Accessing Sources with Diverse Ontologies
Project Award Date: 0000-00-00
Individual agents lack a commitment to a common, pre-defined ontology but share a distributed, collective memory of objects. We are developing a system where each agent creates and learns conceptualizations, or ontologies, which are useful for its individual purposes, but it also shares its knowledge to improve group problem solving performance. Our work shows that multiagent learning of ontologies among individual agents with diverse conceptualizations is feasible and these learned ontologies can be used by the agents to improve group search performance for related semantic concepts through experience in the problem domain.
In our work we use supervised inductive learning to learn semantic concept descriptions in the form of interpretation rules. Each source agent contains a description of the information and knowledge contents of the resident source and a representation of the concept hierarchy it is familiar with. A representation hierarchy can contain bookmarks, or URLs, pointing to a semantic concept object, or Web page. Each set of bookmarks in a hierarchy is used as training instances for the semantic concept learner. The semantic concept learner learns a set of interpretation rules for all of the agent's known semantic concept objects. An entire set of these types of semantic concept descriptions can then be used for future semantic concept interpretation.