SLUE Phase-2

SLUE Phase-2是一个包含四个新任务的口语理解评估数据集，旨在补充现有的SLU数据集或基准。这些任务包括对话行为分类（DAC）、问题回答（QA）、摘要（SUMM）和命名实体定位（NEL），应用于英语口语数据。SLUE Phase-2的优势在于其任务的多样性和挑战性，使用自然对话或较长的论述作为输入，输出不仅限于标签或文本，还包括语音跨度时间戳。数据集的创建过程涉及新的人工注释，确保了数据的高质量。该数据集适用于推动口语理解和自然语言处理领域的研究，特别是在处理复杂口语任务和提高模型性能方面。

SLUE Phase-2 is a spoken language understanding evaluation dataset containing four new tasks, designed to complement existing SLU datasets or benchmarks. These tasks include Dialogue Act Classification (DAC), Question Answering (QA), Summarization (SUMM), and Named Entity Localization (NEL), all applied to spoken English data. The strengths of SLUE Phase-2 lie in its task diversity and challenging nature, which uses natural dialogues or longer discourses as inputs, with outputs not limited to labels or text but also encompassing speech span timestamps. The dataset was developed with novel manual annotations to ensure high data quality. This dataset supports advancing research in the fields of spoken language understanding and natural language processing, particularly in addressing complex spoken language tasks and improving model performance.