[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-273699-105":59,"doc-detail-273699-en":135},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":128,"head_meta":130,"extra_data":132,"updated_unix":134},105,"en","beyond-raw-transcripts-structured-persona-extraction-for-llm-based-digital-twins-preprint","Beyond Raw Transcripts - Structured Persona Extraction for LLM-Based Digital Twins - Preprint","","LLM-based digital twins simulate individual behavior in new environments using a representation built from prior responses. Existing methods rely on survey transcripts or their LLM-generated summaries and score accuracy by predicting holdout responses, yet compression alone does not meaningfully reduce performance. This work argues the core limitation is structural: persona information organization before it reaches the simulator. It introduces a hand-crafted BDE schema that improves accuracy on a homogeneous benchmark, but not on heterogeneous tasks, motivating an automatic structure-discovery pipeline that restores accuracy on diverse sub-studies.",{"@graph":69,"@context":127},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/beyond-raw-transcripts-structured-persona-extraction-for-llm-based-digital-twins-preprint/273699/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/beyond-raw-transcripts-structured-persona-extraction-for-llm-based-digital-twins-preprint/273699.png","ImageObject",300,407,{"name":92,"@type":93},"Skyler","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-06","2026-09-15",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",17,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119,123],{"name":110,"@type":111,"acceptedAnswer":112},"Why do summary-compressed transcripts not significantly reduce digital-twin accuracy?","Question",{"text":113,"@type":114},"Because the information volume is not the primary bottleneck; the work shows performance is more sensitive to how persona information is organized than to how much information is provided.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"What is the proposed BDE structure and how does it help?",{"text":118,"@type":114},"BDE (Background, Decision procedure, Evaluation) structures persona into sub-profiles for identity, reasoning, and preference, and maps transcript evidence into each part via an allocation rule, improving accuracy on a homogeneous benchmark.",{"name":120,"@type":111,"acceptedAnswer":121},"Why doesn’t a fixed structure generalize to heterogeneous tasks?",{"text":122,"@type":114},"The fixed schema yields gains on a homogeneous task family but becomes statistically indistinguishable from the raw transcript baseline on more heterogeneous tasks.",{"name":124,"@type":111,"acceptedAnswer":125},"How does the automatic structure-discovery pipeline address the limitation?",{"text":126,"@type":114},"It iteratively proposes and refines task-specific persona structures and extraction prompts with an LLM, restoring performance across 13 diverse sub-studies and preventing significant losses seen with the fixed schema.","https://schema.org",{"og:url":83,"og:type":129,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":131,"canonical":83},"index,follow",{"doc_id":133,"site_id":62},273699,1790375899,{"code":4,"msg":5,"data":136},{"doc_id":133,"user_id":137,"nickname":92,"user_avatar":138,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":139,"file_id":140,"file_url":141,"file_type":142,"file_size":143,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":144,"language":145,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":146,"faqs":147,"seo_title":148,"seo_description":67,"update_tm":149,"read_time":150,"preview_page_count":4},2336464648746,"https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c","Beyond Raw Transcripts: Structured Persona Extraction for LLM-Based Digital Twins  \narXiv : 2608 .20344v 1 [cs .CL] 13 Jun 2026  \nIris Ye∗  \nBooth School of Business University of Chicago [iris.ye@chicagobooth.edu](iris.ye@chicagobooth.edu)  \nTianze Deng∗  \nUniversity of Chicago [tianze1@uchicago.edu](tianze1@uchicago.edu)  \nOzan Candogan†  \nBooth School of Business  \nUniversity of Chicago  \n[Ozan.Candogan@chicagobooth.edu](Ozan.Candogan@chicagobooth.edu)  \nAbstract  \nLLM-based \"digital twins\" aim to simulate how an individual would behave in new environments or respond to novel questions, given some representation of that individual’s prior responses. A common approach constructs this representation from survey transcripts or summaries derived from them, and evaluates performance by predicting holdout responses. Prior work shows that compressing long transcripts into shorter LLM-generated summaries does not significantly reduce predictive accuracy, suggesting that information volume is not the primary bottleneck.  \nIn this work, we argue that the key limitation is instead structural: how persona information is organized before being provided to the simulator model. We study this by comparing unstructured summaries with structured persona representations. First, we introduce a hand-crafted schema (BDE: Background, Decision procedure, Evaluation), grounded in consumer-behavior theory, and show that it improves predictive accuracy over raw transcripts by +1 .91 percentage points on a homogeneous benchmark (Twin-2K-500), with similar gains on gpt-5 .4-mini and Qwen3-8Bas robustness checks. However, this fixed structure does not generalize across more heterogeneous tasks, where performance is statistically indistinguishable from the raw transcript baseline.  \nTo address this limitation, we propose an automatic structure-discovery pipeline in which an LLM iteratively proposes and refines task-specific persona structures and extraction prompts. On a benchmark of 13 diverse sub-studies, this approach restores performance, improving mean accuracy by +1 .91 percentage points over the raw transcript baseline and eliminating significant losses observed with the fixed schema.  \nOverall, our results suggest that the main constraint in LLM-based digital twins is not how much information is provided, but how it is structured—and that the optimal structure depends on the task.  \n∗ Equal contribution.  \n†Corresponding author.  \nPreprint.  \nFigure 1: Pipeline: from raw transcript to accuracy via the structured persona.  \n1 Introduction  \n1.1 The persona-structure gap  \nLLM-based digital twins aim to simulate how an individual would behave or respond in new settings, given a representation of their prior data. A common instantiation elicits a long survey from each respondent, passes the resulting transcript to a language model, and asks it to predict holdout responses as that respondent would; this is the setting we study. The Twin-2K-500 dataset Toubia et al. [2025b] pairs 500 input questions with 88 holdout questions across 17 distinct prediction tasks for over two thousand respondents, offering substantially broader per-individual coverage than prior digital-twin benchmarks. A follow-up Mega-Study Toubia et al. [2025a]—19 pre-registered sub-studies spanning diverse decision domains—reports that a 13K-character LLM-generated persona summary performs comparably to the 128K-character raw transcript(same as 500 input questions in twins-2k-500 Toubia et al. [2025b]) on prediction accuracy. This robustness to compression implies that information volume is not the binding constraint on digital-twin accuracy. What remains open is how that information should be organized before it reaches the simulator.  \nIn this paper, we focus on this question and explore how to structure the persona information to improve the accuracy of persona based simulators. In our approach, for each respondent, an extractor LLM takes the raw transcript and an extraction promp","cbCaioqqaplvBjwo","https://ap.wps.com/l/cbCaioqqaplvBjwo","pdf",619964,22,"English","# Abstract\n# 1 Introduction\n## 1.1 The persona-structure gap\n## 1.2 Contributions","[{\"question\":\"Why do summary-compressed transcripts not significantly reduce digital-twin accuracy?\",\"answer\":\"Because the information volume is not the primary bottleneck; the work shows performance is more sensitive to how persona information is organized than to how much information is provided.\"},{\"question\":\"What is the proposed BDE structure and how does it help?\",\"answer\":\"BDE (Background, Decision procedure, Evaluation) structures persona into sub-profiles for identity, reasoning, and preference, and maps transcript evidence into each part via an allocation rule, improving accuracy on a homogeneous benchmark.\"},{\"question\":\"Why doesn’t a fixed structure generalize to heterogeneous tasks?\",\"answer\":\"The fixed schema yields gains on a homogeneous task family but becomes statistically indistinguishable from the raw transcript baseline on more heterogeneous tasks.\"},{\"question\":\"How does the automatic structure-discovery pipeline address the limitation?\",\"answer\":\"It iteratively proposes and refines task-specific persona structures and extraction prompts with an LLM, restoring performance across 13 diverse sub-studies and preventing significant losses seen with the fixed schema.\"}]","Beyond Raw Transcripts - Structured Persona Extraction for LLM-Based Digital Twins - Preprint | PDF",1789472551,55]