This chapter is from the book
This chapter is from the book
This chapter is from the book
Anatomy of a Concept Model
Concept models illuminate domains of information, subjecting them to structured classification and relationships. The advantage of a concept model is that it doesn't force a hierarchical relationship among the domain's concepts. To be effective, the concept model must be clear about what's included and how the concepts and domains all relate to each other.
This section discusses the layout and style of concept models. Subsequent sections discuss choosing appropriate concepts and some of the weird things that happen when we try to boil the world down into circles connected by lines.
Table 4.3. Three layers of concept models and the components of each.
|
Layer 1: The Basics |
Layer 2: Added Complexity |
Layer 3: From Model to Illustration |
|
Nodes |
Styles |
Amplify a comparison |
|
Links |
Backdrops and grouping |
Find a metaphor |
|
Branched relationships |
Concepts as backdrops |
|
|
Indirect object relationships |
Simplify the story |
Layer 1: The Basics
With concept models, we have two basic building blocks to work with: nodes and links. The format for a concept model suggests only that the nodes represent nouns and the links represent verbs describing the relationships between them.
Nodes
Use circles to represent nodes. The advantages of circles:
- They're easy to connect with each other. Without sides (or with infinite sides, depending on your relationship with Euclid), it's easy to attach links to a circle. A line that connects to a circle's center is clean and easy to understand. The same thing with a square or other shape can be inelegant and yield "visual noise" that could confuse the diagram.
- They don't crave a grid. Four-sided shapes want to live on a grid. Circles don't need a grid to ensure legibility and visual appeal, which means they can get close to each other pretty easily, and you're not dependent on layout to convey a sense of hierarchy. A nice corollary to this is that aligning circles to a grid is a powerful communication device. Without doing anything besides lining them up, you can imply a particularly strong or central relationship.
- Squares and rectangles imply screens or pages. This may be appropriate in some instances—when your concept model actually represents a network of templates—but if you're laying conceptual groundwork, circles are good for concepts.
)
Figure 4.3 Nodes should be circles. Squares just don't connect well.
Label your circles. A couple of tips:
- Remember, these are the nouns. The next section covers how to choose appropriate concepts.
- Use the plural. It's easier to craft appropriate verbs in the links, and the purpose of these concepts is to generalize about the domain of information. Plurals don't require an article ("a", "an" or "the"), making it easy to construct sentences.
- Specificity is good, but use examples in a subtitle. Note that the sample concept model uses "Genres" as the main concept and "Fantasy" as the example.
)
Figure 4.4 Give your concepts specificity, under the main concept of "Genres" a specific example is given—"Fantasy."
Links
Use lines to represent the connections between nodes. In concept models, the links always have directionality. That is, they're always read as one-way from one node to another. Always label your links; I always regret it when I leave them off. Some ideas for good labels:
- Remember, these are the verbs. They should describe the relationship between one concept and the other by describing how the one acts on the other.
- Avoid "to be." This isn't a Strunk and White–like grudge against the passive voice. "To be" implies belonging, a hierarchical relationship. It doesn't illuminate the relationship. You're using a concept model because the relationships between the concepts are more interesting than a series of nested categories.
- Avoid wishy-washy modifiers. If you try to hedge your relationships by using "may" and "might", your whole model will be one giant apology. A good concept model is like good writing—authoritative and confident. Readers understand implicitly that not every relationship will hold true in every instance.
- Avoid constructing multinode sentences. You might be tempted to concoct a chain of relationships to create a more complete picture. (The next layer talks about how to deal with more complex relationships.) Generally speaking, I try to keep my node-link-node relationships self-contained. Although it's important to capture the complexity of reality, it's more important to create an artifact that serves the design process. Standing at any node, you should understand how it relates to the concepts around it, outside the context of concepts two or more degrees away.
A set of nodes and their relationships may break this last rule, but only when they represent the "backbone" of the domain—the essential theme that ties everything together. The section Creating Concept Models describes different starting points and basic structures for concept models.
)
Figure 4.5 The challenges of multinode sentences: They lack of immediate context. Though the bottom version feels a little disjointed, the reader can start at any node and apprehend a complete thought. Using prepositions alone, in the top example, doesn't illuminate the concepts.
Layer 2: More Detail
It won't take long for you to run into severe constraints with the simple concept models described with layer 1 elements. Indeed, reality frequently demands richer visualizations. The devices in the second layer extend basic node-link-node relationships to afford us more power in describing these complexities.
Table 4.4. Techniques for showing nodes, how they are similar or different in a concept model.
|
Use... |
To show or highlight... |
Be careful of... |
|
Size |
Importance |
Hampering your ability to lay out the diagram effectively. Too many different sizes can make it difficult to create a good layout for the diagram. |
|
Color |
Additional relationships |
Using more than three or four colors. Too much color can distract from your message and make your diagram look inelegant. Adding one color to an otherwise black-and-white picture is ideal to call readers' attention to something specific. |
|
Value (how intense the color is) |
Importance |
Oversaturating individual nodes. Too much contrast can be distracting. Remember the principle of the "smallest effective difference," from Edward Tufte. The difference between two nodes doesn't need to be dramatic. |
|
Line weight |
Importance |
Making it difficult to follow the relationships. Increasing the line weight of nodes can make it hard to perceive the connections between nodes, as all the lines start to jumble together. |
|
Backgrounds |
Additional relationships |
Backgrounds can be useful to group together related nodes, showing how they're related at a more macro level without drawing explicit relationships between them. That said, make sure that in an effort to use a background you don't compromise the overall layout of the diagram. |
Distinguishing nodes with styles and backgrounds
Not every concept in your model is necessarily weighted the same. Some concepts are more central to the story. Others are more important to emphasize. Further, you might identify patterns in the nodes that are not immediately apparent from the relationships between them. You might want to highlight that some nodes represent people while others represent physical objects. In the case of the comic book model, for example, I could distinguish between concepts that allow me to classify and identify comics and those that have to do with the business of comics.
Determining appropriate distinctions and groupings will be discussed in Creating Concept Models, a few pages ahead. There are a few techniques for styling nodes to highlight the differences between them.
Similarly, you can use the same techniques to style the links between nodes. Of course, styling links has different implications. (See Table 4.5.)
Table 4.5. Techniques for distinguishing types of links between nodes in a concept model.
|
Use... |
To show or highlight... |
Be careful of... |
|
Line weight |
Importance |
Obscuring the nodes. Large connecting lines can make it difficult to discern the nodes themselves. This may be acceptable if you want to emphasize the relationships. Generally, however, I compensate for the size of the node with making the color more washed out. I usually use only one line weight, unless I've crafted a "value proposition" model with a central sentence. In this case, the connecting line is substantial to establish it as the foundation of the model. |
|
Color (for links) |
Specific sets of relationships |
Using more than two or three colors. There may be a set of relationships that you want to call attention to. Typically, this is a chain of three or four nodes that tell a story-within-a-story. I use fat arrows with a distinct color to call attention to these. Readers may not know what to focus on if you use more than two of these devices. |
)
Figure 4.6 Differently styled nodes and links, even just using circles, fills, and lines, offer a lot of flexibility to communicating differences.
How much is too much? Readers need to know where to start and what to focus on. They need to come to a particular conclusion. Judiciously applied, styles can enhance the model's ability to tell the story. But too many can further confuse the issue.
In the comics model, I used the heavy arrows around the perimeter to establish a framework for the whole diagram with the main relationships connecting the primary concepts.
The branched relationship
Sometimes, when I get down to the leaves of the tree, so to speak, describing the smaller details of a model, I'm capturing various facets of those small concepts. A single concept is linked to several smaller ones that, in and of themselves, serve only to describe the larger concept. Identifying unique relationships to each of these smaller concepts may be more trouble than it's worth.
Enter, the branched relationship. A single label identifies all the links between the larger concept and the smaller ones. It looks somewhat different: instead of a single line between the concepts, the larger concept has a line that branches and the label sits on that single "trunk" or at the intersection of all the branched lines.
)
Figure 4.7 A branched relationship connects one concept to two others with the same verb. In this case, the value of a comic depends on two things—its condition and important milestones in the story.
There are any number of reasons why you might use this approach, but two specifically come to mind:
- The smaller concepts are descriptive of the larger concept. In this case, it's useful to identify specific aspects, perhaps because you want to highlight particular properties of the larger concept.
- The smaller concepts represent various possible states or conditions for the larger concept.
Direct and indirect objects
The grammarians among you might start to wonder about the simplicity of sentences based on noun-verb-noun. The domains you'll be modeling will likely demand more complex relationships. Breaking down the language of a concept model, the two nodes act as a subject and object: Subject acts on object.
This may be appropriate for most relationships you need to describe. In some instances, however, the relationship makes sense only with the addition of an indirect object.
In my own models, I've found that central ideas need to appear throughout the model. They, in a sense, mediate relationships between other nodes. In our running example, concepts like "Comic Books" and "Stories" are central concepts that may be indirect objects in many of the other relationships.
In some cases, you can change the perspective to frame the same relationship in a different way:
- Sellers sell comic books in Stores ->
- Stores [are] owned by Owners
In other cases, you can phrase the verb to swap the object and indirect object:
- Distributors distribute comic books to Stores ->
- Distributors supply Stores (with comic books)
Such an approach is grammatically and stylistically questionable, but appropriate for the purposes of the model.
If you find you can't effectively describe a relationship without mentioning another noun, consider embedding it in the link's label. You can give it a distinct color, matching the color of the original concept.
Layer 3: From Model to Illustration
You can stop at layer 2. Seriously, if you don't take your model beyond circles and lines, that's OK, especially if you're not likely to show it to anyone else.
But if you're anything like me, you sometimes want a little more. You want a meaningful picture that's fun to look at. If your project allows (or you have nothing better to do on the weekends) you might take some time to refine your model.
There's no easy way to teach this in the space of a few pages, but here are some techniques that I use to refine my models.
Amplify a particular comparison
Successful diagrams let readers compare ideas. You might pick two concepts, or two groups of concepts, and massage the visualization to show the similarities and differences between them.
)
Figure 4.8 Comparing the business and the consumption of comics. This comparison emerged as I was evolving the model, distinguishing the experience of comics (bottom arc) from the business of making them (top arc).
Find a metaphor
Like a concept model, a metaphor is a way of organizing relationships between ideas. By using a metaphor, we're mapping one domain to another and making inferences about the target domain based on knowledge we have about the mapped domain. In a less abstract way: When comparing arguments to battle, I'm mapping the elements of an argument (participants, discussion) to elements of battles (opponents, fighting, winners and losers). This comparison frames the concept of an argument in a particular way, and implies certain conclusions.
Look through the key relationships in your concept model. Is there one set that lends itself to a visual metaphor? Starting with that foundation, extend the metaphor to encompass other concepts in your original model. How does the metaphor explain relationships to other concepts?
)
Figure 4.9 Transforming simple models into beautiful illustrations is a hallmark of Stephen Anderson's work. Driven by a small handful of concepts, Stephen elaborated the basic idea—a definition of user experience—into a visual metaphor that drives the point home. You can learn more about Stephen's work at poetpainter.com.
Concepts as backdrops
A mature model will have a number of midsized concepts. Ideas that are crucial to the story, but serve as a bridge between the central ideas and the meaty details. Instead of using these concepts as a bridge, have them recede, using them as a backdrop for the more detailed concepts.
)
Figure 4.10 Receding concepts into backdrops can add depth to the model as well as communicate simple relationships with less visual noise.
Simplified story
Another technique to facilitate the transition from circles and lines to something more sophisticated is to strip out some concepts. With a smaller number of concepts, you can focus on showing the relationships between them instead of describing them through labels.
There are two ways to get to a simpler story:
- Focus on one part: A good model can show how seemingly unrelated things have important connections. That said, the conceptual leap from one end of your diagram to the other may not be the central part of the story. To generate a more visually appealing illustration, you can zoom in on one part of your model. This may yield one or two concepts serving as the central focus, and you can show how the more detailed and specific concepts relate to them.
- Focus on higher levels: Starting with the model's primary concepts, keep only those nodes that are one or two degrees away, shaving off everything "lower" in the model. With trimmed details, the model becomes easier to shape into a picture. You can play up certain relationships and identify other visual devices for describing the major concepts. As you zero-in on how to tell the high-level story, you can decide whether it's appropriate and feasible to incorporate the stripped-out details.