{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/on-generalization-in-coreference-resolution","title":"On Generalization in Coreference Resolution","arxiv_id":"2109.09667","date":"2021-09-20","proceeding":"CRAC (ACL) 2021 11","authors":["Shubham Toshniwal","Patrick Xia","Sam Wiseman","Karen Livescu","Kevin Gimpel"],"abstract":"While coreference resolution is defined independently of dataset domain, most models for performing coreference resolution do not transfer well to unseen domains. We consolidate a set of 8 coreference resolution datasets targeting different domains to evaluate the off-the-shelf performance of models. We then mix three datasets for training; even though their domain, annotation guidelines, and metadata differ, we propose a method for jointly training a single model on this heterogeneous data mixture by using data augmentation to account for annotation differences and sampling to balance the data quantities. We find that in a zero-shot setting, models trained on a single dataset transfer poorly while joint training yields improved overall performance, leading to better generalization in coreference resolution models. This work contributes a new benchmark for robust coreference resolution and multiple new state-of-the-art results.","url_abs":"https://arxiv.org/abs/2109.09667v1","url_pdf":"https://arxiv.org/pdf/2109.09667v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"on-generalization-in-coreference-resolution","repo_url":"https://github.com/shtoshni92/fast-coref","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"on-generalization-in-coreference-resolution","repo_url":"https://github.com/shtoshni/fast-coref","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"coreference-resolution","task_name":"Coreference Resolution"},{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"coreference-resolution-1","task_name":"coreference-resolution"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/coreference-resolution-on-litbank","task":"Coreference Resolution","dataset":"LitBank","model":"longdoc S (OntoNotes + PreCo + LitBank)","rank_in_archive_order":2,"of":2,"metrics":{"F1":"78.2"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-ontonotes","task":"Coreference Resolution","dataset":"OntoNotes","model":"longdoc S (OntoNotes + 60k pseudo-singletons)","rank_in_archive_order":8,"of":26,"metrics":{"F1":"80.6"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-ontonotes","task":"Coreference Resolution","dataset":"OntoNotes","model":"longdoc S (ON + PreCo + LitBank + 30k pseudo-singletons)","rank_in_archive_order":13,"of":26,"metrics":{"F1":"79.6"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-ontonotes","task":"Coreference Resolution","dataset":"OntoNotes","model":"longdoc S (OntoNotes + PreCo + LitBank)","rank_in_archive_order":14,"of":26,"metrics":{"F1":"79.2"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-preco","task":"Coreference Resolution","dataset":"PreCo","model":"longdoc S (OntoNotes + PreCo + LitBank)","rank_in_archive_order":2,"of":2,"metrics":{"F1":"87.6"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-quizbowl","task":"Coreference Resolution","dataset":"Quizbowl","model":"longdoc S (OntoNotes + PreCo + LitBank)","rank_in_archive_order":1,"of":1,"metrics":{"F1":"42.9"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-wikicoref","task":"Coreference Resolution","dataset":"WikiCoref","model":"longdoc S (ON + PreCo + LitBank + 30k pseudo-singletons)","rank_in_archive_order":2,"of":3,"metrics":{"F1":"62.5"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-wikicoref","task":"Coreference Resolution","dataset":"WikiCoref","model":"longdoc S (OntoNotes + PreCo + LitBank)","rank_in_archive_order":3,"of":3,"metrics":{"F1":"60.3"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-winograd-schema","task":"Coreference Resolution","dataset":"Winograd Schema Challenge","model":"longdoc S (OntoNotes + PreCo + LitBank)","rank_in_archive_order":58,"of":82,"metrics":{"Accuracy":"60.1"},"uses_additional_data":false},{"leaderboard":"/sota/coreference-resolution-on-winograd-schema","task":"Coreference Resolution","dataset":"Winograd Schema Challenge","model":"longdoc S (ON + PreCo + LitBank + 30k pseudo-singletons)","rank_in_archive_order":59,"of":82,"metrics":{"Accuracy":"59.4"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2109.09667","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}