Generating the Numerical Control (NC) tool path for machining a complex shaped component is highly dependent on the proficiency of a Computer-Aided Manufacturing (CAM) programmer in manufacturing field, although the CAM systems now are highly integrated. A Computer-Aided Process Planning (CAPP) system, which can automatically extract the manufacturing features from the Computer-Aided Design (CAD) model and generate the machining process planning, has been expected for a long time. In this research, a graph-based CAPP system was proposed. It mainly includes four modules, data conversion module, feature classification module, feature combination module and process planning module. The first two modules claim a graph-based feature recognition method, output the recognized manufacturing features which are classified into four classes and defined as specific types. The feature combination module generates different paths to combine manufacturing features from a goal model into raw material shape by four kinds of combination methods corresponding to the four classes. Finally, the process planning module will give a cost estimation of all those paths with the consideration of manufacturing resources and time cost. A relatively optimized machining method and machining sequence will be generated as the output of this proposed system.