|
Faceted interfaces |
Why
do we create hierachies? |
We need hierarchies to create an overview. For example: we can
point to a lot of countries by calling them 'Europe'. Another example:
we can point to a lot of equipment by calling them 'tools'. |
The problem with hierachies |
However, there is a problem with hierarchies. Sometimes we are
not sure what is meant by the more general term. Our information
might be under there but on the other hand, it might not. For example,
I am looking for a manual and the site says: "publications"
but also "support". Where should I look?
Another problem arises when the person who organizes the information,
thinks in a different way than the reader. For example, a supermarket
organized its products according to the type of meal the consumer
wants to make. However, consumers expected the supermarket to be
organized according to the type of products.
That's when hierachies do not help but actually cause a problem
instead. |
|
So, we need an overview. However, we do not always want a hierarchy
because the hierachy does not really 'fit'. Is there an alternative?
Yes, the alternative is so-called faceted metadata.
Like the facets on the diamond, faceted metadata allow you to look
from different sides to the same information. Faceted
metadata have the information organized on several dimensions (facets).
The information has values on each of the facets. Readers find information
by chosing values on each of the facets. |
|
Faceted interfaces have the following attributes
- the interface shows several facets
- behind each value it says how many 'information elements' are
still left when the reader makes this choice
- each choice changes the entire view:
- the search result is updated to show the information elements
that still comply to the choices made
- all the facets and choices on those facets are updated to
show the choices that still can be made. Also, the values
behind the choices are updated to show the number of information
elements that will be left when that choice is made.
- all visible choices contain information elements. Choices
that lead to no information elements, are not shown.
- a facet can have a hierachy. For example: continents > countries
> cities
- when facets have too many values, some kind of summary is displayed.
For example ranges or a groups of values
The most important difference between the different faceted interfaces
is whether an information element can have multiple values on one
facet or not. |
Examples |
The most sophisticated interface comes from the University of
Berkeley where professor Marti Hearst runs usability research on
faceted interfaces. The project is called the Flamenco
project.

A webshop using a faceted interface: bizrate

Facetmap.com
has a wine demo:

A faceted interface where the facets are depicted as sliders comes
from shopping.yahoo.com
|
Another alternative to hierarchies |
Another alternative for hierachies are so-called thesauri. A
thesaurus shows:
- what is related (to a term)
- the type of relation
- sometimes also the strength of the relation (e.g. slightly related,
somewhat related, strongly related)
|
Examples of thesauri |
Touchgraph
uses the "similar pages"-function of Google to display
what is related:
In the Netherlands we have the AquaBrowser:
|
The future: Topic Maps |
Topic Maps is another standard also based on XML.
Topic
Maps have:
- topics
- relations (between the topics)
- instances
Topic maps
allow us both to build thesauri and faceted metadata. However, we
still have to wait for the interface that benefits from browsing
through Topic Maps. The Omnigator from Ontopia is a Topic Map browser.
However, the Omnigator does not enable the user much more than any
other website. |