Topic Modeling 251
mate.alpha = TRUE) and as Griffiths and Steyvers suggested [1], this value should be
set at 50/k, where k is the number of topics. Where α is not set as estimated, it will
stay at the initial value. β, the term distribution over topics, is set at estimate.beta
= TRUE. If however, the term distribution have been fitted by a previous model,
setting estimate.beta = FALSE allows hold-out data or testing data to be applied to
the fitted model.
• verbose: verbose is set to 0 as default which means during the algorithm, nothing
will be printed, if this value changes to a positive integer all iterations during the
algorithm will be printed.
• save & prefix: save is default to zero meaning no intermediate results are saved to
file with prefix
• k