{"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/graph-convolution-over-pruned-dependency","title":"Graph Convolution over Pruned Dependency Trees Improves Relation Extraction","arxiv_id":"1809.10185","date":"2018-09-26","proceeding":"EMNLP 2018 10","authors":["Yuhao Zhang","Peng Qi","Christopher D. Manning"],"abstract":"Dependency trees help relation extraction models capture long-range relations\nbetween words. However, existing dependency-based models either neglect crucial\ninformation (e.g., negation) by pruning the dependency trees too aggressively,\nor are computationally inefficient because it is difficult to parallelize over\ndifferent tree structures. We propose an extension of graph convolutional\nnetworks that is tailored for relation extraction, which pools information over\narbitrary dependency structures efficiently in parallel. To incorporate\nrelevant information while maximally removing irrelevant content, we further\napply a novel pruning strategy to the input trees by keeping words immediately\naround the shortest path between the two entities among which a relation might\nhold. The resulting model achieves state-of-the-art performance on the\nlarge-scale TACRED dataset, outperforming existing sequence and\ndependency-based neural models. We also show through detailed analysis that\nthis model has complementary strengths to sequence models, and combining them\nfurther improves the state of the art.","url_abs":"http://arxiv.org/abs/1809.10185v1","url_pdf":"http://arxiv.org/pdf/1809.10185v1.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":"graph-convolution-over-pruned-dependency","repo_url":"https://github.com/qipeng/gcn-over-pruned-trees","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"negation","task_name":"Negation"},{"task_slug":null,"task_name":"Relation"},{"task_slug":"relation-classification","task_name":"Relation Classification"},{"task_slug":"relation-extraction","task_name":"Relation Extraction"}],"methods":[{"method_slug":"pruning","method_name":"Pruning"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/relation-classification-on-tacred-1","task":"Relation Classification","dataset":"TACRED","model":"C-GCN","rank_in_archive_order":5,"of":17,"metrics":{"F1":"66.4"},"uses_additional_data":false},{"leaderboard":"/sota/relation-extraction-on-re-tacred","task":"Relation Extraction","dataset":"Re-TACRED","model":"C-GCN","rank_in_archive_order":6,"of":9,"metrics":{"F1":"80.3"},"uses_additional_data":false},{"leaderboard":"/sota/relation-extraction-on-tacred","task":"Relation Extraction","dataset":"TACRED","model":"C-GCN + PA-LSTM","rank_in_archive_order":26,"of":40,"metrics":{"F1":"68.2"},"uses_additional_data":false},{"leaderboard":"/sota/relation-extraction-on-tacred","task":"Relation Extraction","dataset":"TACRED","model":"GCN + PA-LSTM","rank_in_archive_order":33,"of":40,"metrics":{"F1":"67.1"},"uses_additional_data":false},{"leaderboard":"/sota/relation-extraction-on-tacred","task":"Relation Extraction","dataset":"TACRED","model":"C-GCN","rank_in_archive_order":35,"of":40,"metrics":{"F1":"66.4"},"uses_additional_data":false},{"leaderboard":"/sota/relation-extraction-on-tacred","task":"Relation Extraction","dataset":"TACRED","model":"GCN","rank_in_archive_order":38,"of":40,"metrics":{"F1":"64.0"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1809.10185","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}