I’m not entirely sure what this article is about, but the start of the title sounds very intriguing. There are times when I think that some paper I’ve just reviewed was written by a perl script, using text cut-and-pasted (or only slightly permuted) from the PI’s previous work. In fact, that would be a relatively fun project: use google scholar to find a molecule at random from the literature. You could then automate a set of electronic structure calculations on this molecule for a wide selection of basis sets and levels of theory, automatically tabulate the energy data or some incorrect solvation free energies, and then automatically generate the paper.
(It should be obvious that I’m not serious. I work on methods development and not applications. Applications-only papers come out 10 times as fast as methods papers, and this fact tends to annoy methods people.)
Literature-based Generation of Hypotheses. Otherwise known as getting an idea in journal club.