16/25 𝗔𝗻 𝗜𝗻𝗳𝗲𝗰𝘁𝗶𝗼𝘂𝘀 𝗗𝗶𝘀𝗲𝗮𝘀𝗲 𝗦𝗽𝗿𝗲𝗮𝗱 𝗦𝗶𝗺𝘂𝗹𝗮𝘁𝗶𝗼𝗻 𝗕𝗮𝘀𝗲𝗱 𝗼𝗻 𝗟𝗮𝗿𝗴𝗲 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹 𝗗𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝗠𝗮𝗸𝗶𝗻𝗴 This paper proposes a spatially grounded, agent-based simulation framework that integrates LLM-generated decisions about self-reported influenza-like illness into a census-based synthetic population. By assigning agents to spatial units in cities using real-world census data, it models geographically diverse behaviour. Simulations in San Francisco and Atlanta reveal income and education as dominant drivers of reporting rate variation, with smaller effects from geography, LLM choice, and message framing, supporting spatial epidemiological modelling and bias-aware behavioural analysis. #LLMApplications #AgentBasedModeling #Epidemiology #PublicHealth #BehavioralModeling #SpatialModeling Paper Link: arxiv.org/abs/2606.06360

Model behavioral patterns before they become market signals. Price reacts to outcomes. We analyze behavior. Coordination leaves traces. Narratives leave structure. Our agents model both. #OnChainAnalytics #BehavioralModeling #AI #Crypto #x402

🏆 We have the #WINNERS of the Synerise ACM RecSys Challenge 2025! 🎉 - 4,106 submissions - 436 teams (303 active, 598 participants) - Over 100 academic teams joined us! This year’s challenge focused on Universal Behavioral Profiles, pushing the boundaries of recommender systems research and practice. 👏 Congratulations to the winning teams 🏆 and a huge thank you to everyone who participated, competed, and contributed to advancing the field! Check who finished on the podium - link in comments Authors of the challenge: Łukasz Sienkiewicz, Gergely Stomfai, Maria Janicka, Jacek Dąbrowski Synerise. Dietmar Jannach, Marco Polignano, Claudio Pomo, Abhishek Srivastava Francesco Barile #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles #SyneriseRecSysChallenge2025 #RecSys2025 #RecSys25

#BioDigitalConvergence #StealthTowers #InvisibleInfrastructure #SurveillanceMesh #CovertSignals #AmbientSurveillance #DigitalTwins #MeshBackbone #ConsentBypassed #SyntheticSkyline ⸻ 🧠 Critical Thought Hashtags: #PredictiveControl #BehavioralModeling #RFHarvesting #EdgeAI #WBANs #DataWithoutConsent #TechnocraticMesh #SmartCitySecrets #WeaponizedInvisibility #YourBodyIsTheNode

🎉 Great news! Paper submissions for the Synerise @ACMRecSys 2025 Challenge #Workshop are now open! 📢 If you were taking part in the challenge and think your solution is especially effective, new, interesting, or makes good use of the data, we’d love to see your paper. 👉 Check out the submission details in the link in comments! 🧠📄 Looking forward to seeing what you come up with! Let’s make this a great workshop! #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles #SyneriseRecSysChallenge2025 #RecSys2025 #RecSys25

🏆 We have the winners of the Synerise @ACMRecSys Challenge 2025! 🏆 After months of intense competition, innovation, and thousands of submissions, the top teams have emerged. Check out the #Final and #Academic #Leaderboards to see who made it to the top! 🧨 ➡️ Results are available via the link in the comments ➡️ Codabench link in the comments 🎉 Congratulations to all winners and participants – you’ve made this challenge unforgettable! 🚀 Authors of the challenge: Łukasz Sienkiewicz, Gergely Stomfai, Maria Janicka Jacek Dąbrowski Synerise. Dietmar Jannach, Marco Polignano, Claudio Pomo, Abhishek Srivastava Francesco Barile #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles #SyneriseRecSysChallenge2025 #RecSys2025 #RecSys25

🎉 Synerise @ACMRecSys Challenge 2025 has officially ended! 🎉 🤖 Over 416 #AI teams from around the world, 3486 submissions, 🌍 40 universities. One challenge - Universal Behavioral Modeling Data Challenge. You made Synerise RecSys Challenge 2025 a ride to remember! 🏆 Final Leaderboard & Winners will be announced on June 30th 💻 Don’t forget to upload your final code by July 5th to stay eligible for prizes! 📝 And don’t miss the paper submission deadline - July 12th! 🎉 Congratulations to all participants! You’re awesome! Jacek Dąbrowski, Maria Janicka, Łukasz Sienkiewicz, Gergely Stomfai Dietmar Jannach, Marco Polignano, Claudio Pomo, Abhishek Srivastava Francesco Barile Jaroslaw Krolewski #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles #RecSys2025 #SyneriseRecSys2025

I couldn’t be prouder — once again, the @Synerise 🇵🇱 team has raised the bar for what’s possible in global AI. 🚀 The Synerise ACM #RecSys Challenge 2025 has officially concluded — and it wasn’t just a competition. It was a statement. A bold vision. A global stage where the world’s best minds met to push the boundaries of behavioral modeling and recommender systems. 🌍 416 AI teams 🧠 3486 submissions 🏛️ 40 universities And one unified goal: to solve the Universal Behavioral Modeling Data Challenge — one of the most ambitious tasks in the field. Together with ACM RecSys, the world’s most respected scientific body in this domain, we created a challenge that wasn’t just hard — it demanded excellence, creativity, and vision. And the world responded. 🔬 This was about more than algorithms. It was about ideas, courage, and raising the standards of what AI can be. 💡 It was about opening doors to future breakthroughs that will redefine personalization and understanding of human behavior. To every team that participated — you inspired us. To the scientists, engineers, and dreamers — you made history. And to the Synerise team — I am, once again, deeply proud. You dared to lead. You set the standard. You made space for the future. This is what bold science looks like. This is what leadership feels like. 📅 Winners revealed: June 30 💻 Final code due: July 5 📝 Paper submissions: July 12 Thank you for being part of this incredible journey. The future of AI is brighter because of what we build — together. #AI #RecommenderSystems #BehavioralModeling #SyneriseChallenge #ACM #RecSys2025 #Innovation #Leadership #ProudMoment #FutureIsNow

As we enter the final #24hours of the Synerise ACM @ACMRecSys Challenge ⏳, the countdown to the leaderboard freeze begins 🚀. Every submission could make a difference ⚡️. We're rooting for all participants 🤞 good luck on the final stretch! 💪 Jacek Dąbrowski, Maria Janicka, Łukasz Sienkiewicz, Gergely Stomfai Dietmar Jannach, Marco Polignano, Claudio Pomo, Abhishek Srivastava Francesco Barile #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles #RecSys2025 #SyneriseRecSys2025

🧨 Nearly 200 #AI teams from across the globe have already joined the @ACMRecSys Challenge 2025, authored by Synerise! This year’s edition introduces the concept of Universal Behavioral Profiles 📊 — compact, transferable user representations that capture essential behavioral patterns 🧠. These profiles are designed to power multiple predictive tasks, such as churn prediction 🔄 and product recommendations 🛍️, enabling models to generalize effectively across domains 🌐. Jacek Dąbrowski, Maria Janicka, Łukasz Sienkiewicz, Gergely Stomfai Dietmar Jannach, Marco Polignano, Claudio Pomo, Abhishek Srivastava Francesco Barile #Synerise #RecSysChallenge2025 #RecSys25 #RecSys2025 #AI #BehavioralModeling #UniversalBehavioralProfiles

🧨 Submission system open. Leaderboard goes live! 🚀 Link in comments The @ACMRecSys Challenge is one of the most important global competitions in the field of recommender systems and enjoys significant prestige, especially within academic circles and industries related to data analysis and artificial intelligence. This year, Synerise is not only co-organizing the challenge but also authoring it, bringing together the best #AI teams from around the world to compete for the top position. 🚀 Let the challenge begin—good luck to all! 🚀 Huge thanks to everyone who made this challenge possible: Jacek Dąbrowski Maria Janicka Łukasz Sienkiewicz Gergely Stomfai Dietmar Jannach, Marco Polignano, Claudio Pomo, Abhishek Srivastava Francesco Barile #Synerise #RecSysChallenge2025 #RecommendationSystems #AI #BehavioralModeling #RecSys #AIResearch #MachineLearning

🚀 Join the @Synerise Workshop on the @ACMRecSys Challenge! This workshop is the ideal chance to deepen your knowledge and have your questions answered by the challenge authors! 🗓️ Date: April 15 ⏰ Time: 3:00 PM – 5:00 PM CEST 🌍 Location: Online ➡️ Register now: recsys.synerise.com/online-w… 🎙️ Hosted by Synerise Synerise AI Team: Jacek Dąbrowski, Maria Janicka, Łukasz Sienkiewicz and Gergely Stomfai. During the workshop, we'll cover the challenge details, explore the dataset, and answer all your questions! Don’t miss out—register today! #Synerise #RecSysChallenge2025 #RecommendationSystems #AI #BehavioralModeling #RecSys #WorkshopRecSys #AIResearch #MachineLearning

Are you participating in the @Synerise @ACMRecSys Challenge 2025? 🔥 The challenge kicks off on April 10! Join the #WORKSHOP by Synerise to explore the dataset and get your questions answered by the challenge authors: Jacek Dąbrowski, Maria Janicka, Łukasz Sienkiewicz, and Gergely Stomfai.   📅 Workshop by Synerise date: April 15 ⏰ Time: 3:00–5:00 PM CEST 🌍 Location: Online 🔗 Register link in the comments! 👇 🚀 Don’t miss this expert-led session—sign up today! #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles

Are you participating in the Synerise ACM #RecSys Challenge 2025? 🔥 The challenge kicks off on April 10! Join the #WORKSHOP by Synerise to explore the dataset and get your questions answered by the challenge authors: Jacek Dąbrowski, Maria Janicka, Łukasz Sienkiewicz, and Gergely Stomfai. 📅 Workshop by Synerise date: April 15 ⏰ Time: 3:00–5:00 PM CEST 🌍 Location: Online 🔗 Register link in the comments! 👇 🚀 Don’t miss this expert-led session—sign up today! #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles

🚀 Join the WORKSHOP by Synerise on the RecSys Challenge. Link in comments. Are you participating in the Synerise RecSys Challenge 2025 or simply passionate about #recommendation systems? This is your chance to learn more and get your questions answered directly by the authors of the challenge. 📅 Date: April 15 ⏰Time: 3:00 PM – 5:00 PM CEST 📍Location: Online The WORKSHOP by Synerise will be hosted by: Jacek Dąbrowski – Chief AI Officer Maria Janicka – AI Researcher Łukasz Sienkiewicz – AI Research Scientist Gergely Stomfai – AI Researcher During the #workshop, we’ll go over the challenge details, #dataset, and answer your questions. #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #WorkshopRecSys

🚀 AI is changing the way businesses make decisions, from personalized recommendations to predicting customer churn. But what if we could create a universal approach to understanding user behavior? 🌟That’s exactly what we’re aiming for in the Synerise RecSys Challenge 2025! 💡 What’s the challenge about? Instead of crafting separate customer representations for different tasks, contestants will work on creating one universal user representation — a way to describe past user interactions (like 🛒 purchases, 🔍 searches, or 📄 page visits) so that various downstream models can predict multiple behaviors at once. 🎯 Why does it matter? A single, well-designed Universal Behavioral Profile can power smarter AI systems that adapt across different use cases in industries like e-commerce 🛍️, banking 💳, entertainment 🎬, and more! #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfile

💡We’re introducing a new perspective for #RecSysChallange. 🚀The Synerise #RecSys Challenge 2025 focuses on an innovative approach to Universal Behavioral Modeling. A non-standard competition format. Focus on the universality of different aspects of behavioral modeling. 🌐Submissions in the form of Universal Behavioral Profiles instead of simple model predictions. 📅Launch: April 10 – the leaderboard goes live! Dataset & code repository available now! Check in comments. Jacek Dąbrowski, Maria Janicka, Łukasz Sienkiewicz, Gergely Stomfai Dietmar Jannach, Marco Polignano, Claudio Pomo, Abhishek Srivastava, Francesco Barile @ACMRecSys #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles

🧨 On April 10, 𝐭𝐡𝐞 𝐒𝐲𝐧𝐞𝐫𝐢𝐬𝐞 𝐑𝐞𝐜𝐒𝐲𝐬 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞 2025 begins—an essential competition focused on recommendation systems and behavioral modeling. 📈 This year, Synerise is both a co-organizer and the author of the challenge, 𝐛𝐫𝐢𝐧𝐠𝐢𝐧𝐠 𝐭𝐨𝐠𝐞𝐭𝐡𝐞𝐫 𝐭𝐡𝐞 𝐰𝐨𝐫𝐥𝐝’𝐬 𝐭𝐨𝐩 hashtag#𝐀𝐈 𝐭𝐞𝐚𝐦𝐬 to compete for the top spot. 🔑 Starting today, we are 𝐩𝐫𝐨𝐯𝐢𝐝𝐢𝐧𝐠 𝐚𝐜𝐜𝐞𝐬𝐬 𝐭𝐨 𝐭𝐡𝐞 𝐝𝐚𝐭𝐚𝐬𝐞𝐭 𝐚𝐧𝐝 𝐜𝐨𝐝𝐞 𝐫𝐞𝐩𝐨𝐬𝐢𝐭𝐨𝐫𝐲. On April 10, the submission system will be opened together with the official leaderboard. The hashtag#challenge redefines behavior modeling with a unified approach to tasks like predicting user churn, identifying category interests, and forecasting product purchases. Instead of treating these tasks separately, we introduce 𝐔𝐧𝐢𝐯𝐞𝐫𝐬𝐚𝐥 𝐁𝐞𝐡𝐚𝐯𝐢𝐨𝐫𝐚𝐥 𝐏𝐫𝐨𝐟𝐢𝐥𝐞𝐬—user representations built from past interactions that generalize across multiple predictive applications. ➡️ Synerise RecSys Challenge 2025 recsys.synerise.com/ 🛠 Repository github.com/Synerise/recsys20… #Synerise #RecSysChallenge2025 #RecSys #AI #BehavioralModeling #UniversalBehavioralProfiles

Exploring the depths of human decision-making, researchers from MIT and the University of Washington have unveiled a method that can predict a person’s actions by modeling their computational constraints. This groundbreaking study not only provides insights into the erratic nature of human choices but also offers a scalable solution to improve AI-human collaboration. The 'inference budget', which quantifies how long someone plans for a problem, serves as a key indicator of behavior. Such insights could vastly improve how AI systems understand and interact with their human counterparts, making them invaluable assets. #CognitiveScience #AIResearch #BehavioralModeling

Exploring Temporal Data Splits in BaseModel ➡ bit.ly/3vcSASh @SAIRblog latest post spotlights @basemodelai data-splitting approach and how it connects with our vision of #BehavioralModeling. We recognize the importance of the temporal aspect that is shaping data in real-life settings, and we believe it should be reflected in model training and testing procedures. With BaseModel we make it not only feasible but also super easy thanks to our innovative approach to creating users' behavioral representations. With temporal splits, we are able to: 🔒 Prevent data leakage between train, validation, and test set. 📊 Provide more reliable results than standard testing methods. 🔄 Augment data by creating multiple training points from a single user history. #SAIR #SyneriseAIResearch #TemporalDataSplits