Graph mining based knowledge discovery in designing decision-making context models

At present, researches on design rationale focus on the model representation and retrieval, which lack deeply mining in the design rationale model and cannot support innovative design. This paper proposes a decision-making context model and a graph mining based method to mine the model. This method could get large amount of tacit design rules, design consensus and design evidences in decision-making context model, which has great significance for innovative design. At first we calculate the similarities among the nodes in the models, and then find the similar nodes in different graphs and then unify the graphs, nodes and edges. Second, generate frequent edge set based on support degree, select the edge with the highest degree as the start, add outer edges in frequent edge set and generate frequent subgraph set. At last, modify and explain the graphs in the frequent sungraph set which results in final knowledge. At the end of this paper, we take the design process of automatic marking machine as example and get knowledge about cam design, transmission mechanism and feed mechanism, which substantiates the effectiveness of the mehod.