On collocations and topic models

It is shown how the Akaike information criterion is a more appropriate measure, which suggests that using a modest number of top-ranked bigrams is the optimal topic modelling configuration, and that using these 1000 bigrams results in improved topic quality over unigram tokenization.