SILS MRAT: A Multi-Agent Decision-Support System for Shipboard Integration of Logistics Systems
CDM Technical Report: CDM-15-04, January 1, 2004, pages 1-141.
This report describes work performed by CDM Technologies Inc. on subcontract to ManTech Advanced Systems International, Inc. (Fairmont, West Virginia), and under sponsorship of the Office of Naval Research (ONR). The principal aim of the SILS (Shipboard Integration of Logistics Systems) project is to provide a decision-support capability for Navy ships that integrates shipboard logistical and tactical systems within a near real-time, automated, computer-based shipboard readiness and situation awareness facility. Specifically, SILS is intended to provide the captain of a ship and his staff with an accurate evaluation of the current condition of the ship, based on the ability of all of its equipment, services and personnel to perform their intended functions.
The SILS software system consists of two main subsystems, namely: the SILS IE (Interface Engine) subsystem for information interchange with heterogeneous external applications, developed by ManTech Advanced Systems International; and, the SILS MRAT (Mission Readiness Analysis Toolkit) subsystem for intelligent decision-support with collaborative software agents, developed by CDM Technologies. This report is focused specifically on the technical aspects of the SILS MRAT subsystem.
The automated reasoning capabilities of SILS MRAT are supported by a knowledge management architecture that is based on information-centric principles. Such an architecture utilizes a virtual model of the real world problem situation, consisting of data objects with characteristics and a rich set of relationships. Commonly referred to as an ontology, this internal information model provides a common vocabulary and context for software agents with reasoning capabilities. The concurrent need for incremental capability increases implies a steadily increasing data load from diverse operational (dynamic) and historical (static) data sources, ranging from free text messages and Web content to highly structured data contained in consolidated operational data stores, Data Warehouses, and Data Marts. In order to provide useful high-level capabilities the architecture is required to support the transformation of these data flows into information and knowledge relevant to the concerns and operational context of individual shipboard users. Accordingly, the system must be capable of not only storing data but also the relationships and higher level concepts that place the data into context. For this reason, to manage an increasing number of relationships and concepts over time, the SILS MRAT subsystem was designed to employ a formalized ontological framework.
There were four additional considerations in the selection of the overall SILS architecture. First, utility to support a useful level of automated information management (i.e., the ability to collaboratively analyze data, monitor dynamic operational context, formulate warnings and alerts, and generate recommendations). Second, flexibility to accommodate contributions from multiple team members that may employ differing technologies and implementation paradigms. Third, scalability to allow a progressive increase in the breadth and diversity of the data sources, the volume of data processed, the number of validated components, and the intelligence of the tools (i.e., agents). Fourth, adaptability to facilitate the tailoring of the information management capabilities to different data sources and existing data environments. The current SILS architecture addresses these desirable characteristics by partitioning the system into a lower-level data collection and integration layer, a higher-level information management layer (SILS MRAT), and a translation facility that is capable of mapping the data schema of the lower layer to the information representation (i.e., ontology) of the upper layer (SILS IE). The higher-level information management layer provides a collaborative, distributed communication facility that supports the development of semi-autonomous modules of capability referred to as agents. The agents employ the formalized ontology supported by the communication facility to collaborate with each other and the human users in a meaningful manner.