Certus DSL uses fuzzy sets to quantify software assurance confidence
Fuzzy logic meets assurance cases: new DSL tackles subjectivity and scalability.
Assurance cases (ACs) are structured arguments that a system meets critical quality attributes like safety or security. While quantitative confidence methods exist, they suffer from poor interpretation, subjectivity, scalability limits, and lack of dialectic reasoning, hindering adoption.
Certus solves this by introducing a domain-specific language (DSL) where users express confidence using fuzzy sets—mathematical tools that capture linguistic vagueness (e.g., "high confidence," "moderate"). The language provides transparent syntax for propagating confidence through expressions, making the reasoning inspectable. The team validated Certus on a real-world automotive assurance scenario, showing it can produce reproducible, scalable confidence assessments that are easier to understand and trust.
- Certus uses fuzzy sets to represent confidence with linguistically meaningful terms like "high" or "moderate," reducing subjectivity.
- The DSL includes explicit syntax for confidence propagation, enabling users to inspect and audit the reasoning chain.
- Demonstrated on an automotive domain example, showing improved scalability and interpretability over existing quantitative methods.
Why It Matters
Certus makes confidence in software assurance measurable, transparent, and scalable—critical for safety-critical systems like autonomous vehicles.