The candidate builder feature defines criteria the system uses to add an existing entity to the candidate list as part of the entity resolution process.
Typical candidate builder settings include address, unique numbers, and other numbers. These are the data types that the system compares to determine which existing entities might resolve to an incoming identity. When a new identity record enters the system, if an existing entity has a matching value for any of the data types identified by the candidate builder, that entity is added to the candidate list.
Candidate builder settings are organized by groups called candidate builder configurations. Only one candidate builder configuration can be used within a resolution configuration.
Generics directly affect whether a value is considered as part of the Candidate builder process. After a value is considered a generic value, it is no longer used to generate candidate lists.
Candidate builder settings directly affect system performance. When the system uses index lookups to compare an incoming identity to each and every entity in the entity database, it is only comparing data types that are configured in the candidate builder feature. This allows candidate lists to be generated very quickly. As the entity database grows and includes more entities there is more for the candidate builder to compare. For example, if your entity database contains 100,000 entities and the candidate builder is set to compare three data types when creating the candidate list, then whenever a new identity enters the system, the system can make up to 300,000 comparisons just to generate the candidate list. If our entity database contains 1,000,000 entities and the candidate builder is set to compare three data types when creating the candidate list, then whenever a new identity enters the system, the system can make up to 3,000,000 comparisons just to generate the candidate list. If you add a single candidate builder criteria, the system can make up to an additional 1,000,000 comparisons just to generate the candidate list. That is up to 1,000,000 additional comparisons per identity record loaded into the system. If the candidate lists are too large because they consider too many types of data, the entity resolution process will run much slower than if the candidate builder settings only contain the data types necessary to build effective candidate lists.
When considering whether to use the Default or the Default with name only configuration setting, remember that if you choose Default with name only, you are adding comparisons at an order of magnitude greater than those required by the Default configuration.