New ML pipeline detects self-introductions in legislative testimony

Researchers have developed a machine learning pipeline to automatically detect self-introductions and extract speaker names from legislative testimonies. The system, trained on data from five state legislative sessions, utilizes a combination of features including bag-of-words, positional context, and discourse context. An XGBoost classifier achieved a high F1 score of 0.9747, which was further improved to 0.9782 by augmenting it with fine-tuned BERT probability features. AI

IMPACT This research could improve the efficiency of analyzing legislative proceedings and aid in speaker identification tasks.

RANK_REASON Academic paper detailing a new machine learning pipeline for text analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• BERT
• CatalyzeX
• Connected Papers
• DagsHub
• Gotit.pub
• Hugging Face
• Litmaps
• ScienceCast
• scite Smart Citations
• XGBoost

• paper
• other

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