While modern search engines are excellent resources for finding information on the web, in order to put together that information into a useful mental model for learning or making a decision – such as picking a new car or choosing a JavaScript library – people often need to collect information about the options available and the criteria on which to evaluate the options, synthesize such information from various sources into a meaningful structure, and share and justify the results with others. This sensemaking process, often highly iterative and cyclical, puts a significant cognitive burden on users, and often requires them to externalize their evolving mental models rather than keeping everything in their working memory. However, the tools that people use for externalization – such as browser tabs, documents, spreadsheets, or notetaking apps – poorly support the constant shifts between collecting, extracting, organizing, and reorganizing that are needed. Worse yet, even if people do put in the work to externalize and share a summary of their sensemaking outcome (such as creating a list of suitable cars or a table of front-end libraries), it can still be difficult for subsequent users to evaluate whether they can or should trust and reuse that work so that they don't have to start from scratch.

In this thesis, I aim to bridge the gap between the rapidly evolving mental models in peoples' heads and the externalization of those models. Specifically, I design and build interactive systems to reduce the costs and increase the benefits of externalization, thereby capturing more of the cognitive work that users engage in while making sense of information in order to help them as well as subsequent people who might benefit from their work.

To help the initial users collect and structure information, this thesis first describes Unakite, a browser extension that enables people to easily collect and organize information into a comparison table in a sidebar as they are searching and browsing, which significantly lowered the friction of externalizing mental models compared to conventional approaches like taking notes and saving screenshots in a separate Google Doc. In addition, the knowledge captured in the comparison tables helped subsequent users better understand previous authors' sensemaking processes and rationale. Building on Unakite, we explored approaches to further reduce the cost of externalization and help people focus on their main activity of reading and making sense of web content, such as by intelligently keeping track of key information and evidence on behalf of a user (the Crystalline system) and leveraging novel lightweight interaction techniques (the Wigglite system). To help subsequent users explore and evaluate previous users' work, I developed both a framework and the Strata system that collects and visualizes key signals about the context, trustworthiness, and thoroughness of previous design decisions and rationale. Finally, after gaining a deeper understanding of people's information needs and processes when sensemaking, I circle back to the beginning to help people with reading and understanding information based on those needs. Through the Selenite system, I provide people with a top-down comprehensive overview of the information landscape, enabling users to jumpstart their sensemaking processes and receive guidance for future exploration.

The series of work introduced in this thesis points to the importance of having tool support that helps users efficiently organize and manage information as they find it in a way that could also be beneficial to others, and therefore bootstrapping the virtuous cycle of people being able to build on each other’s sensemaking results, fostering efficient collaboration and knowledge reuse.

211 pages

Thesis Committee:
Brad A. Myers (Co-Chair)
Aniket Kittur (Co-Chair)
Kenneth Holstein
Daniel M. Russell (Google, Inc.)

Brad A. Myers, Head, Human-Computer Interaction Institute
Martial Hebert, Dean, School of Computer Science

Return to: SCS Technical Report Collection
School of Computer Science homepage

This page maintained by reports@cs.cmu.edu