Glossary

Term	Definition
Campaign / CampaignContext	Isolated runtime unit identified by campaign (and often predictor) name; owns plugins, offer matrix, feature lookup
ContactLog	Mongo document written on `/invocations` — scoring snapshot keyed by `uuid`
CustomerFeatureDocument	Open-schema customer/context document from feature lookup
DeploymentStep	Workbench project step that packages config for runtime
Dynamic engagement	Online learning / multi-armed bandit configuration
Entities ontology	Turtle module for Workbench `ecosystem_meta` concepts
Final result	Ranked offer list returned by `/invocations` (`final_result` by `rank`)
Frame	Training / reference dataset (e.g. H2O frame)
Indicative ontology	Draft Turtle inferred from sampling a Mongo collection
Invocation	`POST /invocations` scoring request
ModelingProject	ecosystem-server UML / generation project
Offer matrix	Closed six-field catalog of scoreable offers
OntologyContext	Structured ontology blob on a ModelingProject
params (runtime)	JSON string field on invocations (`"{}"` or `{"input":[…],"value":[…]}`)
Platform ontology	Umbrella Turtle for Ecosystem product concepts (this suite)
Predictor	Static prediction case / model configuration
Project	Workbench hub grouping assets for a use case
Response / ResponseLog	Feedback via `POST /response`; logged with embedded contact
SHACL	Shapes Constraint Language — structural constraints on RDF
Solution case	Productized pack (e.g. Spend Personality, Two Tower)
Synthetic data (Workbench)	LLM-generated Mongo documents, optionally guided by open TTL
uuid	Shared id linking invocation, ContactLog, and response

For class-level catalogs see Entities, Runtime, Server, Workbench.