{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,11]],"date-time":"2026-07-11T02:49:01Z","timestamp":1783738141608,"version":"3.55.0"},"reference-count":16,"publisher":"Association for Computing Machinery (ACM)","issue":"12","content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["Proc. VLDB Endow."],"published-print":{"date-parts":[[2024,8]]},"abstract":"<jats:p>The \"right-to-be-forgotten\" requires the removal of personal data from trained machine learning (ML) models with machine unlearning. Conducting such unlearning with low latency is crucial for responsible data management. Low-latency unlearning is challenging, but possible for certain classes of ML models when treating them as \"materialised views\" over training data, with carefully chosen operations and data structures for computing updates.<\/jats:p>\n          <jats:p>\n            We present Snapcase, a recommender system that can unlearn user interactions with sub-second latency on a large grocery shopping dataset with 33 million purchases and 200 thousand users. Its implementation is based on incremental view maintenance with Differential Dataflow and a custom algorithm and data structure for maintaining a top-\n            <jats:italic>k<\/jats:italic>\n            aggregation over the result of a sparse matrix-matrix multiplication. We demonstrate how interactive low-latency unlearning empowers users in critical scenarios to get rid of sensitive items in their recommendations and to drastically reduce their data's negative influence on other users' predictions.\n          <\/jats:p>","DOI":"10.14778\/3685800.3685853","type":"journal-article","created":{"date-parts":[[2024,11,8]],"date-time":"2024-11-08T17:25:21Z","timestamp":1731086721000},"page":"4273-4276","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":3,"title":["Snarcase - Regain Control over Your Predictions with Low-Latency Machine Unlearning"],"prefix":"10.14778","volume":"17","author":[{"given":"Sebastian","family":"Schelter","sequence":"first","affiliation":[{"name":"BIFOLD &amp; TU Berlin"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Stefan","family":"Grafberger","sequence":"additional","affiliation":[{"name":"BIFOLD &amp; TU Berlin"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maarten","family":"de Rijke","sequence":"additional","affiliation":[{"name":"University of Amsterdam"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2024,11,8]]},"reference":[{"key":"e_1_2_1_1_1","volume-title":"Van Land tot Klant: Onze Ketens. Retrieved","author":"Heijn Albert","year":"2024","unstructured":"Albert Heijn. 2023. Van Land tot Klant: Onze Ketens. Retrieved July 5, 2024 from https:\/\/www.ah.nl\/over-ah\/duurzaamheid\/onze-ketens"},{"key":"e_1_2_1_2_1","unstructured":"Maurizio Ferrari et al. 2019. Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approaches. RecSys (2019)."},{"key":"e_1_2_1_3_1","volume-title":"Inside the Secret World of Trader Joe's. Retrieved","year":"2024","unstructured":"Fortune. 2010. Inside the Secret World of Trader Joe's. Retrieved July 5, 2024 from https:\/\/fortune.com\/2010\/08\/23\/inside-the-secret-world-of-trader-joes\/"},{"key":"e_1_2_1_4_1","unstructured":"Haoji Hu et al. 2020. Modeling Personalized Item Frequency Information for Next-Basket Recommendation. SIGIR (2020)."},{"key":"e_1_2_1_5_1","doi-asserted-by":"publisher","DOI":"10.1145\/3587153"},{"key":"e_1_2_1_6_1","unstructured":"Frank McSherry et al. 2013. Differential Dataflow. CIDR (2013)."},{"key":"e_1_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.14778\/3401960.3401974"},{"key":"e_1_2_1_8_1","volume-title":"\u00dcber das Unternehmen - Struktur und Vertriebslinien. Retrieved","author":"REWE.","year":"2024","unstructured":"REWE. 2024. \u00dcber das Unternehmen - Struktur und Vertriebslinien. Retrieved July 5, 2024 from https:\/\/www.rewe-group.com\/de\/unternehmen\/struktur-und-vertriebslinien\/rewe\/"},{"key":"e_1_2_1_9_1","volume-title":"Amnesia - A Selection of Machine Learning Models That Can Forget User Data Very Fast. CIDR","author":"Schelter Sebastian","year":"2020","unstructured":"Sebastian Schelter. 2020. Amnesia - A Selection of Machine Learning Models That Can Forget User Data Very Fast. CIDR (2020)."},{"key":"e_1_2_1_10_1","volume-title":"Workshop on Challenges in Deploying and Monitoring ML Systems@ICML","author":"Schelter Sebastian","year":"2021","unstructured":"Sebastian Schelter. 2021. Towards Efficient Machine Unlearning via Incremental View Maintenance. Workshop on Challenges in Deploying and Monitoring ML Systems@ICML (2021)."},{"key":"e_1_2_1_11_1","volume-title":"Hedgecut: Maintaining Randomised Trees for Low-latency Machine Unlearning. SIGMOD","author":"Sebastian Schelter","year":"2021","unstructured":"Sebastian Schelter et al. 2021. Hedgecut: Maintaining Randomised Trees for Low-latency Machine Unlearning. SIGMOD (2021)."},{"key":"e_1_2_1_12_1","doi-asserted-by":"crossref","unstructured":"Sebastian Schelter et al. 2023. Forget Me Now: Fast and Exact Unlearning in Neighborhood-based Recommendation. SIGIR (2023).","DOI":"10.1145\/3539618.3591989"},{"key":"e_1_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.1145\/3488717"},{"key":"e_1_2_1_14_1","unstructured":"Maria Vechtomova et al. 2023. Databricks AI Summit: Streamlining API Deploy ML Models Across Multiple Brands: Ahold Delhaize's Experience on Serverless. Retrieved July 5 2024 from https:\/\/www.youtube.com\/watch?v=GSJFyoBiCXk"},{"key":"e_1_2_1_15_1","unstructured":"Vincent Warmerdam. 2021. Beyond Broken. Retrieved July 5 2024 from https:\/\/koaning.io\/posts\/beyond-broken\/"},{"key":"e_1_2_1_16_1","volume-title":"Deltaboost: Gradient Boosting Decision Trees with Efficient Machine Unlearning. SIGMOD","author":"Zhaomin Wu","year":"2023","unstructured":"Zhaomin Wu et al. 2023. Deltaboost: Gradient Boosting Decision Trees with Efficient Machine Unlearning. SIGMOD (2023)."}],"container-title":["Proceedings of the VLDB Endowment"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.14778\/3685800.3685853","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,12,31]],"date-time":"2024-12-31T05:28:08Z","timestamp":1735622888000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.14778\/3685800.3685853"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,8]]},"references-count":16,"journal-issue":{"issue":"12","published-print":{"date-parts":[[2024,8]]}},"alternative-id":["10.14778\/3685800.3685853"],"URL":"https:\/\/doi.org\/10.14778\/3685800.3685853","relation":{},"ISSN":["2150-8097"],"issn-type":[{"value":"2150-8097","type":"print"}],"subject":[],"published":{"date-parts":[[2024,8]]},"assertion":[{"value":"2024-11-08","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}