The Ruminator now knows about words and phrases. From now on, it can tell you about the “district court”, “helen clark”, and the “human rights commission”, instead of jumbling those words up into clark commission court district helen human rights. This took a bit of work, because even the best statistical analysis produces somewhat dodgy results. I have jiggered things so that I get a list of suggestions with each day’s run, but it’s just not safe to automate it.

I have also written a bit more about how the whole thing works, and where I need help.