span-labeling-datasets

🪐 Weasel Project: Span labeling datasets

This project compiles various NER and more general spancat datasets and their converters into the spaCy format. You can use this to try out experiment with ner and spancat or to potentially pre-train them for your application.

📋 project.yml

The project.yml defines the data assets required by the project, as well as the available commands and workflows. For details, see the Weasel documentation.

⏯ Commands

The following commands are defined by the project. They can be executed using weasel run [name]. Commands are only re-run if their inputs have changed.

Command	Description
preprocess-wnut17	Canonicalize the WNUT2017 data set for conversion to .spacy.
convert-wnut17-ents	Convert WNUT17 dataset into the spaCy format
convert-wnut17-spans	Convert WNUT17 dataset into the spaCy format
inspect-wnut17	Analyze span-characteristics
unpack-conll	Decompress ConLL 2002, remove temporary files and change encoding.
preprocess-conll	Canonicalize the Dutch ConLL data set for conversion to .spacy.
convert-conll-spans	Convert CoNLL dataset (de, en, es, nl) into the spaCy format
convert-conll-ents	Convert CoNLL dataset (de, en, es, nl) into the spaCy format
inspect-conll	Analyze span-characteristics
convert-archaeo-spans	Convert Dutch Archaeology dataset into the spaCy format
convert-archaeo-ents	Convert Dutch Archaeology dataset into the spaCy format
inspect-archaeo	Analyze span-characteristics
clean-archaeo
convert-anem-spans	Convert AnEM dataset into the spaCy format
convert-anem-ents	Convert AnEM dataset into the spaCy format
inspect-anem	Analyze span-characteristics
preprocess-restaurant	Make MIT Restaurant Review data set format comply with convert.
convert-restaurant-ents	Convert MIT Restaurant Review data to .spacy format
convert-restaurant-spans	Convert MIT Restaurant Review dataset into the spaCy format
inspect-restaurant	Analyze span-characteristics
generate-unseen	Create unseen entities splits for all preprocessed datasets.

⏭ Workflows

The following workflows are defined by the project. They can be executed using weasel run [name] and will run the specified commands in order. Commands are only re-run if their inputs have changed.

Workflow	Steps
wnut17	preprocess-wnut17 → convert-wnut17-ents → convert-wnut17-spans → inspect-wnut17
conll	unpack-conll → preprocess-conll → convert-conll-ents → convert-conll-spans → inspect-conll
archaeo	convert-archaeo-ents → convert-archaeo-spans → clean-archaeo → inspect-archaeo
anem	convert-anem-ents → convert-anem-spans → inspect-anem
restaurant	preprocess-restaurant → convert-restaurant-ents → convert-restaurant-spans → inspect-restaurant
all	preprocess-wnut17 → convert-wnut17-ents → convert-wnut17-spans → unpack-conll → convert-conll-spans → convert-conll-ents → convert-archaeo-ents → convert-archaeo-spans → convert-anem-ents → convert-anem-spans → preprocess-restaurant → convert-restaurant-ents → convert-restaurant-spans → inspect-restaurant

🗂 Assets

The following assets are defined by the project. They can be fetched by running weasel assets in the project directory.

File	Source	Description
assets/wnut17-train.iob	URL	WNUT17 training dataset for Emerging and Rare Entities Task from Derczynski et al. (ACL 2017)
assets/wnut17-test.iob	URL	WNUT17 test dataset for Emerging and Rare Entities Task from Derczynski et al. (ACL 2017)
assets/wnut17-dev.iob	URL	WNUT17 dev dataset for Emerging and Rare Entities Task from Derczynski et al. (ACL 2017)
assets/conll.tgz	URL	ConLL 2002 shared task data from Tjong Kim Sang (ACL 2002)
assets/archaeo.bio	URL	Dutch Archaeological NER dataset by Alex Brandsen (LREC 2020)
assets/anem-train.iob	URL	Anatomical Entity Mention Detection training dataset from Ohta et al. (ACL 2012)
assets/anem-test.iob	URL	Anatomical Entity Mention Detection test dataset from Ohta et al. (ACL 2012)
assets/restaurant-train_raw.iob	URL	Training data from the MIT Restaurants Review dataset
assets/restaurant-test_raw.iob	URL	Test data from the MIT Restaurants Review dataset