The dataset contains a selection of works of art stored in Wikidata. It has been extracted via Wikidata API with action "wbgetentities"in date 01/05/2022. The dataset comprises literary productions, visual artworks, cultural sites, operatic works, performing arts, journals etc. We semiautomatically surveyed the existing Wikidata subclasses of work of art and removed all elements belonging to music and cinematographic domains as well as statistical rumor. In total, the dataset contains 2'156'363 works of art (wikidata entities) described by 18'924'333 statements and belonging to 6'612 selected classes. The dataset is stored in a zip folder organised in 10'782 json files, each storing 200 entities and their metadata.