other,37-3-P05-1069,ak | identities </term> themselves , e.g. <term> | block bigram features | </term> . Our <term> training algorithm </term> | #10002 We use a maximum likelihood criterion to train a log-linear block bigram model which uses real-valued features (e.g. a language model score) as well as binary features based on the block identities themselves, e.g. block bigram features . |