12.1 Guides: Orientation Assistants

Guides provide a familiar interface to readers by simulating a virtual creature on the screen that assists them in their orientation task. Guides can be as simple as Simon, the storyteller, in Brøderbounds animated children's book of Esop's fable "The Tortoise and the Hare" (figure I.61).

Figure I.61 The guide Simon in [Brø94]

The different pictures in figure I.61 display Simon introducing the different animated pages of the children's book. Simon exhibits hardwired behavior; in the opening sequence to each page he introduces the story of the page in a canned movie. Nevertheless, Simon offers an easy-to-use navigation feature that can be understood even by three-year-olds.

Oren, Salomon, Kreitman and Don describe an implementation of guides as main interface to an educational hypermedia database [Ore90] [Ore90b]. The database contains material about American history from 1800 to 1850. The goal of Oren et al. was to find a way of structuring and providing information that avoided navigational overhead. The reader should always have a next, obvious choice instead of having to choose manually out of a large list of possible selections. The easy solution to this problem would have been to implement some sort of guided tour, but Oren et al. considered this approach too sequential and limiting for the user. Instead, they decided to offer guidance through guides. Their guides are virtual personalities of the appropriate time period, as, e.g., preacher, slave, miner, settler, Indian, etc.. This historical look and feel of the guides has the additional advantage of putting the reader into the right historical context.

Oren et al. describe two versions of guides; the first version had simple, static guides. In a second implementation they added video guides, consisting of video clips where an actor dressed in contemporary clothing would give the navigation tips.

Figure I.62 Selection card for static guides of first implementation (ExFig. 6 p. 371 from [Ore90])

The first guide implementation (fig. I.62) was based on a hierarchical topic index of the articles. Each guide was characterized by a collection of index entries. The "Sea Captain" guide was described, among other entries, by "Transportation", "Ships and Shipping", Clipper Ships", etc.. For each guide a ranked list of articles of interest was compiled. The first guides therefore only consisted of a nested list of topics. Later each guide also got a life history to help choose topics and explain why certain topics had been selected.

In informal field testing of the first guide version, users preferred this form of navigation aid over more conventional ones like tours or maps. But users soon projected too many expectations onto the guides. They assumed a characterization behind the iconic figures. For example, they expected a personalized presentation of a story. They anticipated that the story about Andrew Jackson would be presented differently depending on whether they were guided by the slave or the Indian.

The second version corrected two main deficiencies of the first version. Users wanted to know why the guide had brought them to a particular article and they also wanted to know whether the article would be presented from the guide's point of view. Storytelling guides included these two additional features. Navigation tips could be called from the guide card (fig. I.63). Storytelling guides were then brought up as video clips where a character told a short first-person story explaining why he or she would take a user to a particular article (fig. I.64).

Figure I.63 Video guide: Daily life on the Oregon trail http://www.abbedon.com/Project/guides.html

Fig. III.64 Video guide: the settler gives a navigation tip (ExFig. 12 p. 376 from [Ore90])

In comparison, video guides differ from the simple guides in form, content and function [Ore90 p. 374]:

Table I.2 Simple guides versus video guides

The main goal of guides in Oren et al.'s project was to provide a simpler navigation interface than a complex hypertext map or a conventional query interface. Guides also succeeded in merging browsing and search into one single metaphor. Oren et al. conclude that the deliberate personification of the navigation interface seems to result in increased engagement of the user. Guides also offer an excellent way of exposing the biased view inherent in each article. By letting different guides with different viewpoints (e.g. Indian and settler) tell the same story, the reader gets a more unbiased view of this particular story. This approach is very demanding for the document author, because the stories have to be broken down into single facts which can then be reassembled by the guides.

Particular care has to be taken not to annoy more experienced users, because once users know what they are looking for, they often see a guide more as an obstruction than as a help [Lau92]. This problem can be addressed by monitoring the user's behavior. Once the system has guessed the user's intention, guides can offer themselves to provide the information for which they think the user is searching.

User testing also revealed the need for a different type of guide. Frame guides lead the user through the system functions by providing assistance for all those functions that are system related, as for example, how to use the system. The concept of a frame guide is analogous to a TV anchorperson who has an overview over the whole broadcast and who introduces reporters having more in-depth knowledge about their particular story.

The guide metaphor solves the problem of where to go next in a large document collection. The implementation, on the other hand, is rather complicated. The guides implemented by Oren et al. have been hardwired for this particular application. To build more generally usable guides, a broader framework is needed. Agents, as described in the next section, generalize the notion of guide.