A canonical form for PROV documents and its application to equality, signature, and validation

We present a canonical form for PROV that is a normalized way of representing PROV documents as mathematical expressions. As opposed to the normal form specified by the PROV-CONSTRAINTS recommendation, the canonical form we present is defined for all PROV documents, irrespective of their validity, and it can be serialized in a unique way. The article makes the case for a canonical form for PROV and its potential uses, namely: comparison of PROV documents in different formats, validation, and signature of PROV documents. A signature of a PROV document allows the integrity and the author of provenance to be ascertained; since the signature is based on the canonical form, these checks are not tied to a particular encoding, but can be performed on any representation of PROV.