MinosEval: Distinguishing Factoid and Non-Factoid for Tailored Open-Ended QA Evaluation with LLMs
This addresses the problem of refined and interpretable evaluation for open-ended QA in AI, though it appears incremental by building on existing LLM-based approaches.
The paper tackles the challenge of automatically evaluating open-ended question answering by proposing MinosEval, a method that distinguishes between factoid and non-factoid questions to apply tailored evaluation strategies, resulting in better alignment with human annotations and more interpretable outcomes.
Open-ended question answering (QA) is a key task for evaluating the capabilities of large language models (LLMs). Compared to closed-ended QA, it demands longer answer statements, more nuanced reasoning processes, and diverse expressions, making refined and interpretable automatic evaluation both crucial and challenging. Traditional metrics like ROUGE and BERTScore struggle to capture semantic similarities due to different patterns between model responses and reference answers. Current LLM-based evaluation approaches, such as pairwise or listwise comparisons of candidate answers, lack intuitive interpretability. While pointwise scoring of each response provides some descriptions, it fails to adapt across different question contents. Most notably, existing methods overlook the distinction between factoid and non-factoid questions. To address these challenges, we propose \textbf{MinosEval}, a novel evaluation method that first distinguishes open-ended questions and then ranks candidate answers using different evaluation strategies. For factoid questions, it applies an adaptive key-point scoring strategy, while for non-factoid questions, it uses an instance-aware listwise ranking strategy. Experiments on multiple open-ended QA datasets, including self-built ones with more candidate responses to complement community resources, show that MinosEval better aligns with human annotations and offers more interpretable results.