Trustworthiness Model Representation
The trust measure must also be explainable,
i.e., the user should be able to determine
what contributed to a particular Trust Score.
This makes the Trust Score actionable. If
required, a user can take corrective action by
addressing the cause of a particular Trust
Score.
dependency relation of components and
characteristics that define a solution can be
represented by a directed acyclic graph. A
node in the graph represents a component
or an observed attribute. A directed edge
identifies the dependency of the
relationship. An edge from a source node to
Figure 3: The relationship of attribute Trust Scores to calculating an overall system Trust Score
a destination node indicates that the Trust
Score of the source node influences the
Trust Score of the destination node. The
Trust Score of a node (component or system)
is computed as a weighted combination of
the components or attributes that affect its
score. A user or a system designer can
determine how much weight each
component or attribute has on a Trust Score
of a system. The weight of the Trust Score of
a node on its dependent node is defined as
the weight on the edge between the nodes.
The above figure illustrates an example of a
function that maps observed attributes to a
linear Trust Score in a range of 0 – 100%. A
Trust Score of 100% indicates that the
system is highly trustworthy, while a score of
0% indicates that the system cannot be
trusted.
System Trustworthiness
A System or solution typically involves
multiple
components.
The
overall
trustworthiness of the system or solution
depends on the trustworthiness of each of
its individual constituent components. The
September 2018
The following figure illustrates an example of
a system and its components represented by
- 34 -