The Application of Bayesian Belief Networks to Assurance Case Preparation

Belief Functions for Safety Arguments Confidence Estimation: A Comparative Study

“…Some quantitative approaches have been proposed to assess uncertainties in such arguments [13]. In [15,6,14], Belief Bayesian Networks (BBN) are used to assess confidence in safety case structure and pieces of evidence. They measure confidence by computing probabilities from evidence to conclusion.…”

“…Several works have proposed methods based on Bayesian Networks (BN) to model uncertainty in a safety argument [15,6,14]. Nevertheless, these approaches have a major impediment, which is the need for data.…”

An investigation of proposed techniques for quantifying confidence in assurance arguments

“…For example, safety cases communicate safety design intent to those who will modify existing systems so that safety can be maintained during and after the change [15]. Safety cases also explain the safety rationale and evidence to an assessor-a customer, regulatory agency, or third party-who uses that information to decide whether a system is adequately safe [18][19][20][21][22][23][24][25]. Common definitions of assurance cases call for them to be compelling [3,15].…”

A Model for Safety Case Confidence Assessment

“…Some authors directly use BBNs for modeling arguments and confidence. For instance, in [18], they only use BBNs and commercial tools to calculate "trustworthy", which is actually a conditional probability. With a similar approach, authors of [21] particularly focus on the diversification in argumentation, calculating how a "multilegged" argument (a claim is supported by two evidences) impacts the probability (interpreted as a confidence level) of achieving the main claim.…”

“…For instance, in the "Hazard Avoidance Pattern", some hazards may have a less impact on the overall safety, and the lack of confidence in their treatment, may induce less reduction in the main confidence, than for other more severe hazards. Several models are used in the literature for such arguments, such as simple And-gate [26], weighted mean [7], or Noisy-And [18]. In our case, after several simulations, we decided to define our own Noisy-And, to obtain the trends that are relevant for complementary argumentation.…”