One technique modify the affirms and denies weights reflect the uncertainty
/* Rule 2.7 */
IF temperature high AND NOT(water level low) THEN pressure high
A Bayesian version of this rule might be:
• the evidence could be an assertion generated by another uncertain rule, and which therefore has a probability associated with it;
• the evidence may be in the form of data which are not totally reliable, such as the output from a sensor.
In terms of probabilities, we wish to calculate P(HE), where E is uncertain. We can handle this problem by assuming that E was asserted by another rule whose evidence was B, where B is certain (has probability 1). Given the evidence B, the probability of E is P(EB). Our problem then becomes one of calculating P(HB). An expression for this has been derived by Duda et al. [3]:


D' = [2(1 – D) � P(E)] + D (3.17)
3.2.6 Combining evidence