{"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/zero-shot-information-extraction-as-a-unified","title":"Zero-Shot Information Extraction as a Unified Text-to-Triple Translation","arxiv_id":"2109.11171","date":"2021-09-23","proceeding":"EMNLP 2021 11","authors":["Chenguang Wang","Xiao Liu","Zui Chen","Haoyun Hong","Jie Tang","Dawn Song"],"abstract":"We cast a suite of information extraction tasks into a text-to-triple translation framework. Instead of solving each task relying on task-specific datasets and models, we formalize the task as a translation between task-specific input text and output triples. By taking the task-specific input, we enable a task-agnostic translation by leveraging the latent knowledge that a pre-trained language model has about the task. We further demonstrate that a simple pre-training task of predicting which relational information corresponds to which input text is an effective way to produce task-specific outputs. This enables the zero-shot transfer of our framework to downstream tasks. We study the zero-shot performance of this framework on open information extraction (OIE2016, NYT, WEB, PENN), relation classification (FewRel and TACRED), and factual probe (Google-RE and T-REx). The model transfers non-trivially to most tasks and is often competitive with a fully supervised method without the need for any task-specific training. For instance, we significantly outperform the F1 score of the supervised open information extraction without needing to use its training set.","url_abs":"https://arxiv.org/abs/2109.11171v1","url_pdf":"https://arxiv.org/pdf/2109.11171v1.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":"zero-shot-information-extraction-as-a-unified","repo_url":"https://github.com/cgraywang/deepex","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"factual-probe","task_name":"Factual probe"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"open-information-extraction","task_name":"Open Information Extraction"},{"task_slug":"relation-classification","task_name":"Relation Classification"},{"task_slug":"translation","task_name":"Translation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/open-information-extraction-on-nyt","task":"Open Information Extraction","dataset":"NYT","model":"DeepEx (zero-shot)","rank_in_archive_order":4,"of":4,"metrics":{"AUC":"72.5","F1":"85.5"},"uses_additional_data":false},{"leaderboard":"/sota/open-information-extraction-on-oie2016","task":"Open Information Extraction","dataset":"OIE2016","model":"DeepEx (zero-shot)","rank_in_archive_order":1,"of":12,"metrics":{"AUC":"58.6","F1":"72.6"},"uses_additional_data":true},{"leaderboard":"/sota/open-information-extraction-on-penn-treebank","task":"Open Information Extraction","dataset":"Penn Treebank","model":"DeepEx (zero-shot)","rank_in_archive_order":3,"of":4,"metrics":{"AUC":"81.5","F1":"88.5"},"uses_additional_data":false},{"leaderboard":"/sota/open-information-extraction-on-web","task":"Open Information Extraction","dataset":"Web","model":"DeepEx (zero-shot)","rank_in_archive_order":4,"of":4,"metrics":{"AUC":"82.4","F1":"91.2"},"uses_additional_data":false},{"leaderboard":"/sota/relation-classification-on-fewrel-1","task":"Relation Classification","dataset":"FewRel","model":"DeepEx (zero-shot top-1)","rank_in_archive_order":4,"of":5,"metrics":{"F1":"48.8"},"uses_additional_data":false},{"leaderboard":"/sota/relation-classification-on-fewrel-1","task":"Relation Classification","dataset":"FewRel","model":"DeepEx (zero-shot top-10)","rank_in_archive_order":5,"of":5,"metrics":{"F1":"92.9"},"uses_additional_data":false},{"leaderboard":"/sota/relation-classification-on-tacred-1","task":"Relation Classification","dataset":"TACRED","model":"DeepEx (zero-shot top-1)","rank_in_archive_order":2,"of":17,"metrics":{"F1":"49.2"},"uses_additional_data":false},{"leaderboard":"/sota/relation-classification-on-tacred-1","task":"Relation Classification","dataset":"TACRED","model":"DeepEx (zero-shot top-10)","rank_in_archive_order":16,"of":17,"metrics":{"F1":"76.4"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2109.11171","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}