International Business Machines Corporation

Brian A O'Crowley

Liam S Harpur

Jonathan D Dunne

Trudy L Hewitt

Robert Huntington Grant

Gary Denner

Daniel C Gurney

An approach is provided in which a system analyzes a first set of data to derive a first distribution output that is based on a first conjugated distribution corresponding to the first set of data and a domain class model. The system utilizes the first distribution output as a baseline input to generate a second conjugated distribution corresponding to a second set of data and the domain class model. Next, the system derives a second distribution output of the second set of data based on the second conjugated distribution. The second distribution output identifies at least one personally identifiable information (PII) data field corresponding to the second set of data that was not identified as a PII data field in the domain class model. In turn, the system tags at least a portion of the second set of data as PII based on the derived second distribution output.

Personally identifiable information determination

An injection engine monitors whether a pattern of a current selection of web address requests in outbound web traffic for a user matches one or more web usage patterns that allow for identifying the user based on the current selection of web address requests. The injection engine, responsive to detecting the pattern of the current selection of web address requests by the user matches the one or more web usage patterns, injects one or more random valid web address requests into the outbound web traffic for the user, wherein the one or more random valid web address requests obscure the current selection of web address requests from using the one or more web usage patterns.

Obscuring user web usage patterns

Growing community of inventors

Dublin, Ireland

Brian A O'Crowley