I wrote and defended a thesis proposal on this topic in which the goal was to automate, to the greatest degree possible, merging nodes which mean the same thing. One example is "CO2 causes climate change" and "Climate change is caused by carbon dioxide", both saying the same thing, but neither of which would submit to a simple string comparison. As it turns out, there are Watson-like agents beginning to emerge in the open source arena, including OpenSherlock, my project, which is taking forever to complete owing to the enormous amount of experimental work that must be accomplished.
I'm a bit baffled about this:
> merge argument maps, efficiently unifying nodes that are semantically similar
Is the merging done automatically? If so, this together with some other comments in the thread seems to imply that the software has an ontology mapped onto the arguments—which to my knowledge was so far a pipe dream for most people, on the order a bit below generic AI and somewhere around Watson. Am I mistaken, is there anything practical in this space? Or does the semantic mapping just mean lotsa handiwork?