Abstract: Completion time and accuracy are the most important attributes of processing products' business, however, single attribute optimization algorithm only can optimize one attribute, let alone balancing these two attributes dynamically. In order to fix this problem, a serial reduction algorithm based on directed acyclic graph is proposed. The algorithm constraints each tasks' activity section by using backwards iterative eduction to choose appropriate service for achieving the optimization of these two attributes. Statistical data shows that serial reduction algorithm does get an optimized path with time and accuracy balance, but its optimization efficiency is limited by the completion deadline and the number of tasks. The research conclusion provides a certain theoretical reference for the optimization of production scheduling.