Pure Data Provenance

The point of this use case is to highlight that data provenance is not the same as assertions about truth. The goal in this scenario is to accurately document the origins of different purported facts, even though the person assembling the final report does not necessarily believe all of them.

Deepak is a forensic accountant working for Interpol. He is studying a complex pattern of behavior by Malfoy, an individual who's been accused of money laundering. Malfoy is associated with many different shell corporations.

Deepak assembles a report that details Malfoy's activities. Each fact in the report is meticulously sourced. In many cases, the source is a published document (simple URL). One source is a personal interview with an anonymous informant known as X. Another source is a series of personal interviews with Tanya, an officer of a public company, who digitally signs the transcripts of her testimony. X and Tanya both cite other individuals – sometimes as hearsay, sometimes in ways that can be independently corroborated.

For many facts, there is more than one source.

For some facts, the sources disagree (e.g., about the particular date or amount of a financial transaction). Some of the sources are more reliable than others, so Deepak would like to express a degree of credibility on some of his provenance.

In a few cases, Deepak makes assumptions. These are not facts, but they are likely to be true. Deepak would like to attribute himself with varying degrees of confidence to the provenance of these assumptions (potential/likely facts).