"Automating Commonsense reasoning for Elementary Physical Science" PI: Ernest S. Davis New York University
This goal of this project is to develop a symbolic knowledge base that integrates commonsense understanding of solids, liquids, and gasses with high-school level physics. The knowledge base is intended to support simple reasoning--projection, planning, and design--about physical systems such as barometers, pistons, beakers, pipes, valves, pumps, boxes, and doors. In addition to commonsensical and textbook physics knowledge, the knowledge base will incorporate a theory of action, a representation of partial spatial information, and multiple levels of abstraction. In the near future, the broader impact of this project includes training doctoral students and writing a semi-popular book discussing how commonsense knowledge interacts with mathematical calculations. In the long run, this research could substantially contribute to a large-scale knowledge base of scientific knowledge; autonomous robotics in open environments; automated design for physical tools; automated tutoring systems for elementary science; and cognitive theories of how the understanding of formal science is grounded in ordinary experience of the physical world.