Reproduce “From Counterfactual Edits to Human Narratives: LLM-Driven Interpretations for Recommender System Explanations” 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
Vojtech Cahlik, Amir Reza Mohammadi, Rodrigo Alves, Pavel Kordik, Eva Zangerle
The authors present a language-model framework that turns recommendation-changing interaction sets into summaries of preference patterns. It samples metadata from removed items and remaining history, then describes their contrast. On MovieLens-1M, DPO raises ACCENT–Gemma 3 12B correctness from 0.43 to 0.49 and informativeness from 0.16 to 0.21; classifier scores do not establish user understanding or real-world benefit.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.