A Review of Interoperability Techniques for Models, Data, and Knowledge in Environmental Software

Abstract: An analysis of large and complex systems such as environmental systems linked to socio-economic systems usually requires several simulation models. These simulation models must be able to in-terface with each other in the conceptual level but there may also be some overlap in their application do-mains. The way simulation models are used normally generates large amounts of data, which need to be ex-plored or mined for the analysis and possible decision process. We propose a set of knowledge-based agents, which share a common ontology to aid in the modeling and analysis in large and complex systems. An agent is an active software entity, which is designed for a certain task or tasks. The main requirements for such knowledge-based agents include ontology representation and storage, communication capabilities, and data inspection tools. For example an agent should be able to actively study, e.g., a hard disk volume for available information, discuss the findings with a knowledgeable person, and then produce and communicate a report. The paper discusses the theoretical, technological, and other implications of the proposed approach versus other paradigms. In particular we are interested in using XML as the main tool for ontology representation and in supporting more than one level of conceptualization of the problem domain. Some technologies, nota-bly application servers, also implement some services, which could be utilized for ontology sharing and agent communication. Benefits and constraints of these are discussed