This page describes the new rule actions and attributes associated with KANA Classify.
KANA Classify enhances the message handling capabilities of KANA Response. KANA Response and KANA Classify use the following attributes and actions, which are configured using properties files, to work together:
Rule attribute: classify probability
This rule attribute queries the KANA Classify Server for the confidence level of the specified category ("returns" in this case). The rule attribute returns true if the returned confidence level meets the percentage specified.
Example:
If message classified as <category1> with a probability greater than <90%>, then...
Rule attribute: autoclassify probability
This rule attribute returns true if any category confidence level returned by KANA Classify Server satisfies the specified percentage.
Example:
If the message is autoclassified for any categories with a probability greater than <90%>, then...
See Attributes Reference for more information.
Rule action: autocategorize
This rule action takes the categories returned by KANA Classify, filters out the confidence level or the top number of categories, and applies the categories to the message. A note is attached to the message indicating the confidence level returned by KANA Classify.
Example:
...Autocategorize the message with all categories having a confidence level greater than <90%>.
...Autocategorize the message with the top <3> categories.
See Actions Reference for more information.
Learn about rules and rule groups.
How do I use a regular expression in a rule?