Cognitive Knowledge Graph Ontology Specification

Neural Memory models your digital work and personal life as a semantic property graph. This document specifies the entity node schemas, relationship edges, indexing properties, and query patterns.

1. Core Node Schemas

1.1 `Decision`

Represents an explicit approval, rejection, architectural choice, or agreement reached by the user or their team.

Property	Type	Description	Example
`id`	`String (UUID)`	Unique decision identifier	`"d-a812f0c1"`
`title`	`String`	Concise summary of the decision	`"Approve Cloud Q4 Quote for 5000 EUR"`
`verdict`	`String`	`APPROVED` \| `REJECTED` \| `PENDING`	`"APPROVED"`
`rationale`	`String`	Context or reasoning behind the choice	`"User confirmed email proposal within budget"`
`subject`	`String`	Topic or artifact decided upon	`"Cloud Q4 Proposal"`
`timestamp`	`DateTime (ISO8601)`	Timestamp when decision occurred	`"2026-09-03T14:10:00Z"`
`project_id`	`String`	Associated project partition	`"default"`

1.2 `Commitment`

Represents an action item, deliverable, or promise made by the user to someone else, or to the user by a collaborator.

Property	Type	Description	Example
`id`	`String (UUID)`	Unique commitment identifier	`"c-4b9981e2"`
`title`	`String`	Brief description of the task	`"Send updated slides by Friday 18:00"`
`task`	`String`	Detailed deliverable description	`"Prepare presentation slides and email them"`
`due_date`	`DateTime (ISO8601)`	Resolved deadline	`"2026-09-05T18:00:00Z"`
`status`	`String`	`open` \| `in_progress` \| `completed` \| `cancelled`	`"open"`
`debtor`	`String`	Person responsible for fulfilling (`user` or name)	`"user"`
`creditor`	`String`	Person expecting the deliverable	`"Giulia Bianchi"`
`project_id`	`String`	Associated project partition	`"default"`

1.3 `Meeting`

Represents a collaborative video call, voice sync, or calendar event detected through meeting applications (Google Meet, Zoom, Microsoft Teams, Slack Huddle).

Property	Type	Description	Example
`id`	`String (UUID)`	Unique meeting identifier	`"m-90a1bc33"`
`title`	`String`	Title of the meeting	`"Architecture Sync with Roberto & Francesca"`
`start_time`	`DateTime (ISO8601)`	Start timestamp	`"2026-09-03T10:00:00Z"`
`end_time`	`DateTime (ISO8601)`	End timestamp	`"2026-09-03T10:45:00Z"`
`attendees`	`List[String]`	Names of identified participants	`["Francesca", "Roberto"]`
`summary`	`String`	AI-generated discussion summary	`"Decided on hybrid Neo4j/vector architecture"`

1.4 `Reflection` / `Insight`

Higher-order cognitive abstractions generated during Dream Mode memory consolidation.

Property	Type	Description	Example
`id`	`String (UUID)`	Unique reflection identifier	`"r-77e812d0"`
`category`	`String`	`productivity` \| `architecture` \| `milestone`	`"architecture"`
`topic`	`String`	Main conceptual focus	`"Storage Scalability"`
`synthesis`	`String`	Consolidated insight derived from multiple slices	`"Transition to SQLite eliminates Docker friction"`
`created_at`	`DateTime (ISO8601)`	Consolidation run timestamp	`"2026-09-03T02:00:00Z"`

1.5 `ActivitySlice`

A 15-to-30 minute consolidated block of user workflow combining multiple raw interaction bundles.

Property	Type	Description	Example
`id`	`String (UUID)`	Unique slice identifier	`"as-312019ab"`
`start_time`	`DateTime (ISO8601)`	Slice start	`"2026-09-03T09:00:00Z"`
`end_time`	`DateTime (ISO8601)`	Slice end	`"2026-09-03T09:30:00Z"`
`summary`	`String`	Executive summary of activity	`"Researching zero-config Mac packaging"`
`apps`	`List[String]`	Applications used	`["Safari", "Xcode", "Terminal"]`
`confidence`	`Float`	Semantic parsing confidence score	`0.92`

2. Relationship Semantics (Edges)

graph LR
    Decision([Decision]) -->|TOWARDS_PERSON| Person([Person])
    Decision -->|ABOUT_TOPIC| Topic([Topic])
    Decision -->|ON_ARTIFACT| Artifact([Artifact])

    Commitment([Commitment]) -->|PROMISED_TO| Person
    Commitment -->|PROMISED_BY| Person
    Commitment -->|ABOUT_TOPIC| Topic

    Meeting([Meeting]) -->|ATTENDED| Person
    Meeting -->|DISCUSSED| Topic

    ActivitySlice([ActivitySlice]) -->|CONTAINS_EVENT| RawEvent([RawEvent])
    ActivitySlice -->|INVOLVED_TOPIC| Topic

    Reflection([Reflection]) -->|DERIVED_FROM| ActivitySlice
    Reflection -->|SUMMARIZES| Topic

Relationship	Source Node	Target Node	Semantic Meaning
`[:TOWARDS_PERSON]`	`Decision`	`Person`	The person to whom an approval or decision was directed.
`[:ABOUT_TOPIC]`	`Decision`, `Commitment`	`Topic`	The conceptual subject matter of the item.
`[:ON_ARTIFACT]`	`Decision`	`Artifact`	The specific quote, document, pull request, or file decided upon.
`[:PROMISED_TO]`	`Commitment`	`Person`	The creditor expecting delivery of the promise.
`[:PROMISED_BY]`	`Commitment`	`Person`	The person responsible for completing the task.
`[:ATTENDED]`	`Meeting`	`Person`	A participant present in the collaborative meeting.
`[:DISCUSSED]`	`Meeting`	`Topic`	An agenda topic or subject debated during the call.
`[:DERIVED_FROM]`	`Reflection`	`ActivitySlice`	Slices used by the consolidator to deduce the insight.

3. Query Examples

Finding All Decisions Involving a Specific Colleague (Cypher / Neo4j)

MATCH (d:Decision)-[:TOWARDS_PERSON]->(p:Person {name: "Marco Rossi"})
RETURN d.title, d.verdict, d.rationale, d.timestamp
ORDER BY d.timestamp DESC
LIMIT 10;

Finding All Open Commitments with Upcoming Deadlines (Cypher / Neo4j)

MATCH (c:Commitment)
WHERE c.status = "open" AND c.due_date IS NOT NULL
OPTIONAL MATCH (c)-[:PROMISED_TO]->(p:Person)
RETURN c.title, c.due_date, c.debtor, coalesce(p.name, c.creditor) as creditor
ORDER BY c.due_date ASC;

Querying Graph Data via Embedded SQLite Engine

In Standalone mode, relations are stored in the SQLite tables nodes and links:

SELECT n.id, n.label, n.properties, n.timestamp
FROM nodes n
JOIN links l ON (n.id = l.target_id)
WHERE l.source_id = "dec_001" AND l.rel_type = "ABOUT_TOPIC";