The purpose of this thesis is to improve our understanding of sensemaking process as a basis for building better systems to assist sensemaking. Sensemaking is the task of creating an understanding of a problem or task so that further actions may be taken in an informed manner. Sensemaking is a pre-requisite for many other tasks such as decision making and problem solving. An important part of sensemaking involves making clear the interrelated concepts and their relationships in a problem or task space.

This research investigated the question of how users create and use structured representations for sensemaking. It proposed and refined an iterative sensemaking model building upon previous sensemaking research, learning theories, cognitive psychology and task-based information seeking and use. In particular, the study focused on the processes, conceptual changes, and cognitive mechanisms used during users' sensemaking tasks. The qualitative, multi-case user study investigated how a sample of fifteen users working with news writing and business analysis tasks structure their conceptual space with the assistance of note-taking and concept mapping tools. Data on the sensemaking process were collected from multiple sources including think-aloud protocols, screen movement recordings, interviews, and intermediate and final work products.

Using the iterative sensemaking model as an analytical and descriptive framework, the study captured the often idiosyncratic paths sensemakers took, ranging from planned, systematic to rather random, ad hoc patterns of "search--sensemaking" iterations. Findings also revealed various ways in which the iterations started and exited, which suggested that the heterogeneous patterns of sensemaking lie in the shifts from one iteration to the next, rather than in the iterations themselves. The knowledge structure was updated by accretion, tuning, and restructuring to produce the final knowledge representation and sensemaking product. Several cognitive mechanisms were used in processing new information, examining concepts and relationships, and examining anomalies and inconsistencies. They were used in bottom-up, top-down, and combined fashions to move the processes along and to trigger conceptual changes to the knowledge structure of users.

Based on these findings, the study argues that information system that aimed to assist sensemaking should provide an architecture that links structure, data, and sources that can be represented and manipulated in multiple formats. It should also provide integrated assistances at the task and cognitive mechanism levels.

The research contributes to sensemaking research by extending existing descriptive sensemaking models with an analytical framework that incorporates conceptual changes to the knowledge structure and cognitive mechanisms that trigger the processes and conceptual changes. Furthermore, the research identified core issues in designing information systems to assist sensemaking tasks and suggested design implications for sensemaking tools that may be useful in many settings such as learning, knowledge creation, organization, and sharing.