Tristan Luiggi, Laure Soulier, Vincent Guigue, Siwar Jendoubi, Aurélien Baelde
TL;DR本研究介绍了一项新任务：Dynamic Named Entity Recognition（DNER），提供了一个框架，以更好地利用上下文来评估算法提取实体的能力。DNER基于两个数据集，DNER-RotoWire和DNER-IMDb，我们评估了基线模型并提出了与此新任务相关的问题和研究方向的实验。
Abstract
named entity recognition (NER) is a challenging and widely studied task that involves detecting and typing entities in text. So far,NER still approaches entity typing as a task of classification into universal classes (e.g. date, person, or location). Recent advances innatural language