What symbolic analyses we can define with ontologies and formal methods, and how to make generative AI drive them to produce engineering answers you can trust.
Generative AI can answer questions about a system in seconds. What it cannot do on its own is guarantee that the answer is right. Symbolic analyses can: queries, rules, constraints, and reasoning over ontological models compute an answer exactly, the same way every time, and show how it was reached. The opportunity is to put the two together, with generative AI choosing the analysis a question calls for, running it, and explaining the result, while the analysis itself stays symbolic and exact.
This forum takes up two questions. First, what symbolic analyses can we define over engineering models with ontologies and formal methods, and which engineering questions do they answer, from consistency and completeness checks to requirement verification, trade studies, design space exploration, and impact analysis? Second, how do we make generative AI drive them: find the right analysis for a question, understand what it computes and when it applies, supply its inputs, and read its results, while the engineer stays in control of the method?
This edition also takes up SysML v2: how SysML v2 models will be supported, so that the same symbolic analyses, driven by generative AI, can run over them.
The building blocks are mature. The semantic web stack has grown over 25 years into a family of W3C standards, OWL for reasoning, SHACL for validation, and SPARQL for queries, with proven tooling and production use in industry and government. Alongside it stand decades of formal methods with proven tools of their own: model checkers, SMT and constraint solvers, theorem provers, and simulation and optimization engines. These need to be integrated, so that an analysis over an ontological model can draw on whichever method fits the question. Driving them with generative AI means building on foundations that have already stood the test of time.
This one-day forum brings practitioners and researchers together to share the analyses they have defined, how they have made generative AI drive them, and lessons from putting them to work. It is the third onto:Nexus Forum, following the 2024 and 2025 forums at NASA's Jet Propulsion Laboratory in Pasadena.
Tentative program, Pacific Time. Speakers will be announced here.
Admission is free, in person or online. Registration is required so we can plan seating, and your attendance is confirmed only once you receive a confirmation email from us.
No. The forum is a separate event, open to anyone with an interest in the topic. It is held the day after INCOSE IW 2027 so attendees can stay in Los Angeles one more day.
Yes. The forum is hybrid: join us in person at UCLA or online. Choose how you will attend when you register, and we will send online attendees the link to join.
In Room 289 of Engineering VI, on the UCLA campus in Los Angeles. Parking details will be sent to registered attendees and posted on this page.
We welcome proposals for full session talks and for lightning talks. When you register, add a title and a few sentences on what you would present, and say which kind of talk you have in mind. Proposals are reviewed, and your talk is confirmed only once you receive a confirmation email from us.
February 3, 2027 at UCLA and online. Free, registration required. Seats in the room are limited, so attendees and speakers are confirmed by email.