2010, Sofia Kontaxaki, "Development of Methods and Techniques for Geographic Knowledge Extraction" Print
Greek version / Ελληνική έκδοση


Title: Development of Methods and Techniques for Geographic Knowledge Extraction


Author: Sofia Kontaxaki




The topic of the dissertation is the development of methods and techniques for geographic knowledge acquisition. Initially, a survey is performed to review the work in progress regarding related fields such as geographical knowledge representation, natural language processing and controlled languages, information visualization and especially spatialization. In following, three novel methods and techniques are introduced. The first involves the creation of a new controlled language, which is used to derive spatial knowledge from geospatial databases in both human familiar and effective way. The purpose is to overcome the difficulties arising from the learning of specialized and hard to manipulate languages in order to formulate search queries. The second method consists in the creation of semantic labels for clusters of geographic concepts described by natural language definitions. The aim is to create labels by taking advantage of the semantic information immanent in the definitions of geographic concepts instead of resorting to external information. The resulting labels are structured so as to epitomize the meaning of the clusters and be readily understood by users in the context of semantic-based applications. The forth section of the dissertation concerns the field of information visualization and spatialization. It tackles both the issue of granularity and the way it is handled by existing spatialization methods. A new spatialization technique is introduced and integrated into a prototyping spatialization environment for representing, exploring, and extracting knowledge from large sets of multidimensional data at different levels of granularity, by making use of kernel density estimation and spatial metaphors.


[ Back ]





Do you like our new website?