Abstract: Aiming at the resources searching problem in distributed workflow management system, an improved searching strategy was presented. The nodes were clustered by using clustering algorithm, thus the logical topology and the physical topology were matched during search process. On this basis, ant colony optimization combined with interest was used to find the clues for search efficiency. Experiment results indicated that the proposed method could improve the workflow engine response times remarkably, and have advantages than flooding search and traditional ant colony optimization in search success rate, average response time and average message numbers.

Key words: workflow, resources searching, clustering, ant colony optimization

摘要： 针对分布式工作流管理系统中的资源搜索问题，提出一种新的搜索策略。该方法利用聚类算法将网分簇管理，使得查找过程中的逻辑拓扑和物理拓扑相互匹配，在此基础上采用结合了兴趣度的蚁群算法提供查找线索。实验表明，该方法提高了工作流引擎服务的整体响应时间，在查找成功率、平均响应时间和平均消息数上较泛洪查找和传统蚁群算法都更具优势。

关键词: 工作流, 资源搜索, 分簇, 蚁群算法