This model is a BERT-based Location Mention Recognition model that is adopted from the TLLMR4CM GitHub. The model identifies the toponyms' spans in the text and predicts their location types. The location type can be coarse-grained (e.g., country, city, etc.) and fine-grained (e.g., street, POI, etc.)
The model is trained using the training splits of all events from IDRISI-R dataset under the Type-based
LMR mode and using the Time-based
version of the data. You can download this data in BILOU
format from here. More details about the models are available here.
Different variants of the model are available through HuggingFace:
Arabic models are also available:
To cite the models:
@article{suwaileh2022tlLMR4disaster,
title={When a Disaster Happens, We Are Ready: Location Mention Recognition from Crisis Tweets},
author={Suwaileh, Reem and Elsayed, Tamer and Imran, Muhammad and Sajjad, Hassan},
journal={International Journal of Disaster Risk Reduction},
year={2022}
}
@inproceedings{suwaileh2020tlLMR4disaster,
title={Are We Ready for this Disaster? Towards Location Mention Recognition from Crisis Tweets},
author={Suwaileh, Reem and Imran, Muhammad and Elsayed, Tamer and Sajjad, Hassan},
booktitle={Proceedings of the 28th International Conference on Computational Linguistics},
pages={6252--6263},
year={2020}
}
To cite the IDRISI-R dataset:
@article{rsuwaileh2022Idrisi-r,
title={IDRISI-R: Large-scale English and Arabic Location Mention Recognition Datasets for Disaster Response over Twitter},
author={Suwaileh, Reem and Elsayed, Tamer and Imran, Muhammad},
journal={...},
volume={...},
pages={...},
year={2022},
publisher={...}
}
- Downloads last month
- 45
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.