Reltio match, merge, and survivorship

Learn about match, merge, and survivorship functionality and how Reltio uses these common, yet powerful features.

These master data management (MDM) features can help you get your data to that golden state, where it's the single source of truth.

Go for the gold!

The different ways that you can process data entities are a critical part of mastering data. These processes identify and reconcile variations of the same or similar entities. This potential duplication is typically caused by pulling data in from different sources and systems. The endgame you’re working toward is one where you can collect, compare, and consolidate entities that are very close in nature, so you end up with what’s commonly called a golden copy of the record. Say, for example, multiple entities for the Name type exist in your data. These entities use variations of a name, like Bill Fuller, William Fuller, W. A. Fuller. These are possibly the same person. The match and merge process consolidates and merges entities into a single record.

Typically, and for Reltio, those data management processes are:
  • Matching -- Identifies and collects the occurrences of the same or similar entities
  • Merging -- Compares and applies rules to combine entities when possible
  • Survivorship -- Enables certain entities to persist, even when there are similar entities

Now let’s look at these processes more closely

You can use the power of matching functionality to support different needs. The classic usage scenario includes matching and merging entities; these are often called profiles in Reltio. The system always maintains the loaded entities; that is, the loaded entities persist. The software runs these processes automatically.

Where can I use this stuff?

Reltio matching is versatile and ready to meet your business needs with agility and flexibility. You can use Reltio matching on profiles both within a tenant to deduplicate them or externally from the tenant on records in a file. The software identifies records within that file that match to the profiles within a tenant. Matching is also used within the Reltio data tenant subscription service (DTSS) to match profiles stored within a data tenant to those within your tenant.

Get ready for the power -- the Reltio way!

The Reltio matching process matches and merges the appropriate data, and then uses the final results from the survivorship rules processing. Matching isn’t involved in the actual survivorship rules, it just uses the final results.

The Reltio match architecture handles the identification of profiles within the tenant that are semantically the same or similar.

You follow this high-level process:
  • Create match rules using the match configuration framework.
  • Identify the attributes from your entity types that you want to use for matching.
  • Write a comparison formula within each match rule, which is responsible for doing the actual work of comparing one profile to another.
  • Map the token generator classes that are to be responsible for creating match candidates.
    • For basic match rule functionality, you use the match rule builder. It's located within the data modeler of your Reltio console.
    • For more complex rules, you use the advanced editor. It's located within the match rule modeler.
    • Optionally, use machine learning techniques to match profiles. For more information about these techniques, see the topic Reltio Match IQ.
  • Use the match rule analyzer. For more details, see Match Rule Analyzer version 2 (Dynamic). The console UI also has information about how to check the performance of the designed comparison and tokenization.
  • Use explanation and tuning APIs to develop the comparison and tokenization. For details, see the topics Verify matches and Explain match tokens and match documents.

To continue this process, the Reltio merge architecture handles the subsequent merging of two or more profiles. The software aggregates what’s called the crosswalks, and the attribute values, relationships, and interactions from all the contributing profiles, and then combines them into the winning, or golden, profile.

Unlike other systems, the Reltio architecture operates in real-time. Before the match and merge processes occur, the profile-level cleansers in the software cleanse every profile created or updated. The three-step sequence of cleanse, match, and merge all occur in real-time. This behavior makes the Reltio platform ideal for real-time operational use within your data systems.

To top it off, the Reltio survivorship architecture then creates the cherished golden record. In Reltio, this golden record is an object called the golden profile. The software returns this view to any API call that fetches it, and along with it, a set of operational values (OV) from the profile. These OVs are selected in real-time based on the survivorship rules defined for the entity type.

Important: Keep in mind that a Reltio profile and a Reltio entity are subtly different, even though we tend to use the words interchangeably.
  • An entity is a record, which:
    • Has attributes, like simple, nested, and reference
    • Is treated as a node in the Reltio graph
  • A profile is a broader concept. It refers to the:
    • Entity of interest
    • Entities (nodes) connected directly to the entity via relationships
    • Entity’s interaction data
  • So these differences make a profile the real 360 degree expression of an entity. We do use the word profile a lot when discussing match rules, because it's appropriate. Reltio match rules easily take advantage of attributes from related entities via the reference attribute architecture.