Information Technology Reference
In-Depth Information
At this point we have a resulting value for trustfulness, that is the main goal of the computa-
tional model. However, the resulting values of the other nodes are also shown: they are useful
for further analysis, where thresholds for each feature are considered.
11.12 Experimental Setting
Our experiments show the choice between a doctor and a medical apparatus in the medical
field. We assume that the choice is mainly driven by trustfulness. We have considered two
situations: a 'Routine Visit' and an 'Emergency Visit'. We have built four FCMs representing
trustfulness for doctors and machines in those two situations. Even if the structure of the
nets is always the same, the values of the nodes and the weights of the edges change in
order to reflect the different situations. For example, in the 'Routine Visit' scenario,
Ability
has a great causal power, while in the 'Emergency Visit' one the most important factors is
Accessibility
.
It is also possible to alter some values in order to reflect the impact of
different trustor
personalities
in the choice. For example, somebody who is very concerned with
Danger
can
set its causal power to
very high
even in the 'Routine Visit' scenario, where its importance is
generally low. In the present work we do not consider those additional factors; however, they
can be easily added without modifying the computational framework.
11.12.1 Routine Visit Scenario
The first scenario represents many possible routine visits; there is the choice between a
doctor
and a
medical apparatus
. In this scenario we have set the initial values (i.e. the beliefs sources)
for the
doctor
hypothesizing some direct experience and common sense beliefs about doctors
and the environment.
Most values are set to zero; the others are:
Ability - Direct Experience:
quite (
+
0.3)
;
Ability - Categorization:
very (
+
0.7)
;
Avaialability - categorization:
quite negative (
−
0.3)
;
Unharmfulness - categorization:
some negative (
−
0.2)
;
Opportunity - Reasoning:
some (
+
0.2)
;
Danger - Reasoning:
some negative (
−
0.2)
For the
machine
we have hypothesized no direct experience. These are the values:
Efficacy - Categorization:
good (
+
0.6)
;
Accessibility - Categorization:
good (
+
0.6)
;
Unharmfulness - Categorization:
quite negative (
−
0.3)
;
Opportunity - Reasoning:
some (
+
0.2)
;
Danger - Categorization:
quite negative (
−
0.3)
;
Danger - Reasoning:
quite negative (-0.3)
Search WWH ::
Custom Search