In a large database environment, ie Teradata, there could be huge number of system/ catalog information. Each read can be costly. If you have a classifier policy with many rules, ie: GDPR template, the cost of reading this static information per rule can be significant and makes the process run longer. If at the beginning of the scan Classifer read the catalog information once and reuse that to evaluate all the catalog search, it will improve the performace quite a bit.
Do not place IBM confidential, company confidential, or personal information into any field.