Reproduce “Evaluating Conversational Recommender Systems via User Simulation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shuo Zhang, Krisztian Balog
The authors develop a simulated movie-seeker to evaluate multi-turn movie recommenders, combining a task-specific interaction model with a preference model. In human judgments, CIR6-PKG dialogues were mistaken for human ones in 36%, versus 27% for QRFA-Single. The simulator matched users’ Reward ranking across three anonymized agents, but another metric produced a different ranking, limiting evidence of general predictive validity.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.