An exploratory study of tag-based visual interfaces for searching folksonomies

Aesthetic features such as animation, 3D interaction, and visual metaphors are becoming commonplace in multimedia search interfaces. However, it is unclear which attributes are needed to encourage people to use these interfaces on an ongoing basis. To design a visual interface that will elicit continual use, we first need to establish a better understanding of users' goals and strategies, in order to determine which features are critical to support those tasks. This paper reports on an exploratory study of individuals engaging with five different image and video search interfaces. Our study helped us to understand users' experiences with a variety of features and design elements, as well as categorize their common search tasks and strategies. We identified four distinct types of search: Search Known Objects + Known Keywords, Search Known Objects + Unknown Keywords, Search Unknown Objects + Known Keywords, and Search Unknown Objects + Unknown Keywords. We also identified common strategies used to accomplish each of these search types. Our findings suggest that search interfaces should maximize screen space used for visual representations of the media, provide on-demand access to titles, tags, and other metadata, and provide contextual information about previously viewed items, current keywords, and alternate keyword possibilities.