Assessing the Challenge of Fine-grained Named Entity Recognition and Classification
Named Entity Recognition and Classi- fication (NERC) is a well-studied NLP task typically focused on coarse-grained named entity (NE) classes. NERC for more fine-grained semantic NE classes has not been systematically studied. This pa- per quantifies the difficulty of fine-grained NERC (FG-NERC) when performed at large scale on the people domain. We apply unsupervised acquisition methods to construct a gold standard dataset for FG-NERC. This dataset is used to bench- mark methods for classifying NEs at var- ious levels of fine-grainedness using clas- sical NERC techniques and global contex- tual information inspired from Word Sense Disambiguation approaches. Our results indicate high difficulty of the task and pro- vide a ‘strong’ baseline for future research.