Can an AutoSuggest dictionary be edited?
Thread poster: Christine Andersen

Christine Andersen  Identity Verified
Denmark
Local time: 11:09
Member (2003)
Danish to English
+ ...
Jul 15, 2010

I have generated an AutoSuggest dictionary for Studio 2009 from my largest TM plus some smaller ones with specific terminology. I imported them all into a new combined TM so that they would be included. So far I am generally very pleased with the results. Given that it is pure statistics, and my combined TM covers a very wide range of subject areas and genres, it is really impressive.

But there are one or two strange things that crop up.

One is the name of a Danish actor - statistically, it must be an idiot-proof 100% match somewhere, but he keeps popping up at the top of the list in an entirely different context. He must have come from a one-off job, and should not be there at all!

Another is a spelling mistake. Thanks, I can make my own, and most of the time AutoSuggest saves typos. This one is doubly dangerous, because I trust most of what comes up, and it is one the Spell checker does not catch (like could/cold).

I would like to remove minor flaws like these if possible.

Can anyone please tell me how?
THANKS in advance!


 

Astrid Elke Witte  Identity Verified
Germany
Local time: 11:09
Member (2002)
German to English
+ ...
As far as I know, it is not possible yet Jul 15, 2010

Hello Christine,

I think it has not got so far yet that the AutoSuggest Dictionary is editable. However, I expect we will get to this stage in the future. Let's wait for Paul to comment, though.

I know what you mean. I also have recurring minor flaws in mine, which are a nuisance.


Astrid

P.S. I think the only way at the moment is to quality-check the underlying TM and then re-generate the AutoSuggest dictionary, once it is absolutely perfect. I have this already on my list of things to do - but it will be very time-consuming.

[Edited at 2010-07-15 12:27 GMT]


 

Walter Blaser  Identity Verified
Switzerland
Local time: 11:09
French to German
+ ...
Correct the source TM and regenerate Jul 15, 2010

Christine

AutoSuggest dictionaries are - compared to a TM - a static thing, means they are not automatically fed and also not editable.

If you discover an error in your AutoSuggest dictionary, it means that the error exists in the TM you used to create the AS dictionary, so the proper procedure is to correct the error there (which makes sense anyway) and then regenerate your AS dictionary (which is BTW not so time consuming, it takes between 10 and 30 minutes, depending on the size of your TM).

Walter


 

SDL Community  Identity Verified
United Kingdom
Local time: 11:09
English
Not possible yet Jul 15, 2010

Hi Christine,

Walter is correct in that the way the ASD is constructed it is currently not possible to edit it, so you do have to regenerate whenever you feel the need. This may be because you have added a few thousand TU's and think you are missing something, or because you are getting innaccurate results that may be due to the content of your TM.

As Astrid alluded to it is always possible that we may make some changes to this in the future to enable some kind of editing capability, particularly for your case, so I would encourage you to raise a request in ideas.sdl.com, or to vote for an existing one on this subject.

Regards

Paul


 

Christine Andersen  Identity Verified
Denmark
Local time: 11:09
Member (2003)
Danish to English
+ ...
TOPIC STARTER
Great, thanks! Jul 15, 2010

Thanks everyone for your comments, and thanks, Paul, for the suggestion about ideas.sdl.com. I will do that.

It would be fine if a future development made it possible simply to remove unwanted entries - the odd things like my Danish actor that statistically are foolproof, but in fact not relevant.

One or two of the others might be hard to find in the TM, but hey, I can live with them.

I am used to editing Multiterm more or less on the spot, but I do appreciate that this is not entirely the same thing.

Thanks again!


 

Edric Barbosa Filho
Local time: 06:09
English to Portuguese
I think editing the memory and regenerating the ASD is better... Jul 15, 2010

Hi

When AutoSuggest dictionary (ASD) shows me a term that I don't want, that probably means that the memory from which ASD was generated has several instances of it. In your example, that flaw is in the main memory, so you'd not want it to stay there, and Studio has great features for editing TMs.

For my work it is better to edit the memory to replace all occurences of the undesirable/wrong term pointed out by ASD, and regenerate the dictionary: I'll never see that term again. If the term is only corrected in ASD, it will be kept in my TM, "contaminating" next ASD regenerations.

Of course, this only applies if you are generating the ASDs yourself: if you receive ASDs from other people, the only remedy is correcting your copy, of course...

Rgds

Edric


 

Christine Andersen  Identity Verified
Denmark
Local time: 11:09
Member (2003)
Danish to English
+ ...
TOPIC STARTER
Thanks, Edric, good point. Jul 15, 2010

The fact is that in the TMs the terms rarely if ever appear, if the segments do not match.

But I had forgotten about editing the TM in Studio 2009 - I am told it is far easier than with the earlier versions. Maybe I will cut away some dead wood, because I know there is a lot that will never come up again.

I will certainly give it a try.

icon_smile.gif


 

Grzegorz Gryc  Identity Verified
Local time: 11:09
French to Polish
+ ...
Well... possible... Jul 15, 2010

SDL Support wrote:

Walter is correct in that the way the ASD is constructed it is currently not possible to edit it (...)


No icon_smile.gif
It's technically possible.
As ASD dictionnaries use SQLite, the response is e.g. the SQLite Browser
http://sourceforge.net/projects/sqlitebrowser
Tables S and T.
It's not very handy but it works.

The problem is the ASD edition makes sense only for closed corpora which will never be changed.

As Astrid alluded to it is always possible that we may make some changes to this in the future to enable some kind of editing capability, particularly for your case, so I would encourage you to raise a request in ideas.sdl.com, or to vote for an existing one on this subject.

Think parallel icon_smile.gif

Cheers
GG

[Edited at 2010-07-15 16:51 GMT]


 

Emma Goldsmith  Identity Verified
Spain
Local time: 11:09
Member (2010)
Spanish to English
Editing TMs Jul 15, 2010

Christine Andersen wrote:

But I had forgotten about editing the TM in Studio 2009 - I am told it is far easier than with the earlier versions.
icon_smile.gif


You need to explore the TM maintenance window in Studio. It's very easy to use. You can filter by date, or a certain word, or by field. Batch editing isn't hard either. You can change the number of segments you see in the window at one time by going to options>TM view>number of TUs per page. The only thing I miss is being able to browse to the end of the TM (for some reason that arrow key is missing).

Do a search for your Danish actor and chop him out if he's really pestering you.icon_smile.gif


 

Christine Andersen  Identity Verified
Denmark
Local time: 11:09
Member (2003)
Danish to English
+ ...
TOPIC STARTER
Actually, I rather like the actor! Jul 15, 2010

He used to play a giant teddy bear on a much-loved children's TV programme when my son was young. The books are still going strong... and I dare say the programmes are broadcast once in a while. They were timeless.

But he was just an example of how a statistically produced dictionary, although very useful most of the time, can come up with some oddities, 100% accurate but quite useless, which a human could weed out.

So now to see what I can do - I am going to generate a new AutoSuggest for my laptop and try out everybody´s good advice icon_smile.gif

[Edited at 2010-07-15 19:57 GMT]


 

SDL Community  Identity Verified
United Kingdom
Local time: 11:09
English
GG@asdedit.com Jul 15, 2010

Grzegorz Gryc wrote:
It's technically possible.
As ASD dictionnaries use SQLite, the response is e.g. the SQLite Browser
http://sourceforge.net/projects/sqlitebrowser
Tables S and T.
It's not very handy but it works.
GG

[Edited at 2010-07-15 16:51 GMT]


Hi Grzegorz,

ok, I stand corrected. I was really referring to the average mortal who is looking for the software to do things for them. The sort of user you refer me to quite oftenicon_wink.gif

Probably just a couple of things to add then;

  • You are on your own – if such edits reduce the ASD’s recall please don't come to Support
  • If you want a permanent fix do it in the TM or you will be doing this again and again


It is a hack, and our recommendation would be for users to email GG@asdedit.com
icon_smile.gif

Regards

Paul


 

Grzegorz Gryc  Identity Verified
Local time: 11:09
French to Polish
+ ...
Being rather stupid and lazy... Jul 15, 2010

SDL Support wrote:

Grzegorz Gryc wrote:
It's technically possible.
As ASD dictionnaries use SQLite, the response is e.g. the SQLite Browser
http://sourceforge.net/projects/sqlitebrowser
Tables S and T.
It's not very handy but it works.


ok, I stand corrected. I was really referring to the average mortal who is looking for the software to do things for them. The sort of user you refer me to quite oftenicon_wink.gif

SQLite Browser does icon_smile.gif
But it is not very handy for an average mortal icon_smile.gif

Probably just a couple of things to add then;

  • You are on your own – if such edits reduce the ASD’s recall please don't come to Support

Of course.
As you said, it's a kind of hack and I never tested it thoroughly.

  • If you want a permanent fix do it in the TM or you will be doing this again and again

  • As I said:
    The problem is the ASD edition makes sense only for closed corpora which will never be changed.
    I'm too lazy to repeat these actions after every regeneration and too stupid to automate 'em icon_smile.gif

    Cheers
    GG

    [Edited at 2010-07-15 20:10 GMT]


     

    Spiros Doikas  Identity Verified
    Local time: 12:09
    Member (2002)
    English to Greek
    + ...
    sqlitebrowser Mar 7, 2014

    I tested sqlitebrowser, but I see that there is no way to export in a source -> target format. Any clues about this?

     

    De Sena Viegas  Identity Verified
    Local time: 11:09
    Member
    English to Portuguese
    + ...
    How to merge TRADOS AutoSuggest dictionaries Jun 25, 2014

    Hello everyone,


    Recently, when attending a TRADOS SDL event in Frankfurt am Main, I asked the TRADOS Staff if it is possible to MERGE several AutoSuggest dictionaries into one, and they replied "although "it's not possible yet, you can use several ASD at once".

    However, after reading the post from Grzegorz Gryc I also arrive to the conclusion that it is not even necessary to change the extension *.bpm into *.db3 in order to be possible to read, compare and... MERGE (?) 2 or more AutoSuggest dictionaries.

    Yes, it is a little bit tricky, but it is possible:

    Please download the SQLiteCompareUtility at

    http://www.codeproject.com/Articles/220018/SQLite-Compare-Utility

    After installing SQLITE Compare, open 2 copies of 2 different ASD and copy all the tables "one by one" "from left to right" or "from right to left"...

    Now I am not sure about the result of all this, but perhaps this is a possible solution... ; )

    One good way to find out if it is working good, is to use the new merged ASD in your next translation.



    [Bearbeitet am 2014-06-25 22:12 GMT]


     


    To report site rules violations or get help, contact a site moderator:


    You can also contact site staff by submitting a support request »

    Can an AutoSuggest dictionary be edited?

    Advanced search







    BaccS – Business Accounting Software
    Modern desktop project management for freelance translators

    BaccS makes it easy for translators to manage their projects, schedule tasks, create invoices, and view highly customizable reports. User-friendly, ProZ.com integration, community-driven development – a few reasons BaccS is trusted by translators!

    More info »
    Anycount & Translation Office 3000
    Translation Office 3000

    Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

    More info »



    Forums
    • All of ProZ.com
    • Term search
    • Jobs
    • Forums
    • Multiple search