A Golden Records Project solves the task of creating a single or golden record from a group or cluster of records that refer to the same entity. This task is also known as entity consolidation.
You create golden records based on clusters obtained from a Mastering Project. You can also create golden records for any dataset containing a grouping key.
Curators and Reviewers of the project compose rules to consolidate the values of specific attributes, and review and edit the resulting golden values before publishing the dataset for downstream consumption.
Once you add a dataset and define the grouping key, a curator composes attribute-specific consolidation rules to begin generating golden records. See Working with Golden Records.
The rules can include different types of filters and conditions that you can set per attribute. Conditions narrow down which set of recors to apply to the rule. For each attribute, Tamr first applies one or more conditions that you create and then applies the rule.
- Most common value. This condition is useful when you want to choose the most common value for a particular attribute.
- Is not empty.
- Dataset. This condition is useful when you want to mark some of your input datasets as preferred. For more information, see Working with Golden Records.
- Most common value.
- Count Distinct
- Expression (custom filter)
After you create and review golden records, you can publish their latest version and make it available to consumers. See Publishing Golden Records.