Recommender systems (RecSys) play a central role in helping users navigate and discover content in large, constantly evolving information spaces. However, many deployed systems still rely on static user profiles and precomputed recommendations, limiting their ability to adapt within and across sessions. As personalization demands intensify, privacy regulations tighten, and user intent grows more dynamic, the need for online and adaptive recommender systems has never been greater.

The Online and Adaptive Recommender Systems (OARS) workshop brings together researchers and practitioners from academia and industry to share advances in real-time and adaptive recommender systems. We invite submission of papers and posters, representing new researches, positions, and proposals for new tools, datasets, and resources. All submitted papers will be peer reviewed by an international program committee of researchers of high repute. Accepted submissions will be presented at the workshop.

Topics

🤖 Agentic & LLM-based RecSys

• Agentic recommender systems, assistant-style interfaces, memory and tool-use (2026 special theme)
• LLMs and foundation models in RecSys: semantic IDs, tokenization, multi-modality, in-context learning

⚡ Online, Adaptive & Interactive Learning

• Online and continual learning, reinforcement learning, bandits, and counterfactual evaluation
• Real-time user intent modeling, session-aware and conversational recommendation

🏗️ Architectures & Systems

• RAG-based, streaming, and event-driven architectures for scalable learning
• Industry deployments, infrastructure, and real-world case studies

🧩 Data Challenges & Robustness

• Cold-start, distribution shift, and robustness under data sparsity

📊 Evaluation, Causality & Explainability

• Evaluation, explanation, off-policy, and continuous learning methods for OARS
• Predictive analytics and causal inference for recommendation

🔒 Privacy, Ethics & User Welfare

• Privacy, ethics, fairness, and user welfare in OARS