Hi Kris,
Thanks for sharing the feedback. I think the second bullet is quite useful. It makes sense to provide best practice advice to tool makers for sure. I can see, especially, how this would be useful for tool makers that provide tools that touch span the content creation, content management, and content translation domains. And I also very much like the third bullet. It is useful for people who do not have a lot of experience in managing their own translations and translation memory to understand the various degrees of TM matching.
I must say I found the first bullet odd. I count myself as somebody who directly supports content creators(this is one of my duties at my company). We have realized great benefit by implementing a process that converts our content to XLIFF before passing it along to our LSP. When we implemented this system, we immediately measured and documented a 30% savings in cost and turn-around time. And that does not include savings from removing DTP from the process. It was also only measured first-time translations. The savings are even more dramatic for updates.
I think the primary reason this works so well for us is that we provide our LSP an exact format that suits their expertise and works natively in their tools (XLIFF), as opposed to asking them to do the conversions and management of the content and TM. To be sure, there are very technically competent LSPs who can do conversions and manage content (and provide a fair price for the overhead), but I don't this is their core strength or core competency. Our LSP will be the first to chime in with agreement and endorsement of our strategy. Also, we are able to manage and leverage our own TM. I could list lots of technical reasons this calculation rings true (but I don't think doing so is a good fit for the article).
That said, I am grateful to your Language Technologies division for providing the feedback. I would be very interested in understanding their thinking behind the first bullet. I am very open to a further dialog.
Thanks,
Bryan
From: Kristen Eberlein [mailto:keberlein@sdl.com]
Sent: Monday, August 29, 2011 7:51 AM
To: DITA Adoption TC
Subject: [dita-adoption] SDL comments about the DITA Adoption article regarding XLIFF
As promised, I solicited feedback about the draft article. Below are excerpts from the SDL reviewer in the Language Technologies division:
· My biggest question is about the purpose/intended audience for the document. It reads as though you are targeting the DITA content creators but I can’t see what value would be gained by the DITA content creators in doing this conversion to XLIFF before sending the data into the translation process – this is a process which only adds value for the entity doing the translation, not the content creator. You mention simplifying administration activities – but that is really a very small saving vs. the large cost of the content creator having to perform this transform. If there is some other value that escapes me that I think the document would have more impact if it contained this information.
· Alternatively, you could view this information as much more relevant to tool vendors – giving them guidance on how they should interpret DITA as something ‘more than XML’. Best practise advice on how these and things like conref’s etc. should be ‘dereferenced’ would be useful.
· My only other comment would be on the use of the phrase ‘perfect match’. Certainly in SDL tools when the document source text exactly matches the TM source text (as your document describes) we just call this a 100% match. Beyond that, where the source text is identical and the previous segment is also identical, we have ‘Context Matches’ . And beyond that, where the translation is taken from an earlier version of the same document, we have ‘Perfect Matches’.
Best regards,
Kris
Kristen James Eberlein l DITA Architect and Technical Specialist l SDL Structured Content Technologies Division l (t) + 1 (919) 682-2290 l keberlein@sdl.com
Join us at DITA Fest 2011 this fall. Visit http://www.sdl.com/ditafest2011 for details.