This project will analyze how individuals make sense of information collected within online health forums and how existing computing platforms facilitate or inhibit this process. Increasingly, individuals of all walks of life go online to collect and share information, collectively make sense of it and negotiate its meaning, reach a consensus, or agree to disagree, and, thus, generate a repository of collective wisdom that can be shared and reused. These social behaviors, often referred to as "sensemaking," are particularly consequential in health and wellness management, as increasing numbers of individuals join online health support communities and rely on their peers for help and advice. Despite the ongoing research on social computing platforms and on attitudes, perceptions and behaviors of their users, the dynamics of computer-mediated sensemaking remain poorly understood. Moreover, most of the information collected within these forums continues to exist in the form of discussion threads that provide little guidance as to the main topics discussed, the relationships between posts within the thread, different attitudes towards the topics of interest, and the overall dynamics of the discussions. The overarching goal of this research is to develop novel summarization, visualization, and interaction mechanisms to help individuals make sense of information and opinions collected in online forums.
Specifically, the research will seek to understand how sensemaking unfolds within discussion threads, describe it through a set of formalizations, or semantic discussion typologies, and identify ways to study it computationally. Participatory design methods will then be used to develop novel visualization and interaction mechanisms for facilitating individual and collective sensemaking within online forums. A combination of computational data analysis and visualization techniques will automatically generate visual summaries of discussions threads that can help users to understand the evolution of meaning in discussions. This research will build upon previous work in natural language processing and data mining, to represent salient properties of the discussion threads, such as discussion topics and their transformation overtime, and the various attitudes towards the topics among the users. Finally, the discussion visualization tools will be evaluated on their impact on individual and collective sensemaking in controlled experiments and in a deployment study within an existing online health community oriented toward the difficult health issues associated with diabetes.
