Abstract: In the Internet of Services (IoS),aiming at the inefficiency of traditional service composition and selection methods under large-scale data,an efficient method for constructing service solutions based on the supply-demand bilateral patterns was proposed.The main idea was to mine the prior knowledge in the domain and use modularization to enhance the matching granularity to improve the efficiency of constructing service solutions.A mathematical model of the problem was given.After that,the intension tree based requirement model and the requirement pattern were proposed to support the rapid declaration requirements.The service pattern and its mining method were proposed to improve the reuse efficiency of historical service solutions.Further,the incidence matrix was proposed to depict the matching degree between the bilateral patterns in different service contexts.The precise algorithm,greedy algorithm and artificial bee colony algorithm based on bilateral patterns were proposed to construct service solutions for scenarios of different problem complexities.The experiments demonstrated the superiority of the algorithm using the bilateral patterns in terms of algorithm efficiency.

Key words: matching supply and demand, requirement pattern, service pattern, incidence matrix, intension tree

摘要： 在服务互联网环境下,针对传统服务组合和服务选择方法在大规模数据量下效率低下的问题,提出一种基于供需双边模式的服务方案高效构建方法。该方法的核心思想是挖掘领域中的先验知识,利用模块化的思想提升匹配单元粒度来提高服务方案构建效率。首先给出了问题的数学模型,然后提出基于意图树的需求模型和需求模式以支持需求快速声明,提出了服务模式及其挖掘方法提高历史服务方案的复用效率。进而,提出供需双边模式关联矩阵,刻画了不同服务情境下双边模式间的匹配度。针对不同问题复杂度的应用场景,分别提出了基于双边模式的搜索算法、贪心算法和人工蜂群算法用以构建服务方案。最后,通过实验证明了使用双边模式的算法在效率和解质量上的优越性。

关键词: 供需匹配, 需求模式, 服务模式, 关联矩阵, 意图树