Documents not meeting MinimumSimilarity getting Classified

strotelli

I have documents that should end up unclassified after the classification step. When i test them they show up as a 0% match to any other document type, however when they are run through classification they end up with a document type.

GrooperGuru

Couple thoughts here... First, do they always end up as the same doc type or do they end up assigned to different doc types? If the same, I would suspect that you have defined a Default Doc Type on your Content Model. If different, I would then ask if you are running classification using a service or through manual task processing. The Classification service can retain old versions of the content model. So if you've been actively training classification, the service may not see the latest and greatest changes you've made. At some point in 2.6, we added a feature to force the Classification service to get changes automatically without having to be restarted. It is a setting named Model Refresh Interval, and you configure it in the classification step of the batch process. The setting is described as follows:

"The interval (in seconds) at which content model information will be refreshed from the repository. WARNING: if set to 0 this feature will be disabled, and Content Model information will not be updated until the service(s) are restarted."