When dealing with large data sets, for example, during data preparation of big data, one usually attempts to apply generic coded rules to every data item. However, it can sometimes be hard to craete rules that aply to everything, for example, to deal with "Henry VIII" or "The Artist Formerly Known as Prince" in ruels fpr processing names. Rather than creating ever more complex, and potentially buggy, rules; one can instead create exception sets that say, for example, that "Henry VIII" is a valid name, or "treat the age field in record 7932 as 88". These exception sets are then applied before generic rules as a form of override. This is especially important when the input data set may beuodated and then re-processed.
Used on page 198