Skip to main content

Module topic_modeler

Module topic_modeler 

Source
Expand description

Semantic Topic Modeller — online clustering approach for latent topic modelling.

Models latent topics from a collection of embeddings using a simple online clustering approach, assigning documents to topics and tracking topic drift over time.

Structs§

DocumentTopics
Topic probability distribution for a single document.
LdaTopic
A latent topic with its top words and coherence score.
ModelDocument
Bag-of-words representation of a document.
ModellerConfig
Configuration for SemanticTopicModeller.
SemanticTopicModeller
Online topic modeller that clusters embeddings into latent topics.
TopicAssignment
Records the assignment of a document to a topic at a point in time.
TopicModel
A latent topic represented as a running centroid of assigned embeddings.
TopicModelConfig
Hyperparameters for the LDA topic model.
TopicModelResult
Full output of a fitted topic model.
TopicModeler
LDA-based topic modeler using collapsed Gibbs sampling.
TopicModelerStats
Aggregate statistics derived from a fitted model result.
TopicModellerStats
Aggregate statistics for a SemanticTopicModeller.
TopicWord
A word and its probability within a topic.

Enums§

TopicModelError
Error type for LDA topic modeling operations.

Functions§

cosine_sim
Computes the cosine similarity between two vectors.