From: Hans Aberg (haberg@math.su.se)
Date: Thu Sep 21 2006 - 08:32:10 CDT
On 21 Sep 2006, at 14:26, Doug Ewell wrote:
> Don't forget you need to store the frequency table along with the
> compressed data, so the reader can reconstruct the table. That
> could mitigate your compression somewhat.
One can turn this problem around: First do the statistical analysis,
then choose a translation table that admits a compact representation,
which is not necessarily the one that gives best compression of the
text body. Then from this point of view, a byte-compression scheme
just chooses a more limited set of translation tables. (A binary
translation table can be viewed, at least roughly so :-), as a code
point translation table via suitable character encodings.) So when
playing around in this code point compression picture, one has access
to everything that can be done in the byte-compression picture.
Hans Aberg
This archive was generated by hypermail 2.1.5 : Thu Sep 21 2006 - 08:35:23 CDT