arXiv paper synthesizes keyphrase extraction methods from statistical to LLMs

A new paper on arXiv provides a comprehensive overview of automatic keyphrase extraction (KE) techniques for both text and speech. It synthesizes various research strands into a unified framework, tracing the field's evolution from statistical methods to large language models (LLMs) and end-to-end speech processing. The paper details different KE methodologies, including statistical, hybrid, machine learning-based, and structural approaches, with a focus on emerging paradigms like generative KE using LLMs and challenges in automatic corpus annotation. It also specifically analyzes current directions and difficulties in spoken language KE, comparing transcript-based and end-to-end acoustic methods. The authors conclude that the field is moving towards a more integrated, context-aware paradigm, with future progress dependent on addressing data scarcity for low-resource languages, multimodal fusion, and nuanced evaluation of generative KE systems. AI

IMPACT Provides a unified framework for understanding keyphrase extraction, highlighting the shift towards LLMs and multimodal approaches.

RANK_REASON The item is an academic paper published on arXiv detailing theoretical approaches and practical solutions for keyphrase extraction. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• Influence Flower
• large language models
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• other

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