Database Reference
In-Depth Information
f
th
k
-quantile,
(
v
i
)
is probability computed using the approximation method and
f
(
v
i
)
is the real probability of
v
i
.
7.8 Summary
In this chapter, we investigate ranking queries evaluation on probabilistic linkages.
In contrast to the traditional methods that use simple probability thresholds to ob-
tain a set of deterministic linkages, we fully utilize the probabilities produced by
the record linkage methods and consider the linked records as a distribution over
possible worlds. By preserving the distribution information, we can provide more
meaningful answers to aggregate queries, as illustrated in Section 7.7.1.
Moreover, we extend the ranking query evaluation method to answer aggregate
queries. Efficient approximation and pruning techniques are developed.
