Opened 15 years ago

Closed 14 years ago

#631 closed defect (worksforme)

incremental classifiers

Reported by: kjdon Owned by: kjdon
Priority: moderate Milestone: 2.84 Release
Component: Collection Building Severity: major
Keywords: Cc:


When incrementally updating classifiers, we first reconstruct the old document, then classify the new version with edit mode delete/update. This leads to it being in hte classifier twice in some cases. Can we just not reconstruct a doc if we know it will be deleted or updated?

part of the output:

Adding reconstructed HASH30aa188f4d8ddaef558fc9 into classify structures
Adding reconstructed HASH163120a9a8b21602ebb36c into classify structures
ArchivesInfPlugin: processing /research/kjdon/home/testing/2.83/gsdl/collect/lucenedemo/archives/archiveinf-doc.gdb
GreenstoneXMLPlugin: processing HASH1631.dir/doc.xml
Deleting old HASH163120a9a8b21602ebb36c for List
Deleting old HASH163120a9a8b21602ebb36c for Hierarchy

Change History (2)

comment:1 by kjdon, 14 years ago

Owner: changed from nobody to kjdon
Status: newassigned

comment:2 by kjdon, 14 years ago

Resolution: worksforme
Status: assignedclosed

I couldn't reproduce this. Maybe the code has changed since I found this problem originally?

Will close the ticket for now. It can be reopened if we find a collection where we can reproduce the problem.

I tried with some demo documents, and List adn Hierarchy classifiers. And AZlist and AZSectionList.

Note: See TracTickets for help on using tickets.