Modern operating systems allow users to organize and refind information using many contextual keys such as timestamps, content, custom tags, origin and even location. As humans naturally engage in activities with people and groups of people, we want to investigate how we can use the context of people's social interactions to support information archiving and refinding. Past research has tracked and used remote, social interactions through email communication; this work will concentrate on using physical, social interactions (i.e., face-to-face) to support information archiving and refinding. Research questions include: (1) How do we effectively associate one's information with one's social world? (2) How do we design a user interface that supports refinding information based on social contexts? and (3) How does our approach (i.e., system) affect the users information archiving and refinding practices? This dissertation presents results from two user studies, exploring two refinding systems. The first, longitudinal study examines three participants using a custom refinding tool that tags information based on the people physically present with the user. Our second, diary-driven study examines a refinding tool that integrates information activity with a person's calendar. Our contributions are threefold: (1) an exploration of adding physical social interactions as contextual keys for information archiving and refinding (2), an examination of two user interface designs that enable users to refind information through their physical-social interactions (i.e., people and groups), and (3), a diary-driven methodology for studying realistic refinding behaviors while reducing participant interruptions. / Ph. D.

Refinding electronic information is a common problem, yet it has received less study than the problem of how to find information for the first time. In this dissertation, I examine how people approach tasks to refind information they have seen on the Web and factors that may affect refinding. I conducted a controlled, laboratory study in which participants participated in two sessions: one to find information for a set of 18 tasks and a second session, about a week later, to attempt a set of counterpart refinding tasks. Results indicate that finding and refinding do have differences, but not for all types of tasks. The use of Web search engines was not observed to change significantly from the first session to the second. However, for tasks that participants were more familiar with, search engines were used less. Tasks that involved refinding a subset of the information that was found in the first session took longer to complete and were perceived as more difficult. Participants often went directly to known resources on the Web to start their searches. These sets of known resources included many on-line counterparts of paper resources such as telephone directories, dictionaries, and newspapers. For many tasks, participants used the same starting strategy to refind the information that they used find it, indicating strong patterns of access. This work contributes to the base of knowledge about how people refind information and the factors that affect refinding. It also contributes to the research field of information refinding and personal information management by identifying dimensions and factors that affect refinding. The results reported have implications for the design of Web sites and information repositories, the design of tools to help users find and refind information, and for the research community studying personal information management. / Ph. D.