摘要
针对基于内容的发布/订阅系统匹配算法中的计数算法和树形算法在多操作符匹配结构中效率不高的问题,提出了一种基于订阅划分的多维索引并行匹配算法(SPMP-match).该算法利用二叉树原理对订阅进行分组,然后组织成树形索引结构,采用一种边匹配谓词边观察的策略,及时终止无成功匹配订阅的情形.该算法仅对可能和一个事件匹配成功的所有订阅进行匹配,不会产生无价值的时间消耗,并且在查找匹配成功的订阅时,只须要查看所有可能和某事件匹配成功的订阅表,避免了扫描整个订阅集合.
Counting algorithm and searching tree matching algorithm are two chief types of matching algorithm in content-based publish/subscribe systems.In order to improve the inefficient performance of the two kinds of algorithm practiced in the condition of multi-operator structural requirement,subscription partition based multi-index parallel matching algorithm(SPMP-match)was proposed in this paper.Principle of binary tree was used to grouping subscriptions,then organized into a kind of tree index structure.Therefore,the algorithm can promptly terminate the case with a not successful match,with the strategy of matching predicate and observing.SPMP-match just matches the all possible subscriptions matching the coming event without worthless time consuming.When searching the successful matching subscription,it just needs to search the subscription table,including all the possible subscriptions matching a certain event,so as to avoid scanning the entire subscription set.
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2013年第S2期184-187,191,共5页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家科技型中小企业技术创新基金资助项目(13C26213201857)
江苏省科技型企业技术创新基金资助项目(BC2012210)
江苏省教育厅自然科学基金资助项目(11KJB520003)
关键词
事件
索引
谓词
划分
匹配
发布/订阅
event
index
predicate
partition
match
publish/subscribe