Talk:Semantic compression

Hex to binary
Hex to binary reduces the symbol space by a factor of eight, but isn't lossy unless the binary:hex digit ratio is less than 4:1.

Semantic compression is a lossy compression, that is, some data is being discarded, and an original document cannot be reconstructed in a reverse process.

The unstated implication of the lead is that one of the goals of semantic compression is to reduce the vocabulary without suffering a corresponding increase in prolixity.

It would still be lossy in the prolix regime due to inexacitude of the 1:1 mapping function, rather than lossy as a result of heading deep into many:1 mapping space (a kind of semantic bandpass filter).

None of this is explicit in the lead as it presently stands. &mdash; MaxEnt 19:55, 13 July 2018 (UTC)


 * I decided to take a flier:

In most applications, semantic compression is a lossy compression, that is, increased prolixity does not compensate for the lexical compression, and an original document cannot be reconstructed in a reverse process.


 * But I'd rather another editor came along and did this properly, in full scope. &mdash; MaxEnt 19:59, 13 July 2018 (UTC)

"a targed lexicon" > "a target lexicon" or "a targeted lexicon" ?
"a targed lexicon" > "a target lexicon" or "a targeted lexicon" ? R. Henrik Nilsson (talk) 09:50, 1 October 2023 (UTC)