memoQ’s matching defaults for new entries in a term base (50% prefix matching, permissive capitalization) may not be the best choices for you, depending on the languages involved and how you intend to use the term base. For example, the best settings for translation and review work might not serve you as well for QA checks, where too many warnings might be generated for terms which are just fine.
It is very important to understand the properties of memoQ term entries and how these may improve or detract from your results in translation work. I’ve written and lectured about this subject matter many times, and I am likely to do so further in this forum, but the video link below from a webinar in 2018 offers an overview of the essentials as part of about a 70 minute lecture on many aspects of terminology handling in memoQ. The link embedded here starts at 53:40
in the video where I talk about the importance of properties, but you may want to look at other topics in the content index found in the Description field on YouTube.
The real subject of this post is also included in that video, at time 42:48
.
Changing the default settings for languages in a term base can reduce the time and effort you spend fine-tuning the entries to work better in translation, review and QA. The best defaults are likely to depend on whether a given language is the source language or the target language in a project, so try to get out of the habit of using the same term base bidirectionally if you do that.
These default settings can be seen in the term base properties dialog or in the dialog for creating a new term base by clicking on the blue text link for New term defaults at the lower left of the dialog.
To change the match or case sensitivity defaults, use the dropdown menus marked by the red box in the screenshot above. I usually work with German as my source language, and the frequency of irregular nouns with vowel changes, spelling variations, typographical errors and OCR conversion errors in my source texts make a default of fuzzy matching for terms the most useful. Where that is not the case for an individual term (such as an acronym, where exact matching is usually better), I edit the term entry manually.
The changed settings of a particular term base persist, and will be applied to all new terms entered after you change the defaults. But existing terms in the term base will not be affected by a change of defaults. If you have a term base where many changes are needed, look at the video above at time 56:17
to learn how to change the properties of many terms at once.
Certain languages, such as German or Arabic, may have special advantages for matching with a Fuzzy setting, because they can identify matches embedded in compound words or which bear prefixes or suffixes which will cause memoQ’s usual defaults to fail.
My experience as a translator frequently included source texts with a lot of inconsistent spelling, typos and errors from OCR. These pretty much doom term matching with the 50% Prefix default, so I tend to use Fuzzy as the default for source matching in German, English and Portuguese to handle these issues. If I find too many bad matches or QA issues with such a term entry, I make individual adjustments to that entry, but on the whole a Fuzzy default saves me time tuning things and gives better results.
On the target language side though, it pays to be as specific as possible with the match settings, so that 50% Prefix default (50% or more of the first part of the word matches) might be OK for you (but probably not in some languages). Let the results you see in your work be the ultimate guide.
FYI, the similarity criterion for a fuzzy term matching in memoQ is 80% character matching, with some characters in certain languages deemed equivalent. That equivalence in German would include vowels with umlauts being treated the same as those without for purposes of calculating the match. Remembering this will probably help you understand why some matches you might expect don’t happen or why some unexpected ones do.
I never knew what they meant by "permissive" regarding to case sensitiveness. It is never permissive enough for me. I always have this like "No"