From: jamie@chaplin.csd.uwo.ca (J. Blustein)
Newsgroups: alt.hypertext
Subject: Re: Research or Thinking on Classification of Link Types
Date: 24 Nov 1996 00:33:45 GMT
Organization: Computer Science Dept., Univ. of Western Ontario, London, Canada
Lines: 139
Message-ID: <578559$82o@falcon.ccs.uwo.ca>
References: <3ngfi9$ilk@nrcnet0.nrc.ca> <3nghia$341@Venus.mcs.com> \
<D7LELs.7E6@world.std.com> <D7Jn5C.3AE@world.std.com> \
<D7LDoM.KJ2@world.std.com> <3nj781$ak1@Venus.mcs.com> \
<SHIPMAN.95Apr25151048@hrl9.cs.tamu.edu> <3nluts$5ve@Venus.mcs.com>\
<SHIPMAN.95Apr26160354@hrl9.cs.tamu.edu> <276996@tpi.win.net> \
<3156425A.18BD@passage.com> \
<Pine.SUN.3.90.960325103636.2355A-100000@viking.cris.com> \
<3159CDAF.2781E494@adobe.com> <3296F188.24EED832@emanon.net>
Reply-To: jamie@csd.uwo.ca
Summary: A summary: refs to online docs, quotes from previous articles and
some specific advice about HTML.
Keywords: link-type summary
X-Copyright: copyright (c) J. Blustein, 1996. All rights reserved.
Disclaimer: It's people like you what cause unrest!
In article <3296F188.24EED832@emanon.net> Dan Birchall
<shag@emanon.net> wrote:
>Chris Despopoulos wrote:
>
>> I have always thought there could be a way to describe the
>> intent of a link. For example, one intent might be to see the
>> voting record for a given bill, while the other might be to see
>> the economic effect should the bill pass. Why should there be
>> two links to that bill? Why not have one link that executes
>> differently depending on my intent? Then it's simply up to
>> the interface to somehow capture my desired intent.
If you don't insist on using the markup itself then you can
describe the purpose of the links with text before, or near them.
Ruud van Meer wrote a thesis about this. There are some notes about
it at http://www.knot.nl/users/ruud/ .
I would be remiss if I did not also mention James Allan's PhD
dissertation (Cornell tech report TR95-1484) _Automatic Hypertext
Construction_) and his paper at HT '96.
Here are some quotes from old alt.hypertext articles about link typing
that might be of interest. At the end of this message I some suggestions
for how to encode link types in HTML. But if this discussion is going to
be about HTML and the WWW only then it would be more appropriate in one of
the comp.infosystems.www newsfroups.
In article <276996@tpi.win.net> ttrickel@tpi.win.net (Thomas Trickel)
wrote:
<First off I'd like to say that this group is alt.hypertext
<and not comp.infosystems.www as a result it's possible
<that some of the discussion in this group may have no
<applicability to the WWW. :-) (a thinly veiled reference to
<the "SYMPOSIUM: HYPERTEXT THAT MATTERS" discussion).
<
<Second I'd like to say that I've finally managed to compile
<a discussion that took place in this group quite a while
<ago. The discussion was started with my question about
<link types supported by various Hypertext programs. The
<result was quite enlightening to me and changed my
<perception from links being mere pointers to resources to
<links being complex constructs. My compilation of the
<discussion is available via the URL:
<http://www.win.net/~tpi/hypertext/usentlnk.htm.
[NOTE: the files have moved and are now at http://trickel.org/thomas/hypertext/usentlnk.htm
-- (J. Blustein/10 Novemeber 2009)]
< <A related (short) paper on link properties that I've just <completed is also available via URL: <http://www.win.net/~tpi/hypertext/linkprop.htm. The paper
[NOTE: the files have moved and are now at http://trickel.org/thomas/hypertext/linkprop.htm
-- (J. Blustein/10 November 2009)]
<
<proposes three link properties:
<
< 1.The action the link causes in the user interface.
< 2.The resource the link points to.
< 3.The way two, or more, nodes are related.
<
<and looks at how typical WWW and Gopher clients implement
<these properties. I then discuss how WWW and Gopher clients
<could add a relationship property to their links and conclude
<by saying that links need to become smarter to allow easier
<classification.
In article <D7LELs.7E6@world.std.com> Eastgate@world.std.com (Mark
Bernstein) wrote:
<>Alain Desilets <alain@ai.iit.nrc.ca> asks:
<>>Does anyone know of a study about the various types of hyperlinks
<>>people create when they write hypertext?
[...]
<Actually, there has been a good deal of excellent research in this area,
<beginning with Randall Trigg's classic dissertation (c. 1986) on TEXTNET,
<a hypertext system with a detailed typology of links. (See also the
<NoteCards papers by Halasz, Trigg, et al. in CHI 87 and Hypertext 87),
<and the Hypertext 89 papers by DeRose (a typology of links) and by H. van
<Dyke Parunak. Landow's early rhetoric of arrival and departure is
<relevant here too (Hypertext 87; see also parts of his 1994 monograph
<HYPERTEXT: THE CONVERGENCE OF CONTEMPORARY CRITICAL THEORY AND TECHNOLOGY).
<
<Elaborate link typing gradually fell out of favor circa 1989, in part
<because Trigg argued that users were actually inhibited by the link types
<-- instead of gaining expressive power from typed links, users committed
<prematurely to a type declaration that was too restrictive or opted for a
<very broad declaration that became almost untyped.
<
<Link type strategies depend, too, on the expectation that links have a
<single meaning -- that a link serves *one* specific purpose. This
<assumption was questioned by Joyce (see his new collection of essays, OF
<TWO MINDS) and Bernstein et al (Hypertext 91). Multivalent links -- links
<that carry multiple meanings and play multiple roles -- seem to be of
<great importance in hypertext fiction, and their role in non-fiction
<writing is increasing (see David Kolb's SOCRATES IN THE LABYRINTH;
<HYPERTEXT, ARGUMENT, PHILOSOPHY; also George Landow's newest hypertext,
<WRITING AT THE EDGE, for examples).
<
<Typing has recently reappeared in a new generation of experimental
<systems that develop structure incrementally -- often by deploying agents
<to discover structure autonomously as the document grows and changes. See
<especially Streitz's DOLPHIN, Nanard and Nanard's MacWeb (the progenitor
<of this family), and Marshall's VIKI -- all described in Hypertext 93
<and/or ECHT 94.
In article <SHIPMAN.95Apr26160354@hrl9.cs.tamu.edu>
shipman@hrl9.cs.tamu.edu (Frank Shipman) wrote:
<Link-typologies go back a long way prior to hypertext and can only be
<considered irrelevant in relation to a particular task. Toulmin's
<representation for arguments and Rittel's Issue-Based Information
<Systems structure for design information can be applied successfully
<for certain purposes, but they are not general purpose representation
<schemes applicable to the loose form of communication currently seen
<on the Web. It is debatable whether there is a single set of link
<types that can be generally applied in all cases -- experience
<indicates that for many purposes, people prefer to leave such
<information implicit and reject attempts by systems that coerce them
<to do otherwise.
Subject: HTML only
If you only want to do use link types or link identifiers within HTML
then you should:
(1) consider the following attributes of the A element:
(a) TITLE
I don't think many browsers use it (I know of only one) but
it fits the bill.
(b) REL and REV
If you are interested in link types in HTML then look at
http://www.csd.uwo.ca/~jamie/.Refs/LinkTypes/README.html .
(2) use dynamically generated HTML (e.g., CGI, Java, Penguin) as Dan
Birchall suggested;
(3) use graphics, etc. as a form of advance organizer (see reference to
van Meer's work above);
(4) take this discussion to comp.infosystems.www.<mumble>. (However, if
the discussion is meant to be more general then by all means let's
discuss it in alt.hypertext.)
--
J. Blustein http://www.csd.uwo.ca/~jamie <jamie@csd.uwo.ca>
`We are all in the gutter, but some of us are looking at the stars'
-- Oscar Wilde
This document is copyright by its author, J. Blustein.