This is a tag that was introduced quite recently. Although the term seems to be quite accepted today, a few experts (one of which is an editor of an IEEE transaction journal) claim that is more of a misnomer.
Plagiarism generally means stealing the written work of another by copying it in your publishable work without necessary attribution of the source. But re-publishing your own work cannot be referred to as stealing. This can be recalled as double-publication or content-recycling. However due to widely used literature, the word 'self-plagiarism' do refer to double-publication.
Should we continue to keep the tag as it is or rename it to duplicate-publication and make any references to self-plagiarism suggest double-publication instead?