is promoted. From a logical point of view, this is very convincing and I would like to implement it.

In this approach, each concept has a a termbase entry, so "stain" (the discoloration) has an entry and "stain" (the penetrative dye) has one = two entries (more if the respective verbs are added).

When I organize my TB like this, I get warnings (or errors, whatever is defined in the settings) for each homonym. The white paper mentioned above uses "monitor" as an example (the screen and the activity of observing).

Since the concept-oriented approach is so much promoted by SDL, I am wondering if I am doing something wrong that following that recommendation seems to render a key feature of MultiTerm integration in Studio practically useless. Is there an option to only warn if no target term at all is used?

Daniel

Parents
  • Hi Daniel,

    the concept-oriented approach is not only promoted by SDL - it has always been the foundation of terminology in theory and in real terminology work for decades.

    Terminology work is knowledge-based and linguistically motivated, and thus structured accordingly in a termbase (unless you do software glossaries etc). Translation memory technology, however, works on pattern matching, and so does term recognition (and term verification), it knows nothing about linguistics and concept-orientation.

    There is no perfect solution to combine the linguistic resources with pattern-matching oriented systems.And those pattern-matching approaches are about to become obsolete in times of deep-learning, content enrichment initiatives etc.
    With SDL Language Cloud Terminology, now some linguistic features have been introduced for term recognition (and more "linguistic AI" seems to be on the  SDL product roadmap). So maybe we can expect solutions to your problem in some future Studio + LC Terminology combination.

    Kind regards

    Christine   

Reply
  • Hi Daniel,

    the concept-oriented approach is not only promoted by SDL - it has always been the foundation of terminology in theory and in real terminology work for decades.

    Terminology work is knowledge-based and linguistically motivated, and thus structured accordingly in a termbase (unless you do software glossaries etc). Translation memory technology, however, works on pattern matching, and so does term recognition (and term verification), it knows nothing about linguistics and concept-orientation.

    There is no perfect solution to combine the linguistic resources with pattern-matching oriented systems.And those pattern-matching approaches are about to become obsolete in times of deep-learning, content enrichment initiatives etc.
    With SDL Language Cloud Terminology, now some linguistic features have been introduced for term recognition (and more "linguistic AI" seems to be on the  SDL product roadmap). So maybe we can expect solutions to your problem in some future Studio + LC Terminology combination.

    Kind regards

    Christine   

Children