Funded projects

NEVAKI

Neuro-symbolic validation architecture for trustworthy AI systems

Diese Maßnahme wird mitfinanziert durch Steuermittel auf der Grundlage des vom Sächsischen Landtag beschlossenen Haushaltes.

Project partners:

Nibiru Systems GmbH (coordinator) and HTW Dresden University of Applied Sciences, Chair of Database Systems, Prof. Dr.-Ing. Maik Thiele

Duration:

05/2026 – 09/2028

Funding:

European Social Fund Plus (ESF Plus) and the Free State of Saxony, MINT skilled workers programme (InnoTeam)

Challenge

AI systems increasingly make decisions in safety-critical and regulated areas. Yet their reliability is still assessed mainly statistically, on test data. Whether a system actually meets domain-specific, legal or safety requirements can hardly be demonstrated this way – especially with regard to the EU AI Act, which requires verifiable validation for high-risk applications.

Project goal

In NEVAKI, we are developing a neuro-symbolic validation architecture: large language models make sense of unstructured texts and data, and symbolic methods then check the results against explicitly modelled rules and requirements. A deterministic logic solver makes these checks reproducible, and an explanation component translates the results into comprehensible justifications in natural language. This makes AI systems not only powerful, but also verifiable and controllable.

Planned results

• A reference architecture that couples neural and symbolic components
• Prototypes of an LLM-based semantic parser, formal rule and constraint models, a deterministic logic solver and explainable translation of the check results
• An integrated demonstrator, tested on documents from four application areas: legal briefs, patent specifications, compliance documents and medical descriptions
• Scientific publications and a concept for turning the results into products and services for trustworthy AI, AI governance and compliance

We will report interim and final results here as the project progresses.

InnoTeam

In the InnoTeam, staff from Nibiru Systems GmbH and HTW Dresden work together on the project and build up skills at the intersection of artificial intelligence, knowledge representation and software architecture. The project thus strengthens knowledge transfer between university and industry and Saxony as a location for AI.

The project is co-funded by the European Union through the European Social Fund Plus (ESF Plus) as part of the MINT skilled workers programme of the Free State of Saxony. This measure is co-financed with tax funds on the basis of the budget passed by the Saxon state parliament.