We implemented a new merge/match rule as an auto rule and we are finding that the fuzzy logic is not applying to all the addresses in the same way. Below in the "expected results" and "actual results" sections are some examples of Potential Matches that were caught and merged and also ones that were not caught by the new rule implemented.
The merge/match rule that was implemented is as follows: ExactFirstNameLastNameDOBUID2FuzzyAddress
We have run the API to verify the matches and everything for our exact logic is coming up as 'true' but the fuzzy logic of address line 1 is coming up as 'false.'
We believe that this is occurring because our address line 1 field is not tokenized into separate pieces, instead all of the address line 1 is being tokenized together. Further, our logic then goes on to state that for the value we have "ExactMatchToken."
The below examples (in Actual Results) show that the address line 1 for each is not matching based on them being different but the fuzzy logic is also not picking it up because of the ExactMatchToken value and the address line 1 not being tokenized separately.
These two members merged when they had two different apt numbers.
Why these two value does not fall in fuzzy
"value": "58 Summer St",
"value": "58 Summer St Apt 1"
We have a certain match token and comparator class is our match rule, then why that pattern is catching them.
the current match rule can bring up the following attributes together.
58 Summer St
Summer St 58
Summer 58 St
Summer St 5 8
Also, the fuzzy match rules purely depend upon the type of class that we use. For the address, you can try to use AddressLineMatchToken and its comparator.
You can also try to use AddressLineMatchToken and its comparator.
- Splits the attribute value into separate words.
- Removes all numbers. If the address consists only of numbers, then "ZERO-ADDRESS" is returned.
- Removes all garbage words (such as ste, blvd, ave). If the address consists only of garbage words, then all garbage words are preserved.
- Sorts the remaining words alphabetically.