AI技术保持用户数据删除后的推荐质量

September 8, 2026 AI technique preserves recommendation quality after user data is deleted by Daegu Gyeongbuk Institute of Science and Technology edited by Swati Mestri, reviewed by Robert Egan Swati Mestri Scientific Editor Meet our editorial team Behind our editorial process Robert Egan Senior Editor Meet our editorial team Behind our editorial process Editors' notes This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: fact-checked trusted source proofread The GIST Add as preferred source Credit: Daegu Gyeongbuk Institute of Science and Technology A research team led by principal researcher Sang-Chul Lee of the Division of Nanotechnology at DGIST has developed AI technology that automatically identifies and compensates for user groups whose recommendation performance deteriorates significantly after user data is deleted. Undergraduate researcher Eugene Jeong participated as the first author in the study, which was one of only 28 papers selected for the ACM KDD Undergraduate Consortium (KDD-UC '26) and was presented in Jeju in August. With the recent increase in demand for personal data protection, "machine unlearning," a technology that removes the influence of a user's data from an already trained AI model when the user requests its deletion, has received attention. Although this is a crucial technology for implementing the "right to be forgotten" in AI systems, it has a limitation: Deleting a particular user's data can degrade the quality of recommendations for other users. The research team focused on the fact that, in graph neural network (GNN)-based recommender systems, some users play a structural role in connecting different user groups and that deleting their data can affect the recommendation performance of surrounding users. Based on this observation, the team proposed "Ada-Comp (Adaptive Compensation)," a compensation technique that automatically identifies the user group most affected after data deletion and selectively provides additional training for that group. This study is significant because it expanded the perspective of machine unlearning from the conventional focus on "how effectively the information of users targeted for deletion has been removed" to considering "how the process affects other users." The team proposed a new approach that faithfully reflects the deletion of personal data while preventing unexpected performance degradation from being concentrated among specific user groups. In particular, this achievement is significant because undergraduate-led research was developed into an international research outcome through interdisciplinary research between DGIST's undergraduate and research divisions. Jeong served as the first author, while Lee participated as the corresponding author, thereby linking undergraduate research with the experience of professional researchers. "For trustworthy AI, it is important not only to delete personal data safely but also to ensure that the quality of service for other users does not deteriorate in the process," Lee said. "We expect this study to present a new direction for reducing performance disparities among user groups that may arise after machine unlearning." Key concepts Trustworthy machine learningMachine learning methodologies Provided by Daegu Gyeongbuk Institute of Science and Technology Who's behind this story? Swati Mestri Swati Mestri holds a bachelor's degree in Electronics Engineering and has worked as a content editor since 2019. She has experience editing research documents across technology, health care, and materials science, and has a particular interest in technology and space. Full profile → Robert Egan Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile → Citation: AI technique preserves recommendation quality after user data is deleted (2026, September 8) retrieved 8 September 2026 from https://techxplore.com/news/2026-09-ai-technique-quality-user-deleted.html This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only. New method enables AI models to forget private and copyrighted data Aug 29, 2025 Learning to forget—a weapon in the arsenal against harmful AI Nov 2, 2023 Researchers develop key technology to make personalized AI safer Jul 15, 2026 Extracting personal information from anonymous cell phone data using machine learning Oct 12, 2022 Black-box forgetting: A new method for tailoring large AI models Dec 9, 2024 AI gets a private tutor for learning human preferences more accurately Dec 17, 2025 Memristor chip breaks the capacity limit of brain-inspired associative memory 5 hours ago Invention may bring smart bionic eye into sight 5 hours ago Whole-body expansion and contraction make a robot seem more alive 6 hours ago A tiny crystal tilt may bring greener magnetic memory closer to everyday devices 7 hours ago Tracking energy flow reveals voltage pattern that slashes waste heat by over 60% 8 hours ago Editorial proposes limits on generative AI's role in scientific research 13 hours ago