This is a really promising avenue. I can see how bioinformatics techniques could be used to identify passages as if they were genes. So some of the hard problems are solved. But the kind of knowledge that bioinformatics produces is highly statistical and one of the biggest challenges will be reducing the results of, for instance, combinatorial pattern matching. It seems like they're taking the right approach in identifying which phrases are common, essentially building the Genbank of legalese.
