Focus Research Areas: 1. Information Fusion a. Text Analysis and Mining b. Genetic Algorithms c. Bayesian Networks d. HPC and Peer-to-Peer Computing Currently, there is a tremendous need for intelligent information retrieval and analysis. As sensor networks become larger and dependencies on these networks increase, this need will only become greater. In addition to this need, there is a huge challenge of creating useful knowledge based on raw data obtained from a variety of sources. There is no single source of information that will provide all the required details or “pieces of the puzzle.” Different sources provide specific and unique pieces of information. There is a challenge to find the need sources. There is a challenge to collect the needed information from these sources. Finally, there is a challenge to “connect” the information from these sources together to form actionable knowledge. Consequently, the focal research area is Information Fusion. This is a broad area with plenty of potential for new research opportunities and is of primary importance to both DOD and DHS.e. Agent Technology Within this area of Information Fusion, there are a number of enabling technologies, each with its own research area and challenges. Perhaps the most predominant of these is Text Analysis. A vast amount of information available via the Internet is in text format. This format is readily and easily available, and is much more “processing-friendly” than other formats such as images, video, and audio. However, the most significant challenge for Text Analysis is the semantic understanding of the data. Current analysis is dependent on statistical measures of the text. This limits the useful of text information. New semantically oriented techniques will be needed to further extend the usefulness of text. The next enabling technology for information fusion is Genetic Algorithms. This type of algorithm is specifically intended to be a searching algorithm for very large search spaces. One of the challenges posed by Information Fusion is searching for pertinent information (or a combination of pertinent information) amid an overwhelmingly large collection of data. Genetic Algorithms have been proven to be invaluable for a variety of problem domains. However, two primary challenges still remain in the use of Genetic Algorithms for Information Fusion. These are the problem domain encoding and the fitness function. Addressing these challenges would bring the power and flexibility of Genetic Algorithms to Information Fusion. Not all available information is trustworthy or reliable. There is often some variable level of reliability that can be associated with information. The use of Bayesian statistics provides a means for quantifying this reliability. By quantifying, the information can be effectively used and accepted within an understood context. Therefore, Bayesian statistics is the next enabling technology that will need to be leveraged to make information fusion effective to the end user. Finally, the next enabling technologies are HPC, Peer-to-Peer computing, and Agent technology. There is simply too much information to be processed and analyzed by a single computer. Therefore, it is essential to leverage distributed and parallel processing systems. |
 |
 |
 |
 |
