What is an ontology, and why do you need one?

An ontology is "The hierarchical structuring of knowledge about things by subcategorizing them according to their essential qualities." It is essentially the same thing as a detailed data model, database design, or data architecture for a given subject area, and can "... enable communication between computer systems ... independent of the individual system technologies ..."

Typical benefits of using an ontology in software include:

# Typical (sample) "Ontology-Encouraged" Software Design Features Technical & Economic Benefits 1 Data sourced from a single place, wherever possible Provide more consistent analysis results Easier and cheaper to maintain More time available for critical tasks (rather than resolving program update anomalies) 2 Data structures are documented and more freely available Easier and cheaper to support software Provide greater opportunities for asset re-use (cheaper in the long run, and improves analytic consistency) 3 Data structures isolated from software (e.g. APIs or data access layers) Easier to streamline and consolidate code, and thus easier and cheaper to maintain Easier and cheaper to migrate code to newer platforms and technologies, thus easier to maintain/increase system stability & performance /li> 4 Data available in multiple formats, as needed Easier to streamline, consolidate, and migrate code, thus easier and cheaper to maintain

Based on work for FTI’s clients, plus our own R&D, we conclude the following:

• Technologies, tools, and standards for ontologies are evolving, but are still several years behind those of other technologies (e.g. less work done, less stable, less adherence to common standards, less commercial adoption and interest to date), meaning that it is more difficult to move ontologies from a research to production environment.
• However, we can use "lessons learned" from our other research to identify the most promising avenues of ontology research (most promising=most likely to result in commercially-viable products and real operational results):
• Open source software generates wider acceptance
• Solutions and tools that incorporate widely-adopted standards (e.g. XML, etc.) offer a better chance of success
• Utilizing results from advanced university and government research can help "jump-start" projects
• Select projects very carefully - Not all projects will have a positive return on investment
• Targeted, "piece at a time" efforts (rather than an "all-in-one" approach) is likely to be more successful
• Subject area knowledge, technical experience, and "best practices" are still keys to implementation success