Different results from the 2 methods

Re: Not surprising -- Jean Ichbiah
Posted by Jon Knowles , Mon, Nov 03, 2003, 13:48:50 Reply Top Forum

Unless I'm missing something, there do seem to be differences in the results of the two methods. The following examples were done with "Sort by short code length" checked. Only the Repeto glossary(ies) is/are opened (that is, no ABCZ or other glossaries).

Repeto 2 10 (one glossary): typing "tt"
tenosynovial thickening
the tendon
tenosynovial thickening, no
tenosynovial thickening but
tenosynovial thickening without
the tendon slip
tenosynovial thickening, but
tenosynovial thickening, chronic
tenosynovial thickening, mild
tenosynovial thickening, moderate

Repeto 2 5 plus 6 10 (two Included glossaries):
tenosynovial thickening
the tendon
the tendons
the thumb
the tourniquet
the transverse
tenosynovial thickening, no
tenosynovial thickening without
the tendon was
tenosynovial thickening, but

Repeto 2 10, typing "al":
and lunotriquetral
adherence left small
Advanced left wrist
and ligaments were
and lunotriquetral intervals
at least with
at length, the
and long radial lunate
all loose bone fragments
and ligated with 4-0

Repeto 2 5 plus 6 10
abductor longus
and lunotriquetral
and lateral
and left
and lunate
adherence left small
adherent lateral antebrachial
and lunotriquetral interosseous
and lunotriquetral intervals
at length, the

Repeto 2 10, typing "ala":
adherent lateral antebrachial cutaneous nerve
and left anterior iliac crest
and lunotriquetral intervals appeared fairly
Advanced left wrist SLAC pattern, post-traumatic arthritis
and left upper extremity surgery, in an attempt to
and ligaments were closed as a single layer using over-and-over
at least with a good fibrous union to the underlying

Repeto 2 5 plus 6 10 typig "ala":
adherent lateral antebrachial
adherent lateral antebrachial cutaneous nerve
and left anterior iliac crest
and ligaments were closed as
and lunotriquetral intervals appeared fairly
at least with a good

*****
The 2 5 plus 6 10 does have more entries in toto (words c. 5600 vs. 3715; phrases: 23,700 vs. 17,179. KB: 1200KB vs. 997KB. I don't know how many of these are duplicates.

The combo of a 2 5 plus a 6 10 seems to produce richer choices, with more short phrases - which is what I thought might be missing and seemed like they should be there. With some letters typed (e.g. qo, or other rare letters) there is little or no difference between the two methods but with frequent letter combinations, there being more entries, there can be quite a few differences.

I have tried the high-low approach (2 8, e.g.) before, but it seems to me 2-5 plus 6-10 does give the better results. I'll keep working with them and perhaps try other word freq and max words per phrase combos too.

Jon
Related link: ABCZ Software


Edit | Reply | Where? | | Original Message | Top | Current page | Author
Followups