Self-archiving is the practice of scholars making academic publications freely available online, most often on a personal site or in an institutional repository. However, there are also a number of disciplinary repositories such as arXiv.org and the Social Science Research Network and other websites, such as ResearchGate, Academia.edu and Mendeley that allow you to upload your papers to make them freely available. The versions of the articles uploaded are mostly pre-print or post-print (final versions after peer review) versions, and only rarely publisher pdf. Self-archiving is also known as Green Open Access and has grown in popularity with the advent of more and more funders, universities, and research institutes adopting an open access mandate.
Academia.edu vs Elsevier
At the end of 2013 one of the largest academic publishers, Elsevier, issued a flurry of DMCA take-down notices to Academia.edu summoning them to remove PDFs of articles published in their journals. There was a lot of upset in the academic community over this, partly stemming from a wider discussion on Open Access versus traditional publishing, partly from the feeling that the greedy publishing behemoth Elsevier is trying to have their cake and eat it too through also offering expensive (for the authors) Gold Open Access options. On the other side, Elsevier claims it was only protecting their (and their authors') best interests and focusing on publisher PDFs being uploaded. They also pointed out that most Elsevier journals allow for post-print self-archiving, just not the publisher PDFs.
So do ResearchGate and Academia.edu qualify as self-archiving locations?
ResearchGate, Academia.edu and Mendeley all encourage their users to upload their own papers, explicitly citing self-archiving and increased ease of access as motivation for doing so. The question is: do these academic social networks qualify as locations for self-archiving? The long and short of it is it all depends on the publisher and your copyright agreement. Some publishers allow archiving of the pre-print version on any website, some only on personal or institutional websites. Some allow post-print author versions to be archived, but only on personal and institutional sites and in institutional or disciplinary repositories. Whether, as ResearchGate argues, a profile on an academic social network can be defined as a personal website is debatable, as you are dependent on an external, third-party service, whose actions you can't control. The wisest course would be to err on the side of safety and ask for clarification from your publisher as to how and where you can self-archive your papers.