BERTopic: Neural topic modeling with a class-based TF-IDF procedure
By Maarten Grootendorst. arXiv.org, 2022
2,855 citations recorded by the source corpus.
Topic models can be useful tools to discover latent topics in collections of documents. Recent studies have shown the feasibility of approach topic modeling as a clustering task. We present BERTopic, a topic model that extends this process by extracting coherent topic representation through the development of a class-based variation of TF-IDF. More specifically, BERTopic generates document embedding with pre-trained transformer-based language models, clusters these embeddings, and finally, generates topic representations with the class-based TF-IDF procedure. BERTopic generates coherent topics and remains competitive across a variety of benchmarks involving classical models and those that follow the more recent clustering approach of topic modeling.
Subfields: Artificial Intelligence
View Paper: https://www.semanticscholar.org/paper/5493aefc07e331045be76003ea516aef57f76cb9
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Ion Stoica
- Joseph E. Gonzalez
- Mikke Tavast
- Anastasios Nikolas Angelopoulos
- Walter Quattrociocchi
- Lianmin Zheng
- Ying Sheng
- Wei-Lin Chiang
Citations in the corpus
- Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference (2024, methodology): “To study the prompt diversity, we build a topic modeling pipeline with BERTopic 3 (Grootendorst, 2022).”
- Evaluating Large Language Models in Generating Synthetic HCI Research Data: a Case Study (2023, result): “completions to PROMPT 3) is based on the observation The embedding, dimensionality reduction, and clustering steps that GPT-3 can be prompted to perform a form of qualitative in- are the same as in the BerTopic topic mining approach [26].”
- Growing polarization around climate change on social media (2021, methodology): “Figure 4 shows the ideological position of topics in the COP26 discussion extracted using BERT topic modelling [29], see Methods.”
- KG-Anchored RAG: Retrieval-Augmented Generation for Power System Professional Documents Integrating Topic Modeling and Knowledge Graphs (2026, unknown): “The topic mapping was implemented using BERTopic [18].”
- Bisociative Pivoting as a Method for Moderating User Response to Text Similarity in Next-item News Recommendation (2026, unknown): “Current transformer-based systems represent a paradigm shift: they import rich semantic knowledge from massive general corpora [9], enabling nuanced modeling of news texts.”
- Mapping Scientific Knowledge in Systematic Literature Reviews Through Embeddings and Bibliometric Techniques (2026, unknown): “More recent neural topic-modeling and density-based approaches - such as BERTopic [73], Top2Vec [74], and HDBSCAN-based clustering [75] combined with UMAP dimensionality reduction [76] - share our reliance on trans-former embeddings but differ in the clustering and topic-representation stages.”