[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-125313-en":3,"doc-seo-125313-105":30,"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":4,"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":27,"seo_description":14,"update_tm":28,"read_time":29},125313,1374391974468,"Eden","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","Privacy-Preserving Machine Learning on Web Browsing for Public Opinion","A secure multiparty computation system is deployed in a real-world setting to predict political preferences from private web browsing data while keeping user information confidential. Secure, secret-shared inputs from nearly 8,000 users across Aug 2024 to Feb 2025 are processed with a custom Chrome extension and CrypTen MPC. The work provides an MPC implementation for learning from label proportions, using public polling and election results as ground truth. Results support candidate preference estimation with only aggregate outputs revealed.","Privacy-Preserving Machine Learning on Web Browsing for Public Opinion  \nSam Buxbaum 1 , Lucas M. Tassis 1 , Lucas Boschelli2 , Giovanni Comarela3 , Mayank Varia 1 , Mark Crovella 1 , and Dino P. Christenson4  \n1 Boston University {sambux,ltassis,varia,[crovella}bu.edu](crovella}bu.edu)  \n2 Maritz [boschellil@wustl.edu](boschellil@wustl.edu)  \n3 Universidade Federal do Espirito Santo [gc@inf.ufes.br](gc@inf.ufes.br)  \n4 Washington University [dinopc@wustl.edu](dinopc@wustl.edu)  \nAbstract. We present a real-world deployment of secure multiparty computation to predict political preference from private web browsing data. To estimate aggregate preferences for the 2024 U.S. presidential election candidates, we collect and analyze secret-shared data from nearly 8000 users from August 2024 through February 2025, with over 2000 daily active users sustained throughout the bulk of the survey. The use of MPC allows us to compute over sensitive web browsing data that users would otherwise be more hesitant to provide. We collect data using a custom-built Chrome browser extension and perform our analysis using the CrypTen MPC library. To our knowledge, we provide the first implementation under MPC of a model for the learning from label proportions (LLP) problem in machine learning, which allows us to train on unlabeled web browsing data using publicly available polling and election results as the ground truth. The client code is open source,5 and the remaining code will be open source in the future.  \n1 Introduction  \nSecure multi-party computation (MPC) is a cryptographic protocol that allows several people to contribute their data toward a collective data analysis without ever exposing their personal data to any other party. MPC has been a topic of research for decades [10, 29, 62, 71], and through a variety of algorithmic improvements and software implementations (e.g., [13, 39, 47, 51, 63]), it has seen deployments over the past decade in the commercial and public sectors (e.g., [1,6,11,14,24,44,58]) . Recently, there has been a focus on specialized MPC algorithms for machine learning operations like gradient descent and logistic regression (e.g., [4,28,49,50,52]) with corresponding software implementations like CrypTen [40] .  \nIn this work, we develop and deploy privacy-preserving machine learning ina real-world application based on political science: namely, the estimation of political preferences in the United States during the months around the 2024 U.S.  \n5 [https://github.com/sambux1/opps-client](https://github.com/sambux1/opps-client)  \n2 Buxbaum et al.  \npresidential election. To accomplish this goal, we combine MPC-based privacypreserving machine learning with the work of Comarela et al. [18], which demonstrates that web browsing patterns can be used to assess aggregate preferences in political candidates.  \nBackground. In more detail, Comarela et al. [18] start from a dataset of web browsing records of a cross-section of people provided by a media measurement company. They design a machine learning algorithm based on Learning from Label Proportions (LLP), shown in Algorithm 1, that uses web browsing data to infer the same type of information that is generally produced by political polls: the fraction of voters in each region (e.g., county or state) that prefer a given candidate at a particular point in time. Hence, this work shows potential to augment traditional methods of political polling, since its lower cost enables more frequent and precise polling—an opinion poll can form the initial “ground truth” in one location at one moment in time, and then LLP on web browsing data can be used to predict political preferences at other times and/or locations.  \nThis LLP algorithm is remarkably effective at estimating political preferences, even when given only browsing visits to the most popular websites rather than the long tail of smaller blogs and personal sites. Nevertheless, data privacy concerns make it challenging and risky","cbCail6vcL8D0Fxw","https://ap.wps.com/l/cbCail6vcL8D0Fxw","pdf",1772639,1,20,"English","en",105,"# Introduction\n## Background\n## Our contributions\n## Ethics","[{\"question\":\"What problem does the document address in machine learning for political opinion?\",\"answer\":\"It targets estimating aggregate political preferences from private web browsing data without exposing users’ sensitive browsing histories.\"},{\"question\":\"How is privacy preserved when computing political preferences?\",\"answer\":\"Secure multiparty computation (MPC) is used so browsing data remain secret-shared in client browsers and are never released in the clear.\"},{\"question\":\"What data and tools support the real-world deployment?\",\"answer\":\"A custom-built Chrome browser extension collects daily visit histograms to popular websites, and analysis runs on a cloud MPC backend built with the CrypTen library.\"}]","Privacy-Preserving Machine Learning on Web Browsing for Public Opinion | PDF",1785898113,50,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"privacy-preserving-machine-learning-on-web-browsing-for-public-opinion","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/privacy-preserving-machine-learning-on-web-browsing-for-public-opinion/125313/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the document address in machine learning for political opinion?","Question",{"text":75,"@type":76},"It targets estimating aggregate political preferences from private web browsing data without exposing users’ sensitive browsing histories.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is privacy preserved when computing political preferences?",{"text":80,"@type":76},"Secure multiparty computation (MPC) is used so browsing data remain secret-shared in client browsers and are never released in the clear.",{"name":82,"@type":73,"acceptedAnswer":83},"What data and tools support the real-world deployment?",{"text":84,"@type":76},"A custom-built Chrome browser extension collects daily visit histograms to popular websites, and analysis runs on a cloud MPC backend built with the CrypTen library.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":29,"slug":113},6,"Technology","technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":21,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":127,"show_sort_weight":21,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":46,"category_name":135,"show_sort_weight":106,"slug":136},19,"General","general"]