Database Reference
In-Depth Information
What about more general queries, say FO queries? We know that in the case of CWA,
we can always answer them (and the complexity is CO NP-complete). In the case of OWA,
finding certain answers is an undecidable problem. Annotated mappings give us some ad-
ditional cases, compared to OWA, when query answering is decidable.
The parameter that we use is the following:
# op (
) - the maximum number of open positions per atom in an st-tgd in the anno-
tated mapping.
M
,
α
In our example, the value of # op (
M
,
α
) is 1, as no st-tgd uses more than one open position.
Theorem 8.35 The data complexity of finding certain answers for FO queries under
annotated mappings is as follows:
1. in CO NP if # op (
)=0 (which is the CWA case);
2. decidable (in CO NEXPTIME )if # op (
M
,
α
M
,
α
)=1 .
3. undecidable for some mappings (
M
,
α
) with # op (
M
,
α
) > 1 .
Thus, if more than one position in a rule is marked op , query answering behaves like
in OWA. With no op positions, we have the CWA case. The new intermediate case is one
open annotation per rule, and there we retain decidability of query answering.
Search WWH ::




Custom Search