conv < total.dict then iii, nnn file will be generated.

in iii file, I deleted two word of which irregular information
is wrong.

conv-abbe < dict.abbre will generate eee file of which
including exceptional word, contraction-recovery information.
