English

Pre-training is a Hot Topic: Contextualized Document Embeddings Improve Topic Coherence

Computation and Language 2021-06-18 v2

Abstract

Topic models extract groups of words from documents, whose interpretation as a topic hopefully allows for a better understanding of the data. However, the resulting word groups are often not coherent, making them harder to interpret. Recently, neural topic models have shown improvements in overall coherence. Concurrently, contextual embeddings have advanced the state of the art of neural models in general. In this paper, we combine contextualized representations with neural topic models. We find that our approach produces more meaningful and coherent topics than traditional bag-of-words topic models and recent neural models. Our results indicate that future improvements in language models will translate into better topic models.

Keywords

Cite

@article{arxiv.2004.03974,
  title  = {Pre-training is a Hot Topic: Contextualized Document Embeddings Improve Topic Coherence},
  author = {Federico Bianchi and Silvia Terragni and Dirk Hovy},
  journal= {arXiv preprint arXiv:2004.03974},
  year   = {2021}
}

Comments

Updated version. Published as a conference paper at ACL-IJCNLP 2021

R2 v1 2026-06-23T14:44:12.375Z