Catalogue of Artificial Intelligence Techniques
Keywords: ABSTRIPS, abstract space, criticality, ground space, learning, preconditions
Categories: Inference and Reasoning , Planning , Problem Solving
Author(s): Toby Walsh
Abstraction is a powerful heuristic used in many types of problem solving including planning, common-sense reasoning, theorem proving and learning. It usually involves a mapping from the original problem representation (often confusingly called the ground space) onto a new and simpler representation (often called the abstract space). One of the most famous uses of abstraction was in the ABSTRIPS planning system where the preconditions of Operators were dropped according to their criticality, a numerical measure of their importance.
- Giunchiglia, F. and Walsh, T., A theory of abstraction Artificial Intelligence 56 (1992), 323--390.