Another use case for Sci-Hub is bulk downloads. This is especially important for machine learning. Building a dataset of academic papers across domains would be

Another use case for Sci-Hub is bulk downloads. This is especially important for machine learning. Building a dataset of academic papers across domains would be a insurmountable task, even with APIs like PubMed and using "interal" university networks with good journal access. Rather than paste all the details here, take a look at the opendata stackexchange site were we share details about bulk downloads from sci-hub, including torrents: https://opendata.stackexchange.com/questions/7084/bulk-download-sci-hub-papers https://opendata.stackexchange.com/questions/7084/bulk-downl... Note that some other publishing platforms have bulk access, such as arxiv.org: https://arxiv.org/help/bulk_data https://arxiv.org/help/bulk_data And platforms like https://unpaywall.org https://unpaywall.org try to link open access papers. - And if you are a researcher sitting on your PDFs, consider self-hosting, and making an archive.org backup. These will get indexed by Google Scholar. I did so for my papers, and they were indexed in a couple weeks: https://smalldata.dev/posts/open-access-research/ https://smalldata.dev/posts/open-access-research/