[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82775-en":3,"doc-seo-82775-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82775,137441390410,"Hazel","https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984",8,"Research & Report","Beyond Item Order Temporal Gap Tokenization for Generative Recommendation with Semantic IDs","Semantic-ID based generative recommendation represents each item as compact discrete codes and formulates next-item prediction as code generation, but user histories are often built as static item-ID sequences that omit elapsed time between interactions. This temporal blindness harms modeling of interest continuity and preference drift. The paper introduces ChronoSID, a lightweight temporal augmentation framework injecting time signals into the semantic-ID pipeline via time-aware masked auto-encoding and log-scale gap token discretization. Experiments on Amazon benchmarks show consistent improvements over ReSID, with larger gains for long-gap scenarios.","Beyond Item Order: Temporal Gap Tokenization for Generative Recommendation with Semantic IDs  \nChengkai Huang∗ University of New South Wales, Macquarie University Sydney, Australia [chengkai.huang1@unsw.edu.au](chengkai.huang1@unsw.edu.au)  \nTianqi Gao  \nIndependent Researcher Hangzhou, China [tianqig358@gmail.com](tianqig358@gmail.com)  \nHongtao Huang  \nUniversity of New South Wales Sydney, Australia [frickhuang@foxmail.com](frickhuang@foxmail.com)  \narXiv :2607 .039 18v 1 [ cs .IR] 4 Jul 2026  \nQuan Z. Sheng  \nMacquarie University Sydney, Australia [michael.sheng@mq.edu.au](michael.sheng@mq.edu.au)  \nAbstract  \nSemantic-ID-based generative recommendation has recently emerged as a scalable paradigm for sequential recommendation, where each item is represented by a compact sequence of discrete codes and next-item prediction is formulated as code generation. Existing methods, however, typically construct user histories as sequences of static item identifiers, leaving the elapsed time between consecutive interactions outside the generative input. This temporal blindness is problematic because inter-interaction gaps provide useful cues about interest continuity and preference drift. In this paper, we propose ChronoSID, a lightweight temporal augmentation framework for semantic-ID-based generative recommendation. ChronoSID injects temporal signals into the standard three-stage semantic-ID pipeline from two complementary perspectives. First, we introduce Time-Aware Field-Aware Masked Auto-Encoding (TAFAMAE), which regularizes item representation learning with an auxiliary time-gap prediction objective. Second, we discretize historical interaction intervals into fixed log-scale gap tokens and interleave them with semantic ID tuples as the encoder input of the sequence-to-sequence generator. This design preserves the compact SID generation paradigm while enabling the model to capture timeaware transition patterns. Experiments on Amazon review benchmarks show that ChronoSID consistently improves over ReSID and other competitive generative recommendation baselines. Ablation studies further verify the contribution of both temporal components, and diagnostic analyses show clearer gains under long-gap scenarios where user interests are more likely to drift.  \nCCS Concepts  \n• Information systems → Recommender systems.  \nKeywords  \nRecommender Systems, Generative Recommendation, Temporal Modeling  \n1 Introduction  \nSequential recommendation aims to predict a user’s next interaction from her historical behaviors and has long been a central problem  \n∗ Corresponding author.  \nLina Yao  \nCSIRO’s Data61  \nUniversity of New South Wales Sydney, Australia [lina.yao@unsw.edu.au](lina.yao@unsw.edu.au)  \nSame-category rate (%)  \nFigure 1: Inter-interaction time gaps provide useful cues about interest continuity. Panel (a) shows that the samecategory purchase rate generally decreases as the time gap between consecutive interactions increases. Panel (b) reports the fitted temporal decay coefficient 􀁖 across product domains, suggesting that this decay pattern is broadly observed. Since existing SID-based generative recommenders mainly operate on static item-code sequences, such temporal information is not explicitly represented during sequence construction.  \nin recommender systems [6–8, 11, 34, 35] . Recent advances in generative recommendation reformulate this task as an autoregressive generation problem: instead of retrieving the next item by scoring a large candidate set, the model directly generates an identifier of the target item. To make this paradigm scalable, a growing line of work represents each item as a compact sequence of discrete semantic IDs (SIDs), so that next-item prediction becomes the generation of a short code sequence rather than classification over the entire item vocabulary [4, 13, 22] . Such SID-based recommenders have shown strong potential for large-scale recommendation, as they reduce the output space, enable sequence-to-s","cbCaijNPg04L03vh","https://ap.wps.com/l/cbCaijNPg04L03vh","pdf",1120060,1,12,"English","en",105,"# Introduction\n## Background and motivation\n## Proposed method: ChronoSID\n## Experiments and results\n## Ablation and diagnostic analysis","[{\"question\":\"What problem does the paper address in semantic-ID based generative recommendation?\",\"answer\":\"It addresses the lack of temporal information: interaction histories are typically encoded as static item-ID sequences, leaving elapsed time between interactions outside the generative input.\"},{\"question\":\"How does ChronoSID incorporate temporal signals into the semantic-ID pipeline?\",\"answer\":\"ChronoSID injects temporal signals from two angles: it adds an auxiliary time-gap prediction objective through Time-Aware Field-Aware Masked Auto-Encoding, and it discretizes interaction intervals into fixed log-scale gap tokens interleaved with semantic ID tuples for sequence generation.\"},{\"question\":\"What do the experiments on Amazon review benchmarks show?\",\"answer\":\"ChronoSID consistently improves over ReSID and other generative recommendation baselines, and ablation plus diagnostic analyses confirm that both temporal components contribute, with clearer gains under long-gap scenarios.\"}]",1784182855,30,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"beyond-item-order-temporal-gap-tokenization-for-generative-recommendation-with-semantic-ids","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/beyond-item-order-temporal-gap-tokenization-for-generative-recommendation-with-semantic-ids/82775/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the paper address in semantic-ID based generative recommendation?","Question",{"text":75,"@type":76},"It addresses the lack of temporal information: interaction histories are typically encoded as static item-ID sequences, leaving elapsed time between interactions outside the generative input.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does ChronoSID incorporate temporal signals into the semantic-ID pipeline?",{"text":80,"@type":76},"ChronoSID injects temporal signals from two angles: it adds an auxiliary time-gap prediction objective through Time-Aware Field-Aware Masked Auto-Encoding, and it discretizes interaction intervals into fixed log-scale gap tokens interleaved with semantic ID tuples for sequence generation.",{"name":82,"@type":73,"acceptedAnswer":83},"What do the experiments on Amazon review benchmarks show?",{"text":84,"@type":76},"ChronoSID consistently improves over ReSID and other generative recommendation baselines, and ablation plus diagnostic analyses confirm that both temporal components contribute, with clearer gains under long-gap scenarios.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":28,"slug":121},"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]