From the introduction: The amount of detailed and spatial diverse data available has increased dramatically in recent years, due to the advent of novel technologies, which facilitate acquisition, sharing, and storage of spatial data. Well-known examples of such technologies include global positioning systems, remote sensing, geosensor networks, spatial data infrastructures, location-based services, and volunteered geographic information. This increase can be considered as “the most dramatic shift in the environment for geographic research in the history of science”. Additionally, today’s computing power is greater than is has been ever before, and, according to Moore’s Law, is even expected to increase further. Geographic information science (GIScience) has moved from a data-poor and computation-poor to a data-rich and computation-rich environment. However, the richness of spatial data contrasts with a lack of profound theories and hypotheses.
Large spatial databases presumably contain hidden and unexpected information, which cannot be discovered using traditional statistical methods that typically require a priori hypotheses and also cannot be scaled to handle large amounts of data. However, the wealth of spatial data cannot be fully realized, when such information is neglected. Therefore, GIScience urgently needs new methods and tools to reveal hidden and unexpected information within large spatial databases, and to transform it finally into new and potentially useful knowledge.
To address this need, geographic knowledge discovery (GKD) has emerged as a research field. GDK is a subdomain of the more general knowledge discovery in databases (KDD). Generally speaking, it can be considered as a fusion of spatial analysis and KDD. In this sense, important issues of spatial analysis persist also in the GDK domain.
This thesis develops novel methods applications of spatial data mining, an important step of GDK, and thereby addresses important research challenges of spatial analysis. Therefore, the following two sections give a theoretical background of GDK and spatial data mining, and introduce three important analytical issues involved in this context.