Multilingual Knowledge Questions and Answers (MKQA)

MKQA是一个多语言开放领域问答评估数据集，由苹果公司创建，包含26种语言的10,000个问题-答案对，总计260,000个问题-答案对。数据集的答案基于精心策划的、语言无关的数据表示，使得不同语言的结果可比较且独立于特定语言的段落。MKQA覆盖了广泛的语系分支，旨在为跨语言问答系统提供一个公平、可比较的评估平台，特别关注低资源语言的挑战。MKQA is a multilingual open-domain question answering evaluation dataset developed by Apple Inc. It consists of 10,000 question-answer pairs per language across 26 languages, totaling 260,000 question-answer pairs. The answers in the dataset are based on a carefully curated, language-agnostic data representation, which enables comparable results across different languages and is independent of language-specific paragraphs. MKQA covers a wide range of language families, and aims to provide a fair and comparable evaluation platform for cross-lingual question answering systems, with particular focus on the challenges faced by low-resource languages.