A Gamma-Poisson Mixture Topic Model for Short Text

Algorithm 1

Collapsed Gibbs sampler for GPM.

	Input: Corpus,
	Output: Topic labels for each document,
	Begin
	Initialise and to 0 for each topic
	for each document ,
	randomly sample a topic for
	and
	for each word frequency in
	for iterations
	for each document ,
	record the current topic of document :
	and
	for each word frequency in
	sample a new topic for
	equation (5)
	and
	for each word frequency in