When: Wednesday, September 06, 10:30 a.m.- 1:00 p.m.

Where: 4625 Wean Hall

Alex Yahja, COS PHD Student

Thesis Oral

Abstract:
Simulation models are growing in size and complexity which requires significant time and resources to validate. This dissertation describes a knowledge-based, ontological, and simulation-aided approach for automating model-validation of simulations. A tool implementing this approach (WIZER) is capable of automating the validation of simulations. WIZER can symbolically characterize simulation outputs, perform knowledge-based, causal, and ontological reasoning, and perform simple hypothesis search. Once a simulation model is validated, WIZER shows a way to improve it by perturbations of the model using ontological and knowledge-based inference. Several scenarios on two testbeds are used to show how WIZER works. The testbeds are BioWar and CONSTRUCT. BioWar is a city-scale multi-agent social-network model of the spread of diseases. CONSTRUCT is a model of co-evolution of structure and cognition constituting social networks. The extensions of WIZER could pave a way to the Simulation and Knowledge Web. The implications of WIZER cover several fields including modeling and simulation, organization modeling, management, policy analysis, knowledge management, and computational history.

<< Back