In this paper we will be looking at 2 and 1/2 classification systems*, where they excel, where they fail and how they measure up to our ideas about WhyDoPeopleOrganize and the PrinciplesOfGrok.

Hierarchy An organizational structure, often referred to as a tree, comprised of items and containers of items where both items and containers are organized into a fixed structure of parent-child relationships. As a result,

1. items and containers can only live in a single location in the tree
2. Hierarchies can only express two kinds of relationships: Parent-child and Sibling. Examples include: File system, IMAP email folders, Dewey Decimal System

Faceted classification system An anti-organization, classification system comprised of items and independent Facets or Attributes and Attribute values.

Items can be thought of as object nouns or things (ie. a person, a document, a song, a message). Attribute values are adjectives or characteristics of the noun-items (ie. red, 56KB, 36 years-old) and Attributes describe the characteristics, provide them with specific semantics (ie. hair color, file size, age). A natural language statement such as "Bilbo has furry feet." might be stored in a Faceted system in the following way: The item Bilbo has Feet covering: Fur.

Items in Faceted systems are expected to be multi-faceted, to have a potentially unlimited number of Attribute: Attribute value pairs (ie. Height: 6 feet, Weight: 135 lbs, Age: 35 years-old, Gender: Female, Marital status: Single). This seems beyond obvious, but the fall out of multi-faceted items is that unlike their counterparts in Hierarchies, items in Faceted systems can live in more than one Attribute-based groupings (ie. Six foot tall people, 135lb-ers, 35 year-olds, Single people). Depending on the system, a single item could also belong to multiple Attribute values in the same Facet (ie. a song that is belongs to both the R&B and Hip-hop Genres). This is an item-centric way of describing Faceted systems.

A more eagle-eye perspective reveals the flat, a-structural nature of Faceted systems, where there is only a single level of fixed Parent-child relationships. Attribute values are grouped by Attribute. Attributes can have multiple Attribute values. But a particular Attribute value can only belong to one Attribute. For example, the Attribute Hair color: can be paired with any of a number of different Attribute values: Red, Blonde, Brunette, Black. But Black as an Attribute value for Hair color is unique and is different from Black as an Attribute value for the Attribute Mood.

This single layer of organization is what makes Faceted systems, classification systems. Items aren't just described by characteristics. The characteristics themselves are described or typed or classified into Facets. In this way, Faceted systems are actually more orderly than Hierarchies, which do not attempt to classify their containers into container types. In Hierarchies, all containers are created equal and have no specific semantics (ie. Hair color container or a Marital status container).

However, beyond the single layer of fixed Parent-child relationships between Attributes and Attribute values however, the items lie in a soup. Faceted systems don't try to organize data into structures the way Hierarchies do, they simply attempt to classify items by characteristics or Facets that have no structural, organizational relationship to one another, Parent-child, Sibling or any other kind.

In Faceted systems, there are no explicit relationships or relationship types between items or groups of items, only generic incidental "these are related" relationships:

1. Items are related to each other if they have they share the same Attribute:Attribute value pairs (ie. Six foot tall people)
2. Attribute:Attribute value-based groupings are related to each other if they share the same items (ie. 15 year-olds also happen to be mostly Single).

Examples include: Chandler data model, Spreadsheets and tables and iTunes / iPod. The Browser in iTunes is an example of a Browser UI for Faceted systems.

Tagsonomy The anti-classification, anti-system comprised of items and independent labels or tags (the equivalent of Attribute values in Faceted systems), but no label types or tag types (or Attributes). Tagsonomies do not attempt to organize data into a structure of relationships nor do they try to classify tags into Facets. Istead, items are merely described by a-semantic labels (ie. the tag "football" on a picture could mean that the photo is a picture of a football, a football game, a guinea pig named football, etc..)

Similar to Faceted systems and unlike Hierarchies, Tag-based groupings are simply related to one another only insofar as they share some of the same items. Or conversely, two items can be said to be related because they share some of the same tags. The more items two tags share, the more related the tags are to each other. The more tags two items share, the more related the items are to each other.

In sum, Tagsonomies are the bastard children of Hierarchies and Facets. They do away with the organizing and classifying aspects and retain the generic, a-semantic, a-structural qualities of these two systems, the end result of which is the most free-form anti-system possible.

Examples include: del.icio.us, flickr, Gmail.

	Hierarchy	Facets	Tags
Grouping mechanism	Containers	Attribute value-based groupings	Tag-based groupings
Classification	None. Hierarchy containers are generic.	Attribute value-based groupings are classified by Attribute.	None. Tag-based groupings are generic.
Structure	Strict, fixed structure, capable of growing infinitely wide and infinitely deep	1 level of fixed structure of Attributes and Attribute values. No structure beyond that.	No structure whatsoever
Relationships	Parent-child and Sibling	Incidental, generic "this is related" relationships resulting from overlapping Attribute-based groupings that share the same item and items that share the same Attribute: Attribute value pairs.	Incidental, generic "this is related" relationships resulting from overlapping Tag-based groupings that share the same item and items that share the same Tags.

• 3-systems.png:
  

Caveat These systems do not exist in mutual exclusive isolation. In fact all of the systems I've given are hybrids of at least 2 of the 2 and 1/2 systems described here.

The moral of the story

In the end this paper is going to be a story of the Second Law of Thermodynamics** as it applies to human efforts to represent knowledge in ontologies or classification systems...But with a twist...Because knowledge rarely exists in a closed system, because all things are never equal, the best laid plans for orderly arrangements of information disintegrate into disorder. The very structures we build to make Sense of our data turns into Nonsense when, as The Dude (aka Jeffrey Lebowski) would say, "New sh#$ has come to light! Man." And in our unending quest for more information and more knowledge, "New sh#$ is always coming to light...Man" faster than our stiff-legged classification systems can handle. As a result, the systems ultimately reject the new data and what we're left with is a regurgitated mess of bits and pieces strewn across the landscape of your variegated and uncoordinated information gathering and storage devices: email clients, web mail clients, documents, pdas, paper calendars, sticky notes, notepads, envelopes, napkins, the back of your hand and last but not least, your brain.

*We're treating Tagging systems as a variant of Faceted systems given that Tags can be considered just another Facet of an item.

**The Second Law of Thermodynamics states that in a closed system, entropy or the measure of disorder, always increases. In other words, the end is nigh, so stop filing your email!