{"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/usr-an-unsupervised-and-reference-free","title":"USR: An Unsupervised and Reference Free Evaluation Metric for Dialog Generation","arxiv_id":"2005.00456","date":"2020-05-01","proceeding":"ACL 2020 6","authors":["Shikib Mehri","Maxine Eskenazi"],"abstract":"The lack of meaningful automatic evaluation metrics for dialog has impeded open-domain dialog research. Standard language generation metrics have been shown to be ineffective for evaluating dialog models. To this end, this paper presents USR, an UnSupervised and Reference-free evaluation metric for dialog. USR is a reference-free metric that trains unsupervised models to measure several desirable qualities of dialog. USR is shown to strongly correlate with human judgment on both Topical-Chat (turn-level: 0.42, system-level: 1.0) and PersonaChat (turn-level: 0.48 and system-level: 1.0). USR additionally produces interpretable measures for several desirable properties of dialog.","url_abs":"https://arxiv.org/abs/2005.00456v1","url_pdf":"https://arxiv.org/pdf/2005.00456v1.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":"usr-an-unsupervised-and-reference-free","repo_url":"https://github.com/shikib/usr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"dialogue-evaluation","task_name":"Dialogue Evaluation"},{"task_slug":"open-domain-dialog","task_name":"Open-Domain Dialog"},{"task_slug":"text-generation","task_name":"Text Generation"}],"methods":[],"datasets_introduced":[{"slug":"usr-personachat","name":"USR-PersonaChat","full_name":""},{"slug":"usr-topicalchat","name":"USR-TopicalChat","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/dialogue-evaluation-on-usr-personachat","task":"Dialogue Evaluation","dataset":"USR-PersonaChat","model":"USR - DR (x = c)","rank_in_archive_order":2,"of":5,"metrics":{"Pearson Correlation":"0.6087","Spearman Correlation":"0.4814"},"uses_additional_data":false},{"leaderboard":"/sota/dialogue-evaluation-on-usr-personachat","task":"Dialogue Evaluation","dataset":"USR-PersonaChat","model":"USR","rank_in_archive_order":3,"of":5,"metrics":{"Pearson Correlation":"0.4115","Spearman Correlation":"0.4693"},"uses_additional_data":false},{"leaderboard":"/sota/dialogue-evaluation-on-usr-personachat","task":"Dialogue Evaluation","dataset":"USR-PersonaChat","model":"USR - MLM","rank_in_archive_order":4,"of":5,"metrics":{"Pearson Correlation":"0.0788","Spearman Correlation":"0.0795"},"uses_additional_data":false},{"leaderboard":"/sota/dialogue-evaluation-on-usr-personachat","task":"Dialogue Evaluation","dataset":"USR-PersonaChat","model":"USR - DR (x = f)","rank_in_archive_order":5,"of":5,"metrics":{"Pearson Correlation":"-0.0454","Spearman Correlation":"-0.0495"},"uses_additional_data":false},{"leaderboard":"/sota/dialogue-evaluation-on-usr-topicalchat","task":"Dialogue Evaluation","dataset":"USR-TopicalChat","model":"USR","rank_in_archive_order":3,"of":6,"metrics":{"Pearson Correlation":"0.4220","Spearman Correlation":"0.4192"},"uses_additional_data":false},{"leaderboard":"/sota/dialogue-evaluation-on-usr-topicalchat","task":"Dialogue Evaluation","dataset":"USR-TopicalChat","model":"USR - DR (x = c)","rank_in_archive_order":4,"of":6,"metrics":{"Pearson Correlation":"0.4068","Spearman Correlation":"0.3245"},"uses_additional_data":false},{"leaderboard":"/sota/dialogue-evaluation-on-usr-topicalchat","task":"Dialogue Evaluation","dataset":"USR-TopicalChat","model":"USR - MLM","rank_in_archive_order":5,"of":6,"metrics":{"Pearson Correlation":"0.3345","Spearman Correlation":"0.3086"},"uses_additional_data":false},{"leaderboard":"/sota/dialogue-evaluation-on-usr-topicalchat","task":"Dialogue Evaluation","dataset":"USR-TopicalChat","model":"USR - DR (x = f)","rank_in_archive_order":6,"of":6,"metrics":{"Pearson Correlation":"0.3221","Spearman Correlation":"0.1419"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2005.00456","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}