To create classification profiles

  1. On the Data Discovery menu, select Classification Profiles. The Classification Profile screen appears.

  2. Click the Create New button. The Create Classification Profile screen appears.

  3. Provide a new name and description for the profile.
  4. For the Athena AI field, manage your preference to run discovery patterns.
  5. For the Document Classification field, manage your preference to run the classification on your documents.


  6. In the Advanced Settings section, enable the following settings at your preference to conduct post-processing logic.

    About Post-Processing Logic: there are many cases with structured databases that the column name is a clear indicator of what resides in that field. However, not every indication can be correctly assessed through just Lookup patterns as those names are possibly contextual and multifunctional (e.g. FirstName, LastName, City, and Country could all match the same data element).

    These advanced settings leverage logic to better capture the data elements from obvious fields and use confidence score recommendations to intelligently apply metadata matching.

    Note:

    Advanced Settings is currently under preview.



    Advanced Setting Name

    Description

    Logistic Regression

    If turned off and there is a majority of matching classifiers, the system will apply a base confidence score.

    Important:

    OneTrust recommends to disable this setting by default unless users are testing more advanced discovery use cases.

    If turned on and there is a secondary classifier match within the column, the system will adjust the confidence score(s) based on the overall probability of each classifier present in the column.

    Real-World Interaction

    Change the predicted labels to the highest attribute if there is a real-world interaction between two recommendations, meaning they share the same pattern or overlap in a way that could be misconstrued as another PII.

    Example scenario: Social Security Numbers has hyphens, Canadian Social Insurance Number (SIN) with no hyphens/special characters.

    Unknown Remapping

    If 50% or more of the column matches a recommended classifier, or if the column name matches to a glossary term, this will change all unknowns (i.e. no classifier matches) to that majority classifier.

    Example scenario: An uncommon name, which cannot be matched by another classifier, will be given Name if the majority of the column is in fact individual names.
  7. Click the Next button. The Select Classifier Groups screen appears.

  8. Search and select for the desired System Classifier Groups which need to be added to the profile. Custom classifiers will be available for selection in the Custom Classifier Groups.


  9. When you've selected all the appropriate classifier options, click the Next button. The Classifier Groups Settings screen appears.

  10. For each selected classifier, enter the sample size count. You can also bulk select classifiers to assign the sample size to all selected.


  11. Click the Apply to Selected button to assign the sample size count.
  12. Click the Save button. The new Classification Profile will be created and available on the main Classification Profiles screen.