Entities and linking styles

In the realm of data management an entity is the logical relationship between records.

Entity types provide for a distinction between the way members are viewed and linked. Common examples of entity types include identity, household, and organization. An identity entity is made up of records that reflect a single individual. Household entities represent records that link based on the comparison of a location attribute, such as home address. Organizational entities are associated based on organization or company name.
Note: Hybrid MDM supports only the transitive linking style. The non-transitive, best match, and trusted source linking styles are not available.

The MDM operational server manages entities that are based on the configured style or method of entity linking; transitive, non-transitive, best match, or trusted source.

Transitive
Transitive entities operate on the glue member concept, which means that a record must match only one other record in the entity set. For example, if Member A and Member B match and Member B and Member C match, it is assumed that Member A and Member C match. This match can be achieved by comparing greater than the auto-link threshold or by manual linking. In this example, the matching of Member A and Member C occurs independently of the score between them. In transitive entities, a member can belong to only one entity at a time and thus have only one Entity ID assignment. Most identity entities are based on transitive entity processing.
Non-transitive

Non-transitive or "strong" entities require that all records match greater than the auto-link threshold with all other records in the entity set. As with transitive, members of a non-transitive or strong entity can belong to only one entity set at a time (one Entity ID).

Most household entities are based on non-transitive entity processing. The single entity membership rule applies only within an entity type. If you are a member of a household entity, you can also be a member of an identity entity.

In both transitive and non-transitive configurations, if the comparison scores meet the threshold settings, records can link across all sources.

Transitive and non-transitive styles are defined by a true or false setting in the mpi_enttype table; true sets transitive and false sets non-transitive.

Trusted source
In trusted source entity management, records created in the MDM database are compared and linked to records across systems. However, you can have only one member from the trusted source in an entity. For example, of your three sources (A, B, and C), A is the trusted source. Your entity can contain multiple records from B and C, but only one record from A. The true or false setting in the trusted field (istrusted) in the mpi_srcxent table defines a source as trusted.
Best match

Best match operates by linking the best-matching trusted source member. If at least one source in your data set has the istrusted field in the mpi_srcxent (source-to-entity cross-reference) table set to ‘yes' for an entity type, then best match linking occurs when members from that source are involved in a linkage interaction. The setting that is defined in the mpi_srcxent.srcprior (source priority) field determines the confidence level in accuracy of source data that is used in linkage creation. The higher the numeric value of the source priority, the higher the confidence level.

The best match process first takes the istrusted setting into account. When a trusted source member is updated, the MDM operational server then uses the source priority of all trusted source members to find the best match. Members who have the highest source priorities are considered for linkage.

When an update is made to a member from a non-trusted source, the comparison score is used to determine the best match. The operational server searches for the trusted source member that compares best (highest score) with the updated (non-trusted source) member. The operational server then links the updated member into the trusted source member entity. However, if the updated member scores equally with multiple members, the source priority setting is used to determine the linkage.

The common approach for most implementations is to use the default value of 1 as the priority setting. If 0 is used, members from the source are not linked.

Important: Updates to trusted source records should be done only when the entity type is configured in asynchronous mode (mpi_enttype.isasync). If you are running in synchronous mode and an update to a trusted source record occurs, the records that are currently linked with the updated trusted source record are not relinked. In this case, an error is logged.