基于结构关系的过程模型推荐方法

Business process recommendation based on structural relations

目前已有的大部分过程模型推荐方法无法处理诸如循环相互交叠和不可见任务等复杂结构,也无法处理具有大量过程模型和活动节点的大数据集。此外,这些推荐方法在正确率和效率方面仍有很大的提升空间。由此,提出一种基于前驱活动序列和后继活动节点之间结构关系的推荐方法。该方法首先对所有业务过程模型中的不可见任务进行打标。然后,在指定关系集合中所需要的前驱活动序列长度之后,提取3种记录前驱活动序列和后继活动节点之间关系的关系集合。最后,使用提取出的关系集合推荐适当候选活动节点。实验结果证明,该方法能够处理复杂结构、不可见任务和大数据集,且在正确率和效率方面均领先于目前最好的方法。

Traditional business process recommendation methods cannot deal with silent transitions and complex structures such as interacting loops.They cannot handle big datasets,and there is room to improve accuracy and efficiency.A recommendation method based on structural relations between precursor activity sequences and successor active nodes was proposed.All silent transitions in business process models were relabeled.After specifying the length of precursor activity sequences,the relations between precursor activity sequences and successor active nodes were extracted using three strategies.With the relation sets,the proper nodes were recommended.The experimental results showed that the method could handle the complex structures,silent transitions and large datasets,and it was outperformed all other state-of-the-art methods.