2019
ACL
ACL 2019
Large-Scale Multi-Label Text Classification on EU Legislation
Abstract
AbstractWe consider Large-Scale Multi-Label Text Classification (LMTC) in the legal domain. We release a new dataset of 57k legislative documents from EUR-LEX, annotated with βΌ4.3k EUROVOC labels, which is suitable for LMTC, few- and zero-shot learning. Experimenting with several neural classifiers, we show that BIGRUs with label-wise attention perform better than other current state of the art methods. Domain-specific WORD2VEC and context-sensitive ELMO embeddings further improve performance. We also find that considering only particular zones of the documents is sufficient. This allows us to bypass BERTβs maximum text length limit and fine-tune BERT, obtaining the best results in all but zero-shot learning cases.
π
Interdisciplinary Bridge
β Artificial Intelligence and Natural Language Processing
π
Trend Setter
β Few-Shot Learning
π§
Keyword Pioneer
β bigru network
π£
Hot Topic Early Bird
β few-shot learning
π
Cross-Pollinator
β Artificial Intelligence, Computer Vision, Data Science & Analytics, Deep Learning, Healthcare & Medicine, Interdisciplinary, Machine Learning, Mathematics & Optimization, Natural Language Processing, Reinforcement Learning, Robotics, Speech & Audio
Topics
Artificial Intelligence > Learning Paradigms > Few-Shot Learning
Machine Learning > Learning Types > Zero-Shot Learning
Natural Language Processing > Applications > Text Classification
Natural Language Processing > Resources & Methods > Multilingual NLP
Machine Learning > Learning Paradigms > Few-Shot Learning
Machine Learning > Learning Types > Multi-Label Learning
Deep Learning > Learning Types > Transfer Learning
Deep Learning > Learning Types > Multi-Label Classification