Using machine cognition — a theory, a research programme, and a completed four-scenario study.
Two software systems that must exchange information rarely share a model of the world. The classical remedy is a standard agreed in advance. This project investigates the alternative that opens up once the systems on each side can reason: they reconcile their divergent models ad hoc, for the occasion, machine-to-machine, with no standard settled beforehand. The through-line is a single thesis — it is cognition that completes a reconciliation — with a thin, published reference as the practical anchor that reduces the work, and the pragmatic layer as the frontier that carries the operative meaning. Drawn out, a lifted model is a knowledge graph — concepts and relations as its schema, instances as its assertions — with pragmatics and provenance layered on; reconciliation is then ontology matching plus entity resolution over that graph, and a thin shared reference is the interoperable anchor such graphs can bind to.
The work is in two parts. Pick a door:
proposal/ — Part I develops the theory (semantic models, the lift, the cognition spectrum, shared
references) and Part II sets out the research programme (questions, hypotheses, method, benchmark,
plan). Both as papers and slide decks, with four interactive demonstrations.
study/ — the empirical realisation of the programme across four operational scenarios (configuration,
intent, standard-free, observability), each built, run against a validated gold standard, and reported,
with a master report (and a companion slide deck) drawing them into one synthesis — plus the harness, benchmark, and per-run
data behind them.
demos/ — self-contained, interactive walk-throughs of the four scenarios (configuration, intent,
cross-domain, observability), referenced by both the proposal and the results. They run live once
GitHub Pages is enabled (see each door’s README for the links).
Christopher Janz. The proposal papers build on work in the IETF/IRTF (NMOP, NMRG, TEAS/CCAMP), the ONF Transport API, and TM Forum intent modelling; the study operationalizes and empirically tests the programme’s central claims.