Names from a high order Markov Process and a simplified Katz back-off scheme

From RogueBasin
Revision as of 17:47, 14 March 2012 by Jlund3 (talk | contribs) (Incomplete article, a preview of things to come.)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Various articles on RogueBasin have suggested that a Markov Process might be a good way to generate random names for a game. However, most of them have a weakness or two which can easily be solved. In this article, I will describe a basic first order Markov process seen in other articles on name generation, and discuss some of the short comings. I will then explain how to overcome these weakness using a higher order model, with a simplify Katz back-off scheme. This model is still fairly simple, but will allow you to come up with a data based name generator, even with a somewhat limited amount of data, which will produce pronounceable names with a similar feel to your training data. That is to say...I will be writing this article very soon! :)