Agents with adjustable autonomy for scheduling in the competitive environment

This study addresses the problem of constructing an adequate strategy for multiagent scheduling in situations with high uncertainty and inconsistent goals. An automatic scheduling agent with adjustable autonomy builds the preliminary schedule and assists the human agents in modifying the schedule towards their specific goals. The degree of autonomy of an automatic agent is controlled via the syntax limitations on formulas, which are the basis of reasoning about the mental states of the other (human) agents. Autonomy adjustment therefore could be implemented during the scheduling session. We use the reasoning tool of metalanguage support (MS) with the embedded metapredicates of mental state and action to provide the automatic agent with the ability to model the human agents with various degrees of reflection. This tool gains the inference flexibility to represent the formal scenarios of multiagent interaction. Such MS capabilities as reasoning about action and change, belief, and temporal reasoning are illustrated by the scheduling system for the broadcast advertising industry. The automatic, incremental, interactive and multiagent modes of the automatic agent demonstrate the increasing complexity of the scheduler operation.