for estimating the topic model

act -est -ntopics 50 -twords 20 -dfile train.txt

50 is the topic number, you can change according to your app.

the output files include: 
.tassign, the topic assigned to each word
.theta, the probability P(z|d), z is topic, d is document
.theta_ak, the probability p(z|a), a is author, if you use document id as author, this should be the same as p(z|d)
.theta_ck, the probability p(z|c), c is the conference, 
.phi, the probability p(w|z), w is a word
.twords, the popular (top 20) words and conferences (rates) under each topic
