Visual Information Foraging on the Desktop

A considerable amount of knowledge work deals with information foraging and the subsequent sensemaking of the collected information. Knowledge workers gather information from various sources, such as web pages, email messages, or presentations, but currently have little software-based support to easily store, organize, and interactively explore pieces of discovered information from these sources. The goal of this project is to design a novel technology to support knowledge workers in visual analysis and sensemaking of their casually collected pieces of information. The project will employ a user-centered design approach to clearly understand current deficiencies and knowledge workers needs across multiple disciplines, as well as to guide the iterative development of a prototype implementation. The proposed information foraging environment is based on three core ideas: First, the fundamental organization structure will be a user-maintained collage of application window snippets, to exploit users rich spatial cognition for gradually externalizing their insights obtained from the collected information. Second, casual information visualization techniques will be explored to analyze automatically extracted relations in the collected information snippets. Third, I envision visual information foraging support to be directly embedded into the users desktop interface to allow for seamless transitions between information seeking, organization, and analysis. The main contribution of this proposed project is the design and validation of this novel visual information foraging environment utilizing visual analysis techniques to guide the user through an informal collection of information snippets obtained from arbitrary application windows. As secondary contribution, field and lab studies will lead to in-depth understanding of knowledge workers information collection and spatial organization strategies on modern desktops. Findings from this project can help to shape the future of personal desktops to better support the collection and organization of information snippets, as well as the subsequent analysis of the collected information, compared to existing tools.

Knowledge workers, such as scientists, journalists, or consultants, adaptively seek, gather, and consume information. Initially, knowledge workers may not even know what exactly they are looking for. Only after investigating more and more information, they develop a more concrete mental picture. These "information foraging" and "sensemaking" processes are often inefficient as existing user interfaces provide limited possibilities to combine information from various sources and different formats into a common knowledge representation. The goal of this project was to facilitate such knowledge-intensive tasks through user interfaces that combine manual organization strategies, such as piling of papers and text annotations, with powerful automatic data processing to reveal hidden relations in the collected data. This project led to multiple novel user interface concepts that support knowledge-intensive tasks. For instance, "visual query expansion" is an extension to a search engine that visualizes the effects of suggested variations on an online query result to help users to refine their query terms. The "information collage" lets users capture information fragments from any online sources, which can then be freely recomposed into a common knowledge representation. Natural language processing helps users to identify common topics in their collage. "BiCFlows" helps users to discover hidden relations in very large information collections. For instance, they can explore which authors tend to contribute to similar topics. The user interfaces served as foundation for empirical research with users from different knowledge work domains. This research provided new insights into the strengths but also the limitations of interactive visualization for information foraging and sensemaking. On the one hand, the studies show that, in some situations, interactive visualization can be less efficient than working with unprocessed text-based information. On the other hand, the studies also show that combining automatic data analysis with interactive visualization can lead to more unexpected findings, especially when exploring large data. The extent of manual information organization varies with the task: the less users know about the gathered information, the more automatic organization they expect. In summary, this project contributed novel user interface concepts and design guidelines for user interfaces supporting information foraging and sensemaking. The techniques and study results were published at high-ranking peer-reviewed venues, and it can be expected that the findings of this project shape the way how users seek and organize their information in the future.