A person said four words and you're not sure what the third word was: (w1 = yes, w2 = no, w3 = unknown, w4 = no). The words were generated from this first-order Markov model: p(yes|yes) = 0.7 p(no|yes) = 0.3 p(yes|no) = 0.5 p(no|no) = 0.5 so using bayesian theory What is the probability of whether the missing word is "yes" or "no