FAQs - Expert Center

What's the difference between source, labeled, derived, and inferred entities?

Source Entities are the original entities that exist in your building's data sources (connectors). These are the raw entities pulled directly from connectors - for example, AHU-01_SAT or VAV1.ZNT are BACnet points pulled in from the Mapped UG connector.

Derived Entities are new entities that don’t exist in the original data but are implied by Source Entities. Expert Center extracts these from patterns in Source Entity names and metadata during both labeling and inference. For example, from the Source Entity VAV1.ZNT, Expert Center derives that there must be a VAV box called “VAV1” even though only the temperature point exists in the BMS data.

Labeled Entities are Source Entities that have been manually classified by human experts during the Label step. Experts assign type classifications, identify derived entities, and establish relationships. For example, a human labels AHU-01_SAT as a "Supply Air Temperature Sensor" and identifies it belongs to "AHU-01."

Inferred Entities are Source Entities that have been automatically classified by the AI model during the Process step. The model learns from Labeled Entities to predict classifications for the remaining unlabeled Source Entities. For example, after training on labeled examples, the model infers that AHU-02_SAT is also a "Supply Air Temperature Sensor."

Key differences:

• Source: Original entities from connectors
• Derived: New entities extracted from Source Entity metadata
• Labeled: Source Entities manually classified by humans
• Inferred: Source Entities automatically classified by AI

Source Entities can be Labeled or Inferred, while Derived Entities are created from both.