XLEnt

XLEnt数据集由约翰斯·霍普金斯大学等机构创建，包含1.64亿对跨语言实体，覆盖120种语言与英语的对应关系。该数据集通过自动挖掘网络数据，利用词义音位对齐技术LSP-Align构建，旨在改善低资源语言的跨语言实体识别。XLEnt不仅支持机器翻译和跨语言信息链接，还为多语言自然语言处理任务提供重要资源，有助于推动全球语言技术的进步。

The XLEnt dataset was created by Johns Hopkins University and other institutions. It contains 164 million cross-lingual entity pairs, covering cross-lingual correspondences between English and 120 languages. This dataset was constructed by automatically mining web data and leveraging the lexical-phonetic alignment technique LSP-Align, aiming to improve cross-lingual entity recognition for low-resource languages. Besides supporting machine translation and cross-lingual information linking, XLEnt also serves as a critical resource for multilingual natural language processing tasks, contributing to the advancement of global language technology.