New DARTopic framework enhances neural topic modeling with graph networks

Researchers have developed DARTopic, a novel framework for domain-agnostic neural topic modeling. This approach utilizes a graph neural network operating on token-level embeddings from pre-trained language models to capture corpus-specific semantic structures. DARTopic demonstrates superior performance in topic coherence and document clustering across general, biomedical, and legal domains compared to existing methods, without requiring any fine-tuning of the underlying encoder. AI

IMPACT Enhances topic interpretability and clustering for specialized corpora without encoder fine-tuning.

RANK_REASON Paper describing a new model/framework for topic modeling. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• CatalyzeX
• Connected Papers
• CORE Recommender
• DagsHub
• DARTopic
• Gotit.pub
• graph neural network
• Hugging Face
• Influence Flower
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →