Beyond Success and Failure

We study a new programming framework based on logic programming where success and failure are replaced by predicates for adequacy and inadequacy. Adequacy allows to extract a result from a partial computation, and inadequacy allows to exibly constrain the search space. In this parameterized setting, the classical result of independence of the selection rule does not hold. We show that, under certain conditions, whenever there exists an adequate derivation there is one in which only so-called needed atoms are selected. This result is applied in a practical setting where adequacy is expressed using a notion of request.