Q: How do I tell the effectiveness of my Match Rules based on this?
A: The results of this analysis are color coded, visually showing the outcomes, along with comments on how the rules might be improved to the right. You can find the analyze match rules tool in the data modeler, under match rules in the top right corner. It will generate a report showing which rules are good (in green), which rules have large overlap (in yellow) and which rules have problems (in blue).
Making Changes
The Data Modeler is also where you make Match Rule changes. Pick one of your entity types, and in the header navigate to Match Rules. You can see a list of all your existing Rules. After making changes to a rule, it is best practice to rebuild the match tables. This is available in the console under Tenant Management > Jobs > Rebuild Match Table. The job recreates the tokens used to match records to one another.
Q: If you forget to rebuild, what happens?
A: Your existing data may not match on your modified rule as you might expect. As new records are loaded they will have tokens generated against your modified rule, but be compared to the old tokens of records loaded before the rule changed.
Internal and External Matching
In Reltio, there are two types of matching: internal and external. Let’s assume you wanted to bring on a new source. You want to see how the new records will integrate with your tenant, but don’t want to experience the potentially painful process of unmerging them. External match allows you to upload a file through an API and match your tenant data to the file data. It’s nice to see how a new source will play out before you actually load it.
Internal matching is the process of comparing and merging entities already loaded to Relto. There are currently two types (and soon to be a third type) available: Suspect, cued up for a data steward to review via a workflow; Automatic, records automatically merged and Relation, (coming soon) which creates a relationship between records rather than merging.
Q: If there was a potential match, and an update to a birthdate made the match rule no longer true, will the potential match go away?
A: They will no longer be potential matches and it will remove them from the potential match workflow. If they have already been merged and an update comes through to one of the crosswalks, they will stay merged. We will not unmerge for you. It is a pretty important distinction. There is nothing at this time that monitors and unmerges records.
Q: How do you differentiate between match rules? How do you know which rule was triggered?
A:
Each rule has a name, which acts as an ID for the rule and a label, which is displayed in the UI. For labels, I like to list all the key attributes that are used in the rule; it will show in the UI and help data stewards understand what rule brought these together.