Research On Some Problems Of Spatio-temporal Reasoning

Abstract Contents: Recently, the successful research of qualitative representation and reasoning in the individual field of space and time and the urgent requests for representation and reasoning of the integration of space and time in application lead to the research of spatio-temporal reasoning which simultaneously supports space and timeSpatio-temporal reasoning is the reasoning of the objects that take the space and change with time. Its research includes the representation, modeling and reasoning of spatial, temporal and spatio-temporal information. Spatio-temporal reasoning has now become a hot spot in the relative research fields such as Artificial Intelligence, GIS, GPS, Advanced Autonomous Navigation of Robots, Advanced Vision, Image Understanding and Spatio-temporal database.This thesis is focused on some hot spot problems in spatio-temporal reasoning, especially in spatial reasoning. The research works in spatial, temporal and spatio-temporal reasoning, especially the works of spatio-temporal logic, are surveyed, the methods to construct better spatio-temporal models are discussed, and some feasible plans are proposed. The relativity and similarity of Allen's temporal interval calculus and topological region connect calculus RCC-8 are compared and analyzed, the combinations of the two calculi are studied and discussed, and more reasonable plans for the better combinations of the two calculi are proposed. A few hot spot questions such as the reasoning based on the topological relations between indeterminate regions, reasoning of the spatial region in dynamic environment and hybrid spatial reasoning approach that combines qualitative spatial reasoning with quantitative ones are analyzed and improved here, and new models and algorithms are put forward. The representation of spatial topological relation with rough set theory and a rough set extension method for spatial association rules are put forward, and spatial query predicates based on rough set are defined. These results have some significance to the research works in GIS, spatial data mining, spatial database, spatial and spatio-temporal query language.The main contributions and results included in the thesis are as follows: Firstly, the thesis summarizes and analyses the state of arts in spatio-temporal reasoningbased on the integration of spatial and temporal theory. The research works in spatial, temporal and spatio-temporal reasoning are surveyed and analyzed. It is pointed out that the knowledge representation of integrating spatial and temporal information is currently the main task.Secondly, the overview of state of arts in spatio-temporal logic is given. The international research works in spatio-temporal logic are summarized, which include spatial topological logic, spatial metric logic, the integration of metric and topological logic, Allen's temporal interval logic, propositional temporal logic, and spatio-temporal logic based on linear time, branching time, interval time. Their computational properties are analyzed here, and the existent problems and future emphasizing research works are pointed out.Thirdly, the relativity of Allen's temporal interval calculus and topological region connection calculus RCC-8 and research works in the combinations of the two calculi are analyzed, and reasonable plans for the better combinations of the two calculi are proposed. The famous region connection calculus RCC is proposed in spatial domain in terms of Allen's temporal interval calculus, the relation of them is very close. Based on the introduction of Allen's temporal interval calculus and topological region connection calculus RCC-8, their relativity in JEPD set, neighborhood graph, composition tables, CSR, path-consistency propagation algorithm and computational complexity are compared and analyzed. To further study Allen's temporal interval calculus and RCC-8 spatial calculus and explore their better combination methods, the typical methods for integration of the two is analyzed, and reasonable solving plans are derived through analysis and discussion.Fourthly, based on conceptual neighborhood structure in RCC and "fuzzy sets theory, the thesis presents extensional egg-yolk model and region movement model. Egg-yolk model is an extension of RCC, which can express topological relations between indeterminate regions. To represent vagueness in the relations between regions, our extensional egg-yolk model is presented, and to deal with regions that can move or change over time in dynamic environment, our region movement model is put forward. Based on the path consistency algorithm proposed by Allen, the fuzzy version algorithm is presented for the two models to reason with.Fifthly, based on rough set theory, the model of fuzzy region relation in RCC is constructed. Uncertainty management is necessary for real world application, especially for data mining. To improve uncertainty management of spatial data system, and to provide better representation of spatial data, the thesis presents a rough set representation method for expressing topological relations previously defined by RCC and egg-yolk model, and clustering model of the 46 relations...