gtfintechlab/bank_of_canada

这个数据集是从加拿大银行的会议记录中提取的1000个句子，用于立场检测、时间分类和不确定性估计三个任务。立场检测有四个标签：鹰派（支持紧缩性货币政策）、鸽派（支持扩张性货币政策）、中立（既不鹰派也不鸽派或两者都有）和不相关（与货币政策无关）。时间分类有两个标签：向前看（讨论未来的经济事件或决策）和非向前看（讨论过去或当前的经济事件或决策）。不确定性估计也有两个标签：确定（句子明确地呈现信息）和不确定（句子以推测、可能性或怀疑的方式呈现信息）。

This dataset consists of 1,000 sentences extracted from the meeting minutes of the Bank of Canada, annotated for three tasks: Stance Detection, Temporal Classification, and Uncertainty Estimation. Stance Detection has four labels: Hawkish (supports contractionary monetary policy), Dovish (supports expansionary monetary policy), Neutral (contains neither hawkish nor dovish sentiment, or both), and Irrelevant (not related to monetary policy). Temporal Classification has two labels: Forward-looking (discusses future economic events or decisions) and Not Forward-looking (discusses past or current economic events or decisions). Uncertainty Estimation also has two labels: Certain (presents information definitively) and Uncertain (presents information with speculation, possibility, or doubt).