Manhattan Based Hybrid Semantic Similarity Algorithm for Geospatial Ontologies

摘要

The interest on the geo-spatial information system is increasing swiftly, which leads to the development of the competent information retrieval system. Among the several semantic similarity models, the existing models such as Geometric Model characterizes the geo-spatial concept using their dimensions (i.e. properties) and the Network Model, using their spatial relations which has yielded less precision. For retrieving the geo-spatial information efficiently, the dimensions and the spatial relations between the geo-spatial concepts must be considered. Hence this paper proposes the Hybrid Model which is the concoction of the Geometric Model's dimensions and the Network Model's relations using the Manhattan distance method for computing semantic distance between geo-spatial query concept and the related geo-spatial concept in the data sources. The results and analysis illustrates that the Hybrid Model using Manhattan distance method could yield better precision, recall and the relevant information retrieval. Further the Manhattan Based Similarity Measure (MBSM) algorithm is proposed which uses the Manhattan Distance Method for computing the semantic similarity among the geo-spatial concepts which yields 10% increase in precision compared to the existing semantic similarity models.