Evaluating Category Membership for Information Architecture

We examine two approaches for evaluating category membership within an information taxonomy: card sorting and a simple statistical approach based on the word frequencies of content items. To assess these two approaches, we developed a category structure for a web site based on card sorting. Metrics from the card sorting were obtained to predict problematic tasks. Potentially problematic tasks were also identified using word frequencies of the content items under each category. We then conducted a web navigation study to compare actual user performance of tasks identified by the two approaches. Despite its simplicity, the word-based statistical approach produced marginally better predictions. In particular, it predicted problematic tasks that took nearly twice as much time on average to complete than the remaining tasks in the web navigation study. We review these results and discuss the general usefulness of card sorting and word-based statistics for designing information taxonomies such as those used in web sites and menu systems. Author