Sensemaking in Collaborative Web Search

Sharoda A. Paul, Meredith Ringel MorrisHum. Comput. Interact 2011 · DOI

TL;DR

This paper introduces CoSense, a tool designed to enhance "sensemaking" (understanding and interpreting information) during collaborative web searches. It addresses the challenges of shared understanding and information synthesis in group search tasks, offering insights into how to design better collaborative search systems.

What Problem Does This Paper Address?

Collaborative web search, where groups work together to find information, is becoming increasingly common. However, existing tools primarily focus on the mechanics of finding and retrieving information, neglecting the crucial aspect of "sensemaking" – the process of understanding, interpreting, and synthesizing that information to build a shared understanding. This is particularly challenging in collaborative settings because information is often distributed, and individuals may interpret it differently. The paper aims to understand how sensemaking occurs in collaborative web search, identify the challenges users face, and propose design implications for tools that can support this process.

Methods

The authors conducted a two-phase study. First, a formative study using a tool called SearchTogether helped identify sensemaking challenges in collaborative web search. Based on these findings, they developed CoSense, an extension to SearchTogether. CoSense provides multiple views (search strategies, timeline, workspace, chat-centric) to help users make sense of both the information found and the collaborative process itself. The effectiveness of CoSense was then evaluated through an experiment where participants performed collaborative web search tasks using both SearchTogether and CoSense. Data was collected through system logs, questionnaires, and interviews.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Collaborative sensemaking is more than just information retrieval: It involves understanding group dynamics, what others have found, and how they've interpreted it, adding layers of complexity beyond individual sensemaking.
  • Sensemaking has a strong temporal component: The products of sensemaking are passed between group members over time, making it crucial to visualize these "sensemaking trajectories."
  • Handoffs are critical points: When one group member takes over a task from another (asynchronous search), making sense of the previously gathered information and decisions is a significant challenge.
  • Two distinct search strategies emerge: Participants adopted either a "search-lead" strategy (prioritizing finding new information) or a "sensemaking-lead" strategy (prioritizing understanding existing information). Both strategies have implications for tool design.
  • The "why" behind information is key: Users need to understand not just what information was found, but why it was considered important or useful by others.

Key Results

  • CoSense enhanced sensemaking: The evaluation showed that CoSense significantly improved participants' ability to make sense of information during collaborative search tasks.
  • Search-lead vs. Sensemaking-lead: In synchronous search, one group adopted a sensemaking-lead strategy, spending over 50% of their time on sensemaking initially. Another group used a search-lead strategy, with search dominating over 75% of their activity.
  • CoSense usage varied by strategy: Search-lead strategists primarily used the search strategies and chat-centric views, while sensemaking-lead strategists found the workspace and timeline views more useful during asynchronous search.
  • Understanding group dynamics is challenging: Participants found it difficult to answer questions about group contributions, skills, and roles, suggesting this is a less prioritized aspect of sensemaking in certain tasks.
  • Task state and progress were easiest to understand: Questions about decisions made and steps remaining were the easiest for participants to answer, indicating a clearer grasp of task progress.
  • Search-lead strategists perceived greater success: While both strategies had similar actual success rates, search-lead strategists felt more successful in completing the task.

Limitations and Future Work

  • Overwhelming temporal information: While visualizing sensemaking trajectories is beneficial, presenting too much information can be overwhelming. Future tools need to provide selective displays and filtering options.
  • Prioritization of information: Current tools struggle to help users prioritize information effectively. Participants desired finer-grained rating scales and better integration of ratings with chat.
  • Structure in representations: While free-form text entry is flexible, users also desired more structure for creating representations of information, suggesting a need for tools that balance flexibility with guidance.
  • Future work: The authors plan to further investigate the differences between search-lead and sensemaking-lead strategists, study subsequent rounds of handoffs, and explore different task types and group structures.

Summary updated

BibTeX
@article{paul2011sensemaking,
  author = {Sharoda A. Paul and Meredith Ringel Morris},
  title = {{Sensemaking in Collaborative Web Search}},
  journal = {Hum. Comput. Interact.},
  year = {2011},
  doi = {10.1080/07370024.2011.559410},
}
Figure