COVID-19 News NER dataset

Description

This dataset contains annotated corona-related news text for fine-grained named entity recognition (NER). The corpus was constructed from Corona news articles published by the German news channel Tagesschau between December 2020 and June 2022, translated into English, cleaned, and annotated with 23 generic and domain-specific entity types. The annotation pipeline combines gold seeds from domain experts, silver seeds from Wikidata, and generic entity annotations from an OntoNotes-trained NER model. The dataset supports research on COVID-19 information extraction, named entity recognition, and downstream relation extraction from news text. The accompanying paper reports training, validation, and test splits of 89,986, 4,999, and 1,000 sentences, respectively, with the test set manually annotated by domain experts.

Please cite

Efeoglu, S., & Paschke, A. (2024). Fine-Grained Named Entities for Corona News. SWAT4HCLS 2023, Basel, Switzerland.

Acknowledgement:

The research presented in this paper was supported in part by the German Federal Ministry of Education and Research (BMBF) project "PANQURA" under grant 03COV03F, in part by the European Union project "FAST-LISA" under grant 101049342.