LiveDocs corpus curation and master memories
A different approach to the idea of "Master TMs"
Master translation memories. What are they? And how are they different from working translation memories, which I suppose must be the slaves? And what has all this got to do with LiveDocs? Nothing really, unless we decide to make LiveDocs relevant to this master/slave… uh, master/working memory thing.
But first, indulge me as I give a little background in the logic of differing roles for memoQ translation memories, something that those of us who spend most of our time working with other tools or who may be new to memoQ may be a little fuzzy on.
There are basically three possible roles for a memoQ translation memory in a project: working, master, and reference.
First, the Working translation memory. I capitalized that word and made it bold, because in a sense, it is the most important translation memory in a project, and in fact it may be the only one. For a lot of people it probably is, and in that case, if you look on the Translation memories page of the Project home, you’ll see it labeled as the Working/Master memory:
I suppose the product engineers of memoQ chose that label as a way of telling you that this translation memory is fulfilling the roles of both, roles that I haven’t described yet….
The Working translation memory is the one to which content is written when you confirm a segment while you work. A memoQ project can have any number of translation memories, local (as is the case in the example above, recognizable by the computer icon at the right) or remote (indicated by a clous icon, see below). The Working TM can be either local or remote (or even a locally cached copy of a remote TM, but that’s a subject for another day).
Now when I select a second TM to use in the project (a remote one in the screenshot above), nothing changes with the first translation memory. The second TM is simply a reference TM; no content is written to it as you work (that only happens to the Working TM), but if there are useful matches these will be shown in the Translation results pane of the working interface for translation and editing if they meet the criteria defined in your settings. And of course the content of reference TMs is available for concordance searching, as is the case for all TMs and LiveDocs corpora attached to a project.
But if there are two matches, one from each TM, the match from the TM labeled Master will always take priority. In the example above that TM also happens to be the Working TM. But let’s say that second TM has the authoritative, client-approved content from past work in it, content which should override anything else. Well, select that TM (it is selected in the screenshot) and set it as the Master TM:
Then the translation memory list looks like this for the TMs in the project:
And as you add some more memories for reference:
The Working and Master TMs are marked in bold fond because they are the most important. With equal match percentages, the Master has top priority, the Working TM has secondary priority, and then come the reference TMs and any LiveDocs corpora. If one of the latter has a match of a higher percentage score, it will have top priority, of course.
Notice that the names of some of the reference TMs are written in italics. That’s because they are reverse language pair TMs. The project in the example has English as the source language and German as the target language. The italicized reference TMs all have a German as the source language and an English as the target language. Reverse language pair TMs can have a reference role only; they cannot be used as Working or Master TMs.
Now here’s the point of this post:
But what if the most authoritative translation content to be used for reference is in a LiveDocs corpus? With equal matches, that preferred content is not prioritized! Well, that’s easy to fix as follows:
Select the preferred content, i.e. the relevant documents (maybe you don’t want all the documents in a given corpus) in your LiveDocs corpora, then
Export it to a TM!
That could be some master TM, or it could be a new, empty TM which you will set as the Master once the LiveDocs content export is finished.
Voila! Your preferred reference content from the LiveDocs corpus or corpora is now in a Master TM and will take priority over everything else if you are in a situation of equal matches. And if you exported to an exiting master TM that has a match of equal value for some reason, that “newer” content from the transfer takes priority. This is especially important in PRETRANSLATION.
But keep your LiveDocs corpus attached! Why? Because although you get “redundant” hits from the corpus and the Master TM, only the document in the LiveDocs corpus can be opened from the Translation results or Concordance (via the right-click context menu) if you need to read more context. “Context is everything” we translators say and thus, in this context, LiveDocs is king!
Excellent piece. I am prompted to rethink my LiveDocs strategy.
I wait in eager anticipation for this year's Sustainability Report from a favourite client (130-odd pages in Word) and exporting that LiveDocs to a TM is a great way to start.