Translate
Hi there
I've written a machine translation translation-provider plugin for 2011 which submits segments to an MT web service and retrieves translations.
A customer of ours who is trying to pre-translate documents using both a file-based TM with a minimum segment quality of 85% and our machine translation plugin, but they find that the file-based segments are never used by the pretranslation process, and all their segments end up being MT'd with scores of 100%.
In the code for our plugin, we retrieve an MT translation for every segment in a document, regardless of its current translation status, and assign it a score of 100% which is what the online Trados SDK recommended, but I'm wondering if that will prevent segments from other sources like TMs being used. I tried appying a penalty of 20% to the MT segments to bring their quality score below the 85% threshold used for the file-based TM, effectively forcing Trados to only use the TM-based segments, but it still seems to bring in only MT segments. We don't apply penalties anywhere in our own code, as I assumed Trados would apply the penalties itself when processing all segments once pretranslation had completed, but this appears not to be the case.
I'm just wondering if you know what the best practices are in this case, and if we are going about things the wrong way, or if you don't, if you could advise me on who could tell me?
