--- In dejavu-l@yahoogroups.com, "Ineke" <maillijsten@...> wrote:
>
> Picture this:
> I am quitly sitting behind my pc,
> minding my own business,
> trying to work a bit with DV,
> and meanwhile deleting useless entries in the autosearch field.

If you've just imported a big Trados TM into a DVX MDB, first thing
you'll probably want to do is weed out all the very short "junk"
entries (references, single words out of context, short portions
beginning with or ending with a number, etc.).
This will save you a lot of time "deleting useless entries" and make
DVX assemble "useful" portions much, much better.

1. Make a copy of your MDB
2. Open the MDB in the TM editing environment.
3. Enter one or more the following type of SQL commands in the SQL
select box:

Sentence not like "*[a-z]*"
(selects any sentences without any letters in them)

Sentence not like "*[a-z]* *[a-z]*"
(selects any sentences not containing at least two words)

Sentence not like "* *[!0-9]"
(selects entries without any spaces in them and ending with a
number)

(Sentence like "*[0-9]" or sentence like "[0-9]*") and len(sentence)<15
(select sentences beginning or ending with a number and less than 15
characters long)

LEN(sentence) <= 10
(selects any sentences less than 10 characters long)

(Sentence like "*[0-9]" or sentence like "[0-9]*") and len(sentence)<15
(Sentences beginning or ending with a number and less than 15
characters long)

Very short portions beginning with "the", "a" and "an" will also cause
problems with assemble. You can select and delete them using:

(Sentence like "the *" or sentence like "a *" or sentence like "an *")
and len(sentence)<15

You can sort the portions found in alphabetical order to give you a
better idea of the ones you want to delete.

HTH,
Dave