The authors said in the beginning that this paper is meant to be a critical review of approaches to capture provenance for RDF data, demonstrate their use in examples, and compare different methods. IMHO, the paper achieved these goals. It is easy to read and contains various important information about provenance in semantic systems: data models, annotation frameworks, KOS, serialization/syntaxes, algebras, mechanisms, examples, evaluation, and limitations.
The paper identifies and describes variants of PROV methods and components that are available in the Semantic Web environments. These include PROV data representation, formal semantics, and serialization (including various provenance granularities). Furthermore, it discusses existing data models, ontology, and vocabularies, which goes beyond the standard PROV-DM and PROV-O. It also introduces several provenance data manipulation tools, libraries, storage solutions, and workflow engines currently available.
An important takeaway from the paper is written in the conclusion part of the paper: “a variety of use case scenarios require a trust mechanism that can be supported by capturing data provenance. Such trust mechanisms are important for a semantic system. Intelligent systems implementing SW standards need provenance manipulating capabilities to be viable, especially in settings where RDF triples are derived from diverse sources and are generated and processed on the fly, or modified via update queries”.
The statement resonates very well with our group, as we are currently working on several projects dealing with Auditability of (RDF-based) information systems (i.e., OBARIS, WellFort, and VasQua) – I am looking forward to checking out these methods and tools for our projects! Sikos, Leslie F., and Dean Philp. 2020. “Provenance-Aware Knowledge Representation: A Survey of Data Models and Contextualized Knowledge Graphs.” Data Science and Engineering. http://link.springer.com/10.1007/s41019-020-00118-0.