Combinative Reasoning With Rcc5 And Cardinal Direction Relations

Juan Chen,Dayou Liu,Changhai Zhang,Qi Xie

DOI: https://doi.org/10.1007/978-3-540-76719-0_13

2007-01-01

Abstract:It is inadequate considering only one aspect of spatial information in practical problems, where several aspects are usually involved together. Reasoning with multi-aspect spatial information has become the focus of qualitative spatial reasoning. Most previous works of combing topological and directional information center on the combination with MBR based direction model or single-tile directions. The directional description is too approximate to do precise reasoning. Different from above, cardinal direction relations and RCC5 are introduced to represent directional and topological information. We investigate the mutual dependencies between basic relations of two formalisms, discuss the heterogeneous composition and give the detail composing rules. Then point out that only checking the consistency of topological and directional constraints before and after entailing by the constraints of each other respectively will result mistakes. Based on this, an improved constraint propagation algorithm is presented to enforce path consistency. And the computation complexities of checking the consistency of the hybrid constraints over various subsets of RCC5 and cardinal direction relations are analyzed at the end.