NatCat

홈페이지 · 논문 1편

A general purpose text categorization dataset (NatCat) from three online resources: Wikipedia, Reddit, and Stack Exchange. These datasets consist of document-category pairs derived from manual curation that occurs naturally by their communities. Source: [Natcat: Weakly Supervised Text Classification with Naturally Annotated Datasets](/paper/natcat-weakly-supervised-text-classification)