: Vocabulario de PLN: correlated topics model

NLP methods and tools > NLP algorithm > topic modelling algorithm > correlated topics model

correlated topics model

Correlated topic models (Blei and Lafferty 2005 2007) are LDA extensions that attempt to learn the structure of topic associations within a document. (Byrne, Horak, Moilanen & Mabona, 2022)

Correlated Topic Model (Blei and Lafferty 2007) replaces Dirichlet prior with logistic Normal prior for topic distribution in each document in order to capture the correlation between the topics. (Wang, Zhang & Zhai, 2011)
Correlated topic models and their neural extensions learn a flat topic structure while adding scalar associations whereas our method induces a topic graph. (Byrne, Horak, Moilanen & Mabona, 2022)
Correlated topic models were introduced 2006 (Blei and Lafferty 2006; Li and McCallum 2006) to improve topic coherence and to provide graph visualizations based on topics as nodes and their correlations as edges. (Hagerer, Kirchhoff, Danner, Pesch, Ghosh, Roy, Zhao & Groh, 2021)
The Correlated Topic Model (Blei and Lafferty 2006) induces a correlation structure between topics by using the logistic normal distribution instead of the Dirichlet. (Bairi, Iyer, Ramakrishnan & Bilmes, 2015)
We therefore use the correlated topic model (CTM) specifically designed to model topical structures (Blei and Lafferty 2006). (Deng, Kuleshov & Rush, 2022)

http://data.loterre.fr/ark:/67375/8LP-W6H2PH3G-0

RDF/XML TURTLE JSON-LD última modificación 30/9/24

Loterre