> This justification doesn't sound terribly good. If we can't infer the
> meaning of the font changes, perhaps translating the scanned document to
> HTML is inappropriate; perhaps the scanned document should be stored in
> TeX or Postscript or GIF, depending on what other image characteristics
> of the document seem important. Or perhaps the font change information
> should simply be discarded on conversion to HTML, as it does not
> translate to meaning.
I think this is perhaps a bit extreme, and in practice most people would
prefer some preservation of italic emphasis etc. (an empirical question)
> Or perhaps the right way to think of these is to apply the output rules,
> in reverse. That is, the user can specify how <EMPH> is to be formatted
> on output; similarly, perhaps the user could specify how bold font is to
> be interpreted on input.
Perhaps we ought to use <EMPH> for all inline emphasis! This gets around
the problem in dealing with every growing categories for different needs.
You then would see elements like:
<EMPH tag="author">H. G. Wells</EMPH>
The Web could then have an evolving set of well known categories.
Unrecognised categories would simply be ignored by the browser. Users
would also be able to change how the browser renders each category.
Dave Raggett