Q. What is the purpose of the latent factors weight matrix Λ in LSA?

View solution

Q. In LSA, what is the primary purpose of dimensionality reduction?

View solution

Q. What characteristic do both matrices U and VT have in LSA?

View solution

Q. What is the outcome of LSA analysis typically based on?

View solution

Q. In the context of LSA, what is the benefit of using a subset of singular values for dimensionality reduction?

View solution

Q. What does the "folding in" method allow LSA/LSI to do?

View solution

Q. What is a key limitation of techniques like latent semantic analysis (LSA) as compared to generative approaches like LDA?

View solution

Q. What is the primary advantage of latent Dirichlet analysis (LDA) over techniques like LSA?

View solution

Q. In the context of LDA, what are anchor terms?

View solution

Q. What role does the Dirichlet distribution play in LDA?

View solution

Q. How is the overall number of topics for a given LDA analysis determined?

View solution

Q. What does the generative model in LDA describe?

View solution

Q. What is the primary purpose of using Gibbs sampling or variational inference methods in LDA?

View solution

Q. How are the results of LDA typically evaluated?

View solution

Q. What is a useful indication for selecting appropriate numbers of topics when evaluating LDA from a domain perspective?

View solution

Q. How does LDA handle multilabel classification in the context of topic modeling?

View solution

Q. What does LDA's ability to uncover scientific topics "hidden" in documents provide from a university point of view?

View solution

Q. What is a challenge in introducing a flexible number of topics in LDA, as compared to clustering methods allowing for an adaptable number of clusters?

View solution

Q. What motivated the extension of the LDA approach to allow for an infinite number of topics?

View solution

Q. What does the Dirichlet process (DP) primarily model when used in the context of hierarchical Dirichlet process (HDP)?

View solution