From Provenance WG Wiki
Jump to: navigation, search

Comment on Monotonicity

Paolo writes: "resource r in state s. Its provenance prov(r_s) is a subset of prov(r_{s'}) for any s' that temporally follows s."

It's an interesting property, but is it true?

prov( yesterday) is a subset of prov( today)?

potentially, there is nothing common between the states of these resources. Why would the provenance of the latter resource state include the provenance of the former?

--Luc Moreau 09:34, 1 June 2011 (UTC)


Can you really "retract" provenance (history) of a piece of data? not in the material world, i.e., the archetypical provenance of a painting. It moves on, but the history is immutable.

In general I don't think you can say that the states are unrelated to each other. Maybe I think too functional. To me state s+1 comes about after state s through a transformation f. So prov(r_{s+1}) is a composition of prov(r_s) plus the derivation induced by f.


I think the real issue is that you just require temporal order between the two resource states, without requiring causal order (I know what do we mean by causal order, but let's assume for now we understand). In your example of a painting, there is causal order.

--Luc Moreau 11:26, 1 June 2011 (UTC)

In fact, looking at my paper with Natalia and Jan, derivation implies temporal order, but not the other way round.

--Luc Moreau 11:44, 1 June 2011 (UTC)

This view does not seem to consider concurrency. On the web, two clients may have separate sessions, with a same resource. I don't think that r_s generated in session 1 is necessarily dependent on r_s' generated in session 2 at an earlier time.

An illustration is a web page that returns the number of times a given client has downloaded this page.

--Luc Moreau 20:28, 1 June 2011 (UTC)

I raised a different example at

Responding to Paolo's comment about retracting the provenance of some data, then sure. But what if the data itself is removed from some larger structure, does its provenance still apply to the larger entity?

-- Graham Klyne 2011-06-02 12:45 (BST)