SOAR User's Manual.
Rept. for 1 Jan 82-15 Jun 85,
XEROX PALO ALTO RESEARCH CENTER CA INTELLIGENT SYSTEMS LAB
Pagination or Media Count:
Soar is an architecture for general intelligence that has been applied to a variety of tasks many of the classic artificial intelligence AI toy tasks such as the Tower of Hanoi, and the Blocks World tasks that appear to involve complex, non-search reasoning, such as syllogisms, the three wise men puzzle, and sequence extrapolation and large tasks requiring expert-level knowledge, such as the R1 computer-configuration task. This chapter provides a brief overview of the Soar architecture. In Soar, every task or problem is formulated as heuristic search in a problem space to achieve a goal. A problem space consists of a set of states and a set of operators that transform one state into another. Problem solving is the process of moving from a given initial state in the problem space through immediate states generated by operators until a desired state is reached that is recognized as attaining the goal. For each goal, there is always a single current problem space, state, and operator. The current problem space, state and operator, together with the goal, form a context. Goals and their contexts can have subgoals and associated contexts, which form a strict goal-subgoal hierarchy. Soar is an architecture for problem solving and learning, based on heuristic search and chunking. Keywords Cognitive architecture, Problem solving, Learning, Production system, Problem spaces, Goals.