Qualitative Graphical Inference with Enhanced Knowledge Fusion

Qualitative reasoning has been widely applied in the analysis of complex networks, but the loss of quantitative information easily result in reasoning conflict. In this article, we adopt a hierarchical decomposition strategy of the network, and present a general framework of qualitative graphical inference based on enhanced multi-level knowledge fusion so as to avoid reasoning conflict. The framework can effectively obtain local knowledge from multilevel analysis and experts experience. Then the method combines the local experience with global structural features, as much as possible to eliminate information loss brought by structural decomposition. Finally, the case also illustrates that the method can effectively solve the trade-off problems in complex networks.