Wiki

Clone wiki

OYSTER / Index_Fewer_Attributes_Strategy

It is not necessary to index every attribute used in a rule. For some situations it may be better for the match-key generator to use fewer inputs in order to attain alignment. The left-most box of Figure 26 shows the same misalignment of match rules and index as in Figure 24. The right-most box of Figure 26 shows how this misalignment can be corrected by dropping the hash on the FirstName attribute and only generating the match-key from the LastName attribute.

Screen Shot 2019-09-10 at 5.21.23 PM.png As illustrated in Figure 27, for a given set of attributes and their corresponding hash algorithms, it will always be true deleting one or more of the terms (i.e. an attribute and its hash algorithm) will never decrease the number of pairs of references generating the same match-key, and in most cases will increase that number. Although indexing on fewer attributes may help attain rule-to-index alignment, the fact that it returns more candidates for matching may result in a performance problem. Another strategy that can help to address the performance issue while still attaining alignment is to implement more than one index.

Back to OYSTER Reference Guide page

Click Prev Alignment and Scenarios page

Click Next Multiple Index Strategy page

Updated