It is very important to ensure overall temporal correctness which is usually essential for the usefulness of execution results in a scientific workflow. Through checking temporal consistency to perform temporal verification is a crucial step to guarantee the on-time completion of scientific workflow. In order to improve the efficiency of temporal verification, many efforts have been dedicated to the selection of more effective and efficient checkpoints along scientific workflow execution for verifying temporal consistency so that temporal violations can be found and handled in time. In recent years, several checkpoint selection strategies for such a purpose have been proposed. Most of these strategies involve temporal constraints, but the relations among the temporal constraints of a scientific workflow have been ignored. The only existing strategy considering relations among temporal constraints is for the nested situation only. However, the intersectant, i.e. sequential twisted/joint, situation has never been taken into account. In this paper, a novel strategy for efficient temporal verification in scientific workflow based on temporal dependency by constraint adjustment is presented for the intersectant situation. Simulations show that the temporal verification efficiency of our strategy is significantly improved by over 90% than its representative counterpart.
