Hierarchical knowledge integration using layered conceptual graphs

Abstract

We describe the 'Hierarchical as View' approach to knowledge integration from heterogeneous sources. This is based on a novel representation called Layered Conceptual Graphs (LCGs), a hierarchical extension of conceptual graphs that address interoperability issues. We introduce LCGs on the basis of a new graph transformation system, which could be an appropriate hierarchical graph model for applications that require consistent transformations. We highlight a new type of rendering based on the additional expansion of relation nodes. Both the querying and integration capabilities of our approach are based on projection, an operation also defined for simple conceptual graphs. Integration is performed in a way that neither depends on the order in which the sources are combined nor on their physical availability. © Springer-Verlag Berlin Heidelberg 2005.