数据集:
McGill-NLP/FaithDial
子任务:
dialogue-modeling语言:
en计算机处理:
monolingual大小:
10K<n<100K批注创建人:
crowdsourced预印本库:
arxiv:2204.10757许可:
mitFaithDial 是一个忠实的知识引导对话基准,包含5649个对话、50761个对话轮次。通过通过Amazon Mechanical Turk进行了问卷调查,要求注解员修改了 Wizard of Wikipedia (WoW) 中的不真实话语。在我们的对话设置中,我们模拟了两个说话者之间的交互:信息搜索者和机器人向导。搜索者较为自由,而机器人向导在其沟通中有一些约束。事实上,机器人必须遵守以下规则:
英语
'train' 的一个例子如下所示:
[ { "utterances": [ ... // prior utterances, { "history": [ "Have you ever been to a concert? They're so fun!", "No I cannot as a bot. However, have you been to Madonna's? Her 10th concert was used to help her 13th album called \"Rebel Heart\".", "Yeah I've heard of it but never went or what it was for. Can you tell me more about it?" ], "speaker": "Wizard", "knowledge": "It began on September 9, 2015, in Montreal, Canada, at the Bell Centre and concluded on March 20, 2016, in Sydney, Australia at Allphones Arena.", "original_response": "It started in September of 2015 and ran all the way through March of 2016. Can you imagine being on the road that long?", "response": "Sure. The concert started in September 9th of 2015 at Montreal, Canada. It continued till 20th of March of 2016, where it ended at Sydney, Australia.", "BEGIN": [ "Hallucination", "Entailment" ], "VRM": [ "Disclosure", "Question" ] }, ... // more utterances ] }, ... // more dialogues ]
如果 'original_response' 是空的,意味着回答对源信息忠实,我们将其视为 FaithDial 的回答。WoW 中的忠实回答也会经过轻微的编辑,以解决一些语法问题或错误。
验证集包括 WoW 中的 'seen' 和 'unseen' 数据划分。测试集也是如此。我们还包括 FaithDial 验证集和测试集的这些划分。
根据 Sheehan. 2018 中规定的伦理众包指南,我们雇佣了Amazon Mechanical Turk(AMT)工人来编辑在WoW对话中的不忠实回答。为了确保任务定义的清晰,我们为术语提供了详细的示例。此外,我们在几个月的时间内进行了几轮分阶段的任务。
要参与任务,工人必须位于美国和加拿大,并且在资格测试中成功回答了20个问题。在启动主要注释任务之前,我们进行了一小轮试点任务(60个HITS),以检查工人的表现。我们向犯错误的工人发送电子邮件,提供示例以帮助他们在未来的任务中纠正错误。
FaithDial 中的搜索者话语可能包含个人和敏感信息。
近年来,对话式人工智能市场出现了各种应用程序的大量增长,这些应用程序由大型预训练语言模型驱动,涵盖了广泛的领域,例如客户支持、教育、电子商务、健康、娱乐等。确保这些系统是可信的,对于在高风险领域的真实应用大规模部署至关重要。FaithDial 有望鼓励信息搜索对话的忠实性,使虚拟助手更加安全可靠。
MIT
@article{dziri2022faithdial, title={FaithDial: A Faithful Benchmark for Information-Seeking Dialogue}, author={Dziri, Nouha and Kamalloo, Ehsan and Milton, Sivan and Zaiane, Osmar and Yu, Mo and Ponti, Edoardo and Reddy, Siva}, journal={arXiv preprint, arXiv:2204.10757}, year={2022}, url={https://arxiv.org/abs/2204.10757} }