ODIF automates the sharing and integration of information among different organizations. ODIF applies a common ontology for integrating diverse data sources and uses knowledge extraction techniques and ontology analysis methods to extract knowledge from distributed, unstructured text sources.
An important challenge for any large enterprise is effectively sharing and integrating information among the organizations and diverse systems that make up the enterprise. If you’re the U.S. Air Force or the Missile Defense Agency (MDA), enterprises of staggering size, sweep, and application and systems diversity, this challenge is particularly complex.
The computing boom of the last 30 years has created a proliferation of specialized and sophisticated software tools for systems design, analysis, and decision support. Enterprises engaged in large-scale engineering projects have taken advantage of these innovations, dividing project work among smaller, more specialized teams, each with a specific focus and each with unique software support. The data produced by these specialized software systems is typically maintained in closed architecture databases, and, in the majority of cases, each software tool has its own private data repository. How can these different contexts—these different ontologies—effectively share the information they produce and maintain in support of the overarching goals of the enterprise?
KBSI, working in conjunction with the Air Force and MDA, has developed and demonstrated an Ontology-driven Integration Framework (ODIF) that automates the sharing and integration of information among different organizations by establishing a common ontology for integrating their diverse data sources. ODIF combines knowledge extraction techniques with ontology analysis methods to extract knowledge from distributed, unstructured text sources. This is the first step in facilitating the sharing of this knowledge among large-scale application project teams.
ODIF takes a novel approach to knowledge extraction and semantic mapping by applying ontology-assisted text mining and natural language processing methods. This Knowledge Extraction and Semantic Mapping (KESM) method facilitates text pre-processing from multiple, unstructured data sources, automated ontology extraction, and ontology learning during CCAFS systems development and operational testing. The ODIF architecture provides automated support for the KESM method and the targeted use of extracted ontology information to perform semantic mapping that facilitates operations enterprise integration.
Extending these capabilities is ODIF’s learning and self-adaptation mechanisms. ODIF enables automatic revisions to ontology knowledge by automatically learning new ontology concepts using the extracted information contained within multi-source text documents. This ontology learning capability is based on an approach that uses statistical and machine learning methods. A key outcome of applying the ontology learning process is a set of candidate “enhancements” to the application domain ontologies that include new ontology classes (types), characteristics (properties), and relations (associations).
The ODIF tool and framework represent a major advancement in the ability of large-scale organizations to identify, describe, organize, and, most importantly, share knowledge across systems and teams. In addition, ODIF’s scalable component-based software design strategy makes the tool amenable to rapid and cost effective integration and deployment into a host of military applications. While ODIF is successfully facilitating semantic information sharing and knowledge management for the Air Force and MDA, the ODIF concepts and technologies can also improve enterprise application integration for any large, distributed organizations performing knowledge management, collaborative planning and scheduling, supply chain management, and business intelligence.