DSXchange

In the midst of our integration, we are getting results that might be explicible in some way but since we don't know the internal comparison algorithm, I don't know if this is correct.

Using a cutoff of 800, all these comparisons were given full disagreement weights:
ANDERSON and ANDERSEN
AITKIN and AITKEN
BOOBY and BOBBY
AUGER and AUGUR
BLUEMINK anad BLUCMINK
BRITTEN and BRITTIN
BARTEAU and BARTEAN
BRACKEN and BRAGKEN
BEDDIE and BEEDIE


Can these diagreements be explained? Any suggestions on how not to get QS to do this?

Are you using special variable handling such as CRITICAL MISSING OK or are you using disagreement weight override?

How are you testing for this?
- Have you taken those records out and tested them specifically to see that the that field gives the disagreement weight?
- Are you interpreting QSMATCHPATTERN to see this?
- Or are you saying that the those names are in records that you would have expected to match, and didn't.

Just trying to make sure you mean what I think you do: I would have expected most if not all of those examples to be close enough to give a partial agreement weight at 800.

An alternative is to try them at 750, but you might be getting a bit too loose for your needs. Worth having a look though.

Hello,

Thanks for the replies so far.

We are not using disagreement weights, nor declared it CRITICAL.

We are seeing them because we are looking at the match pairs in the Test Environment and seeing the weight contribution of that variable. (First name and sex are typo free and getting full agreement weights.)

I would expect these to match at 800 as well. I'd prefer to use 850 (and still would expect these to match).

It's pretty surprising. I wouldn't have thought it was that picky at 800.
There aren't any control characters or trailing spaces, etc? How do they look with a hex editor? Maybe there is more than the 1 character difference you can easily see.
Grasping at straws a bit here, TBH.