KB Articles
KBID: KB0008915
Topic/Category: Search Classification, Recostar, tif, Learn
Issue: Very Large Sample Set Unable to Learn
Analysis:
Recostar may not be able to process the files with respect to the following. Please check your files and try the suggestions:
RecoStar has a pixel size limitation: Maximum document height : 24000 pixels; Maximum document width: 32000 pixels
Ideally tif’s used are ones normalized by ephesoft “Folder Import” process and retrieved for use from [SharedFolders\ephesoft-system-folder\BIxx]
Learn files should all be single page case sensitive tif files
Corrupted files may be preventing the learn from occurring successfully.
A large sample set of files may require temporarily changing the timeout periods in Ephesoft.
Solution:
- Delete all learn temp files in [SharedFolders\BCx\lucene-search-classification-sample] – bd, xml, rsbatch
- Delete all learn index files located in [SharedFolders\BCx\learn-index]
- Increase timeout period for UI-session-expire:
http://wiki.ephesoft.com/session-timeout - Increase timeout period for Recostar-wait
Since number of learn files are large, Recostar Learn process times-out after a specific interval of time which can be configured from application.properties.
To learn such a large set of files user will require to increase the wait time in application.properties file.
Please follow the steps to increase the wait time.
1.) Stop Ephesoft Server.
2.) Open the application.properties file under <Ephesoft-Installation-Directory>/WEB-INF/classes/META-INF/application.properties
3.) Provide a new value for the property recostar.command.wait.time=<new-time-in-minutes>
Once Learn Index is created and batch class definition is finalized, it is recommended to set the timeout period back to default.
