Tag-based Information Flow Analysis for Document Classification in Provenance

A crucial aspect of certain applications such as the ones pertaining to Intelligence domain or Health-care, is to manage and protect sensitive information effectively and efficiently. In this paper, we propose a tagging mechanism to track the flow of sensitive or valuable information in a provenance graph and automate the process of document classification. When provenance is initially recorded, the documents of a provenance graph are assumed to be annotated with tags representing their sensitivity or priority. We then propagate the tags appropriately on the newly generated documents using additional inference rules defined in this paper. This approach enables users to conveniently query to identify sensitive or valuable information, which can now be efficiently managed or protected once identified.