Text classification

When you enable DLP document classification options on this page, documents will be transmitted to the server and classified according to the categories specified here.
The classification result can be viewed in the "Files Classification" report.
In the current version, classification is performed through LLM AI (settings are here).

Each category consists of a unique name (English letters only!) and a description.
It is advisable to use the word "probability" in the description.
If you need the "other" category, it is better to place it last in the list.

Condition for event formation
If a condition is specified, then when it is met, the corresponding DLP-event is recorded in the "Events: user" report. See events settings here.
The condition itself is a logical expression of comparisons of greater/less for category values (in percents), with the possible use of AND/OR operators, as well as parentheses.
For example, condition (confidentiality >= 70) AND ((agreement >= 60) OR (statement >= 60)) means "if the document's confidentiality more than 70% and at the same time one of two conditions (or both): the probability of an agreement is more than 60% or the probability of an statement is more than 60%".