Applying social network analysis to genetic algorithm in optimizing project risk response decisions

Risk interaction changes the probability of occurring a risk and also the impact of the risk, which calls for new approaches for making risk response decision (RRD). In this work, a simulation-based network model of abstracting the risk interactions is built for evaluating the RRDs. Meanwhile, genetic algorithm (GA) is tailored and improved for optimizing the RRDs, whose crossover operator is designed and enhanced by the social network analysis (SNA). Specifically, the application of SNA is two-fold: transforming the network for changing risk and risk interaction into the same network element; designing a new index to quantify the element's significance in the network view. Double-sorting map crossover is proposed for tailoring GA, and accordingly, multi-sorting map crossover is designed by integrating the quantified significance. An application example of the proposed approach is provided to illustrate its process and utility. Furthermore, contrastive analysis is conducted based on three different size cases, and the result demonstrates that the improvement in the GA is effective.