一句话：提出知识图谱驱动的推理测试方法，用于发现问答软件中的缺陷行为。
问答软件在数据驱动编程范式下可能包含缺陷，即在真实应用中行为异常。现有问答软件测试技术主要包括基于参考的测试与蜕变测试两类。该工作关注数据驱动问答软件的质量保障，论文题目表明作者提出知识图谱驱动的推理测试，其核心是以知识图谱支撑对问答软件推理行为进行测试，从而检测缺陷。所给摘要未提供方法步骤、实验设置、评价指标与关键结果，因此无法进一步概括技术细节与效果。
In the wake of developments in the field of Natural Language Processing, Question Answering (QA) software has penetrated our daily lives. Due to the data-driven programming paradigm, QA software inevitably contains bugs, i.e., misbehaving in real-world applications. Current testing techniques for testing QA software include two folds, reference-based testing and metamorphic testing.
@inproceedings{paperbot11387,
title = {Knowledge Graph Driven Inference Testing for Question Answering Software},
author = {Jun Wang and Yanhui Li and Zhifei Chen and Lin Chen and Xiaofang Zhang and Yuming Zhou},
booktitle = {ICSE},
year = {2024},
doi = {10.1145/3597503.3639109}
}