Universal subgoaling
John E. Laird · 1984
The goal of this thesis is to develop a problem-solving architecture where all appropriate knowledge is brought to bear to control all aspects of problem-solving behavior. Such an architecture allows the creation of completely reflective problem solvers. We identify a specific capability called universal subgoaling that together with previous work on a universal weak method makes this possible. With universal subgoaling, subgoals arise whenever there is a difficulty in performing the problem-solving functions. In a subgoal, the problem solver brings its knowledge to bear to reason about and eliminate the difficulty. We identify a set of requirements that must be met by any problem-solving paradigm and architecture that realizes universal subgoaling. We then describe an implementation of universal subgoaling within Soar, a production system based on search in a problem space. We provide two dimensions of universal subgoaling: (1) Soar creates subgoals whenever difficulties arise in any aspect of problem solving, (2) it is possible to encode the knowledge required to produce the weak methods so that the knowledge is used whenever it is needed. As part of the second demonstration, we provide a useful taxonomy of the weak methods based on the knowledge required to encode the in Soar. *This research was sponsored by the Defense Advanced Research Projects Agency (DOD), ARPA Order No. 3597, monitored by the Air Force Avionics Laboratory Under Contract F33615-78-C-1551. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Defense Advanced Research Projects Agency or the US Government.