Provenance
MIF provenance has two layers. The core is a lightweight, always-available
set of fields — sourceType, confidence, trustLevel, plus optional
sourceRef / agent / agentVersion — that captures how a unit came to exist
and how much to trust it. Layered on top is an optional W3C-PROV-aligned set
of fields (wasGeneratedBy, wasAttributedTo, wasDerivedFrom) that lets a
unit carry a real PROV entity/activity/agent lineage when richer provenance is
needed. The entity/activity/agent roles are expressed through the @type
values (prov:Entity, prov:Activity, prov:SoftwareAgent) on the provenance
object and its nested nodes.
Both layers are OPTIONAL and additive: a conforming unit MAY omit provenance
entirely, MAY use only the lightweight core, or MAY include the PROV-aligned
fields. The PROV fields project to the W3C PROV vocabulary (prov:) through the
JSON-LD context, so a full PROV graph is expressible but never required.
Source Types
Section titled “Source Types”| Type | Description | Confidence Range |
|---|---|---|
user_explicit |
User directly stated | 0.90 - 1.00 |
user_implicit |
Inferred from user actions | 0.70 - 0.89 |
agent_inferred |
AI reasoning from context | 0.50 - 0.69 |
external_import |
From external data source | 0.30 - 0.70 |
system_generated |
Automatically generated | 0.20 - 0.50 |
Trust Levels
Section titled “Trust Levels”| Level | Description |
|---|---|
verified |
Confirmed by multiple sources |
user_stated |
User explicitly provided |
high_confidence |
Strong inference |
moderate_confidence |
Reasonable inference |
low_confidence |
Weak inference |
uncertain |
Unverified |
Provenance Schema
Section titled “Provenance Schema”The lightweight core (all fields here are OPTIONAL within an OPTIONAL block):
provenance: sourceType: user_explicit # how the unit was created sourceRef: conversation:conv-456 # reference to the originating source agent: claude-3-opus # creating agent agentVersion: "20240229" # creating-agent version confidence: 0.95 # confidence score (0-1) trustLevel: user_stated # trust classificationTo attach a W3C-PROV-aligned lineage, add the OPTIONAL PROV fields. Keys stay
plain (no prov: prefix) and are mapped to the W3C PROV vocabulary by the
JSON-LD context; the @type values use the prov: prefix:
provenance: '@type': prov:Entity sourceType: user_explicit confidence: 0.95 trustLevel: user_stated wasGeneratedBy: # prov:wasGeneratedBy '@id': urn:mif:activity:extraction:mem-001 '@type': prov:Activity wasAssociatedWith: # prov:wasAssociatedWith '@id': urn:mif:agent:claude-3-opus '@type': prov:SoftwareAgent wasAttributedTo: # prov:wasAttributedTo '@id': urn:mif:entity:person:jane-doe wasDerivedFrom: # prov:wasDerivedFrom '@id': urn:mif:conversation:conv-456