• State Suggested Answer
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 432 subscribers
  • Views 2115 views
  • Users 0 members are here
Options
Related

AutoSuggest dictionary based on regular SDLTM or UpLifted SDLTM

Toon Theuwis

In Studio 2017, I was just wondering if creating an AutoSuggest dictionary would yield a different result depending on whether I would use my old TM or my UpLifted TM. There are a couple of thoughts crossing my mind, but as I cannot see how everything works in the background I cannot be sure. I'd think that there should be no difference as the dictionary cuts up the TM in little parts similar to UpLift (I guess), but creating an AutoSuggest dictionary from an UpLifted TM might not even be an good idea as the size of that TM is about 3 to 4 times bigger, which means you could end up with a dictionary that is X times bigger for no good reason (or maybe that simply isn't the case). Or would there even be no need for an autosuggest dictionary when Fragment Recall is enabled? In which case I wonder why you would even buy it as an add-on.
Rate translation
×
Suggest better translation
×
Moderator UI
×
  • Get AI Suggestion
Parents
  • 0

    While I don't know for sure how different the internal results are between the two, I can share my experience, as I recently recreated an AutoSuggest dictionary based on a large UpLifted TM. My pre-UpLift TM was 1.7 GB and the post-UpLift TM jumped up to about 4 GB immediately after being created. The old AutoSuggest dictionary is 101 MB and the new one is 120 MB. However, there's a 6-month difference between the two AutoSuggest dictionaries, and a lot of content has been added to the TM in that time, so the size increase seems reasonable to me.

    As far as whether it's worth having an AutoSuggest dictionary in addition to Fragment Recall, I would say yes. The main reason that comes to mind is that the default minimum words for a match in Fragment Recall is set to 3 for whole TUs and 2 for TU fragments, but the AutoSuggest dictionary will give you one-word suggestions as well. While we can (and I often have) decrease the defaults to 1, that can potentially result in some extra noise in the Fragment Recall window. One more thing is that it's not uncommon to see ASD suggestions in the list that are not shown in the Fragment Recall window, so it doesn't look like Fragment Recall entirely duplicates the results of the ASD.

    In any case, I find that having both enabled yields better results and is more flexible.
Reply
  • 0

    While I don't know for sure how different the internal results are between the two, I can share my experience, as I recently recreated an AutoSuggest dictionary based on a large UpLifted TM. My pre-UpLift TM was 1.7 GB and the post-UpLift TM jumped up to about 4 GB immediately after being created. The old AutoSuggest dictionary is 101 MB and the new one is 120 MB. However, there's a 6-month difference between the two AutoSuggest dictionaries, and a lot of content has been added to the TM in that time, so the size increase seems reasonable to me.

    As far as whether it's worth having an AutoSuggest dictionary in addition to Fragment Recall, I would say yes. The main reason that comes to mind is that the default minimum words for a match in Fragment Recall is set to 3 for whole TUs and 2 for TU fragments, but the AutoSuggest dictionary will give you one-word suggestions as well. While we can (and I often have) decrease the defaults to 1, that can potentially result in some extra noise in the Fragment Recall window. One more thing is that it's not uncommon to see ASD suggestions in the list that are not shown in the Fragment Recall window, so it doesn't look like Fragment Recall entirely duplicates the results of the ASD.

    In any case, I find that having both enabled yields better results and is more flexible.
Children