Taxonomies can take many forms

We ended our previous chapter by saying that taxonomies do not always need to be in a tree format. Our objective in this chapter is to understand the variety of forms that a taxonomy can take, and to appreciate the strengths and weaknesses of each. This grounding will give us added versatility in putting together our taxonomy development strategy later on. These are the taxonomy forms that we’ll examine.

■ lists

■ trees

■ hierarchies

■ polyhierarchies

■ matrices

■ facets

■ system maps.

Lists

Lists are the most basic form of taxonomies, and in fact list-making is a foundational activity for most other (more complex) ways of representing taxonomies.

We make lists all the time: when we are going shopping, when we are figuring out how to plan our day, when we want to ‘sort things out’ or make important decisions.

A list is simply a collection of related things; the relationship, usually, is defined by our purpose in making the list. Hence the name of the list (‘shopping list’, ‘people I dislike’, ‘to-do list’) often defines the relationships between the items of the list. Lists typically describe relationships of similarity or of sequence (e.g. the things we need to do to start a project). So the list satisfies our first two requirements for taxonomies: (1) in classifying related things together, and (2) in making their relationships obvious.

Most other, more complex taxonomy structures are simply about putting lists in relation to each other. Tree structures or system maps often grow from lists that got too complicated to handle – either a collection of lists of different things, or lists that have grown too long to be easily managed. As a general rule of thumb from the usability perspective, we start to have difficulty in easily comprehending and navigating a list that goes above 12–15 items. We’ll look at some of the evidence for this in Chapter 8. Long lists require more complex structures to represent the content.

Consider a taxonomy of military rank in the British army. Hierarchy (as power structure) is present, but the rank hierarchy is most straightforwardly represented in the sequence of the list, not in a tree structure.

Field Marshal

General

Lieutenant-General

Brigadier

Colonel

Lieutenant-Colonel

Major

Captain

Lieutenant

Second Lieutenant

A tree structure only becomes relevant in this case when we want to describe an actual command structure with real people at each level – i.e. an organisational structure – or when we want to say numerous types of things about each rank. For example, we may want to define sub-categories for each rank containing lists of their insignias, pay scales, privileges and so on. Then a tree structure might start to emerge, and it will serve an entirely different purpose from the ‘tree as command structure’. Trees can serve many different purposes, but the lists from which they grow can be just the same. Lists are the basic building blocks of taxonomies. The function that the taxonomy is intended to serve will guide the structure you adopt.

Taxonomy lists can be compiled based on the following characteristics or relationships:

■ commonality in attributes or purpose (e.g. motivational factors, competencies required for jobs, engineers in my firm);

■ collocation (e.g. living-room furniture, people in my department);

■ sequence (e.g. project start-up activities, regular duties, activity cycles);

■ chaining (e.g. cause and effect chains, stages in a manufacturing process);

■ genealogy (e.g. parent–child relationships);

■ gradients in attributes (e.g. from private to general, from tall to short, relative percentages of market share).

Tree structures

Using a branching tree structure to represent the transition from general to specific or from whole to part (the ‘broader term/narrower term’ in a thesaurus) is a very powerful way of representing relationships. When we looked at the list of military ranks above, we saw that adopting a tree structure allows us to add a lot of additional information in a structured way to a simple list.

The tree structure mirrors the way we manipulate our ideas of the world. However, it is so common that it often causes problems, because it can serve many different purposes (as we have just seen) and these purposes can easily get confused. For example, tree structures are easily confused with biological taxonomies, which actually work quite differently from knowledge taxonomies. More about that in our next section.

Tree structures are powerful because they reflect the way we think. In human perception we naturally identify ‘basic level categories’. These are not, as we might think, the most granular, atomistic elements in our world. They tend to be whole objects that we can identify and act upon in a direct way. Linguistically we usually have shorter and simpler names for them compared to other objects, and they tend to be accessible things that populate our everyday world.

We generalise upwards from these basic level categories (e.g. from the pine tree in my garden to trees in general, or from ‘monthly project status report’ to ‘reports’) based on similarities or resemblances.

We differentiate downwards from these basic level categories to more specific levels based on discriminating differences, distinctions or whole/part relations.

So, in practice, we don’t usually reason from most general in the direction of most specific, nor from most specific to most general. Human beings start in the middle, and from there we generalise up, and differentiate down (Lakoff, 1987: 31–8). This has important implications for how people will use our taxonomies, and we will have to consider this later when we look at how a taxonomy should be designed.

The tree structure supports this basic cognitive operation beautifully (with the single exception that the use of a tree wrongly implies we start cognitively at the most general level). It clusters families of related things together bound by a common superordinate label or concept. It mirrors the way we see our families and our tribes and how we organise the ‘stuff’ of our lives.

The tree structure also works much better than a list does as a mnemonic for the world around us. If we recall the ‘memory palace’ technique of Matteo Ricci from Chapter 1, you construct your imaginary palace and begin in the entrance hall, which is imprinted with the three or four main classes of things that you want to remember. In each of the rooms branching from this hall, you place further imaginary items representing the more detailed things you want to remember. From each of those rooms, you can have three more doors (in fact, in your own memory palace you can define as many doorways as you want) and so, room by room, you build a memory system of thoughts, all rooted in the first set of ideas represented in your hallway and all branching out from that central node in a classic tree structure (Spence, 1984).

So the tree structure turns out to be a powerful technique for visualising and structuring memory, and it’s also a powerful taxonomic representation device. The metaphor of ‘containers within containers’ with a path leading from one idea to further ideas in a structured and predictable way is exactly mirrored in our ready adoption of the digital folder structure metaphor in our organisations – although the palace metaphor sounds more interesting!

Like the memory palace technique, tree structures are very versatile. They can express several different kinds of relationship: general to specific; concept to example; cause to effect; parent to child; whole to part; group to member; container to contained (see Table 2.1).