Resolution Outcome

Entity resolution outputs three types of results:

1. No match: The source record does not match any other record already in the system.
2. Definitive Match: The source record matches another record already in the system.
3. Possible Match: The source record potentially matches another record already in the system but needs to be reviewed first.

No-match

When a system has not identified a match for an incoming source record, it considers this source record a new real-world entity it needs to track. A new unified record with a unique identifier is automatically created, and the source record is linked to it. This unified record is now part of the pool of unified records used in the entity resolution process.

Definitive Match

The system is 100% confident about the match and will process it automatically. The incoming source record is linked to an existing unified record and joins the existing group of source records associated with it. The source record's properties are added to the aggregate of source record properties and will be used for future comparison.

Possible Match

Possible matches must be reviewed before the system can process the incoming source record further. When this happens, the system creates an item with an issue and places it in the Resolution Queue, where the item will sit until it is picked up for review and the issue is resolved. Once resolved, the system will pick right back where it left off and apply the result of the issue resolution.

Resolution Queue

The resolution queue is part of the Clinia Platform Queue Module and allows users to integrate humans into otherwise automated workflows. It is a system queue whose primary purpose is to enable human users to review entity resolution outcomes and resolve issues that arise from them. Many issues can arise from an entity resolution run for a single source record. In that case, a single queue item is created with all its issues.

Resolving an issue involves examining its associated data (contextual information, source records, relationships, unified records, etc.) and deciding on a resolution. Issues are presented as questions a user needs to answer.

Possible Match Issue

A possible match issue is created when the outcome of entity resolution is a possible match. The question asked to the user is:

Does this source record represent the same real-world entity as this unified record?

If the answer is yes, then the system will link the source record to the unified record.

If the answer is no, then the system will not link the source record to the unified record.

Possible Duplicate Issue

A possible duplicate issue is created when entity resolution outputs more than one possible match. This happens when many existing unified records are flagged as possible matches for an incoming source record. Such events suggest that the system tracks two or more unified records as distinct entities, although they might be the same real-world entity.

The result is that, in addition to the many possible match issues, one or more possible duplicate issues are created to ensure that existing data does not contain duplicates. The question asked to the user is:

Do both unified records represent the same real-world entity?

If the answer is yes, then the system merges the two unified records and links all source records to a single, surviving unified record.

If the answer is no, then the system does not merge the unified record.

📘

Unified Record Merge

When merging unified records, the system prompts the user for a merge direction, which dictates which unified record survives and which is merged. The merged record's unique identifier will be preserved as a legacy identifier for the surviving unified record.