Analysis window merging counts for two classes
Brought to you by:
louburnard
I have a bug in 1.21 where the analysis window is merging together the counts for two classes.
See enclosed PDF.
The corpus has seven texts, each in a class of its own. The search is for the XML element "q".
As you can see class 5 has a much larger count than the others and class 6 doesn't have any at all.
However, looking at the actual concordance shows that the last 120 examples or so are actually in text 6, and thus in class 6 (screenshots of this also in PDF).
screenshots of bug
Logged In: YES
user_id=1036552
Originator: NO
This is the 'hits in final text assigned to wrong class' bug and it has appeared and apparently been cured several times. Okease test in 1.22 and if still there let me have the files.
Logged In: YES
user_id=1460495
Originator: YES
The bug has not gone away in v 1.22, although it has changed. Class 2 is now being combined with class 1. See enclosed PDF. Will send actual files via email.
Andrew.
File Added: Document1.pdf