SLUE (Spoken Language Understanding Evaluation)

口语理解评估 (SLUE) 是一套用于口语理解评估的基准任务。它由有限大小的标记训练集和相应的评估集组成。该资源将允许研究社区跟踪进展，评估高级任务的预训练表示，并研究开放性问题，例如管道与端到端方法的效用。 SLUE 基准套件的第一阶段包括命名实体识别 (NER)、情感分析 (SA) 和相应数据集上的 ASR。语料库包括： SLUE-VoxPopuli：由 ASR 和 NER 任务组成 - CC0 许可证 SLUE-VoxCeleb：由 ASR 和 SA 任务组成 - CCBY 4.0 许可证Spoken Language Understanding Evaluation (SLUE) is a benchmark task suite for spoken language understanding assessment. It comprises limited-sized labeled training datasets and corresponding evaluation sets. This resource will enable the research community to track research progress, evaluate pre-trained representations for advanced tasks, and investigate open questions such as the utility of pipeline versus end-to-end approaches. The first phase of the SLUE benchmark suite includes Named Entity Recognition (NER), Sentiment Analysis (SA), and ASR tasks on corresponding datasets. The corpus includes: SLUE-VoxPopuli: consists of ASR and NER tasks, licensed under CC0 SLUE-VoxCeleb: consists of ASR and SA tasks, licensed under CC BY 4.0

搜集汇总

背景与挑战

背景概述

SLUE是一套用于口语理解评估的基准任务，包含命名实体识别和情感分析等任务。该数据集基于SLUE-VoxPopuli和SLUE-VoxCeleb两个子语料库构建，分别采用CC0和CCBY 4.0许可证，旨在支持研究社区跟踪进展并评估预训练表示。