Use Cases for Ontologies in Information Fusion

Abstract

-Ontologies are becoming increasingly popular due to recent efforts to extend the capabilities of the World Wide Web through the addition of formal semantics. While ontologies have traditionally been used as precise languages to facilitate efficient exchange of information among people, the "Semantic Web" is extending this role to software agents. For this to be possible, ontologies must be formalized in languages processable by computers, such as OWL, the W3C's Web Ontology Language. The purpose of OWL ontologies is to permit software agents to understand web content and to interact intelligently with Web services (which may themselves be software agents). The use of such ontologies, however, need not be constrained to the Web. Recently, ontologies have found their way into higher-level information fusion where they are providing a means for describing and reasoning about sensor data, objects, relations and general domain theories. To the best of our knowledge, there is as of yet no documented effort to capture the main uses of ontologies in information fusion. In this paper we start filling this void by presenting a number of "use cases," i.e., scenarios of the use of ontologies in the context of higher-level information fusion. In this paper we develop use cases in which ontologies are used both for the fusion process itself and for the development of fusion systems. The use cases cover scenarios in which the agent roles are played by people, software or both.

Ontology alignment in geographical hard-soft information fusion systems

“…Kokar [41] provided details of ontology need in SA scenarios. For a surveillance task, it is important to determine the entities of the situation.…”

Tractor: A framework for soft information fusion

“…[4,20,25], a soft information fusion process must be capable of fusing the information in a newly processed message with the background information about that message. This allows future processes to make use of all the relevant information associated with the message, and places the contents of the message in context.…”

“…Since propositional graphs are types of semantic networks [24,34,35], they have all of the advantages of semantic networks, like RDF, for soft information fusion, such as [20]:…”

“…In common sense reasoning the information retrieval component can aid in knowledge retrieval [18,19,26] and situation dependent acting [2]. Contextbased information retrieval can aid information fusion processes that depend on efficient retrieval of information from ontologies [4,20,25].…”

Ontological Considerations for Uncertainty Propagation in High Level Information Fusion

“…There has been significant work in using ontological approaches in developing fusion techniques and some of these approaches have taken uncertainty considerations into account (e.g. [3] [4] [5] [6]). Various techniques exist to model and propagate uncertainty in a fusion process, with varying strengths and difficulties.…”