LGA proposes to create an Intelligent Agent Environment (IAE) that, when deployed with a smart, collaboratory, portal application (e.g., LGA's WhereWeWork) would make available to the NIMA analysts the following capabilities: transparent, user-directed access to heterogeneous data sources via a semantically-smart search and query engine; automatic updating of searches results; the flexibility to organize information in ways useful for analysis; easy sharing of information with colleagues; agent directed or mediated searches that find and display information for the analyst on the basis of the tasks being performed (i.e. information push); tools to support the analytical process (e.g., what would an expert do next), and prompts for other appropriate actions (e.g., initiating a tasking request.). The IAE provides the analyst with a virtual "master sergeant", who knows the ropes, is familiar with the data sources, knows how information should be processed before the analyst sees it, knows who else is working similar tasks or domains, and promptly advises the analyst of any information of interest, and suggests appropriate actions. Moreover, the "master sergeant" learns about the analyst's tasks and preferences, and so can provide more relevant and appropriate information over time. The "master sergeant becomes, in effect, the institutional memory. The Intelligent Agent Environment (IAE) should reduce the amount of time and effort spent on searching for the right information, and allow the analyst to concentrate on the analytic tasks at hand. Because the agents learn as they go, they become advisors to their analysts, and embody the institutional memory on certain tasks.The proposed architecture for IAE will allow the system to be customized for a variety of task domains, including, but not limited to law, law enforcement, medicine, and scientific research of many kinds.
Keywords: INTELLIGENT AGENTS, PORTALS, COLLABORATION, AGENT ARCHITECTURES, ANALYSIS, MACHINE LEARNING
