latent dirichlet allocation (lda) for topic modeling

A generative model that allows sets of observations to be explained by unobserved groups, often used for discovering abstract topics from a collection of documents.