Figure 4 shows an example of an inital world state that has been semi-automatically built by the OWLS-Xplan editor. Otherwise the service execution could fail since checking the service preconditions may reveal that they are not satisfied in the actual world state. OWLS- Xplan would only use a service description during its planning process, if the additional precondition predicate ” agentHasKnowledgeAbout ( Y ) ” on available knowledge about service input data is satisfied such that X = Y holds. Sim- ilarly, each input variable Y is mapped to an input parameter Y of an PDDL action complemented by precondition predicate ” agentHasKnowledgeAbout ( Y ) ”. In particular, every output variable X is described in, and added to the current (physical) planning world state by means of a newly created add-effect predicate in PDDL uniquely named ” agentHasKnowledgeAbout ( X ) ”. This problem can be solved by mapping the service output parameter X to a special type of the service hasEffect pa- rameter. However, PDDL does not allow to describe non- physical knowledge such as train connections produced as an output of a service. The problem is that the service hasEffect condition explicitly describes how the world state will change while this is not nec- essarily the case for an hasOutput parameter value, though it indeed could implicitly influence the composition planning process. the conversion of the output of an individual OWL-S service, that is the information the service offers to the world, to PDDL turns out to be more problematic.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |