Hi Angela,
Thanks for joining! Unfortunately there is no way to mix binary and weighted matching in one rule. However, you can approximate it by setting the weights of the non exact attributes to low values. Attributes with an undefined weight will default to 1.0, which is the highest weight.
Notice in my example below how I set the weight to .2 for non exact attributes. Maybe try something similar and if it is still bring back undesirable results set to to .1?
Let us know how it goes!
{
"uri": "configuration/entityTypes/HCP/matchGroups/ByNames",
"label": "By Names",
"type": "relevance_based",
"rule": {
"and": {
"fuzzy": [
"configuration/entityTypes/HCP/attributes/FirstName",
"configuration/entityTypes/HCP/attributes/LastName"
],
"exact": [
"configuration/entityTypes/HCP/attributes/Addresses/attributes/Addressline1",
"configuration/entityTypes/HCP/attributes/Addresses/attributes/Zip/attribute/Zip5"
],
"exactOrAllNull": [
"configuration/entityTypes/HCP/attributes/Other"
]
},
"weights": [
{
"attribute": "configuration/entityTypes/HCP/attributes/FirstName",
"weight": 0.2
},
{
"attribute": "configuration/entityTypes/HCP/attributes/FirstName",
"weight": 0.2
},
{
"attribute": "configuration/entityTypes/HCP/attributes/Suffix",
"weight": 0.1
}
],
"actionThresholds": [
{
"type": "auto_merge",
"threshold": "0.85-1.0"
},
{
"type": "potential_match",
"threshold": "0.4-0.85"
}
],
"matchTokenClass": "com.reltio.match.token.ExactMatchToken"
},
"matchServiceClass": "com.reltio.businesslogic.match.providers.internal.InternalMatchService"
}
------------------------------
Joel Snipes
------------------------------
Original Message:
Sent: 01-21-2022 16:06
From: Angela Wawrzaszek
Subject: Relevance Based Matching
Great Webinar, with that it brings questions!
We recently configured a Rule for testing this.
In the rule we have:
Exact: Addressline1, state, zip
Fuzzy: Name
other attributes have an Exact or Null logic.
How do we assign weight? Meaning we want the Exact Rule to be 100% this must happen to consider anything else.
the Threshold should be based on the Fuzzy.
What we are seeing is that because all other attributes in the rule meet a true value and therefore the threshold is high however the Exact on Address is false and a lot of bad suggested matches are seen.
How do you ensure the Exact rules are met before the fuzzy for setting a threshold? (meaning if the Exact is False, do not consider a threshold)
------------------------------
Angela Wawrzaszek
Manager, MDM governance and Application Enablement
------------------------------