Find
“What matters for this question?”
A bounded set of relevant objects and direct relationships.
live hereOntology by Stochastic Inference
Build a system that can answer from the evidence.
Ontology turns articles, papers, records, and project state into typed relationships that people and agents can question.
The Parallax project starts with a navigable history experience, processes the research behind Cannae, and shows the difference between a story surface and a source-backed evidence system.
Live public proof · 8 sources → 38 passages → 18 reviewed claims → 101 new ontology objects
Parallax before → after
real source processingQUESTION
What becomes possible after Ontology reads the sources?
- 0 claims
Parallax story surface
Narrative structure, no research evidence graph
- 38 passages
Research packet
Three ancient accounts, geoarchaeology, military reception
- 18 claims
Enriched ontology
Answers carry sources, epistemic status, and limits
- Sources
- 8
- Graph delta
- +101
- Conflicts kept
- 1
The designed system
Model. Select. Prove.
A useful ontology is not just a graph. It is a repeatable contract for turning domain material into context that can carry responsibility.
- 01
Model the domain
Project source material into typed objects and explicit relationships without replacing the source.
- 02
Select what matters
Turn a question into a small, scored neighborhood instead of loading the entire graph.
- 03
Prove the result
Return the objects, paths, snapshot, budget, and boundary that produced the context.
What it can get you
Questions become explicit output contracts.
Three contracts are live in the public Parallax proof. The others show where the same ontology machinery is already designed to travel.
Explain
“Why did this context appear?”
Match reasons, scores, predicates, and neighboring objects.
live hereAudit
“What supports this result?”
Source files, snapshot identity, selection budget, and limits.
live hereCompare
“What changed, and what does it affect?”
A snapshot delta and the dependent surfaces that need review.
application patternOrient
“What is the next defensible action?”
The nearest unresolved constraint and the evidence needed next.
application patternPrepare
“What should another agent know?”
A task-sized context package rather than the whole workspace.
application patternOne contract, different domains
The structure changes. The discipline holds.
Each domain defines its own objects and predicates. Ontology contributes the projection, bounded selection, evidence receipt, and explicit limits.
Parallaxpublic proof
history
“How do these perspectives differ?”
Stories become navigable through chapters, events, places, and points of view.
Produces
A navigable relationship between stories, perspectives, chapters, and events.
Coreapplication pattern
software factory
“What should happen next?”
Projects become explainable through goals, evidence, blockers, and dependencies.
Produces
An evidence-bearing relationship between work, proof, dependencies, and release state.
PK memoryapplication pattern
agent context
“What should the next run remember?”
Past encounters become bounded context rather than an undifferentiated history dump.
Produces
A traceable path from encounter history to relevant future context.
Inspect the complete chain
The public proof is small enough to understand—and real enough to test.
Ask the enriched Cannae graph, inspect the source locators and evidence digests, then explore the complete snapshot. Nothing important is hidden behind a marketing answer.