ailsntua/QEvasion

这是一个包含从政治访谈中提取的问题-答案对的数据集，以及反映回答清晰度或回避程度的注释。数据集包含的问题-答案对来自于政治访谈，并相应地进行了注释。该数据集旨在利用大型语言模型和人类专业知识解决政治访谈中问题回答清晰度的问题。数据集采用两级分类法，既考虑了回答在给定问题中提供的信息的清晰度（高级别），也提供了与不清晰、模棱两可的回答相关的逃避技巧的细粒度分类法（低级别）。

This is a dataset containing question-answer pairs extracted from political interviews, along with annotations reflecting varying degrees of response clarity or evasion. The question-answer pairs in the dataset are drawn from political interviews and are annotated accordingly. The dataset aims to address the issue of response clarity in political interviews using the capabilities of Large Language Models (LLMs) and human expertise. The dataset employs a two-level taxonomy, considering both the clarity of the information provided in a given question (high-level) and a fine-grained taxonomy of evasion techniques related to unclear, ambiguous responses (lower-level).