Topic Modeling

An unsupervised statistical method family that infers recurring latent thematic structures in large text corpora by estimating representations in which each document is expressed as a mixture of topics and each topic is a probability distribution over words (e.g., Latent Dirichlet Allocation).