This thesis proposes a domain-specific approach to compressing large volumes of event log records by exploiting the known or static structure of the data and patterns identified through observation.
Please check the details in the provided link for the thesis.