In the Maximum Entroy Markov Model (MEMM), we model the sequential tagging problem of finding a sequence of tags given a sequence of words with the following independence assumption: Chain rule: p(t_1, t_2, ..., t_n | x_1, ..., x_n) = Πi=1..n p(s_i | s_1, ..., s_(i−1), x_1, ..., x_n) Independence assumption: p(t_i | t_1, ..., t_(i-1), x_1, ..., x_n) = p(t_i | t_(i-1), x_1, ..., x_n) In addition, we model the conditional distribution using a log linear distribution with a feature extraction function Φ and parameters w. What are the parameters of the feature extraction function Φ?