Certus: A domain specific language for confidence assessment in assurance cases

2025-05-03

AI Summary

Existing quantitative confidence assessment methods suffer from poor interpretability, high subjectivity, limited scalability, absence of dialectical reasoning, and insufficient credibility in assurance cases (ACs). This paper introduces Certus, a domain-specific language (DSL) designed for safety-critical system assurance, enabling semantic-precise encoding of expert judgments via fuzzy sets and defining an interpretable syntax for confidence propagation. Its core contribution is the first explainable, quantitative confidence modeling framework that integrates fuzzy logic with DSL design—rigorously grounded in mathematics while ensuring syntactic readability and domain adaptability. Evaluated on an automotive case study, Certus successfully models and visualizes confidence propagation across multi-level evidence chains, significantly enhancing assessment transparency, consistency, and engineering practicality.

Technology Category

• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories
• Reasoning under Uncertainty: Uncertainty Representations
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Security and Privacy: Large-scale security measurements
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Assurance cases (ACs) are prepared to argue that a system has satisfied critical quality attributes. Many methods exist to assess confidence in ACs, including quantitative methods that represent confidence numerically. While quantitative methods are attractive in principle, existing methods suffer from issues related to interpretation, subjectivity, scalability, dialectic reasoning, and trustworthiness, which have limited their adoption. This paper introduces Certus, a domain specific language for quantitative confidence assessment. In Certus, users describe their confidence with fuzzy sets, which allow them to represent their judgment using vague, but linguistically meaningful terminology. Certus includes syntax to specify confidence propagation using expressions that can be easily inspected by users. To demonstrate the concept of the language, Certus is applied to a worked example from the automotive domain.

Problem

• Develops a DSL for quantitative confidence assessment in assurance cases
• Addresses issues of interpretation and subjectivity in existing methods
• Uses fuzzy sets to represent linguistically meaningful confidence judgments

Innovation

• Domain specific language for confidence assessment
• Uses fuzzy sets for vague linguistic judgments
• Specifies confidence propagation with inspectable expressions