Week 4: Ups, downs, and in-betweens

This week's reading: Morville and Rosenfeld, Chapters 7, 8, and 9

While last week’s reading discussed how the interface should be built and named, this week we focused on how the user will interact with the interface. How will he find his way from one organizational chunk of our site to another? Will it be through browsing or through searching? How will we enable him to browse and search effectively without requiring him to become an expert on our website or on information science in general? There’s a lot of meat here, and the answer to each question is almost always “it depends,” but there’s a pattern to best practices that seems to illuminate a fundamental issue in designing IA for the user.

To illustrate this pattern, let me take examples from the chapter on navigation and the chapter on controlled vocabularies; I will draw parallels between them. In designing a classic thesaurus, the designer first identifies a preferred term for a topic (Cougar) and explains the content of the topic using scope notes (Cougar SN Felis concolor, a large predatory cat native to the Western Hemisphere). The thesaurus maps the term’s variants to the preferred term in the manner of an authority file (Mountain Lion Use Cougar); it also links the preferred term to the next broader term in its hierarchy (Cougar BT Cat), as well as to any narrower terms (Cougar NT Florida Panther). The thesaurus may also note terms that are qualitatively associated with the main entry, as determined by the compiler or by software (Cougar RT Jaguar). In sum, the thesaurus traverses several kinds of relationships (equivalence, hierarchy, and association) in trying to make sense of a search input or otherwise guiding the user to his endpoint.

Compare this straightforward if technical design of a thesaurus to the design of a website. When the user sets out to navigate a large website, the website is usually well served to provide him with global, local, and contextual navigation tools. The global tools move the user up in the site hierarchy, not unlike the “broader term” relationship of a thesaurus. Global navigation is unlike BT in that the user of global navigation can quickly navigate to a different area of the site entirely, while BT by design only connects the user to the next most chunked classification of the term he is already viewing, but both tools share the purpose of giving the user a way to see something more general.

Local navigation tools move the user within the subsite or site section he is already viewing; for example, within Amazon.com’s section on video games, a local menu allows the user to view games for the Nintendo Wii or Xbox 360. If the user clicks on Wii, another local menu prompts him to select a genre of video game. The process is analogous to the classic thesaurus’s “narrower term” relationship: both the local navigation system and NT serve to help the user find something more specific.

Contextual navigation tools most often take the form of hypertext links in the content. Users expect that the text of such links describes the content of the linked webpage. Designers insert hyperlinks when they need to escape website hierarchy and move laterally to pages that are related to the current page’s content. This type of navigation is analogous to the classic thesaurus’s “related term,” in that contextual navigation and RT each serve as a catch-all; they take the user to a place that is related nonhierarchically.

(It’s worth noting as an aside that navigation systems do not generally have a precise equivalent to the thesaurus’s equivalence relationship, because organization is unique and language is not. Some advanced navigation designs do admit parallels to equivalence relationships, but this discussion is tangential to the main point of this entry.)

Clearly, website navigation and thesaurus design have something in common. The commonality lies in how information can be related – broader, narrower, associated – and the likelihood that the user will want to move along one or more of these lines from one idea to another. This mostly-hierarchical relationship makes the organization of a website or a thesaurus transparent, allowing the designer to apply labeling skills to express its contents in a way that the user can readily navigate. I am left, however, with the question of whether hierarchy is uniquely suited to these tasks, or whether information can be organized in a way that is nonhierarchical and yet coherent. Morville and Rosenfeld gesture in this direction with their discussion of tag clouds, which resultantly seem to be a rich resource that semantic web software could use to guess what terms are related. Though creating clear and easily navigable hierarchies is obviously central to information architecture, I’ll remain alert to situational alternatives to the hierarchy that may present themselves!