A dataset consists of one or more files (e.g. organized in directories) and descriptions in the form of metadata. Datasets are created and edited by curators and subcurators in the workspace. With RADAR, datasets can either be archived or published. Each dataset is assigned an identifier (DOI or RADAR-ID). To create a new dataset, individual files as well as multiple files in compressed form, for example ZIP, TAR, GZIP or other, can be uploaded. Any directories are automatically adopted for the dataset. RADAR recommends the use of file formats suitable for long-term preservation.