Mapping an information architecture's scent: Evaluating the effects of task and context on judgments of relevance

Using an information-based web site to seek target information depends largely on the usefulness of the site's information architecture (IA). Users must make guesses about the probability that a menu label will lead to target information. They are making judgments about the relevance or value of those labels and following what Pirolli (1997) refers to as an information scent.; This research explored the effects that different tasks and contexts have on information scent. The methodology used in this research proved to be a good tool for mapping the scent trails of an IA and the degree of agreement about a labels scent. The metric used for computing the probability that a menu label will be chosen also proved informative. Providing tools that allow researchers and developers to diagnose the usefulness of an IA and inform their design decisions is one of the main contributions of this research.; In addition, the results showed that providing users with multiple access points to the same target information increases the likelihood of a user finding that information. Results also showed that users identified a greater number of partially relevant menu labels when they were judged from a situational context rather than from a topical context. A final conclusions was that a menu label's information scent depends on the scenario driving the search.