Abstract: In the process of exploratory service composition,a large number of service composition processes are continuously generated,which can be used for learning.By mining these service composition processes and abstracting the mining results to service patterns,the efficiency of service composition can be effectively improved.At present,most of the work focuses on success service composition processes,and there is little research on failure service pattern mining of failure service composition processes.To solve this problem,an incremental failure service pattern mining algorithm was proposed for exploratory service composition,which extended the gSpan algorithm for focusing on the failure track of the failure service composition processes.At the same time,the incremental mining of the new service composition processes was used by the algorithm.Through the extensions,the efficiency of the algorithm was effectively improved.The experimental evaluation was explained to verify the algorithm,and the result showed that the efficiency of the proposed incremental failure service pattern mining algorithm had a significant improvement in mining failure service patterns compared with the failure service pattern mining algorithm and the gSpan algorithm.

Key words: exploratory service composition, service pattern mining, gSpan algorithm, incremental update

摘要： 为了有效提高服务组合效率,对探索式服务组合过程中大量可用于学习的服务组合流程进行挖掘,并将挖掘结果抽象为服务模式,其中针对失败服务组合流程进行的失败服务模式挖掘,提出一种探索式服务组合中的增量式失败服务模式挖掘算法,该算法对gSpan算法进行扩展,将挖掘聚焦于失败服务组合流程的失败轨迹部分,同时对新增服务组合流程进行增量式挖掘,以有效提高失败服务模式挖掘效率。实验评估表明,相比未采用增量式的失败服务模式挖掘算法和原始gSpan算法,所提增量式失败服务模式挖掘算法的挖掘效率均有显著提升。

关键词: 探索式服务组合, 服务模式挖掘, gSpan算法, 增量式更新