Provenance-based Explanations for Automated Decisions

Algorithms play a key role nowadays in many technological systems that control or affect various aspects of our lives. Decisions made by those systems are often driven by data and their processing, both of which are typically shielded from end users that are affected by them. For this reason, various recent regulations require organisations to explain and justify automated decisions affecting their data subjects.

Provenance, and its standard PROV, which describes how a piece of information or data was created and what influenced its production, provides the technological capability to explain complex decision pipelines and their outcomes.

On this website, we present a loan scenario, a credit card application scenario, and a school application scenario, in which we explored ways to explain various aspects of automated decisions to end users from the recorded provenance of those decisions.

The credit card and school application scenarios was developed as part of the PLEAD Project and funded by EPSRC, Grants EP/S027238/1 and EP/S027254/1.

The loan scenario was carried out in collaboration with the Information Commissioner's Office and University of Southampton; it was funded in part by EPSRC Impact Acceleration Account at King's College, London.

KCL logo