Re: Navigational paradigms in very large hypertexts?

[Chris Despopoulos <cud -at- ARRAKIS -dot- ES>]

I'm aware of research in the area of paradigms for large
hypertexts, but I can't cite anything because I have long
since given up those materials.  I attended a few ACM
Hypertext conventions, and there were always papers that at
least touched upon that issue.  Some examples that come to
mind...

   * The OED project...  Putting the Oxford English
     Dictionary online.  The gist of that was explaining how
     the textual structure was translated into a database ER
     diagram, and how that model informed the UI.  As I
     recall, categorization was very important.  Look for
     Frank W. Tompa and Darryl Raymond (spelling not
     guaranteed) from U of Waterloo in Canadia,
     somehwere...  (Yes, I'm US American).
   * Taxonomic Rreasoning in Hypertext...  Pretty occult
     stuff, complete with equations to express
     categorization according to taxonomy rather than
     hierarchy.  I haven't a clue what it means.  Somebody
     named Parynak, or something like that???
   * Flying through Hypertext...  Well, this was an idea
     within the problem space of how to represent the web of
     a very large hypertext, and how to move through it.  In
     the early 90s, people thought it might be useful to
     generate and show maps (webs) of entire hypertexts.  I
     believe this proved practically useless...  If the
     hypertext was big enough to require that level of view,
     there was no display large enough to show it.  So the
     idea of flying was an alternative.  As I recall, you
     sort of paged through the web very quickly, looking for
     visual cues that made you want to slow down and pick a
     page.  From there, you might want to see a TOC.  Of
     academic interest, but I don't think anybody picked it
     up.  I cannot recall even a phoenetic guess at the
     guy's name.
   * I know layering has always been discussed, and for some
     reason it has become a favorite of mine.  Cannot recall
     any references.  But a TOC that can expand and contract
     strikes me as very useful.
   * Categorization seemed to rise up to the top of the test
     tube...Within a unit of information (whatever that is),
     presenting a consistent set of categories, and
     expressing them clearly is good.  Layering can work...
     Something like (maybe) tabs, or a sub-TOC for every
     discussion.

Anyway, look to the ACM, is my suggestion.

For my own opinion, layering and categorization go a long
way.  But I have not worked on huge hypertexts.  Maybe 1,000
pages or so is the max for me.  Nonetheless, one thing I
found is that it pays to avoid unnecessary depth...  To many
clicks to arrive at the desired information is bad.  Making
the navigation to a piece of information as direct and
consistent as possible is good.

As for setting information off via visual cues, I say do
it.  But I also suggest you hire a graphic designer to
implement it.  First determine the categories, and express
the likely ways in which they'll be used.  Then get a
graphic designer.  The problems that can arise online are
the same as with paper...  A poor graphic design comes off
as busy or unclean, and can actually detract from the value
of the text.

In fact, I found that the biggest difference between online
and print (aside from the tactile differences) is that
online text is more sensitive to the same problems that
occur in print.  Readability, organization, aesthetics...
you name it, if it is a concern in print, it is a concern
online - just more so.  Not that you need exquisite control
over display and kerning, etc.  But that in its own way,
online text is very sensitive.

And because of the tactile difference (or lack of tactile
information), you MUST ADD VALUE by going online...  not
just save bucks or be cool.  Well, that's in a perfect
world.  But for a huge text, I think you could argue that
point.  And hypertext adds value when it adds a rhetoric of
behavior.  The way you get to a page, and what you can do
with it also has meaning.  Choose specific behaviors and use
them as though they are design elements.

Yet another rant, brought to you by cud

 From ??? -at- ??? Sun Jan 00 00:00:00 0000=