Wednesday, 5/23/2007
11:45 AM - 12:15 PM
Level: Technical - Intermediate

If only every employee were an ontologist. Every document would be classified, indexed and tagged with consistent, actionable metadata. All managers would have access to 100% of the facts and trends recorded by their companies. And any employee could use concepts like “good customer,” “person of interest” or “local supplier” with repeatable precision. However, 85% of the information companies store is in freeform text, written by people with no interest in recording their thoughts in computer-friendly code. So, how do we solve the problem of making everybody an ontologist? Better still, why build ontologies in a special language? Why not reduce OWL, RDF, SCL, SWRL, SPARQL and other arcane programming languages to an artifact of compilation? Why not write, store, and communicate ontologies in plain English? Thanks to a new breed of natural language processing software, organizations can now automate collecting facts and building terminologies. Lessons Learned:

• Discover how natural language ontologies work
• Learn more about automated approaches to parsing sentences into actors, actions and objects
• Add adjectives, adverbs and prepositions to reliably capture the meaning in most text
• Compare basic tools for automating fact collection (ABox) and terminology building (TBox)
• Explore how to incorporate new tools into existing programs and processes

Chris Moran is the director of technology solutions for Attensity's Government System's team. He works with government agencies to integrate Attensity technology and develop solutions to meet critical missions. He joined Attensity after successfully implementing the company's software at the U.S. Department of Customs as an architect and designer. Before Customs, he worked as a rules designer at NEXTEL. He is the creator of Rules4J Rule, and The Java Developer's Journal recently published his article "Does Your Project Need a Rule Engine."