What’s in a name? Exploiting URIs to enrich provenance explanations in plain English

Provenance allows decision-makers to evaluate the importance of pieces of data. PROV is the standardised model of provenance for use on the web, particularly suited for situations where data is generated by systems under distributed control, such as in coalition operations. If human decision-makers are to make effective use of provenance data, they need to understand it, and this work establishes techniques for explaining PROV graphs to human users in natural English. In this paper, we demonstrate the potential role of exploiting the linguistic information that is informally encoded in the URIs used to denote provenance data resources to generate these more natural English explanations of provenance. We show how this additional linguistic information allows us to generate richer, more readable explanation texts, thus enabling better decisionmaking and increasing the value of preexisting provenance data.