There are loads of duplicate papers. Searching, for example. for "latent dirichlet allocation" shows the same JMLR paper about four times in the top results.