Simulation of Interactive Information Retrieval: A Guided Tour

Interactive information retrieval (IIR) systems, including search engines and conversational systems, are increasingly central to user experiences. However, rigorously evaluating their performance, particularly as interactions become highly personalized, remains a scientific challenge. While user simulation offers a powerful methodology for reproducible evaluation, its adoption is hindered by a steep learning curve and a fragmented landscape of complex tools. This half-day tutorial provides a practical, hands-on introduction to user simulation at varying levels of complexity, from foundational statistical models to advanced, LLM-driven frameworks. Through a series of guided problems, participants will acquire practical skills in using popular libraries, learning user models from data, and applying large language models (LLMs) to simulate user behavior. The tutorial concludes with evaluating the simulators themselves, providing participants with guidance on appropriate use cases and fidelity assessment.