Matching process

The matching process uses the derived data and comparison data. The process determines whether an incoming record is related to existing records in the MDM database, or if an update to an existing record triggers any data issues.

After the operational server constructs the core member data and derived data, the matching process begins. These stages comprise the matching process.

  1. Candidate selection
  2. Comparison
  3. Scoring
  4. Linking

The following graphic illustrates the MDM matching process in a virtual MDM implementation:

Graphic depicts data moving through the matching stages. Inbound broker MDM operational server Core data creation and storage Standardized data Candidate selection Comparison, scoring, and linking Tasks

The entity manager setting in the MDM operational server configuration influences the actual point at which comparison begins after member data is derived. The entity manager is a multi-threaded process within the operational server that facilitates comparison and entity and relationship definition. There is one entity manager process for each entity type defined. For example, if you implement both identity and household entity types, there are two entity manager processes. When a new member record or an update to an existing record is passed to the operational server, that MemRecno (member record number) is flagged. The entity manager periodically polls the tables (the polling interval and the number of records to process at one time are defined in the mpi_enttype table). When it sees a flagged MemRecno, the entity manager processes a cross match (comparison) against other records, linking and unlinking records, and identifying tasks that are based on task threshold settings and comparison scores. If updates are occurring at a frequent pace, less than every 10 seconds, the entity manager operates in a continuous mode.

The entity manager operates in synchronous or asynchronous mode (also defined in mpi_enttype). If entity management is running in synchronous mode, comparison is done between derived data creation and storage in the database. If set to asynchronous, the data is stored in the database before the comparison stage is processed.

Running in asynchronous mode also gives you the ability to prioritize entity management. There are three ways in which you can use priority management to override your normal asynchronous mode. You can prioritize entity management by source priority, by set priority, or by specialty queue. These options are described in detail in the entity management topics.