The {required element} of the system is {interpretable} in line with the safety case requirements, to the {intended audience} in the {intended context}

{Interpretable}

{required element}

{Intended audience}

{Intended context}

Argument over {interpretability requirements}

{interpretability requirements}

Implemented {interpretability method} appropriate for {requirements}, i.e. the correct thing is being explained

Argument over {interpretability methods}

{Explainability method
Evidence}

{interpretability method} is faithful to {model} process

Argument over faithfulness of {interpretability method}

{Explainability method
Evidence}

{ML Model}

{Explanations} produced at appropriate {times}

Argument over appropriateness of {explanations} {time}

{Time Evidence}

{Times that explanations
are provided}

{Explanations} are appropriate for {audience}

Argument over appropriateness of {explanations} to {audience}

{Audience Evidence}

{Audience}

{interpretability method}

{Explanations}

What?
See section 4.1.2

How?
Section 3

When? Section 4.1.3

Who? Section 4.1.4

Why?
See section 4.1.1