How is Suggester Merge Confidence Score Calculated?
H
Hitesh PAU
started a topic
over 1 year ago
I have 8 golden records suggested for merge with a Suggester Merge Confidence Score = 45. If I look at each individual matches the score averages to >97, which should have caused an automatic merge.
Curious to know How is Suggester Merge Confidence Score Calculated?
The B_MACTHSCORE is defined for all effective couples of masters identified in DU.
However the suggested merge conf score is based on all couples of suggestions.
So you'll need first to identify all the masters present in your suggested merge group (B_SUGGMERGEID) and then query the DU table to bring back all those couples of masters to determine the end suggested merge conf score.
The B_SUGGMERGECONFSCORE is the confidence score for the suggested match group.
If you check all the records for this suggestion group, do you have more results that would tend to this score of 45.
You can check this with the column B_SUGGMERGEID.
H
Hitesh PAU
said
over 1 year ago
The screenshot shows all the records for one suggested merge id
A
Alexia SIMOND
said
over 1 year ago
Hello Hitesh,
To understand what is happening here, we will need to exchange further explanations and screenshots.
So I will open a ticket based on this forum and we can take it over there.
If there is anything interesting to share to the community, I will update this forum afterwards.
Thank you and talk to you over there.
A
Alexis SCHAUVING
said
9 months ago
Hello !
Same question here, I'm debugging a matcher and here is my situation :
- in DU table, I get a B_MATCHSCORE of 85 on 5 records => same name, first name and email, so I get the expected B_MATCHSCORE
- but then in MI table, the B_SUGGMERGECONFSCORE decreases to 61
When I debug into the the management > integration batches, I can find all SQL queries updating DU and MI tables, EXCEPT the one which (I think) calculates the B_SUGGMERGECONFSCORE, in the step "Refine Non Singleton Groups" :
Could you please provide the SQL executed in this part of the integration process so I can finally find how the B_SUGGMERGECONFSCORE is calculated ?
Thanks !
A
Alexia SIMOND
said
9 months ago
Answer
Hello Alexis,
The B_MACTHSCORE is defined for all effective couples of masters identified in DU.
However the suggested merge conf score is based on all couples of suggestions.
So you'll need first to identify all the masters present in your suggested merge group (B_SUGGMERGEID) and then query the DU table to bring back all those couples of masters to determine the end suggested merge conf score.
Hitesh PAU
I have 8 golden records suggested for merge with a Suggester Merge Confidence Score = 45. If I look at each individual matches the score averages to >97, which should have caused an automatic merge.
Curious to know How is Suggester Merge Confidence Score Calculated?
Hello Alexis,
The B_MACTHSCORE is defined for all effective couples of masters identified in DU.
However the suggested merge conf score is based on all couples of suggestions.
So you'll need first to identify all the masters present in your suggested merge group (B_SUGGMERGEID) and then query the DU table to bring back all those couples of masters to determine the end suggested merge conf score.
- Oldest First
- Popular
- Newest First
Sorted by PopularAlexia SIMOND
Hello Hitesh,
The B_SUGGMERGECONFSCORE is the confidence score for the suggested match group.
If you check all the records for this suggestion group, do you have more results that would tend to this score of 45.
You can check this with the column B_SUGGMERGEID.
Hitesh PAU
The screenshot shows all the records for one suggested merge id
Alexia SIMOND
Hello Hitesh,
To understand what is happening here, we will need to exchange further explanations and screenshots.
So I will open a ticket based on this forum and we can take it over there.
If there is anything interesting to share to the community, I will update this forum afterwards.
Thank you and talk to you over there.
Alexis SCHAUVING
Hello !
Same question here, I'm debugging a matcher and here is my situation :
- in DU table, I get a B_MATCHSCORE of 85 on 5 records => same name, first name and email, so I get the expected B_MATCHSCORE
- but then in MI table, the B_SUGGMERGECONFSCORE decreases to 61
When I debug into the the management > integration batches, I can find all SQL queries updating DU and MI tables, EXCEPT the one which (I think) calculates the B_SUGGMERGECONFSCORE, in the step "Refine Non Singleton Groups" :
Could you please provide the SQL executed in this part of the integration process so I can finally find how the B_SUGGMERGECONFSCORE is calculated ?
Thanks !
Alexia SIMOND
Hello Alexis,
The B_MACTHSCORE is defined for all effective couples of masters identified in DU.
However the suggested merge conf score is based on all couples of suggestions.
So you'll need first to identify all the masters present in your suggested merge group (B_SUGGMERGEID) and then query the DU table to bring back all those couples of masters to determine the end suggested merge conf score.
-
Deployment History Date
-
Username in Tomcat Access Logs
-
Is It Possible to Perform Automatic Authentication with The User's Windows Account?
-
Where is the documentation for Version 5.2.5?
-
On Prem Semarchy Authentication using Azure AD?
-
Delete old models
-
Sync production model version with dev
-
Recreate a dev environment. Any side effect?
-
Indexes on xDM database tables
-
What logging technology is used in the Semarchy xDM platform?
See all 82 topics