How do the GB-dispute derivations (for computing grounded beliefs) as found in the 'Computing Ideal Sceptical Argumentation' paper differ from the AB-dispute derivations (for computing admissible beliefs)?
They differ in that the AB-dispute derivations have two extra filtering checks. As a result, the GB-dispute derivations are "less complete" and less efficient.
The GB-dispute derivations do have "filtering of potential defence arguments by culprits" and "filtering of culprits by defence assumptions" so that the final assumption-defence set constructed by the derivation does not attack itself. But do not have "filtering of defence assumptions by defences" and "filtering of culprits by culprits" and thus are "highly incomplete" for admissibility and more inefficient.
In what sense do the GB-dispute derivations compute grounded beliefs?
The proof for this is long. In observation, because there is no "filtering of defence assumptions by defences" the procedure will rightly fail (loop) for all non-well-founded frameworks.
No comments:
Post a Comment