Abstract: There is a gap on expressive power and reasoning ability between the action theories which arc based on first or higher-order logics and the action theories which arc only propositional. As a kind of dynamic extensions of descriplion logics, the dynamic description logic DDL provides an approach for describing and reasoning about actions. An aclion theory based on DDI_ was presented and studied systematically. Firstly, based on a representation of static domain knowledge with description logics, the parameterized atomic action definitions and the parameterized complex action definitions were introduced for describing the knowledge of actions; both of these knowledge, and together with the knowledge on the state of the world,are unified as a DDL-based knowledge representation system. Secondly,many reasoning tasks on the knowledge represented in this system were formally defined; corresponding reasoning mechanisms were also provided. Finally,the application of this action theory for the modeling of intelligent agents was discussed. The action theory based on DDL offers not only considerable expressive power but also attractive reasoning services; it is suitable for the description and reasoning of actions in the environment of the semantic Web.

Key words: Keywords Dynamic description logic, Knowledge representation and reasoning, Action theory, Intelligent agent, Semantic Web