PolyNarrative: a multilingual, multilabel, multi-domain dataset for narrative extraction from news articles

Abstract

We present PolyNarrative, a new multilingual dataset of news articles, annotated for narra- tives. Narratives are overt or implicit claims, recurring across articles and languages, promot- ing a specific interpretation or viewpoint on an ongoing topic, often propagating mis/disinfor- mation. We developed two-level taxonomies with coarse- and fine-grained narrative labels for two domains: (i) climate change and (ii) the military conflict between Ukraine and Russia. We collected news articles in four languages (Bulgarian, English, Portuguese, and Russian) related to the two domains and manually anno- tated them at the paragraph level. We make the dataset publicly available, along with experi- mental results of several strong baselines that assign narrative labels to news articles at the paragraph or the document level. We believe that this dataset will foster research in narrative detection and enable new research directions to- wards more multi-domain and highly granular narrative related tasks.