Towards Parallelisation of Qualitative Spatial and Temporal Reasoning

This summary paper is concerned with algorithmic aspects of qualitative reasoning (QR), in particular qualitative spatial and temporal reasoning (QSTR). Fundamental reasoning techniques in this area are constraint propagators and algorithms for deciding core reasoning problems such as the satisfiability of constraint networks. Developing new algorithms for solving these tasks that exploit parallel computing hardware such as multicore processors or GPUs can be expected to lead to improved reasoning performance. We are motivated to study parallelisation of qualitative reasoning algorithms in order to determine these improvements. In the presence of machine learning applications, which often come with special hardware that supports parallelisation, efficient QR implementations that exploit such hardware can have important applications, for example online verification of AI systems performed in real time.