Next: Michael Gelfond
Up: Statements of the panelists
Previous: Statements of the panelists
raised in his comments one more question: In most
approaches, ASP solutions to a
problem are represented by answer sets. While doing answer set
programming, is one still interested in entailment?
His answers to questions 1.-4. are:
- Ad Q1.
- Although most languages/language extensions have a clean
formal characteristics, we still lack of more useful building
block results (like e.g. splitting sets) for proofs.
- Ad Q4.
- In answer set programming solutions are answer sets, while
in prolog, solutions are substitutions. If we deal with lists/sets
of assignments, this makes a major difference.
An interesting point is when we deal with problems in P vs
problems in NP. For problems in P there are nice efficient
solutions in Prolog, while in ASP we have not so much control on
the search and we have to effectively indirectly guess
solutions. A motivating example is List concatenation:
conc([],L,L).
conc([H|L1],L2,[H|L3]) :- conc(L1,L2,L3).
While in Prolog this can be done efficiently, in ASP we can
represent the data structures to some extent but guess all
solutions, i.e. lists in some sense.
Chitta reminded at this point Piero Bonatti's approach for finitary
programs and restricted
use of function symbols in this context.
Next: Michael Gelfond
Up: Statements of the panelists
Previous: Statements of the panelists
Stefan Woltran
2005-08-22