Making Educated Guesses: A Knowledge-level Approach to Commonsense Reasoning
I am interested in understanding the nature of human commonsense reasoning by building computational models of it. Commonsense reasoning provides a solution to overcoming brittleness in Artificial Intelligence (AI) systems. It is also a necessary prerequisite to creating useful software that collaborates with human partners to accomplish tasks like damage control assessment, operations planning, sifting through on-line information for relevant data, teaching and tutoring, and developing complex scientific and engineering models. It also provides a very interesting vantage point to explore cognitive aspects of problem solving, categorization and memory.
In this talk, I will discuss my work on Back-of-the-Envelope reasoning, which involves making estimates when there isn’t enough information for an accurate answer, for questions from both commonsense and scientific domains, e.g., “How much money is spent on healthcare in the US?” A key contribution of this work is that a core set of seven strategies provides broad coverage, and is arguably the complete set of back of the envelope problem solving strategies. Using this work as an example, I will lay out my heuristic reasoning research program. Heuristic reasoning methods exploit the information processing structure of the reasoning system and the environment to produce reasonable answers when knowledge and/or computational resources for finding the perfect correct answer might not exist -- which is analogous to the human ability to make educated guesses.
Praveen Paritosh is a Ph.D. candidate in the Computer Science Department, with a specialization in Cognitive Science,
working with Ken Forbus and Dedre Gentner at Northwestern University. He graduated with a Bachelor of Technology from IIT Bombay,
India, in 1999. His senior thesis on Qualitative Simulation in Process Engineering was awarded the Indian National Academy of
Engineers award for innovative student projects.
Last modified: Fri Aug 8 11:38:17 CST 2006