Abstract: Large manufacturing scale and process complexity of modern manufacturing companies are posing great challenges to production scheduling. On the one hand, it becomes more and more difficult to build accurate mathematical scheduling models for manufacturing processes based on conventional methods; on the other hand, the lack of accurate and real-time parameters of scheduling models may lead to low effectiveness of these models. Therefore, following the information-extraction path of data--information--knowledge--decision, it is necessary to explore innovative theories and methodologies of data-based scheduling for complex manufacturing processes. In this paper, based on systematic research of related works worldwide, a data-based scheduling framework composed of a data layer and a model layer for complex manufacturing processes is proposed. Related theories, methodoloies, and technologies for this scheduling framework are discussed as well.