English

Entry-Wise Eigenvector Analysis and Improved Rates for Topic Modeling on Short Documents

Statistics Theory 2024-05-29 v1 Statistics Theory

Abstract

Topic modeling is a widely utilized tool in text analysis. We investigate the optimal rate for estimating a topic model. Specifically, we consider a scenario with nn documents, a vocabulary of size pp, and document lengths at the order NN. When NcpN\geq c\cdot p, referred to as the long-document case, the optimal rate is established in the literature at p/(Nn)\sqrt{p/(Nn)}. However, when N=o(p)N=o(p), referred to as the short-document case, the optimal rate remains unknown. In this paper, we first provide new entry-wise large-deviation bounds for the empirical singular vectors of a topic model. We then apply these bounds to improve the error rate of a spectral algorithm, Topic-SCORE. Finally, by comparing the improved error rate with the minimax lower bound, we conclude that the optimal rate is still p/(Nn)\sqrt{p/(Nn)} in the short-document case.

Keywords

Cite

@article{arxiv.2405.17806,
  title  = {Entry-Wise Eigenvector Analysis and Improved Rates for Topic Modeling on Short Documents},
  author = {Zheng Tracy Ke and Jingming Wang},
  journal= {arXiv preprint arXiv:2405.17806},
  year   = {2024}
}

Comments

50 pages

R2 v1 2026-06-28T16:43:13.862Z