R/top.topic.words.R

In lda: Collapsed Gibbs Sampling Methods for Topic Models

top.topic.words <-
function (topics, num.words = 20, by.score = FALSE) 
{
    if (by.score) {
        normalized.topics <- topics/(rowSums(topics) + 1e-05)
        scores <- apply(normalized.topics, 2, function(x) x * 
            (log(x + 1e-05) - sum(log(x + 1e-05))/length(x)))
        apply(scores, 1, function(x) colnames(scores)[order(x, 
            decreasing = TRUE)[1:num.words]])
    }
    else {
        apply(topics, 1, function(x) colnames(topics)[order(x, 
            decreasing = TRUE)[1:num.words]])
    }
}