A core calculus for provenance inspection

Abstract

Recent research has been devoting increasing attention to provenance, or information describing the origin, derivation, and history of data, due to its relevance to critical issues including transparency, privacy, and security. Engineering a software system to make it provenance-aware by means of ad-hoc instrumentation requires a substantial effort: the development of general-purpose infrastructure is thus very important to achieve the goal of making provenance widely available. In this article we describe a core functional language equipped with a provenance-aware semantics that is sufficiently generic to accomodate many notions of provenance proposed in the literature. While existing proposals typically treat provenance views and provenance extraction as second-class, extralinguistic mechanisms, in our work provenance views are expressed as standard programs and provenance data can be reflected into the language, allowing for programs that inspect their own provenance.