For what it's worth, you may certainly have my permission to post the little program that I wrote to illustrate fuzzy linkage/matching using a highly simplified scoring method. Some may find my example of how to use the SAS SPEDIS() function (cribbed almost directly from SI documentation) useful as they begin to experiment with fuzzy key linkage. I still feel compelled to warn anyone who does experiment with it that it forms a Cartesian product of a table and its own image. If applied to tables of more than a few thousand rows, it will likely blow up.

A couple of years ago or so I posted a SAS macroprogram that generated weights based on frequencies for elements elements of linkage keys and used the weights to calculate similarity scores per record pair. That program includes a section in which the user can specify blocking variables that SAS SQL can use to form an index. Anyone considering a fuzzy or probabilistic record linkage/matching project needs to understand blocking strategies and how to use them. Sig