arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2205.12420cs.CL
收藏

从说明书手册中学习动作条件以理解指令

Learning Action Conditions from Instructional Manuals for Instruction Understanding

University of California, Los Angeles(加利福尼亚大学洛杉矶分校);University of Cambridge(剑桥大学);Information Sciences Institute, University of Southern California(南加州大学信息科学学院)

Te-Lin Wu, Caiqi Zhang, Qingyuan Hu, Alex Spangher, Nanyun Peng

AI总结:

本文提出动作条件推断任务,利用弱监督从说明书手册中自动构建训练数据,并通过利用完整指令上下文和启发式方法显著提升动作前后置条件推断的F1分数。

摘要:

推断动作的前置条件和后置条件对于理解复杂指令至关重要,并且对于自主指令引导智能体以及辅助人类执行物理任务的辅助型人工智能等应用而言必不可少。在本工作中,我们提出了一项称为动作条件推断的任务,并收集了一个高质量、经人工标注的说明书手册中动作前置条件与后置条件数据集。我们提出了一种弱监督方法,从在线说明书手册中自动构建大规模训练实例,并整理了一个经过密集人工标注和验证的数据集,以研究当前自然语言处理模型在指令文本中推断动作条件依赖关系的能力。我们设计了两种类型的模型,其区别在于是否利用上下文信息和全局信息,并设计了多种启发式组合来构建弱监督信号。我们的实验结果表明,考虑完整指令上下文可使F1分数提升超过20%,而采用所提出的启发式方法可带来超过6%的F1分数收益。

英文摘要:

The ability to infer pre- and postconditions of an action is vital for comprehending complex instructions, and is essential for applications such as autonomous instruction-guided agents and assistive AI that supports humans to perform physical tasks. In this work, we propose a task dubbed action condition inference, and collecting a high-quality, human annotated dataset of preconditions and postconditions of actions in instructional manuals. We propose a weakly supervised approach to automatically construct large-scale training instances from online instructional manuals, and curate a densely human-annotated and validated dataset to study how well the current NLP models can infer action-condition dependencies in the instruction texts. We design two types of models differ by whether contextualized and global information is leveraged, as well as various combinations of heuristics to construct the weak supervisions. Our experimental results show a >20% F1-score improvement with considering the entire instruction contexts and a >6% F1-score benefit with the proposed heuristics.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑