Understanding the topic model typically mentions several of the most basic build models: Unigram model, Mixture of Unigram,plsa, and then briefly describe the difference between them:
1.Unigram model
The diagram on the left shows that a document is generated by words, each word has its probability of appearing, and the product of all word probabilities is the probability of generating a document.
2.Mixture of Unigram
Compared to Unigram, the probability of the probability product of all the words appearing under each topic is the probabilities of generating the document.
3.pLSA
The difference between pLSA and mix Unigram is given a number of possible topics, and the model is a conjugate distribution Dirichlet distribution that has no word distribution and topic distribution corresponding to LDA, i.e.
Several simple topic models (build models)