UXSim: Towards a Hybrid User Search Simulation
2026-02-27 • Information Retrieval
Information RetrievalHuman-Computer Interaction
AI summaryⓘ
The authors created UXSim, a new way to better mimic how people use complex search systems. They combined older methods that use fixed user models with newer large language models that can think and adapt, but might sometimes lack solid facts. By connecting these two, their system more realistically shows how users behave and helps explain why the AI made certain decisions. This approach aims to make user simulations more accurate and understandable.
User simulationInteractive search systemsLarge language modelsUser behavior modelingHuman-computer interactionCognitive processesExplainabilityAdaptive agents
Authors
Saber Zerhoudi, Michael Granitzer
Abstract
Simulating nuanced user experiences within complex interactive search systems poses distinct challenge for traditional methodologies, which often rely on static user proxies or, more recently, on standalone large language model (LLM) agents that may lack deep, verifiable grounding. The true dynamism and personalization inherent in human-computer interaction demand a more integrated approach. This work introduces UXSim, a novel framework that integrates both approaches. It leverages grounded data from traditional simulators to inform and constrain the reasoning of an adaptive LLM agent. This synthesis enables more accurate and dynamic simulations of user behavior while also providing a pathway for the explainable validation of the underlying cognitive processes.