UXSim: Towards a Hybrid User Search Simulation

TL;DR

UXSim is a hybrid framework that combines traditional user simulators with adaptive large language models to create more accurate, dynamic, and explainable user behavior simulations in complex search systems.

Contribution

It introduces UXSim, a novel integrated approach that leverages grounded data and LLM reasoning for improved user experience simulation and validation.

Findings

Enhanced simulation accuracy and dynamism.

Improved explainability of user behavior models.

Potential for better personalization in search systems.

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.