基于散列的语义服务匹配算法

Hashing-based Semantic Service Matchmaking

为提升服务匹配算法的性能,提出在散列过程中进行语义匹配。设计了基于框架的本体描述模型和支持多属性的服务描述模型。针对不同粒度模型子块的"约束结构"或"定义结构"设计散列函数,在按子块粒度递增的多重散列过程中完成服务匹配;设计了子块间的包容关系语义,并基于该语义进行散列冲突消解。理论分析表明该方法的匹配耗时指标为一区间常数。实验证明,该匹配策略相对于当前主流服务发现方法,提升查全率和查准率的同时,降低了匹配耗时;能快速建立支持高效服务组合的服务依赖关系图。

A novel strategy of semantic matchmaking during the process of hashing is proposed to improve the performance of service matchmaking algorithms.A frame-based ontology description model and a multiproperty supported service description model are designed to implement hashing functions corresponding to the constraint structure or definition structure of different granular lexical sub-blocks derived from the service description model.The service matchmaking is accomplished in a process of multilevel hashing with an ascending order of granularity.A collision resolution is constructed on the basis of the subsumption relations between blocks.According to the theoretical analysis,the time for service matchmaking via multilevel hashing can be regarded as a constant varying within a narrow range on the level of millisecond.In contrast to the currently popular service discovery methods,the strategy promotes the precision and recall as well as the time for matchmaking,and can quickly generate the service dependency graph to support an efficient service composition.