• State Suggested Answer
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 434 subscribers
  • Views 2229 views
  • Users 0 members are here
Options
Related

PLEASE, consider disconnecting segmentation rules and TMs making the system more intuitive and logical!

Michael Beijer

I think that SDL really needs to rethink the philosophy underlying the use of Language Templates and TMs. The current system is just too counterintuitive and wastes too much time. every time I come across a new abbreviation that I would like to add to my segmentation rules, or all of a sudden realised that I want all of my TMs to segment on a soft return, I have to go through hours of work and a ridiculous amount of thinking, just to make this possible. Segmentation rules should be unlinked from individual TMs! The current system just doesn't work. I think that in reality most average users will just completely give up on adding anything to or modifying their segmentation rules as it is just too complicated.

Paul, I know that you said that changing this would involve a lot of deep level changes, but I highly recommend your team look into exactly how difficult it would be to achieve because the current system is pretty terrible.

For example, I just realised that SDL is not segmenting on soft returns. Using a very helpful blog post by Nora Diaz (http://noradiaz.blogspot.com.es/2014/01/adding-soft-return-segmentation-rule-to.html), I then managed to add this functionality to a single TM using a regular expression. Yes, I managed to add 8 to my current project TM. I can no longer even remember how to add it to a Language Resource, so I could  use it the next time I create a new TM, so I have to google this once again. Not to mention my 25 other TMs, some of which are very large: how in god's name am I supposed to quickly add this to all these TMs? The simple answer is that I simply can't, and so only this one TM will end up segmenting on soft returns, and the next time it pops up in another project (which I'm sure it will) it will annoy me and remind me to write another post here similar to this one ;-)

So PLEASE, consider disconnecting segmentation rules and TMs making the system more intuitive and logical!

Michael
Rate translation
×
Suggest better translation
×
Moderator UI
×
  • Get AI Suggestion
Parents
  • 0
    Hi Michael,

    I understand your frustration. How about making a set of segmentation-only TMs? In other words, if you have an empty TM and add the segmentation rule(s) you want to it, then add it at the top of your translation providers list, your files will be segmented using those rules and the TM can be easily accessible to quickly add to other projects as well. As an extra benefit, you know where your rules are stored so you can copy them to your work TMs if needed. Not an ideal solution, but at least a workaround to help keep those segmentation rules handy.
  • 0 in reply to Nora Díaz
    Thanks Nora! That actually sounds like a pretty darned good idea! I wonder if someone could use this workaround to create an App for this. I also wonder if the SDL developers might be able to "solve" the problem in a much easier (albeit more superficial) way by bodging it using this trick?

    I didn't know that Studio segments a project using the topmost TM, but since it does, this actually makes it quite easy. Having to use an empty dummy TM (created with the sole purpose of containing my latest segmentation rules) for each project (and remembering to do so !) might be slightly annoying, aesthetically (so to speak), but if it works, it works ;-)

    As I have said elsewhere, I generally speaking add around three of four different abbreviations every workday, so really need a quick and easy way to keep track of them and ensure that they are used in all new projects.
  • 0 in reply to Michael Beijer

    Our Translation Memory Server product called GroupShare has this. You can specify a Language Resource Template and link it to one or more TMs.

     

    Cheers,

    Luis

    Best regards,
    Luis Lopes | Principal Product Manager | RWS | (twitter) @Luis___Lopes |

  • 0 in reply to Michael Beijer
    I remember I once used a similar method to what Nora suggested. One of my monthly projects involved translating programme descriptions (each one had to be within 180 character limit), as well as larger synopses (up to 1000 characters).

    Now, to make sure that I don't exceed the 180 character limit for each description, I needed to apply a specific QA rule that checks the number of characters within one segment, which usually is a single sentence. However, I wanted to see if I don't exceed the character limit for a single description (which consisted of 2-3 senteces), so Trados had to segment the document on paragraphs rather then individual sentences. In this way I could clearly see how many characters there were left.

    The thing is, I didn't want this rule to apply for synopses (they were too long for paragraph segmentation).
    Hence, I created two TMs - one with paragraph segmentation and the other with sentence segmentation.
    First I loaded the descriptions and put the TM with paragraph segmentation at the top of my TMs. As you may guess, the document got segmented on paragraphs.
    Then I changed the order of TMs so that the TM with sentence segmentation was at the top and after that I added the document with synopses. Of course the document got segmented on sentences.

    Hope it helps in a way.
Reply
  • 0 in reply to Michael Beijer
    I remember I once used a similar method to what Nora suggested. One of my monthly projects involved translating programme descriptions (each one had to be within 180 character limit), as well as larger synopses (up to 1000 characters).

    Now, to make sure that I don't exceed the 180 character limit for each description, I needed to apply a specific QA rule that checks the number of characters within one segment, which usually is a single sentence. However, I wanted to see if I don't exceed the character limit for a single description (which consisted of 2-3 senteces), so Trados had to segment the document on paragraphs rather then individual sentences. In this way I could clearly see how many characters there were left.

    The thing is, I didn't want this rule to apply for synopses (they were too long for paragraph segmentation).
    Hence, I created two TMs - one with paragraph segmentation and the other with sentence segmentation.
    First I loaded the descriptions and put the TM with paragraph segmentation at the top of my TMs. As you may guess, the document got segmented on paragraphs.
    Then I changed the order of TMs so that the TM with sentence segmentation was at the top and after that I added the document with synopses. Of course the document got segmented on sentences.

    Hope it helps in a way.
Children
No Data