Research On Indeterminacy Management In Spatiotemporal Databases

In the complex realities, spatiotemporal indeterminacy abound. The uncertainty of entities'knowledge, fuzziness of concepts and the derivation of data introduce the indeterminant characteristics of spatiotemporal objects. The topic of indeterminacy management is a new issue in spatiotemporal database research. To survive in the complex spatiotemporal reality, sound solutions that incorporate the spatiotemporal indeterminacy must be carefully designed. This dissertation focuses on the data models, analysis operations and querey process of indeterminacy spatiotemporal database. The main contributions of this dissertation are summarized as follows:Firstly, a new spatiotemporal abstract model based on grey sets is proposed. It uses a pair of membership functions and the grey region between them to represent the indeterminacy of objects. Compared to fuzzy sets models, this model can manage the objects with high level indeterminacy, for example, with unknown membership functions. Compared to rough set models, this model can give the quantitive description of objects'indeterminacy, provide a unified representation for objects with different levels of indeterminacy. Based on the abstract model presented, a type system is proposed, including the formal definitions for types and different operations on them.Secondly, for indeterminate evolving regions, a new topological relations model based on the RCC calculus is proposed. It checks the correspondence between 3d topological relations and spatiotemporal relations. Then it extends Egg/Yolk model to the higher dimension that can describe the approximate topological relations for indeterminate evolving regions. The result is a collection of formal definitions of relations clusters that have different spatiotemporal nature.Thirdly, a new grey modeling method used for the calculation of the discrete models of indeterminate continuously moving objects is presented. The GM(1,1) model generated from the snapshots sequence reduced the randomness of discrete snapshots and generate the holistic measure of objects'movements. Comparisons to traditional linear models show that when information is limited this model can be used in the interpolation and near future prediction of uncertain continuously moving spatio-temporal objects.Fourthly, a new notation for indeterminate spatiotemporal queries is proposes. The notation is based on the temporal, spatiotemporal and indeterminant spatiotemporal predicates. It is simple and extensible and can provide a consistent way to refer to indeterminate temporal or spatiotemporal queries, which can also be easily applied to prototype implementation and further research. Using this notation, a framework for indeterminant spatiotemporal query processing is introduced, using the Minimum Bounding Rectangle pairs of indeterminant spatiotemporal objects to filtrate unrelevant calculation, querys are reconstructed and the processing efficiency enhanced. Finally, a histogram-based grey estimator for spatiotemporal selectivity estimation is introduced, the basic idea is that although the individual object's movements has much randomness, the overall data distribution varies gradually with time, due to the continuity of movement. Using prediction models on the history and present query results, it will be more accurate to get query estimate than using existing linear extrapolating spatiotemporal selectivity estimation techniques. To enhance the estimation performance, grey prediction model GM(1,1) is used, which can reduce the randomness inside the history query results sequence and generates its holistic measure. Comparisons to traditional approaches show that as randomness of history query results increasing, the near future prediction results of spatiotemporal window queries remain accurate and stable.