{"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/factspotter-evaluating-the-factual","title":"FactSpotter: Evaluating the Factual Faithfulness of Graph-to-Text Generation","arxiv_id":null,"date":"2023-10-25","proceeding":"Findings of EMNLP 2023 10","authors":["Kun Zhang","Oana Balalau","Ioana Manolescu"],"abstract":"Graph-to-text (G2T) generation takes a graph as input and aims to generate a fluent and faith- ful textual representation of the information in the graph. The task has many applications, such as dialogue generation and question an- swering. In this work, we investigate to what extent the G2T generation problem is solved for previously studied datasets, and how pro- posed metrics perform when comparing generated texts. To help address their limitations, we propose a new metric that correctly identifies factual faithfulness, i.e., given a triple (subject, predicate, object), it decides if the triple is present in a generated text. We show that our metric FactSpotter achieves the highest correlation with human annotations on data correct- ness, data coverage, and relevance. In addition, FactSpotter can be used as a plug-in feature to improve the factual faithfulness of existing models. Finally, we investigate if existing G2T datasets are still challenging for state-of-the-art models. Our code is available online: https://github.com/guihuzhang/FactSpotter.","url_abs":"https://dumas.ccsd.cnrs.fr/INRIA-SACLAY/hal-04257838v1","url_pdf":"https://aclanthology.org/2023.findings-emnlp.672.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":"factspotter-evaluating-the-factual","repo_url":"https://github.com/guihuzhang/FactSpotter","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"data-to-text-generation","task_name":"Data-to-Text Generation"},{"task_slug":"dialogue-generation","task_name":"Dialogue Generation"},{"task_slug":"kg-to-text","task_name":"KG-to-Text Generation"},{"task_slug":"question-generation","task_name":"Question Generation"},{"task_slug":"table-to-text-generation","task_name":"Table-to-Text Generation"},{"task_slug":"text-generation","task_name":"Text Generation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/data-to-text-generation-on-dart","task":"Data-to-Text Generation","dataset":"DART","model":"T5B Baseline","rank_in_archive_order":1,"of":5,"metrics":{"BLEU":"48.47","FactSpotter":"96.65","METEOR":"40.74"},"uses_additional_data":false},{"leaderboard":"/sota/data-to-text-generation-on-dart","task":"Data-to-Text Generation","dataset":"DART","model":"FactT5B","rank_in_archive_order":2,"of":5,"metrics":{"BLEU":"48.37","FactSpotter":"97.60","METEOR":"40.72"},"uses_additional_data":false},{"leaderboard":"/sota/data-to-text-generation-on-dart","task":"Data-to-Text Generation","dataset":"DART","model":"JointGT Baseline","rank_in_archive_order":4,"of":5,"metrics":{"BLEU":"47.51","FactSpotter":"95.86","METEOR":"40.43"},"uses_additional_data":false},{"leaderboard":"/sota/data-to-text-generation-on-dart","task":"Data-to-Text Generation","dataset":"DART","model":"FactJointGT","rank_in_archive_order":5,"of":5,"metrics":{"BLEU":"47.39","FactSpotter":"97.25","METEOR":"40.32"},"uses_additional_data":false},{"leaderboard":"/sota/data-to-text-generation-on-webnlg","task":"Data-to-Text Generation","dataset":"WebNLG","model":"JointGT Baseline","rank_in_archive_order":3,"of":20,"metrics":{"BLEU":"67.08","FactSpotter":"99.09","METEOR":"48.34"},"uses_additional_data":false},{"leaderboard":"/sota/data-to-text-generation-on-webnlg","task":"Data-to-Text Generation","dataset":"WebNLG","model":"T5B Baseline","rank_in_archive_order":4,"of":20,"metrics":{"BLEU":"67.04","FactSpotter":"99.44","METEOR":"48.35"},"uses_additional_data":false},{"leaderboard":"/sota/data-to-text-generation-on-webnlg","task":"Data-to-Text Generation","dataset":"WebNLG","model":"FactT5B","rank_in_archive_order":5,"of":20,"metrics":{"BLEU":"67.04","FactSpotter":"99.71","METEOR":"48.22"},"uses_additional_data":false},{"leaderboard":"/sota/data-to-text-generation-on-webnlg","task":"Data-to-Text Generation","dataset":"WebNLG","model":"FactJointGT","rank_in_archive_order":6,"of":20,"metrics":{"BLEU":"66.89","FactSpotter":"99.67","METEOR":"48.19"},"uses_additional_data":false},{"leaderboard":"/sota/kg-to-text-generation-on-webnlg-2-0-1","task":"KG-to-Text Generation","dataset":"WebNLG 2.0 (Constrained)","model":"FactT5B","rank_in_archive_order":1,"of":9,"metrics":{"BLEU":"67.08","FactSpotter":"99.71","METEOR":"48.34"},"uses_additional_data":false},{"leaderboard":"/sota/kg-to-text-generation-on-webnlg-2-0-1","task":"KG-to-Text Generation","dataset":"WebNLG 2.0 (Constrained)","model":"JointGT Baseline","rank_in_archive_order":2,"of":9,"metrics":{"BLEU":"67.08","FactSpotter":"99.44","METEOR":"48.34"},"uses_additional_data":false},{"leaderboard":"/sota/kg-to-text-generation-on-webnlg-2-0-1","task":"KG-to-Text Generation","dataset":"WebNLG 2.0 (Constrained)","model":"T5B Baseline","rank_in_archive_order":3,"of":9,"metrics":{"BLEU":"67.04","FactSpotter":"99.05","METEOR":"48.35"},"uses_additional_data":false},{"leaderboard":"/sota/kg-to-text-generation-on-webnlg-2-0-1","task":"KG-to-Text Generation","dataset":"WebNLG 2.0 (Constrained)","model":"FactJointGT","rank_in_archive_order":4,"of":9,"metrics":{"BLEU":"66.89","FactSpotter":"99.67","METEOR":"48.19"},"uses_additional_data":false},{"leaderboard":"/sota/question-generation-on-grailqa-compositional","task":"Question Generation","dataset":"GrailQA-Compositional","model":"FactJointGT","rank_in_archive_order":1,"of":4,"metrics":{"BLEU":"31.25","FactSpotter":"97.10","METEOR":"36.21"},"uses_additional_data":false},{"leaderboard":"/sota/question-generation-on-grailqa-compositional","task":"Question Generation","dataset":"GrailQA-Compositional","model":"FactT5B","rank_in_archive_order":3,"of":4,"metrics":{"BLEU":"31.66","FactSpotter":"96.53","METEOR":"35.72"},"uses_additional_data":false},{"leaderboard":"/sota/question-generation-on-grailqa-iid","task":"Question Generation","dataset":"GrailQA-IID","model":"FactT5B","rank_in_archive_order":1,"of":4,"metrics":{"BLEU":"46.10","FactSpotter":"99.50","METEOR":"42.67"},"uses_additional_data":false},{"leaderboard":"/sota/question-generation-on-grailqa-iid","task":"Question Generation","dataset":"GrailQA-IID","model":"FactJointGT","rank_in_archive_order":4,"of":4,"metrics":{"BLEU":"43.61","FactSpotter":"99.12","METEOR":"41.65"},"uses_additional_data":false},{"leaderboard":"/sota/question-generation-on-grailqa-zero-shot","task":"Question Generation","dataset":"GrailQA-Zero-Shot","model":"FactJointGT","rank_in_archive_order":2,"of":4,"metrics":{"FactSpotter":"94.95","METEOR":"37.55","bleu":"32.46"},"uses_additional_data":false},{"leaderboard":"/sota/question-generation-on-grailqa-zero-shot","task":"Question Generation","dataset":"GrailQA-Zero-Shot","model":"FactT5B","rank_in_archive_order":3,"of":4,"metrics":{"FactSpotter":"95.61","METEOR":"37.39","bleu":"32.39"},"uses_additional_data":false},{"leaderboard":"/sota/table-to-text-generation-on-dart","task":"Table-to-Text Generation","dataset":"DART","model":"T5B Baseline","rank_in_archive_order":1,"of":6,"metrics":{"BERT":"0.9505","BLEU":"48.47","BLEURT":"0.6749","FactSpotter":"96.65","METEOR":"0.4074"},"uses_additional_data":false},{"leaderboard":"/sota/table-to-text-generation-on-dart","task":"Table-to-Text Generation","dataset":"DART","model":"FactT5B","rank_in_archive_order":2,"of":6,"metrics":{"BERT":"0.9505","BLEU":"48.37","BLEURT":"0.6743","FactSpotter":"97.60","METEOR":"0.4072"},"uses_additional_data":false},{"leaderboard":"/sota/table-to-text-generation-on-dart","task":"Table-to-Text Generation","dataset":"DART","model":"JointGT Baseline","rank_in_archive_order":3,"of":6,"metrics":{"BERT":"0.9492","BLEU":"47.51","BLEURT":"0.6733","FactSpotter":"95.86","METEOR":"0.4043"},"uses_additional_data":false},{"leaderboard":"/sota/table-to-text-generation-on-dart","task":"Table-to-Text Generation","dataset":"DART","model":"FactJointGT","rank_in_archive_order":4,"of":6,"metrics":{"BERT":"0.9492","BLEU":"47.39","BLEURT":"0.6726","FactSpotter":"97.25","METEOR":"0.4032"},"uses_additional_data":false},{"leaderboard":"/sota/text-generation-on-dart","task":"Text Generation","dataset":"DART","model":"T5B Baseline","rank_in_archive_order":1,"of":7,"metrics":{"BLEU":"48.74","FactSpotter":"96.65","METEOR":"0.4074"},"uses_additional_data":false},{"leaderboard":"/sota/text-generation-on-dart","task":"Text Generation","dataset":"DART","model":"FactT5B","rank_in_archive_order":2,"of":7,"metrics":{"BLEU":"48.37","FactSpotter":"97.60","METEOR":"0.4072"},"uses_additional_data":false},{"leaderboard":"/sota/text-generation-on-dart","task":"Text Generation","dataset":"DART","model":"JointGT Baseline","rank_in_archive_order":3,"of":7,"metrics":{"BLEU":"47.51","FactSpotter":"0.9586","METEOR":"0.4043"},"uses_additional_data":false},{"leaderboard":"/sota/text-generation-on-dart","task":"Text Generation","dataset":"DART","model":"FactJointGT","rank_in_archive_order":4,"of":7,"metrics":{"BLEU":"47.39","FactSpotter":"97.25","METEOR":"0.4032"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}