Reproduce “A Large-Scale Empirical Study of LLM Orchestration and Ensemble Strategies for Sentiment Analysis in Recommender Systems” 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
Konstantinos I. Roumeliotis, Dionisis Margaris, Dimitris Spiliotopoulos, Costas Vassilakis
The system predicts existing one-to-five-star ratings from English-language Amazon product reviews by giving a GPT-5-based meta-model both review text and ratings proposed by 12 models. On a balanced test set of 5,000 reviews, GPT-5 aggregation achieved 71.40% accuracy, versus 62.64% for majority voting and 62.96% for rounded mean aggregation. The full system cost $130.45; findings apply to this task and sample.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.