AIMSMART: Adaptive, Interactive MDA Semantic Model-based Articulation

Period of Performance: 07/10/2013 - 07/14/2015

$1000K

Phase 2 STTR

Recipient Firm

Cycorp, Inc.
7718 Wood Hollow Drive Suite 250
Austin, TX 78731
Principal Investigator
Firm POC

Research Institution

George Mason University
4400 University Drive MS 4C6
Fairfax, VA 22030
Institution POC

Abstract

We will develop automatic ontology-aligning software, leveraging the enormous existing Cyc AI system to drive the formulation and ranking (pro/con argumentation) of hypotheses about term-term relationships, especially where the relationship is not simple 1-to-1 correspondence. Multiple Cyc micro-theories (contexts) will be created, to hold incommensurate alternative mappings, and logical consequences within each will be checked for internal inconsistency with each other and scored for their accuracy of external predictions about data not yet integrated. Secondly, we will develop a suite of semi-automatic software power tools to help MDA users control, review/navigate, and correct the output of that automatic aligner. Each part (the automatic aligner and the alignment manager toolkit) will be built by augmenting Cyc's existing knowledge base, ontology, inference engine, and UI's, using MDA sources and MDA ground test planning tasks to drive the development and testing. GMU will spearhead liaising with MDA experts, and Cycorp will spearhead the Cyc extensions.