基于混合遗传算法的工作流重构研究

随着知识抽取、数据挖掘等技术的兴起，基于日志的企业流程重构引起了学术界和工业界的高度重视。本文针对目前重构算法大都采用本地策略因而无法保证最优挖掘以及算法对噪声不强壮的情况，提出了一种新的工作流重构算法。该算法结合遗传算法和模拟退火算法的思想，通过对流程活动依赖关系的度量，构建因果关系矩阵映射流程实例（CASE）作为种群染色体，进而完成初始种群的构建，有效地提高了收敛速度，减少时间开销，同时避免了早熟收敛，因此该算法在有效处理噪声问题的基础上，实现了对流程模型的正确挖掘。

Study of Workfiow Reconstruction Based on Hybrid Genetic Algorithm

With the rise of the technology of knowledge extraction and data mining, it is highly noted in both research and industrial field that the workflow reconstruction based on process log. However, current techniques using local strategy can not guarantee that the outcome of the locally optimal steps will result in a globally optimal process model. And they are not robust when dealing with the presence of noise in the logs. To tackle these problems, this paper pro- poses a completely new approach based on hybrid genetic algorithm. As can be expected, this approach is able to deal with noise and incompleteness. However, it is not easy to represent process model properly in a genetic setting. In this paper, we define activity causal matrix as a representation for individuals. With the combination of genetic algorithm and simulated annealing algorithm, the reconstruction method can effectively deal with noise and incompleteness and correctly discover the process model.