TheFinAI/JF-ICR

JF-ICR（日本金融隐含承诺识别）是一个基准数据集，用于评估日本投资者问答中的隐含承诺和拒绝识别。该数据集包含来自4家公司在3年内的94个单轮问答对，标注为5种隐含立场类别，区分了高语境话语中的同意、模糊表达和拒绝。该数据集解决了日本金融交流中间接表达和隐含承诺理解的挑战，其中拒绝通常通过语用线索而非明确陈述传达。

JF-ICR (Japanese Financial Implicit Commitment Recognition) is a benchmark dataset for evaluating implicit commitment and refusal recognition in Japanese investor-facing Q&A. The dataset consists of 94 single-turn question-answer pairs from 4 companies spanning 3 years, annotated into 5 implicit stance categories that distinguish agreement, hedging, and refusal under high-context discourse. This dataset addresses the challenge of understanding indirect expression and implicit commitment in Japanese financial communication, where refusals are frequently conveyed through pragmatic cues rather than explicit statements.