Latent Dirichlet allocation
Latent Dirichlet allocation is a generative probabilistic topic model representing documents as mixtures of latent topics and topics as distributions over words.
Natural Language Processing in Action, Second Edition
Latent Dirichlet allocation is a probabilistic topic-modeling technique that represents documents as mixtures of latent topics and topics as distributions over words.
- How it is used
- Creates interpretable topic vectors for topic analysis, with demonstrations in toxicity classification and potential use in semantic search and summarization.
- How deep it goes
- Explains document–topic mixtures and Dirichlet priors, then fits and evaluates a scikit-learn model without deriving its statistical algorithm.
- Examples
- Fits 16-topic and 32-topic models to toxic comments, inspects term allocations, and compares classifier performance.
- What you can do after
- Apply it via scikit-learn to build comment topic vectors, inspect their terms, and evaluate them as classifier features.
Machine Learning Algorithms in Depth
Latent Dirichlet allocation is a generative probabilistic model that represents documents as mixtures of topics and topics as distributions over words.
- How it is used
- Uses LDA to discover shared topics and document-level topic proportions in text collections.
- How deep it goes
- Develops the bag-of-words model, latent variables, variational inference, online updates, and a from-scratch implementation.
- Examples
- Applies LDA to 20 Newsgroups and visualizes ELBO progress and inferred topics as word clouds.
- What you can do after
- Build it from scratch with variational Bayes to infer and inspect topics in a document collection.
Read how Machine Learning Algorithms in Depth explains latent Dirichlet allocation →
Related concepts
- topic modelkind of · A topic model is a probabilistic latent-variable model that represents documents as mixtures of unobserved thematic topics inferred from observed words.
- approximate distributionuses · An approximate posterior distribution is a tractable probability distribution selected or optimized to represent an intractable posterior distribution over unknown quantities.
- topic modelingimplements · Topic modeling is an unsupervised statistical or computational technique for inferring recurring latent thematic structures from a collection of documents.
- topic vectorproduces · A numerical vector encoding the strengths of associations between terms or documents and latent topics.
- vector space modeluses · A model that represents documents or other language objects as vectors and uses relationships among those vectors for analysis or retrieval.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.