... expand collocations indicative of each move (6)Develop a hidden Markov model for move tagging Figure 1: Processes used to learn collocation classifiers 3.1 Collecting Training Data In the first four ... ) t1 ,t2 , ,tn In summary, at the beginning of training time, we use a few human move- tagged sentences as seed data Then, collocation-to -move and moveto -move probabilities are employed to build ... number of sentences with collocation in each abstract, move sequence score and collocation score Table 2: The score of bi-gram move sequence (Note that “$” denotes the beginning or the ending of...