[UAI] A varaible that have two states in one case When Modeling Bayesian Network?

From: ruan tong (ruan@otcaix.iscas.ac.cn)
Date: Wed Mar 14 2001 - 09:51:58 PST

  • Next message: Zimmer: "[UAI] CFP: Safety Critical Electronic Systems"

    Hi,

            When I model user interests with Bayesian Network, I find that
    a variable may two states in one case, for example, a document belong
    to two categories, a patient who has many diseases. That is the states
    are not exclusive. How is such kind of suituation normally modeled?
    One way many be make this case two cases. Is it workable and sensible?
            As you know, the
    probabilities should sum to 1, if a variable A has two states, a1 and a2.
    a1 and a2 happen simutaneously. It means that a1 and a2 are compatible,
    It may lead that
    p(a1+a2)!=p(a1)+p(a2). I wonder it contradict with the bayesian rule.

    Tong.

    This archive was generated by hypermail 2b29 : Wed Mar 14 2001 - 10:00:56 PST