语义物联网中基于RDF流的复杂事件处理方法研究

发布时间：2018-05-01 18:36

本文选题：语义物联网 + RDF流　；参考：《大连海事大学》2017年硕士论文

【摘要】：语义物联网作为较新的研究领域是对物联网的一个扩展,其特点是在使用语义技术消除数据异构的基础上,能结合丰富的知识进行语义查询、推理。传感器网络作为语义物联网的基础设施,实时、持续地产生高度动态的数据,为了能良好地表达和处理这种动态数据,RDF流数据以及相应的RDF流处理技术被提出。目前的RDF流处理大多是在扩展了 SPARQL查询的基础上对RDF流数据进行持续地查询处理,但如何在RDF流数据查询的基础上分析、识别其中蕴含的时间、因果关系,即以事件驱动的角度进行更高层次地处理还较少涉及,本文主要针对这一问题进行研究。本文阐述了已有的RDF流处理技术和基于RDF流处理技术的语义复杂事件处理方法,在此基础上,首先提出了基于RDF流的复杂事件处理框架RCEP。然后对RCEP中使用到的事件本体进行建模,在建模过程中考虑了事件模式的特点,并结合使用了现有的SSN本体,使得构建的事件本体具有良好地可表达性;在用户自定义事件过程中,根据用户的设置和传感器本体中的元数据选择合适的传感器;在RDF流处理上,选择Sparkwave作为RDF流处理工具,针对其RETE网络结构不够优化这一问题进行改进,即在RETE网络构建过程中根据事件模式的子图的过滤能力这一因素决定其在RETE网络中的连接顺序,以此达到减少连接比较次数和内存占用的目的。在Sparkwave的RDF流推理过程中,为减少推理所用时间,根据事件模式选择性地从本体中加载背景知识。最后对RCEP的各个模块依据文中提出的方法进行设计与实现。为了验证本文中提出的方法对Sparkwave的改进效果,对比分析了改进后的Sparkwave和原Sparkwave在吞吐量和内存占用上的情况。实验证明,使用本文改进的Sparkwave能够有效地适用于复杂事件的检测。
[Abstract]:The semantic Internet of things, as a new research field, is an extension of the Internet of things. It is characterized by semantic query and reasoning based on the use of semantic technology to eliminate the heterogeneity of data. Sensor networks, as the infrastructure of semantic Internet of things, generate highly dynamic data in real time and continuously. In order to express and process the dynamic data well, RDF-stream data and the corresponding RDF stream processing technology are proposed. Most of the current RDF stream processing is based on extending the SPARQL query, but how to analyze the RDF stream data query and identify the time and causality. That is to say, a higher level of event-driven processing is less involved, this paper mainly focuses on this issue. In this paper, the existing RDF flow processing technology and the semantic complex event processing method based on RDF flow processing technology are described. Based on this, a complex event processing framework based on RDF flow is proposed. Then, the event ontology used in RCEP is modeled, the characteristics of event pattern are considered in the modeling process, and the existing SSN ontology is combined to make the event ontology well expressible. In the process of user-defined event, the appropriate sensor is selected according to the user's setting and the metadata in the sensor body, and the Sparkwave is chosen as the RDF flow processing tool in the RDF stream processing. In order to solve the problem that the RETE network structure is not optimized enough, the connection order in the RETE network is determined by the filtering ability of the sub-graph of the event pattern in the process of constructing the RETE network. In order to reduce the number of connections and memory consumption. In the process of RDF flow reasoning in Sparkwave, background knowledge is selectively loaded from ontology according to event pattern to reduce the time required for reasoning. Finally, each module of RCEP is designed and implemented according to the method proposed in this paper. In order to verify the improved effect of the proposed method on Sparkwave, the throughput and memory usage of the improved Sparkwave and the original Sparkwave are compared and analyzed. Experiments show that the improved Sparkwave can be used to detect complex events effectively.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.44;TN929.5

【参考文献】