Edges

On this page

  1. Passing data through an edge
  2. Edge types
  3. Static vs. dynamic edge types

An edge defines a relation between two records in CluedIn.

  • Each edge connects a source and a target (also referred to as “from” and “to”).

  • Each edge has an edge type that explains the nature of the relation (for example, /WorksFor, /LivesWith).

  • Edges can also include properties, such as:

    • Weight – Indicates the strength of the relation between two Golden Records. Used in the processing pipeline to influence decision-making.

    • Property bag – Stores additional metadata about the relation.

Passing data through an edge

Edges can also be used to propagate data between related records.

  • Often, you will denormalize references to other entities but still want user-friendly values for display.

  • When creating an edge, you can attach extra properties to either the “from” or “to” reference.

  • These properties are propagated to the connected entity and are typically used for passing vocabularies.

Consider this example:

  • A company record references a country by an ID.

  • Through the edge, the company record can also display friendly country values in the UI.

  • If the country record is updated, CluedIn automatically updates all references to that country.

By default, CluedIn copies these properties only to the target reference, and only if that reference exists.

Edge types

An edge type defines the kind of relation between two records, typically expressed in the structure: Object – Verb – Object. For example: John – works at – Lego

In CluedIn, edge types can also store metadata such as weights and properties, but their primary purpose is to describe and query relationships between nodes.

Static vs. dynamic edge types

  • Static edge types – Defined by fixed rules that do not change.

  • Dynamic edge types – More flexible, determined at runtime based on additional conditions.

The best practice it to keep edge types in crawlers as generic as possible. Use processors in the processing server to resolve them dynamically into specific types.

Consider this example:

  • A crawler sets a static edge type: Works At.

  • However, employment is temporal — it may not always be valid.

  • A processor can check job start and end dates and dynamically update the edge type to Worked At once the person leaves.