Jump to content

Talk:Sukhotin's algorithm

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Attention needed

[edit]
  • Expand - the article does not explain what it is, merely whaere it may be used and what category it falls into. Expand to explain what the algorithm is
  • References needed
Chaosdruid (talk) 23:47, 9 January 2011 (UTC)[reply]

The algorithm itself

[edit]

Hi I just read the sukhotin paper, the idea behind the algorithm is:

Hi I just read the sukhotin paper, the idea behind the algorithm is:

it is hard to pronounce a prolonged string of consonants CCCCCC... but few variation seems possible with vowels VVVVV... so most languages alternate vowels and consonants. Consider some function of co-occurence, i.e. being adjacent in a word.

Call D a possible division of letters into disjoint sets of vowels V and consonants C.

Call f(l_i,l_j) the frequency of co-occurence in some corpus.

Then we expect Q(D)=sum over all vowels according to D of (sum over all consonants according to D of (frequency of co-occurence)) should be a maximum.

He alludes to some optimizations but does not give them in the paper... — Preceding unsigned comment added by 83.134.178.100 (talk) 16:26, 26 February 2012 (UTC)[reply]