Article Summary, NIEL90a, 5604n, Group 5
	Shirley Carr
	Mike Joyce
	Bushra Khan
	Zakia Khan
	Vas Madhava
         Article Summary: (NIEL90A) ART OF NAVIGATING HYPERTEXT

        As hypertext becomes more popular, especially in dealing with
large amounts of loosely structured information (such as on-line
documentation) a major problem is user-disorientation.  Studies have
shown that 56% of users were confused on "where they were."  The basis
of this article was to investigate the various user interface options.
The hypertext report was done on a Mac and based on a report of events
that occurred in a workshop.

        With usability testing, the goal was to try out as many
hypertext options as possible.  Thus "discount usability engineering"
was used.  With this method, fast feedback is gotten from the users
and the system is modified very quickly.  This way many iterations of
the system can be tested.  The negative aspect of this approach is
that statistical tests on the different interface designs can't be
done.

        A hypertext system has two levels of navigation: linear and
hypertext.  The former lets one go back and forth as with normal text
pages while the latter lets you jump between various elements of the
document.  One of the problems with a hypertext system is the
homogeneity problem.  On-line text always looks the same and this
makes it difficult for users to recognize information and get a sense
of location.  This was overcome in this system by using various fonts
and screen layouts.  Some people argue for homogeneity.  They say that
electronic text should be consistent with printed text.  But the
authors rebut by stating that people read text on paper and the screen
differently.  They do agree, however, that the interfaces need to be
consistent.

        Overview diagrams were used as navigational aids.  Here they
used two levels, although a particular application might need more or
less.  The top level diagram should not have all the sub links because
it would make it too busy.  Ideally, the diagrams should be generated
using the users relevance judgements.  One way to do this would be to
keep histories of how people have traversed the system and use this
information to generate the overview diagram automatically.

        Backtracking was found to be one of the most important
features of the system.  Many people also used the "landmark" feature
to go to the home screen.  The history feature listed all the nodes
which a person has visited.  Structure-oriented backtracking, in which
a person moved back a certain number of levels, was removed because it
was too fancy and people rarely used it.

        The interaction history was maintained in two ways: All
movements between screens were timestamped.  Also, the amount of time
spent at each location was recorded.  Footprints were used by placing
checkmarks at all the places a user has visited.  But the problem with
this is that eventually all the locations will have checkmarks.  To
overcome this, the checkmarks could be made to "decay" over time.
Another issue related to this is whether to use two different types of
checkmarks, one for the sites a person has activated and one for sites
a person has seen but not activated.  This system used one checkmark
for both.

        Another issue is context-in-the-large versus
context-in-the-small.  The former refers to a user's context within
the entire structure while the latter refers to the context within a
small area of it.  The former can be addressed by navigational aids.
The latter occurs when there is not enough screen space to give the
user a full overview of the text and therefore the text needs to be
broken up.  How this text needs to be broken up becomes problematic.
A good system needs to address both issues.

        The system the authors used was developed on Hypercard using
HyperTalk, an object oriented language.  They had many problems with
it and had to resort to dirty tricks to get it to work right.  The
amount to time needed to develop the various versions of the system
was rather small.  But this was probably because the author was an
experienced HyperTalk programmer.

        The authors conclude by stating that although hypertext
systems are great for certain applications they do have serious
usability problems that need to be addressed.

================================================================

HT Article Summaries by Group I: Fitzgerald, Kalafut, Klein, and Muhlenburg.
          "Through Hypertext" by Jakob Nielsen

     Hypertext is non-sequential writing with direct-link-connected
nodes each containing a smaller part called an anchor.  When a user
activates an anchor, they navigate to its destination node.  More
prominent anchors are called landmarks.  Hypertext is often used for
on-line presentations or general reading.  Usability, especially user
disorientation, is the main concern of hypertext systems.  This
article presents a prototype hypertext system while discussing
usability issues.
     Like the article itself, hypertext has no strict sequential
order.  An author can only suggest a path for novice users while
allowing experts to navigate freely.  The prototype hypertext system
was developed using "discount usability engineering" for rapid
feedback to fix usability problems.  Navigation facilities are either
linear, like going forward and backward in page numbers in a book, or
non- linear, like zooming in and out of hypertext nodes.
     There are several usability issues with hypertext systems.  The
homogeneity problem of all information looking the same can be
facilitated with different graphical designs for different types of
information.  The navigational aids of both coarse and fine overview
diagrams are important to large hypertext systems to try to help steer
the user toward their interests using similarities between links and
multiple user histories.  Though some alternative backtracking
methods, like structured backtracking, have proved useful for experts,
the simple and consistent method of reverse path-following,
implemented with a stack, is the overall best solution especially for
novice users.  Interaction history; including time stamps, time
accumulations, footprints with check marks, and the fading of those
check marks over time, also help the user to navigate through a
hypertext.  Context-in- the-large is usually doable, while the
context-in-the-small problem, like explanatory text being too large to
accompany a figure on the same screen, remains a difficult hypertext
issue.
     The prototype was developed with HyperCard using Hypertalk with
the main purpose of understanding user interface issues of hypertext.
To repeat, the article itself was organized like hypertext with
non-sequential but still significant additions and interjections.
Perhaps the author's aim was to give the reader the feeling that they
had been reading hypertext.  To conclude, hypertext still has several
usability problems, the most difficult being context-in-the-small.