@inproceedings{yoon-etal-2025-using,
    title = "Using tournaments to calculate {AUROC} for zero-shot classification with {LLM}s",
    author = "Yoon, WonJin  and
      Bulovic, Ian  and
      Miller, Timothy A.",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/name-variant-enfa-fane/2025.findings-emnlp.1281/",
    doi = "10.18653/v1/2025.findings-emnlp.1281",
    pages = "23583--23591",
    ISBN = "979-8-89176-335-7",
    abstract = "Large language models perform surprisingly well on many zero-shot classification tasks, but are difficult to fairly compare to supervised classifiers due to the lack of a modifiable decision boundary. In this work, we propose and evaluate a method that transforms binary classification tasks into pairwise comparisons between instances within a dataset, using LLMs to produce relative rankings of those instances. Repeated pairwise comparisons can be used to score instances using the Elo rating system (used in chess and other competitions), inducing a confidence ordering over instances in a dataset. We evaluate scheduling algorithms for their ability to minimize comparisons, and show that our proposed algorithm leads to improved classification performance, while also providing more information than traditional zero-shot classification."
}