Smith, D. E.

Conditional nonlinear planning


Conditional actions have two possible outcomes P or P. During the first planning pass, the planner assumes that all of the conditional actions are unconditional with a single outcome P. Warplan-C attempts to develop a plan using one branch of each conditional and then reinvokes the planner to plan for each dangling'else' branch. Figure 1: Operation of Warplan-C The planner develops an unconditional plan assuming that the conditional action has outcome P (Fig 1.a.). A conditional plan consists of a set of steps, reason and context labels for those steps, a set of ordering constraints, a set of bindings for variables in the plan, a set of causal links and a set of conditioning links. Solid lines, dashed lines and thick lines represent causal links, ordering constraints and conditioning links, respectively.

Controlling recursive inference


"Loosely speaking, recursive inference occurs when an inference procedure generates an infinite sequence of similar subgoals. In general, the control of recursive inference involves demonstrating that recursive portions of a search space will not contribute any new answers to the problem beyond a certain level. We first review a well-known syntactic method for controlling repeating inference (inference where the conjuncts processed are instances of their ancestors), provide a proof that it is correct, and discuss the conditions under which the strategy is optimal. We also derive more powerful pruning theorems for cases involving transitivity axioms and cases involving subsumed subgoals. The treatment of repeating inference is followed by consideration of the more difficult problem of recursive inference that does not repeat. Here we show how knowledge of the properties of the relations involved and knowledge about the contents of the system's database can be used to prove that portions of a search space will not contribute any new answers." Artificial Intelligence, 30 (3), 343-89.

An Overview of Meta-Level Architecture


An Overview of Meta-Level Architecture Michael R. Genesereth Stanford University Computer Science Department Stanford, California 94305 Abstract: One of the biggest problems in AT programming is the difficulty of specifying control. In a data base management system, data base accesses and modifications would be base-level actions, and:;ll query optimiTation would be mcta-level. In an autonralcd deduction system, inf2rencc steps would be base-level, and the activity of deciding the order in which to perform infcrencc steps would be meta-level. 1 tie central idea in meta-level architecture is the use of a declarative language for describing behavior.