8. Filters

In this chapter we examine several filters that can be used for choosing the best match candidates from the list of ranked map pairs returned by the Similarity Flooding algorithm. Usually, for every element in the matched models, the algorithm delivers a large set of match candidates. Hence, the immediate result of the fixpoint computation may still be too voluminous for many matching tasks. For instance, in a schema matching application the choice presented to a human user for every schema element may be overwhelming, even when the presented match candidates are ordered by rank. We refer to the immediate result of the iterative computation as multimapping, since it contains many potentially useful mappings as subsets.