Mining Spatial Temporal Network Data for Threats to Homeland Security
The military is often tasked with knowledge discovery in large spatio-temporal databases.  Spatio-temporal data has been gaining prominence in database applications, partly because technology has increased our ability to assemble very large sets of spatio-temporal data.  As a result, knowledge discovery within large spatio-temporal datasets will be an important task in military and homeland security applications.  The sheer amount of available geospatial data, from sources such as cell phones or car GPS devices, makes spatio-temporal data mining even more useful and important.  However, this large amount of data also makes it difficult to perform knowledge discovery because of the costs.  By developing good approximations for calculations such as distance over a network, the viability of data mining of spatial data is greatly increased. Knowledge discovery using data within a spatial network requires approximating network distance between locations in the network.  The Road Network Embedding (RNE) transforms network locations into a space where the chessboard metric, a simple O(1) metric, is used to calculate an accurate approximation of network distance.  With this approximation various data mining tasks such as clustering, outlier detection, co-location detection become much simpler because the network distance approximation has now low cost. 

METOC GIS Data Integration Research Project
The ability to process, in real-time, and to quickly visualize activity patterns in large multidimensional data sets such as the Navy’s METOC (Meteorological and Oceanographic) data sets is another critical task in military and homeland security applications. Some of the objectives of this project   include the efficient indexing and querying of very large METOC data sets stored in netCDF or GRIB files, and the 4D visualization of their activity patterns using existing GIS visualization tools.

METOC Data Asset and Tactical Decision Aid Integration with Navy Service Oriented Architecture Systems
The objective of this project is to study how to properly facilitate and integrate the vast, heterogeneous data assets in Navy’s Net Centric framework, and design necessary interfaces to enable applications such as Tactical Decision Aids to access data assets and fully realize the Navy METOC (Meteorology and Oceanography) Data Services Framework (NMDSF) for tactical advantage.  At the center of this project is the design and prototype of an interface that improves performance and ease-of-use of existing applications, and assists the METOC data assets consumers, the warfighters and their supporting forces, to make JBML (Joint METOC Broker Language) requests based on semantic and case-based information more effective.

Application Reengineering Technologies Research
The objective of this project research is to identify and select for demonstration state-of-the-art software renovation tools and services such that the large inventory of Navy’s outdated legacy programs can be renovated into the state-of-the-art environment based on the service-oriented architecture (SOA).  In order to achieve more effective use of shrinking IT funding, the tools should promise a 10:1 advantage over re-development.  The outcome of the project will be identified tools and technologies for cost effective renovation of Navy legacy systems, and the ROI estimations for operating and maintaining of the renovation technologies. 

Automatic Orchestration of Geospatial Web Services
Geospatial Web services are distributed systems accessible via the Internet that provide geospatial data products or perform data processing. An orchestration or composition of Web services is made when multiple Web services are combined to perform a task. For example, a service which provides road network data may be connected to a service which provides satellite imagery. By combining these two services, a new composite service is created which creates a new data product of a road map on top of satellite imagery. Creating orchestrations from geospatial Web services will replace functions which used to require multiple data providers and analysts coordinating tasks. Creating Web service orchestrations can be a complicated and technical process. Automatic orchestration of geospatial Web services will provide non-technical users with the ability to create orchestrations with a simple request. Services will be composed in order to fulfill the request and return the appropriate data or perform the desired processing. A combination of syntactic and semantic analysis will be required for automatic orchestration of geospatial Web services. Improved type analysis will be required to determine if input and output parameters are compatible even if their data structure is different. Semantic reasoning is necessary to ensure functionality of services are compatible, and an learning system for semantic annotations will be required to annotate services whose metadata is incomplete or not-standard. Performance will be important to the usability of the system. To that end an index to improve service search must be used to enable fast searching over both syntactic and semantic properties of a service. The resulting geospatial Web services automatic orchestration system will significantly reduce the time and cost of geospatial processing and product creation.