{"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/counterfactual-learning-from-human","title":"Counterfactual Learning from Human Proofreading Feedback for Semantic Parsing","arxiv_id":"1811.12239","date":"2018-11-29","proceeding":null,"authors":["Carolin Lawrence","Stefan Riezler"],"abstract":"In semantic parsing for question-answering, it is often too expensive to\ncollect gold parses or even gold answers as supervision signals. We propose to\nconvert model outputs into a set of human-understandable statements which allow\nnon-expert users to act as proofreaders, providing error markings as learning\nsignals to the parser. Because model outputs were suggested by a historic\nsystem, we operate in a counterfactual, or off-policy, learning setup. We\nintroduce new estimators which can effectively leverage the given feedback and\nwhich avoid known degeneracies in counterfactual learning, while still being\napplicable to stochastic gradient optimization for neural semantic parsing.\nFurthermore, we discuss how our feedback collection method can be seamlessly\nintegrated into deployed virtual personal assistants that embed a semantic\nparser. Our work is the first to show that semantic parsers can be improved\nsignificantly by counterfactual learning from logged human feedback data.","url_abs":"http://arxiv.org/abs/1811.12239v1","url_pdf":"http://arxiv.org/pdf/1811.12239v1.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":"counterfactual-learning-from-human","repo_url":"https://github.com/carolinlawrence/nematus","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"BSD-3-Clause"}}],"tasks":[{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"semantic-parsing","task_name":"Semantic Parsing"},{"task_slug":null,"task_name":"counterfactual"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}