N13A-T024 TITLE: Situational Awareness as a Man-Machine Map Reduce Job

TECHNOLOGY AREAS: Information Systems, Human Systems

ACQUISITION PROGRAM: PMMI, MARCORSYSCOM; EAITE POM 14 Approved EC

OBJECTIVE: Design and implement a mixed initiative (man-machine) distributed fusion capability that allows lower level fusion (entity/event recognition, disambiguation) inferences to be reasoned over in order to increase the relevance and accuracy of higher order fusion services (behavior prediction) that are coupled to warfighter decision tools. The objective will require the development of a distributed cloud based mixed initiative (man, machine) workflow manager.

DESCRIPTION: Solutions and algorithms developed to facilitate large data management, while successful in the acquisition of data, have been constrained by problems of human-system integration as the solutions developed to support military decisions remain based on an information fusion model that assumes that if a person understood the physical space they could make the �right� decision. However this assumption should be challenged as it may not possible to fully understand the physical space as data will be "dirty", inconsistent, or undiscoverable. The goal of the topic is to understand how to place human intuition back into a process that translates large data sets into actionable situational understanding. An offeror is expected to mature a set of map reduce applications that continuously reason about possible future states/paths in a way that allows warfighters to validate both the state/path conclusion as well as the inference confidence level. The system should allow the human user to adjust confidence levels attributed to lower level inferences (the presence of a unique entity or behavior) before higher level fusion applications (intent prediction) use the lower level inference to support a decision aide. The required system must assume that large data will be held at distributed locations throughout a battlespace and the movement of data across data links will not always be possible. The goal of the system will be a more accurate decision tool that has leveraged both automated data analysis as well as human intuition at multiple processing/reasoning steps. Phase 1 performers may work with synthetic data video and text based data sources. The Hadoop framework should be utilized.

The specific challenges of this topic include: 1) Maturing a set of related map reduce jobs that can act on distributed stored images/imagery and unstructured text to find data that supports a prediction of a specific future state/path 2) Development of collaboration environment that allows each entity/behavior recognition event to be reviewed by a human who can increase or decrease confidence levels based on knowledge about data conflicts, sources or missing data 3) Development of a distributed higher level fusion application that translate human vetted lower level information about entities to predictions about higher level future paths (behavior intent) 4) Development of a human-machine collaboration application that allows humans to raise or lower confidences in a future behavior intent predictions based on presented evidence 5) Demonstration that a fusion system build on man-machine collaboration will provide a more accurate input to supported decision tools.

Research in the areas of social and cognitive science, mathematics, computer science, information theory, decision science, operations research as well as multidisciplinary areas that may prove promising are of interest. In addition to the application of research methods and approaches, it is important to evaluate the impact of these efforts areas with regards to the way they change how data is collected, processed, or shared to positively impact decisions.

The OSD is interested in innovative R&D that involves technical risk. Proposed work should have technical and scientific merit. Creative solutions are encouraged.

PHASE I: Complete a plan and detailed approach for developing a system that enables distributed applications to reason about possible future states using information from large cloud based data, allows human examination of the evidence behind lower level inferences and then supports higher level reasoning using the same methodology. Identify the critical technology issues that must be overcome to achieve success. Technical work should focus on the reduction of key risk areas. For a constrained set of possible entity/behavior states and behavior/intent states, using distributed data stores, demonstrate that phase 1 risk reduction work has shown that a full implementation of the approach is technically tractable. Prepare a revised research plan for Phase 2 that addresses critical issues.

PHASE II: Produce a prototype system that is capable of improving the effectiveness of machine supported decision aids by allowing human markup of all inference supporting evidence. The prototype system should assemble information by automated means, perform automated reasoning about futures, accept human assessment, provide performance metrics and offer visualization appropriate to a warfighter. Produce a prototype distributed fusion service that can produce accurate input to a warfighter decision support system even when some data is incomplete, conflicting or wrong. The system should be able to process large distributed and varied data sources. The prototype should enable a demonstration of the capability to be conducted using relevant data sources, some of which may be classified. The prototype should be capable of operating in a real time mode. Identify appropriate test performance dependent variables and make trade-off studies. The prototype should be relevant to both DoD and commercial use cases.

PHASE III: Produce a system capable of deployment in an operational setting. The work should focus on a specific user environment intended for product transition. Test the system in an operational setting in a stand-alone mode and as a component in a cloud processing architecture. The work should work towards a transition to program of record, military organization or commercial product. The system should adhere to open standards and open software where feasible.

PRIVATE SECTOR COMMERCIAL POTENTIAL/DUAL-USE APPLICATIONS: News organizations would greatly benefit from an ability to research a story using both machine data mining and analysis and human intuition, enabled by a mixed initiative system.

REFERENCES:
1. Apache Software Foundation. ActiveMQ. 9/22/2010 http://activemq.apache.org

2. Apache Hadoop: http://hadoop.apache.org/

3. Cloud computing, Wikipedia 2011, http://en.wikipedia.org/wiki/Cloud_computing

4. Open Networking Foundation formed to Speed Network Innovation, March 21, 2011 http://www.openflow.org/wp/2011/03/open-networking-foundation-formed-to-speed-network-innovation/

5. Introduction to Structural Equation Modeling (Path Analysis): http://www.sgim.org/userfiles/file/AMHandouts/AM05/handouts/pa08.pdf

6. Mixed Initiative Interaction: http://wwwhome.cs.utwente.nl/~conagent/Frank%20van%20Es/References/Hea99.pdf