11.12 Machine Man

The EVS will be a hybrid of human intuition and data mining technologies working in tandem on an on-going basis to detect individuals involved in terrorism and criminal activity. The capabilities of machines and people clearly complement each other: Computers are able to process very large amounts of data, whereas analysts and investigators are able to bring to bear large amounts of knowledge about human behavior, criminal activity and alliances, and the politics and cultures of terrorist and criminal organizations.

A hybrid human-knowledge and machine-learning EVS has the potential to be an effective system if the data mining process can be made dynamic as it develops its profiles, where the induction process requires that it be ongoing and incremental under the guidance of knowledgeable human investigators and analysts. Data relevant to assessing perpetrators is received continually over long periods of time. The structure of criminal and terrorist organizations and their methods of operation can change over time, so patterns of behavior may change without any real change in the severity of the terrorist threat. After 9/11, it is highly unlikely that a similar attack could be undertaken; however, other groups and methods exist by which other types of attacks could be attempted.

The hybrid combination of human knowledge and machine learning must go through an incremental process to couple the analyst's insight with the data mining power of algorithms and access to a vast network of data sources. For example, an initial profile may be developed by intelligence analysts to formulate a description of a perpetrator. This profile is then used to guide a more focused search to detect additional confirming or disconfirming evidence via machine-learning techniques. This back-and-forth process of human-computer interaction will continue as part of the development and the refinement of the perpetrator profile. The EVS must be capable of integrating data from a variety of disparate, dynamic, and possibly conflicting data sources. When new data or entirely new data sources become available, the system must be able to analyze the impact of the new data and incrementally reevaluate and revise its conclusions as it validates the identity of individuals.