Distributed Sensemaking: Externalizing and Aggregating Structured Mental Representations
Abstract
People spend an enormous amount of time making sense of the world online, whether patients trying to determine the causes of their conditions, scientists trying to understand an emerging field, or citizens trying to understand the effects of a proposed health care bill. Estimates of the amount of time spent on such complex sensemaking tasks put it at around 70 billion hours a year in the U.S. alone; to put this in perspective, it is about the time required to create Wikipedia ~twice ~ spent every day. However, after each sensemaking episode in which an individual develops a useful mental representation of an information space for themselves, their representations are essentially lost, with no one else benefiting. Not only did the time spent learning likely not help anyone else, but if asked today, that individual may have forgotten much of what they spent so much time learning in the first place. Consider instead if we could capture the work that the individual did such that others interested in similar information could benefit from it, even if those individuals did not know each other. Enabling people to build on each others representations to make sense of large informationspaces has the potential to benefit fields ranging from education to science to policy to health to the enterprise, supporting better decisions, greater innovation, and more literate, knowledgeable and intelligent citizens. To realize this future will require fundamental advances in addressing two key problems: 1)extracting the mental representations that individuals build up while making sense of information, and 2) aggregating them across individuals. To address these challenges of extraction and aggregation we plan to explore approaches in which humans partner with machines to collaboratively build mental models that are immediately useful for decision making while also enabling aggregation and reuse. The intuition is that if a machine could learn about a human s evolving mental model while helping them filter, highlight, and organize their information for them, these models could be helpful for not only the initial individual but could be reused by and aggregated across future individuals with similar needs. In this proposal we explore several approaches that have the potential to incentivize the initial individual in building their mental model, while extracting their structured representations and aggregating them.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 20, 2019
- Source ID
- N000141912454
Entities
People
- Aniket Kittur
Organizations
- Massachusetts Institute of Technology
- Office of Naval Research
- United States Navy