DSXchange

Hi All,

I am currently developing a Match specification to match First Name, Middle Name, Last Name, State, ZIP and Phone Number. I have some queries regarding the actual implementation:

1. This is going to be a reference Match and the both the data as well as the reference set keep getting refreshed in production. Do we have to generate the Match Frequency before each run?

2. When we specify multiple passes do only the unmatched records from the first pass go through the successive passes till it encounters a reasonable match?

3. I would prefer to do a direct lookup on the state code and ZIP rather than include them in the Match spec. Is that advisable or would u recommend I include it as a part of the spec?

Thanks,
Madhumitha

1. You have to have the match frequencies. Whether you generate them for the reference data set immediately ahead of running the match or not depends on (a) whether the reference data set is fairly static and (b) your available time windows.

2. For two-source matches, yes. For one-source matches you can control this.

3. These are good differentiators; I would tend to keep them in the match specification. Are you planning to use the lookup merely for validation or to enrich the locality information? In the latter case I'd suggest doing both.

Hi Ray,

Thanks for the clarification. I need to use the State, Zip and phone numbers for narrowing down the matches in case of duplicates in name match and for validation if there is a single match.

Try blocking also on NYSIIS of primary name and initial of first name.