[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-a-diffusion-model-that-separates-old-habits-from-new-whims":10,"sections":35},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},6710,"a-diffusion-model-that-separates-old-habits-from-new-whims","A Diffusion Model That Separates Old Habits From New Whims","Researchers split diffusion-based recommendation models by long-term and short-term preference, beating rivals by up to 29% on accuracy tests.","A new recommendation framework treats your year-long habits and this morning's impulse click as separate signals instead of one blur.\n\nResearchers built TDPM (Time-aware Diffusion based on Preference disentanglement), a generative recommender that swaps traditional item IDs for semantic indices and runs a diffusion model, the same generative approach behind image tools like Stable Diffusion, adapted to guess what a user wants next. Earlier diffusion-based recommenders treated every past click as equally relevant no matter when it happened. TDPM instead splits user preference into two tracks: a \"period preference\" for slow, steady habits, and a \"point preference\" tied to recent, one-off triggers, then times its diffusion process to weigh the two differently. Tested on three public datasets, TDPM beat existing state-of-the-art baselines by up to 29.21% on a top-20 hit-rate metric and 25.45% on a ranking-quality metric, and ablation tests confirmed the time-aware split itself, not just the diffusion backbone, drove those gains.\n\nRecommendation engines have long lumped a user's years of viewing history in with what they clicked an hour ago, which is part of why feeds can feel stuck in a rut or wildly overreactive to one stray click. Separating slow-burn taste from short-term spikes offers a plausible fix for both problems at once, and the size of the reported gains suggests the time dimension has been underused in generative recommenders specifically.\n\nIt is still an arXiv preprint tested on three benchmark datasets, not a shipped product, and production feeds already use cruder recency weighting, so the real question is whether this disentanglement holds up outside curated benchmarks.","[\"generative-recommenders\",\"diffusion-models\",\"recommendation-systems\",\"ai-research\"]","2026-09-17T04:00:00.000Z","2026-09-18T07:12:48.208Z","2026-09-18T07:13:00.132Z","published",null,[],"ai",[26,27,28,29],"generative-recommenders","diffusion-models","recommendation-systems","ai-research",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.01670",0,{"sections":36},[37,41,45,50,55,59,63,68,73,77,82,87,92,97],{"name":38,"slug":24,"count":39,"latest_published_at":40},"AI",3852,"2026-09-17T08:27:09.000Z",{"name":42,"slug":43,"count":44,"latest_published_at":18},"Security","security",648,{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",338,"2026-09-11T04:00:00.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":18},"Hardware","hardware",154,{"name":60,"slug":61,"count":62,"latest_published_at":18},"Science","science",114,{"name":64,"slug":65,"count":66,"latest_published_at":67},"Consumer Tech","consumer-tech",99,"2026-09-09T17:27:33.000Z",{"name":69,"slug":70,"count":71,"latest_published_at":72},"Software","software",75,"2026-09-10T20:41:21.000Z",{"name":74,"slug":75,"count":76,"latest_published_at":18},"Dev Tools","dev-tools",73,{"name":78,"slug":79,"count":80,"latest_published_at":81},"Startups","startups",55,"2026-09-09T23:14:29.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":86},"Gaming","gaming",43,"2026-09-10T12:18:06.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"General","general",41,"2026-09-08T01:57:23.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]