Understanding the Limiting Factors of Topic Modeling via Posterior Contraction Analysis

By Jian Tang, Zhaoshi Meng, XuanLong Nguyen, Qiaozhu Mei, Ming Zhang. 2014

283 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

PTE: Predictive Text Embedding through Large-scale Heterogeneous Text Networks (2015, neutral): “This is because document-level word co-occurrences suffer from the sparsity in short documents, with similar results observed in statistical topic models [26].”
On posterior contraction of parameters and interpretability in Bayesian mixture modeling (2019, background): “In a number of modern applications of mixture modeling to heterogeneous data, such as in topic modeling, the number of mixture components (the topics) may be very large and not necessarily a meaningful quantity [3, 49].”
Topic modelling for qualitative studies (2017, methodology): “Literature analysis shows that the values as well as methods of optimization for LDA hyperparameters α and β depend on the final research goals and on quality measures used; suggested values vary greatly [39, 46, 47].”
Anchored Correlation Explanation: Topic Modeling with Minimal Domain Knowledge (2016, background): “The anchored topics reflect different aspects of the framing of the “protests” and “riots,” and are generally interpretable, despite the typical difficulty of extracting coherent topics from short documents using LDA (Tang et al., 2014).”
How to make causal inferences using texts (2018, unknown): “In a formal sense we are interested in studying the posterior contraction rates of the model, a problem taken up analytically in Tang et al. (2014) for the related Latent Dirichlet Allocation model.”
Scalable inference of topic evolution via models for latent geometric structures (2018, background): “The topic or population polytope is a fundamental geometric object that underlies the presence of latent topic variables in topic and admixture models [4, 19, 21].”
Streaming dynamic and distributed inference of latent geometric structures (2018, background): “The topic or population polytope (Nguyen, 2015; Tang et al., 2014) is a fundamental geometric object that underlies the presence of latent topic variables in topic and admixture models (Blei et al., 2003; Pritchard et al., 2000).”