Facets differentiate between container types.
Faceted systems on the other hand,
do differentiate between container type. Items are not only grouped around common characteristics (ie. Tall, Short, Wide, Narrow), but the characteristics themselves are typed or classified into facets (ie. Height and Girth).
However, Faceted systems like Hierarchies
do not differentiate between different kinds of relationships. Hierarchies have Parent-child and Sibling relationships, Facets have Sibling and Intersecting-Facets relationships.
Yet again, we run into the problem of an over generic, undifferentiated system that makes it hard for people to discern patterns and make groupings.
Examples of Facet relationships
You can model a whole range of relationships with Facet-based groupings.
- A group is wholly contained in another group: Genuine Parent-child (ie. The Ainu are a subset of all Indigenous peoples OR Sichuan is a province wholly contained inside of China.)
- A group partially overlaps another group in the same Facet: Genuine Siblings (ie. Economics and Sociology are two Areas of study that happen to have a lot of overlap in subject matter.)
- A group partially overlaps another group in a different Facet: Correlation (ie. There is a high correlation between Blue eyes and Blonde hair. There is a high correlation between Sichuanese people and the land area of Sichuan. But not all Sichuanese people live inside of Sichuan or even inside of China.)
However, relationships in Faceted systems are rarely presented in such discrete, coherent forms. You can usually tell the difference between groups in the same Facet that overlap versus groups in different Facets that overlap. But you cannot readily distinguish between:
- groups that cut across many different facets (ie. black hair can be found in most parts of the world) versus
- groups that only cut across a few different facets (ie. red hair is relatively rare and localized to only a few places in the world) versus
- groups that are wholly contained in other groups (ie. redheads are wholly contained in the Caucasian race group)
It is also hard to compare the relative degrees to which groups overlap each other. Few faceted systems give you such feedback. And even if they do (see
CocoonDev's faceted browser.), you can only view relative overlaps with respect to a single facet value at a time.
- Faceted browsers answer questions like: Which languages overlap the most with the Document type Attachments?
- NOT questions like: Are there more English language Attachments? or English language CodeSamples?
Ultimately however, it's up to your brain to:
- synthesize information you experience in temporal succession and then
- try and detect any patterns you may remember
Some people are better at doing this than others. Everyone has a limit as to how much information they can grok over a given period of time.
This is again because Faceted browsers like Hierarchies "fake chunking" by hiding data behind opaque containers rather than overlaying data on top of each other and using other visual mechanisms to chunk down information.
Introduction to Faceted browsers
Below are some examples of Faceted browsers. Most of them use the same underlying machinations. Users either select or are given a set of Attributes or Facets to manipulate. Selecting a value in one Facet changes the list of values in the other Facets such that you are only ever presented with overlapping groupings that will yield result sets.
For example, if I select the genre Hip-Hop in iTunes, I will only be presented with Artists and Albums that have songs in the Hip-Hop genre. That way, I am never led astray to select an Artist or Album that does not overlap with the Hip-Hop genre, thereby yielding a null result set.
- iTunesBrowser.png:
Some other faceted browsers:
Different variations on Faceted browsers include:
- Whether or not they allow you to re-order the Facets
- Whether the Facets you're presented with change as you select values in each Facet