A Behavior Strategy for Agents in the Semantic Web Using Dynamic Description Logics

The paper presents a behavior strategy for agents in the Semantic Web which implements the philosophy underlying the decision process of human beings. It employs dynamic reasoning to simulate the scenario analysis in human decision and introduces a notion of favorable factor to measure each possible choice's utility. We dynamically extend description logic ALCO@K in a proportional dynamic logic manner, and then encode agents' statistic knowledge in ALCO@K, describe their abilities to change the world (by performing available Web applications) in terms of their preconditions (epistemic queries to the knowledgebase) and effects (possibly negated ALCO@K assertions involving only primitive concepts). With these knowledge formalized in the formalism, we propose faithful algorithms to compute the most favorable action.