“…As discussed earlier, this is computationally feasible only for small data sets. In practise, blocking, filtering, indexing, searching, or sorting algorithms [2,9,15,21,23] are used to reduce the number of record pair comparisons as discussed in Section 2.1. The aim of such algorithms is to cheaply remove as many record pairs from the set of non-matches U that are obvious nonmatches, without removing any record pairs from the set of matches M .…”