Wednesday, August 19, 2015

New alignment in SDL Trados Studio 2014 – Easier, faster, smarter.


When developing the new alignment module for SDL Trados Studio 2014 we thought we’d try and take this opportunity to step back and rethink this functionality in a number of ways, leaving no stone unturned. In this blog I’d like to give a few more technical insights into what we did and how you will be able to use alignment in the upcoming Studio release in new ways that depart in many aspects from how alignment was performed in the SDL Trados 2007 days with WinAlign. In fact it’s best to think about alignment using the same tagline as we are using for Studio 2014 – and talk about how alignment has got easier, faster and smarter.
The Old Days
In the previous implementation using WinAlign, users had to set up an alignment “project”, add their existing source and target documents, go through some configuration steps, some of which were non-trivial, and ultimately kick off the alignment process itself. After alignment was complete, you then opened the result in an Editor and had to edit with the alignment result. When you were satisfied, you could then export the result as TMX and then import this into a Studio translation memory, which again was quite a manual process, even if a wizard is available for this. There were quite a few gaps in WinAlign compared to what Studio can do – not all Studio file types were supported, alignment was often a fragile process and many customers gave us feedback that we should really do our homework and replace WinAlign with a new alignment feature. Well – that’s what we did.
The WinAlign days are now gone. In Studio 2014, things have become a lot easier. All you need to do is go to the Translation Memories view and click the Align Documents button on the new Home ribbon. This opens a wizard that then leads you through the process step by step. Here you specify which TM you would like to use to import the alignment results. You can also create a new one at this step:
1
The next step is where you can see that the alignment has got a lot smarter as well. Here you add source and target files – using all file types that Studio supports, so you can now even PDF files if you like -, but you don’t have to add them one by one. Rather you can add all the files you wish to align by folder. This will add all files from all subfolders in one step. All you need to do is specify the folder and let Studio do the rest. In my example, I have added 46 XML files (software RESX format, to be precise) from 11 folders – all it took is two mouse clicks to specify the root source and target folder.
2
The next step is probably the most interesting in terms of settings and options.The first option is highly interesting. In our new alignment module, we would like to come to a place where users can avoid having to post-edit the alignment results. Rather we see alignment as a “no thought” process where you can basically trust the new alignment algorithm to self-evaluate the results and only filter out those that are useful for your daily work with the new TM you create as part of the alignment process. So the all decisive setting is Alignment quality value. For my project, I will leave it at the default value. This will ensure that I only get high-quality alignment results.
At the Alignment and TM Import Options step, you can also apply field values to the alignment so you can add certain meta data such as customer name or document type. I leave those empty but I’m sure that users will appreciate these new integrated options that help to optimize TM contents based on alignments from day 1.3
Now comes the big moment: I’m starting the alignment process of the forty six file pairs.I’m quite impressed with speed – the new alignment is definitely faster than WinAlign as well.
The most interesting question though is – what does the result look like? Again Studio 2014 is smarter than our previous solution in this respect. It opens the resulting translation memory in the Translation Memory maintenance right away for me to check the results; I can also spot edit them if I like.
All the results look correct though – so no editing is actually required. That’s exactly what we’d like to achieve – no editing necessary and only high quality alignments exported to the TM.
4
I am now ready to start my translation with the new TM that I just created based on the legacy documents.Acute eyes will probably notice one thing – all matches have been inserted as context matches. This is one of the key new features of the new, smarter alignment algorithm: it takes document context into account. This means when it can align segments in a sequence, it will mark all segments in such sequences as context matches. Obviously you will want to do this only when you trust the alignment results to be fully correct. If not, an alignment penalty is now available that ensures that you will review alignment results the first time they are used in your document. In my case, the translation would then look like this:
5
In this case all aligned segments have a penalty of 1% applied to them. Once I confirm them that penalty is cleared and I can then work with them as normal translation memory matches. This means you will be easily able to review your alignment results “as you go” when doing new translations.
6
As product manager for SDL Trados Studio I’m very excited about the new alignment capability – at long last we are able to say goodbye to WinAlign and welcome a much easier, smarter and faster successor in SDL Trados Studio 2014!

No comments:

Post a Comment