> In Dr Lichtarge’s view, hypothesis-generation software works in part because science writing tends to be free of humour, sarcasm and “emotive or literary overlay” that could trip it up.
This looks like a decent argument to create some sort of publication language for science that describes hypotheses, methods, results and their linkages. Something a bit more formal and less varied that English that could be easily processed.
This is a very interesting idea. We're already living in times where volume of information grows exponentially almost every year, and I believe that "filters" such as/similar to presented in the article will become more and more important part of our civilization. Partially, Google/HN are such "filters", selecting pieces of information that have some value for us.
These days I'm thinking about AI/"Singularity" -- I noticed that I envisioned Singularity as a single point in time, but now I think it's currently happening (like work on semiconductors or Turing's papers were the begining of information age). Why? Well, we're asking machines to give us clues on what to work on. It's not a formal argument but it's hard to provide one when it comes to predicting the future.
This looks like a decent argument to create some sort of publication language for science that describes hypotheses, methods, results and their linkages. Something a bit more formal and less varied that English that could be easily processed.