The tool is intended to provide a
method for quickly navigating a website whose contents lend themselves to a
hierarchal layout. The star tree representation allows a large hierarchy to be
traversed while maximizing the use of the display space. It's major application
is to websites with a significant amount of content, and the most interesting
method used is the star tree representation.
The first time the software is
installed, it appears that it is on the website administrator to fill it with
knowledge. This can no doubt be automated through the use of scripts and what
not, since it's a matter of replicating the hierarchy of the web site in
whatever format the Star Tree tool uses. Presumably, web site administrators are
the only people with the access needed to add knowledge to the system. The
effort to originally populate and add new knowledge is dependent on the initial
time spent automating the process. New knowledge is readily added to the system,
unless it changes the hierarchy of the existing knowledge somewhere other than
the leaves, although again this is related to the effort put forth automating
the process.
The underlying organization is not
revealed by the company's site or the demos, but I would speculate it is
hierarchal in nature, reflecting its visualization. The objects representing the
knowledge are probably dynamic in the sense that additional nodes can be hung
off of them with relative ease.
The visualization scheme utilized by
the Star Tree Tool is called, amusingly enough, the star tree. The display is
represented by nodes and links, or vertices and edges, if you prefer. The user
can pan the view by clicking and dragging, bringing in pieces of the tree that
are not currently visible, or they can click on nodes to center the display on
the node. The centered node is the focus of user attention, and the links off of
that node hint at the quantity of children nodes, without distracting the user.
The system comes with a set of ways in which the display can be enhanced:
animation, node coloring, link coloring, etc. Used wisely these can make the
display more clear, used poorly they would certainly make it worse. This
visualization technique does a good job allowing the user to view a reasonable
subset of a large hierarchy while maintaining their sense of context within the
hierarchy. The most significant problem I observed in this tool's visualization
interface is that the node's textual labels are often cut off as the node's
textbox is resized in accordance to tree panning. This may be the fault of the
customer's implementation or an overall problem of the software, but I believe
it is significant enough to warrant investigation.
The visualization interface acts as
the platform for the retrieval of knowledge (as it should). The user can
double-click on a particular node to relocate to the web content associated with
it (or whatever is associated with it by the administrators), and they can also
opt for a midrange if the administrators have decided to implement it, which
basically consists of a thumbnail image or some textual data about the node's
associated content. This fits well with the overall scheme being used: the use
first navigates the hierarchy to an appropriate-sounding node, and then follows
that node "down" to the actual content.
Although Inxight's webpage did not
appear to make explicit mention of the ability to monitor knowledge usage, it
strikes me as a feasible feature to implement by the customer or by the company.
The company could track tree traversals and content viewing, and use that
knowledge to revise and optimize their hierarchy.