plsa source code undeer Map-reduce frame the input is doct word pair per-line E_step M_step according to EM-algorithm init folder: map: input source data, output doc-word reduce: init EM parameters EM folder: E-step: update p(topic|doc, word) M-step: update p(topic|doc) p(word|topic)=N(word,topic)|N(topic) run scirpt: map-reduce process