| |
Explanation of eXpertise2Go Conclusions
Minimum confidence factor for accepting a value as a fact: 80.0%
Value 1 of the recommended action is refuel the car with 91.6% confidence
Determined the recommended action is refuel the car with 91.6% confidence from:
| |
[1]Rule 2 (out of gas) is true and assigned the value refuel the car with 91.6% confidence:
| | If the gas tank is empty
Then
the recommended action is refuel the car
Determined the gas tank is empty with 91.6% confidence from:
| |
[1]empty was input with 70.0% confidence
[2]Rule 5 (empty tank) is true and assigned the value empty with 72.0% confidence:
| | If the result of trying the starter is the car cranks normally
and a gas smell is not present when trying the starter
Then
the gas tank is empty with 90% confidence
Determined the result of trying the starter is the car cranks normally with 100.0% confidence from:
| |
[1]the car cranks normally was input with 100.0% confidence
|
Determined a gas smell is not present when trying the starter with 80.0% confidence from:
| |
[1]not present when trying the starter was input with 80.0% confidence
|
|
Multi-source confidence calculation for the gas tank is empty: 70.0% + 72.0% of (100% - 70.0%) = 91.6%
|
|
|
|
All of the values of all of the goals for a consultation are listed first. In this
example, there is only one value of one goal: the recommended action. Then,
a detailed explanation of the logic eXpertise2Go used to derive each
value is provided. A vertical gray bar ties together all of the evidence used to
determine the value provided at the top of the bar. This evidence could include
rules, user input and default settings.
If multiple sources were combined to determine an attribute's value,
the confidence factor (CF) calculations are summarized as the last item beside the
gray bar that delineates all of the evidence. The
Inference Methods and Uncertainty
tutorial explains the techniques used to calculate confidence factors
when values are determined from multiple sources and when the if part
of a rule is evaluated with some degree of uncertainty.
|