D-S Theory for Argument Confidence Assessment

Modelling confidence in railway safety case

“…Once the quantified argument model is built, it only requires evaluating the trustworthiness of the lowest-level of the argument; then these values are aggregated to estimate the trustworthiness in the top goal. The quantitative aggregation rules were developed based on the D-S theory in the previous work Wang et al (2016b). An intuitive evaluation matrix used in the framework is adopted from Cyra and Gorski (2011) for extracting expert opinions on arguments.…”

“…Confidence aggregation is realized through the combination of the estimations (bel,uncer,disb) of lower-level goals into the higher-level goals (see Section 4.2). This step is based on the confidence assessment method derived from the Dempster-Shafer theory (Wang et al, 2016b). As shown in Fig.…”

“…More generally, this framework can be applied to a safety argument with multiple sub-goals and more hierarchical levels, due to the general version of aggregation rules (Wang et al, 2016b).…”

“…We briefly introduce below the formalized concepts and the final aggregation rules. The development of the calculus has been presented by Wang et al (2016b).…”

“…Due to the limited size of this paper, the parameter formalization and the development process of the aggregation rules are not presented. For more details and a general assessment model for N-node arguments, please refer to (Wang et al, 2016b). …”