And when needed can tell you so.
Your organization's knowledge lives in many forms: in documents and conversations, in tables and transactions, in data and in people's experience. Isagog makes it explicit, verifiable and usable by AI assistants and applications, with a vision, a method and a platform.
Text search, even by semantic similarity, only finds relevant passages. An RDF graph, instead, connects entities, facts, sources and relations; the ontology makes explicit the rules for deriving new knowledge.
Illustrative example based on the ontology of a contemporary art museum.
Which works by Italian artists born after 1980 can I see in the exhibition “La luce della migrazione”?
Two of the five works in the exhibition: “Rotta di terra” (2021) by Nadia Ferri and “Attraverso” (2019) by Marco Sabbatini.
Find the exhibition from its title.
ExhibitiontitleCollect the works shown in that exhibition.
exhibited_inexhibit_artworksFrom each work, trace back to its author, when the author is an artist.
authored_byArtistKeep the Italian authors born after 1980.
nationalitybirth_date“Works” was not searched one class at a time: a painting and a video work are included because their classes are subclasses of Artwork.
ArtworkPaintingVideoArtworkThe collective has no declared or derivable Artist type, so it does not satisfy the query's positive condition. This does not prove a negation.
ArtistPersonCollectivePaintingNadia Ferri — Italian, 1988VideoArtworkMarco Sabbatini — Italian, 1985InstallationCollettivo Mareo — the Artist type is neither declared nor derivable from the available dataPaintingGiulio Neri — Italian, 1979: excluded by date of birthPaintingHélène Roux — French, 1962: excluded by nationality and date of birthArtworkPaintingVideoArtwork“Opera creativa, che può essere materiale, immateriale o performativa”
Painting and VideoArtwork are subclasses of Artwork: the question says “works”, and the system collects them by subsumption, without anyone having listed the classes to search for.
ArtistPersonCollective“Un individuo umano”
Artist is a subclass of Person. The Collective type, on its own, does not allow Artist to be derived. The query does not include that result, but the absence of the type is not a negation: a further axiom would be needed.
The question combines relations and filters on the graph with class subsumption. The collective does not satisfy the positive Artist condition; it is not derived to be a non-Person.
Illustrative instances on a real schema: the exhibition, the works and the halls are invented; the classes and properties traversed — Exhibition, Painting, Installation, VideoArtwork, Hall, exhibited_in, located_in, adjacent_to — are those of the MAXXI ontology (v2.8) and of the Isagog top ontology it imports.
Illustrative example based on the schema of a historical newspaper archive.
What do we know about the Parco Nord project?
We retrieve five facts with explicit provenance: an editor, an AI model, DBpedia, Wikipedia and a description tied to the moment of the article.
Coordinated by the city department for public green spaces
HumanDescriptorLaunched in 2019
AIDescriptorLocated in the Nord district
DBPediaDescriptorDescribed in the district's Wikipedia entry
WikipediaDescriptorUnder construction (at the time of this article)
ContextualDescriptorHumanDescriptor“Descrittore di entità creato da annotatori umani”
AIDescriptor“Descrittore di entità generato da sistemi di IA”
DBPediaDescriptor“Descrittore di entità che riferisce risorse DBpedia”
WikipediaDescriptor“Descrittore di entità che riferisce articoli Wikipedia”
ContextualDescriptor“Descrizione di un'entità indicizzata temporalmente, così come caratterizzata al momento di una specifica menzione”
Illustrative instance on a real schema: the case is invented; the descriptor classes — HumanDescriptor, AIDescriptor, DBPediaDescriptor, WikipediaDescriptor, ContextualDescriptor — are those of a newspaper archive's ontology.
Synthetic, pseudonymized case; statements and sources from the clinical graph.
Is the patient allergic to penicillin?
It depends on when you ask: reported but never tested from February to April, then ruled out (de-labelled) in September 2025.
ReportedL1, line 22Reazione cutanea a penicillina in età pediatrica, riferita
reported, not verified
NegatedL1, line 22Test allergologico mai eseguito
explicitly negated in the source
ReportedL2, line 41Allergia riferita a penicillina
reported, not verified
RuledOutL3, line 44Allergia a penicillina esclusa (de-labellazione)
ruled out; supersedes earlier assessments
The September statement explicitly supersedes the earlier ones through the supersedes relation: choosing the most similar or most recent text is not enough.
L3: RuledOutsupersedesL1: Reported, L2: Reported
Negated records an explicitly negated fact; RuledOut an excluded assessment. A missing fact remains unknown and a suspicion remains open: they do not become false for lack of confirmation.
AllergyStatementReportedreported, not verified
“Riferita dal paziente reazione cutanea a penicillina in età pediatrica; episodio non documentato”
Reported by the patient; the episode is not documented.
ProcedureStatementNegatedexplicitly negated in the source
“mai sottoposto a test allergologico”
No allergy test has ever been performed.
AllergyStatementReportedreported, not verified
“In considerazione dell'allergia riferita a penicillina”
The unverified label begins to steer treatment choices.
AllergyStatementRuledOutruled out; supersedes earlier assessments
“Il paziente è stato de-labellato: può assumere beta-lattamici penicillinici. La segnalazione di allergia riportata nelle precedenti lettere di dimissione è pertanto da considerarsi superata.”
The specialist assessment supersedes the two earlier reports.
Real data (synthetic, pseudonymized): the clinical case is synthetic by construction, but the statements, quotations and document-and-line references come from the knowledge graph.
Founder and CEO — formerly Director of the Center for Advanced Studies, IBM Italy
For years Guido led IBM's research in Italy on language, logic and knowledge representation. Today he teaches Artificial Intelligence at Università Marconi, and brings to Isagog the same idea that guided his research: an artificial intelligence that reasons on explicit foundations, not just on statistical correlations.
Co-founder — formerly Health and Research Executive at IBM, medical doctor
Bob has been applying artificial intelligence to clinical practice for forty years. Experience gained where an error has real consequences — and the reason why, at Isagog, the traceability of answers is not a technical detail but a requirement.
Only with your consent, PostHog (EU) and Cloudflare measure pages visited and their order, visit duration, referral, approximate country, browser, device, and performance.
We collect the paths of pages visited and their order, visit times and duration, the referring site, approximate country (derived from the IP address), and basic browser and device information. Cloudflare also measures page performance.
We do not record form contents, individual clicks, or session videos. PostHog does not create a personal profile and keeps an anonymous identifier only for this browser tab. We save your choice in the browser for future visits.
If you disagree, you can leave the site without activating this analytics.