> When viewing an HTML document, I find it a common experience to want
> to print out part of the web, not just this particular document.
> Supporting a model of how retrievable chunks form a larger whole would
> permit this.
I think there are two aspects to this:
(A) Identifying a subset of the Web - a number of asssociated HTML
documents (and other non-text data).
(B) Processing this set in some way - eg printing it, indexing it,
searching it, perhaps mailing it to someone while preserving the
links within the set, etc.
For (A), the set could be determined by the author (eg, someone who
publishes a manual on the web as a set of HTML documents).
The set could also be determined by the reader, who may say "Include all
documents no more than N links away from this one". Or more generally,
you could add an attribute to a link giving a "relevance number"
decrement: the user could specify an initial relevance count, and the
set would be all documents which could be reached from the currently
displayed one without reducing the relevance count to 0. I think
this is appropriate for much of the information presently on the Web,
and is more realistic than expecting readers to write their own TOCs.
The actual processing of the set of documents needs some extra
information - eg, printing requires sequence information (available from
a TOC). In some ways, it would be nice to separate the subsetting (A)
from the extra information needed for processing (B). But its not
obvious how to achieve that.
How about more than one level of TOCs? A manual chapter could have its
own TOC, which would be linked to from the overall manual TOC. Many
Microsoft programming manuals are structured like this. This would
shorten the main TOC and would also give the reader the option of
printing out the whole manual or just one chapter.
> The results so far are very encouraging. Provided you don't mind seeing
> work at an early and incomplete stage, you may be interested in looking
> at "ftp://hplose.hpl.hp.com/pub/dtd2.html". I would be very interested
> in collaborating on this work.
I would like to help - but hplose.hpl.hp.com won't cooperate: it says
"Connection timed out". Can you mail the DTD?
Chris Adie Phone: +44 31 650 3363
Edinburgh University Computing Service Fax: +44 31 662 4809
University Library, George Square Email: C.J.Adie@edinburgh.ac.uk
Edinburgh EH8 9LJ, United Kingdom