U.S. Department of Energy

Pacific Northwest National Laboratory


Data is being generated at a tremendous rate. Classic data handling architectures are falling behind in a growing number of data-intensive applications because they rely on data ingestion systems to feed databases and offline analytics for decision support.
AIM's research activity targets six key questions:

  • Which data is relevant for me to making the right decision in a timely fashion?
  • What has happened in past that provides context to my decision making?
  • Should I be concerned about the events that I'm seeing?
  • Can the streaming analytics alert me to specific trends or patterns that have occurred or may occur in the future?
  • Can the machine recommend an analysis approach or course of action?
  • What’s the right balance between humans and machines?

AIM advanced the state of the art in answering these questions. Our unique contributions can be found in AIM’s research capability areas.

AIM developed ways to provide support for both human and machine decision making based on continuous streams of data. Streaming decision support improves the efficiency, transparency, and effectiveness of decision making as well as provides the ability to transfer work to earlier stages in the analytic workflow.  

AIM developed ways to identify, rank, sort, group, describe, and track useful events when they occur. We can do this both within data streams and across multiple data streams. This provides greater confidence in the analytic workflow and enhances tactical and strategic decisions.

AIM developed ways to increase and optimize machine and human intelligence through strategic acquisition of new data and collective interaction of machines and humans. Joint human-machine intelligence means human feedback is seamlessly integrated into algorithms and machine intelligence adapts to maintain correct, timely, useful system performance. Human-machine intelligence reduces analyst fatigue and increases operational performance.

AIM developed ways to dynamically query streams of data with high confidence in the results. Stream querying reduces false positives, decreases cognitive load on analysts, and streamlines infrastructure for maintaining queries.

AIM developed ways to adjust and adapt stream acquisition, fusion, and corresponding analytics to optimize correct, useful, and timely interpretation of data. Successful steering provides for enhanced, targeted data collection and situation awareness. Bandwidth requirements are reduced. Resiliency is improved. The use of existing downstream resources can be optimized.

AIM developed ways for machines to help summarize and abstract multiple streams and static data sources over time to assist in sensemaking and overall situation awareness. AIM R&D brings greater efficiency and effectiveness in hand-offs, and our summarization capabilities reduce errors.

| Pacific Northwest National Laboratory