SIM-NET: A View-Based SemanticSimilarity Model for Ad Hoc Networks ofGeospatial Databases

摘要

Semantic similarity is a fundamental notion in GlScience for achieving semanticinteroperability among geospatial data. Until now, several semantic similaritymodels have been proposed; however, few of these models address the issues relatedto the assessment of semantic similarity in ad hoc networks. Also, several models arebased on a definition of concepts where features are independent, an assumption thatreduces the richness of the geospatial concept representation. This article presentsthe conceptual basis for Sim-Net, a novel semantic similarity model for ad hocnetworks based on Description Logics (DL). Sim-Net is based on the multi-viewparadigm. This paradigm is used to include inferential knowledge in semanticsimilarity, that is, the knowledge about implicit dependencies between features ofconcepts. In Sim-Net, assessing semantic similarity relies on the notions of SemanticReference Systems and Formal Concept Analysis (FCA), which are combined toestablish a common semantic reference frame for ontologies of the ad hoc networkcalled the view lattice. The Sim-Net semantic similarity measure distinguishesconcepts that belong to different or similar domains and takes into account theneighbours of a concept in the network. An application example is used to show thepositive impact of the properties of Sim-Net.