mlsanityData to 7 Oct 2026

Are Neural Topic Models Broken?

arXiv 2210.16162 ↗First posted 28 Oct 2022cs.CL cs.HCcited by 11

Not on the map. Why?

Abstract (arXiv)

Recently, the relationship between automated and human evaluation of topic models has been called into question. Method developers have staked the efficacy of new topic model variants on automated measures, and their failure to approximate human preferences places these models on uncertain ground. Moreover, existing evaluation paradigms are often divorced from real-world use. Motivated by content analysis as a dominant real-world use case for topic modeling, we analyze two related aspects of topic models that affect their effectiveness and trustworthiness in practice for that purpose: the stability of their estimates and the extent to which the model's discovered categories align with human-determined categories in the data. We find that neural topic models fare worse in both respects compared to an established classical method. We take a step toward addressing both issues in tandem by demonstrating that a straightforward ensembling method can reliably outperform the members of the ensemble.

Key facts. Cited by 11 papers in the map (data through 7 Oct 2026). Builds on 18 of its 135 references that are in the map. First posted on arXiv 28 Oct 2022.

Builds on

18 of 135 references.

References · 18

  1. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,049
  2. 2016Pointer Sentinel Mixture Modelscited by 2,992
  3. 2019Topic Modeling in Embedding Spacescited by 224
  4. 2020Pre-training is a Hot Topic: Contextualized Document Embeddings Improve Topic Coherencecited by 121
  5. 2021Is Automated Topic Model Evaluation Broken?: The Incoherence of Coherencecited by 74
  6. 2020Cross-lingual Contextualized Topic Models with Zero-shot Learningcited by 66
  7. 2021Topic Modelling Meets Deep Neural Networks: A Surveycited by 51
  8. 2017Neural Models for Documents with Metadatacited by 46
  9. 2019Topic-Guided Variational Autoencoders for Text Generationcited by 44
  10. 2020Improving Neural Topic Models using Knowledge Distillationcited by 25
  11. 2014How Many Topics? Stability Analysis for Topic Modelscited by 16
  12. 2016What is Wrong with Topic Modeling? (and How to Fix it Using Search-based Software Engineering)cited by 14
6 more
  1. 2020Context Reinforced Neural Topic Modeling over Short Textscited by 10
  2. 2017Topically Driven Neural Language Modelcited by 8
  3. 2021Fine-tuning Encoders for Improved Monolingual and Zero-shot Polylingual Neural Topic Modelingcited by 8
  4. 2020A Topic Coverage Approach to Evaluation of Topic Modelscited by 7
  5. 2018Measuring LDA Topic Stability from Clusters of Replicated Runscited by 6
  6. 2014A Deep and Autoregressive Approach for Topic Modeling of Multimodal Datacited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 5; 2025: 3; 2026: 2.

11 papers.

  1. 2024Trust No Bot: Discovering Personal Disclosures in Human-LLM Conversations in the Wildcited by 43
  2. 2024FASTopic: Pretrained Transformer is a Fast, Adaptive, Stable, and Transferable Topic Modelcited by 34
  3. 2025Sparse Autoencoders for Hypothesis Generationcited by 24
  4. 2024Explaining Datasets in Words: Statistical Models with Natural Language Parameterscited by 21
  5. 2023Contextualized Topic Coherence Metricscited by 19
  6. 2025Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoderscited by 7
  7. 2024Interactive Topic Models with Optimal Transportcited by 1
  8. 2024Enhancing Topic Interpretability for Neural Topic Modeling through Topic-wise Contrastive Learningcited by 1
  9. 2025Objectifying the Subjective: Cognitive Biases in Topic Interpretationscited by 0
  10. 2026Improving Topic Modeling by Distilling Soft Labels from Language Modelscited by 0
  11. 2026Segment-Level Agentic Topic Modeling for Improved Data Exploration and Resource Efficiencycited by 0