--- dataset_info: - config_name: full features: - name: doc_key dtype: string - name: gutenberg_key dtype: string - name: sentences sequence: sequence: string - name: clusters sequence: sequence: sequence: int64 - name: characters list: - name: name dtype: string - name: mentions sequence: sequence: int64 splits: - name: train num_bytes: 118643409 num_examples: 45 - name: validation num_bytes: 5893208 num_examples: 5 - name: test num_bytes: 2732407 num_examples: 3 download_size: 317560335 dataset_size: 127269024 - config_name: split features: - name: doc_key dtype: string - name: gutenberg_key dtype: string - name: sentences sequence: sequence: string - name: clusters sequence: sequence: sequence: int64 - name: characters list: - name: name dtype: string - name: mentions sequence: sequence: int64 splits: - name: train num_bytes: 118849212 num_examples: 7544 - name: validation num_bytes: 5905814 num_examples: 398 - name: test num_bytes: 2758250 num_examples: 152 download_size: 317560335 dataset_size: 127513276 language: - en pretty_name: BOOKCOREF size_categories: - 10M

This data repository contains the BookCoref dataset, introduced in the paper BookCoref: Coreference Resolution at Book Scale by G. Martinelli, T. Bonomo, P. Huguet Cabot and R. Navigli, presented at the ACL 2025 conference. We release both the manually-annotated `test` split (BookCoref_gold) and the pipeline-generated `train` and `validation` splits (BookCoref_silver). In order to enable the replication of our results, we also release a version of the `train`, `validation`, and `test` partitions split into 1500 tokens under the configuration name `split`. ## ⚠️ Project Gutenberg license disclaimer BookCoref is based on books from Project Gutenberg, which are publicly available under the [Project Gutenberg License](https://www.gutenberg.org/policy/license.html). This license holds for users located in the United States, where the books are in the public domain. We do not distribute the original text of the books, rather our dataset consists of a script that downloads and preprocesses the books from an archived verion of Project Gutenberg through the [Wayback Machine](https://web.archive.org/). Users are responsible for checking the copyright status of each book in their country. ## 📚 Quickstart To use the BookCoref dataset, you need to install the following Python packages in your environment: ```bash pip install "datasets==3.6.0" "deepdiff==8.5.0" "spacy==3.8.7" "nltk==3.9.1" ``` You can then load each configuration through Huggingface's `datasets` library: ```python from datasets import load_dataset bookcoref = load_dataset("sapienzanlp/bookcoref") bookcoref_split = load_dataset("sapienzanlp/bookcoref", name="split") ``` These commands will download and preprocess the books, add the coreference annotations, and return a `DatasetDict` according to the requested configuration. ```python >>> bookcoref DatasetDict({ train: Dataset({ features: ['doc_key', 'gutenberg_key', 'sentences', 'clusters', 'characters'], num_rows: 45 }) validation: Dataset({ features: ['doc_key', 'gutenberg_key', 'sentences', 'clusters', 'characters'], num_rows: 5 }) test: Dataset({ features: ['doc_key', 'gutenberg_key', 'sentences', 'clusters', 'characters'], num_rows: 3 }) }) >>> bookcoref_split DatasetDict({ train: Dataset({ features: ['doc_key', 'gutenberg_key', 'sentences', 'clusters', 'characters'], num_rows: 7544 }) validation: Dataset({ features: ['doc_key', 'gutenberg_key', 'sentences', 'clusters', 'characters'], num_rows: 398 }) test: Dataset({ features: ['doc_key', 'gutenberg_key', 'sentences', 'clusters', 'characters'], num_rows: 152 }) }) ``` ## ℹ️ Data format BookCoref is a collection of annotated books. Each item contains the annotations of one book following the structure of OntoNotes: ```python { doc_id: "pride_and_prejudice_1342", # (str) i.e., ID of the document gutenberg_key: "1342", # (str) i.e., key of the book in Project Gutenberg sentences: [["CHAPTER", "I."], ["It", "is", "a", "truth", "universally", "acknowledged", ...], ...], # list[list[str]] i.e., list of word-tokenized sentences clusters: [[[79,80], [81,82], ...], [[2727,2728]...], ...], # list[list[list[int]]] i.e., list of clusters' mention offsets characters: [ { name: "Mr Bennet", cluster: [[79,80], ...], }, { name: "Mr. Darcy", cluster: [[2727,2728], [2729,2730], ...], } ] # list[character], list of characters objects consisting of name and mentions offsets, i,e., dict[name: str, cluster: list[list[int]]] } ``` We also include character names, which are not exploited in traditional coreference settings but could inspire future directions in Coreference Resolution. ## 📊 Dataset statistics BookCoref has distinctly book-scale characteristics, as summarized in the following table:

## 🖋️ Cite this work This work has been published at ACL 2025 (main conference). If you use any artifact of this dataset, please consider citing our paper as follows: ```bibtex @inproceedings{martinelli-etal-2025-bookcoref, title = "{BOOKCOREF}: Coreference Resolution at Book Scale", author = "Martinelli, Giuliano and Bonomo, Tommaso and Huguet Cabot, Pere-Llu{\'i}s and Navigli, Roberto", editor = "Che, Wanxiang and Nabende, Joyce and Shutova, Ekaterina and Pilehvar, Mohammad Taher", booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", month = jul, year = "2025", address = "Vienna, Austria", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2025.acl-long.1197/", pages = "24526--24544", ISBN = "979-8-89176-251-0", } ``` ## Authors - [Giuliano Martinelli](https://www.linkedin.com/in/giuliano-martinelli-20a9b2193/) - [Tommaso Bonomo](https://www.linkedin.com/in/tommaso-bonomo/) - [Pere-lluis Huguet Cabot](https://www.linkedin.com/in/perelluis/) - [Roberto Navigli](https://www.linkedin.com/in/robertonavigli/) ## ©️ License information All the annotations provided by this repository are licensed under the [ Creative Commons Attribution-NonCommercial-ShareAlike 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license.