Annotation Errors and NER: A Study with OntoNotes 5.0
This work addresses data quality issues for researchers and practitioners using NER datasets, though it is incremental as it applies existing error detection methods to a specific corpus.
The study tackled the problem of annotation errors in the OntoNotes 5.0 corpus for English NER by employing three simple detection techniques, resulting in corrections to ~10% of sentences and ~8% of entity mentions, and showing an average improvement of 1.23% in F-scores when training models on the corrected dataset.
Named Entity Recognition (NER) is a well-studied problem in NLP. However, there is much less focus on studying NER datasets, compared to developing new NER models. In this paper, we employed three simple techniques to detect annotation errors in the OntoNotes 5.0 corpus for English NER, which is the largest available NER corpus for English. Our techniques corrected ~10% of the sentences in train/dev/test data. In terms of entity mentions, we corrected the span and/or type of ~8% of mentions in the dataset, while adding/deleting/splitting/merging a few more. These are large numbers of changes, considering the size of OntoNotes. We used three NER libraries to train, evaluate and compare the models trained with the original and the re-annotated datasets, which showed an average improvement of 1.23% in overall F-scores, with large (>10%) improvements for some of the entity types. While our annotation error detection methods are not exhaustive and there is some manual annotation effort involved, they are largely language agnostic and can be employed with other NER datasets, and other sequence labelling tasks.