“…Those methods include n-gram language models, hidden Markov models, nave Bayes classifiers, Bayesian networks, multilayer perceptrons, decision trees, transformation-based learning, and memory-based learning [3,5,9,[20][21][22][23][24][25]. Most of them, however, utilized language model information, which cannot be used in our problem since the two types, DQs and statements, have the same transcription.…”