[MUD-Dev] Random Placename Generation

[Hans-Henrik Staerfeldt]

On Monday 05 April 2004 22:23, Matt Chatterley wrote:

> The best approach which I have so far formulated to this topic is
> to create lists of 'name fragments' and 'name types' - the 'type'
> (random) would dictate which sets of fragments could be used, and
> how those fragments should be assembled. This might reflect the
> type of government or civilization present upon the planet (or
> lack thereof).

I would propose you experiment with making a hidden markov model,
(order and complexity depending on your training set size), over
some known words which sound like you want your placenames to sound,
thus making the statistical measure of which letters follow
typically eachother in a placename sequence. You may even
incoorperate your beginning-middle-end model into the markov
model. Then ask the model to generate random words.  ... and as
someone else wrote; check for duplicates.... It works for genetic
sequences anyway :-)

--
--Hans-Henrik St=E6rfeldt
_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev