Customer Feature Lookup
Unlike the Offer Matrix, customer / feature store documents are open: any JSON shape the campaign’s feature lookup expects.
Ontology class
CustomerFeatureDocument— runtime namespace
IRI:https://ecosystem.ai/ontology/runtime#CustomerFeatureDocument
Represented in logs as nested scoring_data / feature-store payloads. The platform ontology does not enumerate product-specific fields (spend traits, RFM, embeddings, …). Those belong in a domain or product ontology you upload (e.g. in Workbench2 synthetic).
Configuration concepts
| Class | Role |
|---|---|
FeatureLookupConfig | DB + collection + key for customer/context lookup |
FeatureStoreRef (entities) | Workbench-side pointer for Dynamic Engagement |
ScoringParams | Internal bag after lookup / pre-predict |
Practice
- Keep offer rows fixed (six fields).
- Put customer context in open feature docs.
- Optionally describe that open schema in Turtle for synthetic generation or documentation — without locking the platform ontology to one vertical.
Download: /ontology/runtime.ttl
Last updated on