Reltio Connect

Hi Team,

We all are aware that Match rules can be implemented as per requirements as much based on the data resides in the tenant. 

Thinking about the priority how the match rules works on few scenarios. 

Scenario - Have configured the Match rules as below

Match Rule 1 - Name + Website + Country
Match Rule 2 - Website + Country
Match Rule 3 - Primary Key + Country
Match Rule 4- Primary Key

Data Example :
Match Rule Precedence to be taken - Rule 1 -> Rule 2 -> Rule 3 -> Rule 4
If the record doesn't match based on Rule 1 then Rule 2 should pitch in which goes on till Rule 4.
Every Match rule will have unique match scores, attributes and configuration setup as per requirements.

As per my understanding, whichever gives best score it falls in respective category despite the order, but just thinking does it work precedence way as mentioned above.

Thanks & Regards,

------------------------------
Chetan P
Lead MDM Specialist
Freshworks
Chennai
------------------------------

No, all rules are effectively evaluated independently in parallel.  If any rule is set as an Automatic rule it will trigger the immediate merging of profiles, unless there is a Negative match rule blocking an automatic merge (like because they have different SSN, TaxID or DoB).

Match Score will be calculated as the weighted sum of the rules which produce a positive match.  In defining scores for each rule there is a standAloneScore and an incrementalScore.  The standAloneScore is taken from the positive match rule which has the highest base score.  The incrementalScore is then added for all other rules which generated a positive match.  More details are here.

There is not a definitive science to assigning these scores as they will need to be re-tuned each time a new rule is introduced - but the score primarily serves as guidance to the Data Steward on the confidence of the match if an automatic merge did not take place.  You can use this score in searching and segmenting profiles for stewarding - effectively to prioritize higher confidence matches of those not auto-merged.  

------------------------------
Dan Gage
Principal Solution Consultant
------------------------------

Thank you so much @Daniel Gage for this response. This helps.

Regards,

------------------------------
Chetan P
Lead MDM Specialist
Freshworks
Chennai
------------------------------

Original Message:
Sent: 08-01-2023 15:04
From: Daniel Gage
Subject: Match Rule Functionality | Details

No, all rules are effectively evaluated independently in parallel.  If any rule is set as an Automatic rule it will trigger the immediate merging of profiles, unless there is a Negative match rule blocking an automatic merge (like because they have different SSN, TaxID or DoB).

Match Score will be calculated as the weighted sum of the rules which produce a positive match.  In defining scores for each rule there is a standAloneScore and an incrementalScore.  The standAloneScore is taken from the positive match rule which has the highest base score.  The incrementalScore is then added for all other rules which generated a positive match.  More details are here.

There is not a definitive science to assigning these scores as they will need to be re-tuned each time a new rule is introduced - but the score primarily serves as guidance to the Data Steward on the confidence of the match if an automatic merge did not take place.  You can use this score in searching and segmenting profiles for stewarding - effectively to prioritize higher confidence matches of those not auto-merged.  

------------------------------
Dan Gage
Principal Solution Consultant
------------------------------

one additional thing to be aware of Chetan per your example.  all matches that are identified on match rule 1 will also be caught on match rule 2 because it's effectively just a weaker version of match rule 1 (ie you don't have the name in rule 2).  Same applies to rule 4, all the matches from rule 3 will also be there on rule 4 (plus additional ones obviously that rule 4 detects given it's only focusing on primary key).  ideally you should not be thinking of creating rule x if rule x is effectively a copy of another rule minus one or more attributes because then rule x has all it's matches plus all those from that other rule it was based on.   ideally each rule is looking at a different set of attributes and not just weaker versions of other rules.

------------------------------
Guy Vorster
------------------------------

Original Message:
Sent: 08-03-2023 00:25
From: Chetan P
Subject: Match Rule Functionality | Details

Thank you so much @Daniel Gage for this response. This helps.

Regards,

------------------------------
Chetan P
Lead MDM Specialist
Freshworks
Chennai
------------------------------


Original Message:
Sent: 08-01-2023 15:04
From: Daniel Gage
Subject: Match Rule Functionality | Details

No, all rules are effectively evaluated independently in parallel.  If any rule is set as an Automatic rule it will trigger the immediate merging of profiles, unless there is a Negative match rule blocking an automatic merge (like because they have different SSN, TaxID or DoB).

Match Score will be calculated as the weighted sum of the rules which produce a positive match.  In defining scores for each rule there is a standAloneScore and an incrementalScore.  The standAloneScore is taken from the positive match rule which has the highest base score.  The incrementalScore is then added for all other rules which generated a positive match.  More details are here.

There is not a definitive science to assigning these scores as they will need to be re-tuned each time a new rule is introduced - but the score primarily serves as guidance to the Data Steward on the confidence of the match if an automatic merge did not take place.  You can use this score in searching and segmenting profiles for stewarding - effectively to prioritize higher confidence matches of those not auto-merged.  

------------------------------
Dan Gage
Principal Solution Consultant