US Patent 11023607 Detecting behavioral anomalies in user-data access logs

A method for detecting anomalies in audit logs of database operations performed on databases. The method includes obtaining a first audit log of database operations performed on one or more databases, generating, for each of a plurality of attribute values associated with a designated attribute appearing in the first audit log, a profile of that attribute value that indicates expected attribute characteristics of one or more attributes when that attribute value is associated with the designated attribute, obtaining a second audit log of further database operations performed on the one or more databases, and detecting an anomaly responsive to a determination that a log entry in the second audit log includes an attribute value associated with the designated attribute but attributes in the log entry deviate from the expected attribute characteristics of the one or more attributes indicated by the profile of the attribute value associated with the designated attribute.