PROV-ISSUE-81 (identity-clash-scope): In a given scope, are entities with same identifier but different attributes legal? [Conceptual Model]

PROV-ISSUE-81 (identity-clash-scope): In a given scope, are entities with same identifier but different attributes legal? [Conceptual Model]

http://www.w3.org/2011/prov/track/issues/81

Raised by: Luc Moreau
On product: Conceptual Model




Let us consider two entity assertions, inspired by those discussed in [1].

entity(http://dx.doi.org/10.1007/978-3-642-17819-1_37, [author = "Jim Myers", pagenumber={15-17}])

entity(http://dx.doi.org/10.1007/978-3-642-17819-1_37, [author = "Jim Myers", reviewed={yes}])


Let us note that they have the same identifier but they have different
attributes.


What does it mean to have these two assertions occurring together in
the provenance of something?


1. If they were asserted by the same asserter, I would argue this is
   not well formed provenance. Again, having a scoping construct is
   useful, and we could introduce the following constraint:

   Within an account, two entity assertions with the same identifier
   must have the same attribute-value pairs.

2. Let us now imagine that the two assertions were created in separate
   accounts (alice's and bob's), but now, we decide to "merge" all assertions
   together.

  2.1.  The identifier had a scope that was local to the account in which it occurs.

        Then it's OK again, in a sense, since we could apply an
        alpha-conversion, renaming consistently the identifier in its
        account before merging, so as to avoid a clash.  The two
        entities would be regarded as different, because having
        different attributes (they just happened to have the same
        identifier in their respective scope).

  2.2 The identifier has a global scope. Then again, the same
  constraint as above should apply (replacing account by global
  scope).


[1] http://lists.w3.org/Archives/Public/public-prov-wg/2011Aug/0326.html

Received on Wednesday, 24 August 2011 20:56:12 UTC