| >> |
!EIlelCfNmI 01/19/12(Thu)11:48 No.33403511>>33403383
Nah. Rocket science is what I should do to school right now instead of sitting in MLPG.
This is easy. Imagine it like that: there are cities connected by roads, and for each outgoing road, there is a probability you will go that given road.
That's what this generator does; the cities are words, and the probabilities are just counted via going through the MLPG archive - I counted ALL the "bigrams" (two words near each other), so I know that after word "Pinkie", there is most likely to be the word "Pie", so I go to the word "Pie"; once I am in the word "Pie", though, I ignore the word "Pinkie" and just search for word that is most likely to be after "Pie". And so on, until I reach the end (the end is a special word).
...yeah, it probably sounds still too complicated, but it really isn't. The point is, when I reach some word, I ignore all the previous ones, so it's doable with normal memory and still produces somehow believable results. |
