Trustworthiness Model Representation
Trust Score of a component. An application
can assign different weights to attributes
contributing to the trust measure. Further,
an application can assign weights to logical
sets of attributes.
actions to restore trust. At the same time,
this score will not account for elements such
as policies and procedures that also
contribute to the overall trustworthiness of
an environment. Thus it is a component of an
overall suite of tools that an organization will
employ
to
ensure
end-to-end
trustworthiness, beyond that measured for
its cyber-physical systems.
The qualitative and quantitative parameters
can be further mapped to the five broader
categories of safety, security, privacy,
reliability and resilience of the solution. Each
parameter
contributes
towards
computation of the trustworthiness of each
of these attributes.
A numerical Trust Score is calculated by
using an algorithm that combines individual
scores of all observed attributes.
Representing Trustworthiness Trust Score Function
An
effective
representation
of
trustworthiness is through a numerical score
that provides actionable information to the
user. Based on the Trust Score, the user can A Trust Score Function is responsible for
computing the Trust Score of a solution
based on a set of observed qualitative and
quantitative attributes.
Figure 2: A Trust Score Function takes attributes as input (expected and observed values) and produces a Trust Score
get answers to questions like: (a) Can I trust
an entity? (b) Can I trust a solution? (c) What
is the confidence level of the performance of
an entity or a solution? (d) What factors
were considered in measuring trust? (e)
What affects the trust of an entity or a
solution? (f) What actions should a user take
to restore trust? This quantitative measure
should help the user of a trust model focus
on problem areas by knowing what impacts
trust and enables the user to take corrective
The Trust Score Function defines the range
of input values and their interpretation, and
uses that to calculate the associated Trust
Score. It may be preferred to compute a
Trust Score on a linear scale, as a linear
representation may be easier for users to
interpret. A higher score would represent a
highly trusted solution. A lower score would
represent a lower level of trust in a solution.
- 33 -
IIC Journal of Innovation