Automatic Answerability Evaluation for Question Generation
Zifan Wang, Kotaro Funakoshi, Manabu Okumura
TL;DR
The paper tackles the challenge of evaluating whether generated questions are answerable by reference answers, a property poorly captured by traditional n-gram metrics. It introduces PMAN, a prompting-based metric that leverages large language models with a chain-of-thought prompting design to assess answerability by requiring the model to answer, compare with the reference, and output a YES/NO verdict. Through extensive experiments on manually created and model-generated samples, PMAN demonstrates reliable alignment with human judgments and complements conventional metrics in evaluating MQG on HotpotQA. Additionally, a GPT-4–based MQG model achieves state-of-the-art performance in generating answerable questions, illustrating PMAN's practical utility for advancing QG systems and their evaluation.
Abstract
Conventional automatic evaluation metrics, such as BLEU and ROUGE, developed for natural language generation (NLG) tasks, are based on measuring the n-gram overlap between the generated and reference text. These simple metrics may be insufficient for more complex tasks, such as question generation (QG), which requires generating questions that are answerable by the reference answers. Developing a more sophisticated automatic evaluation metric, thus, remains an urgent problem in QG research. This work proposes PMAN (Prompting-based Metric on ANswerability), a novel automatic evaluation metric to assess whether the generated questions are answerable by the reference answers for the QG tasks. Extensive experiments demonstrate that its evaluation results are reliable and align with human evaluations. We further apply our metric to evaluate the performance of QG models, which shows that our metric complements conventional metrics. Our implementation of a GPT-based QG model achieves state-of-the-art performance in generating answerable questions.