Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.anomalyarmor.ai/llms.txt

Use this file to discover all available pages before exploring further.

Auto-classification scans your column names and metadata to identify potentially sensitive data like emails, phone numbers, and SSNs. This helps you track PII for compliance without manual tagging.

​
How It Works

Auto-classification runs automatically during discovery:
  1. Discovery scans your database metadata
  2. Pattern matching checks column names against known PII patterns
  3. Tags applied automatically to matching columns
  4. Results visible in the asset detail page
Auto-classification analyzes column names and types only, never actual data values. A column named email gets tagged, but we never read the emails inside it.

​
Detected Patterns

PatternExample Column NamesTag Applied
Emailemail, user_email, email_addresspii:email
Phonephone, phone_number, mobilepii:phone
SSNssn, social_security, tax_idpii:ssn
Credit Cardcc_number, card_number, credit_cardpii:credit-card
IP Addressip, ip_address, client_ippii:ip-address
Addressaddress, street, zip_codepii:address
Namefirst_name, last_name, full_namepii:name

​
Viewing Classified Data

To see which columns have been auto-classified:
  1. Go to Assets in the sidebar
  2. Click on any asset to view details
  3. Look for Classification tags on columns
  4. Or filter: Assets → Filter → Classification → select a PII type

​
Enabling Auto-Classification

Auto-classification is enabled by default for all data sources. To verify or change:
  1. Go to Settings → Data Classification
  2. Check that Auto-classification is toggled ON
  3. Select which patterns to detect (all enabled by default)

​
Accuracy and Limitations

Auto-classification uses pattern matching on column names, which means: It will catch:
  • Columns with obvious names (email, phone_number, ssn)
  • Common naming conventions (user_email, billing_address)
It may miss:
  • Abbreviated names (em, ph)
  • Non-English names
  • Ambiguous names (data, value, field1)
False positives:
  • email_sent_count might be tagged as email (it’s not PII)
  • Review classifications and remove incorrect tags manually

​
Customizing Detection

To adjust which patterns are detected:
  1. Go to Settings → Data Classification
  2. Under Auto-classification Patterns, toggle patterns on/off
  3. Changes apply to the next discovery run
Run a manual discovery after changing patterns to see updated classifications immediately.

​
Manual Override

If auto-classification gets it wrong: To remove an incorrect tag:
  1. Go to the asset detail page
  2. Find the column with the incorrect tag
  3. Click the tag → Remove
To add a missing tag:
  1. Go to the asset detail page
  2. Click Add Tag on the column
  3. Select the classification that matches the data type (e.g., pii:email for email addresses, pii:phone for phone numbers)

​
Verification

After discovery runs, verify auto-classification is working:
  1. Go to Assets → Filter → Classification → Has any
  2. You should see assets with classified columns
  3. Click into an asset to see which columns were tagged
If no assets appear:
  • Verify auto-classification is enabled in Settings
  • Check that discovery has run recently
  • Your column names may not match detection patterns

​
Common Questions

​
Does auto-classification read my actual data?

No. Auto-classification looks only at column names, types, and metadata. Column values are never read or transmitted. A column called email is tagged based on its name, not on inspecting what’s inside.

​
What PII types get detected out of the box?

Email, phone, SSN, credit card, IP address, physical address, and name. Detection runs automatically during discovery and tags matching columns with values like pii:email or pii:phone.

​
Why did a non-PII column like email_sent_count get tagged?

Pattern matching is name-based, so anything with email in the name matches. Open the column, click the tag, and remove it manually. Manual overrides persist across future discovery runs.

​
Will it miss PII in abbreviated or non-English column names?

Yes, it’s likely to miss short names like em or ph, non-English names, and ambiguous ones like data or value. Add the appropriate pii:* tag manually from the asset detail page for those columns.

​
Is auto-classification on by default?

Yes, for all data sources. You can toggle individual patterns on or off in Settings → Data Classification. Changes take effect on the next discovery run.

​
Next Steps

Custom Tags

Add your own classification tags

Data Handling

How we protect your data