TEAMMix framework enhances LLM-based hierarchical text classification

Researchers have developed TEAMMix, a novel framework designed to enhance Hierarchical Text Classification (HTC) using LLM-based data augmentation. This method addresses challenges like complex label hierarchies and class imbalance by first semantically enriching the label hierarchy through keyword generation and corpus mining. It then uses LLMs to generate pseudo-samples to combat the long-tail problem and employs a Gaussian mixture model for confidence-based resampling to improve data quality. Experiments show that TEAMMix effectively boosts the reliability of LLM-generated labels and significantly improves classification performance on fine-grained and imbalanced datasets. AI

IMPACT This research offers a new approach to improve the performance of LLMs in hierarchical text classification tasks, particularly for imbalanced datasets.

RANK_REASON The cluster contains an academic paper detailing a new method for text classification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• Gaussian mixture model
• hierarchical text classification
• LLM
• TEAMMix

• paper
• model release

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