Annotation Guidelines for Named Entity Recognition, Entity Linking and Stance Detection

We describe the NewsEye annotation guidelines for named entity recognition, entity linking and stance detection.
While the part of the guidelines on stance detection annotation is new, these guidelines are derived from Impresso NE annotation guidelines which are derived from Quaero guidelines. Originally designed for the annotation of “extended” named entities (i.e. more than the 3 or 4 traditional classes) in French speech transcriptions, Quaero guidelines have furthermore been used on historic press corpora. Impresso guidelines main’s difference with respect to Quaero’s is reduction: only a subset of Quaero entity types and components are considered, as well as a subset of linguistic units eligible as named entities. These adaptations result from what we deemed most relevant to annotate in our context, and from time and resource constraints. Despite these adaptations, impresso annotated corpora will mostly remain compatible with Quaero guidelines.

These guidelines allowed building a multilingual dataset for named entity recognition, entity linking and stance detection in historical newpapers in French, German, Finnish and Swedish. The paper describing the guidelines and the dataset is available here. If you end up using the guidelines or the resource, please cite this paper:

@inproceedings{Hamdi2021multilingual,
  title={A Multilingual Dataset for Named Entity Recognition, Entity Linking and Stance Detection in Historical Newspapers},
  author={Hamdi, Ahmed and Boro{\c{s}}, Emanuela and Pontes, Elvys Linhares and Nguyen, Thi Tuyet Hai and Hackl, G{\"u}nter and Moreno, Jose G and Doucet, Antoine},
  booktitle={Proceedings of the 44rd International ACM SIGIR Conference on Research and Development in Information Retrieval},
  year={2021}
}

Acknowledgments
This work has been supported by the European Union's Horizon 2020 research and innovation programme under grant 770299 [NewsEye](https://www.newseye.eu/).