Causal Video Question Answering (CVidQA) queries not only association or temporal relations but also causal relations in a video. Existing question synthesis methods pre-trained question generation (QG) systems on reading comprehension datasets with text descriptions as inputs. However, QG models only learn to ask association questions (e.g., ``what is someone doing...'') and result in inferior performance due to the poor transfer of association knowledge to CVidQA, which focuses on causal questions like ``why is someone doing ...''. Observing this, we proposed to exploit causal knowledge to generate question-answer pairs, and proposed a novel framework, Causal Knowledge Extraction from Language Models (CaKE-LM), leveraging causal commonsense knowledge from language models to tackle CVidQA. To extract knowledge from LMs, CaKE-LM generates causal questions containing two events with one triggering another (e.g., ``score a goal'' triggers ``soccer player kicking ball'') by prompting LM with the action (soccer player kicking ball) to retrieve the intention (to score a goal). CaKE-LM significantly outperforms conventional methods by 4% to 6% of zero-shot CVidQA accuracy on NExT-QA and Causal-VidQA datasets. We also conduct comprehensive analyses and provide key findings for future research.

翻译：因果视频问答（CVidQA）不仅查询视频中的关联或时序关系，还查询因果关系。现有的问题合成方法在阅读理解数据集上预训练问题生成（QG）系统，并以文本描述作为输入。然而，QG模型仅学习提出关联性问题（例如，“某人在做什么……”），由于将关联性知识迁移到专注于因果性问题（如“某人为什么在做什么……”）的CVidQA中效果不佳，导致性能较差。基于此观察，我们提出利用因果知识生成问答对，并设计了一个新颖框架——从语言模型中提取因果知识（CaKE-LM），利用语言模型中的因果常识知识解决CVidQA任务。为从语言模型（LM）中提取知识，CaKE-LM通过向LM提示某一动作（如“足球运动员踢球”）来检索其意图（如“得分”），从而生成包含两个事件（一个触发另一个）的因果性问题（例如，“得分”触发“足球运动员踢球”）。在NExT-QA和Causal-VidQA数据集上，CaKE-LM的零样本CVidQA准确率显著优于传统方法4%至6%。我们还进行了全面分析，并为未来研究提供了关键发现。