Biomedical Text Categorization Based on Ensemble Pruning and Optimized Topic Modelling

Box 1

The generative process of LDA (Blei et al., 2013; [19, 20]).

For each document in a corpus D:

Choose N Poisson (ξ).

Choose Θ Dir (α).

For each of the N words :

(a) Choose a topic z_n Multinomial (Θ).

Choose a word from p(, β), a multinomial probability conditioned on the topic z_n.