Efficient Real-time Semantic Data Stream Processing Based On Forward And Backward Chain Reasoning

In smart cities,large-scale infrastructure operation monitoring,wearable assistance devices and the Internet of Things,and other emerging application scenarios,Use RDF to describe the data information of various sensors,which can clearly show the relationship between various subjects.At the same time,incoming continuous data streams must be processed within a very short delay,so semantic stream processing has become an important research area.At present,semantic stream processing systems cannot meet the real-time query requirements when processing high-rate streaming data,and most of these systems do not support the real-time reasoning.Therefore,on the basis of summarizing the existing semantic data stream reasoning related research and technology,this paper uses characteristic index and semantic stream reasoning technology to further improve the performance of semantic stream reasoning and reduce the query delay.The main work is as follows:In terms of data-driven reasoning optimization,this paper combines the incremental model of the time point in the incrementally updated data stream processing system with forward reasoning,and based on the characteristics and inference rules of forward reasoning,a step-based incremental results update strategy is proposed.Moreover,in view of the lack of forward reasoning results,a forward real-time reasoning mechanism based on multi-level index is proposed.In terms of query-driven reasoning optimization,this paper adds a semantic reasoning based on query statements to the backward reasoning scheme based on the extended characteristic set,and redesigns the semantic data stream storage strategy based on memory based on the dynamic characteristics of semantic stream data,built an efficient semantic data stream storage and query engine.In this paper,multiple sets of query and comparison experiments are designed on the LUBM dataset.The results show that the incremental result update strategy significantly improves the throughput to reach millions of triples per second;themulti-level indexing mechanism can effectively remove redundant inference results,Its proportion is as high as 99%;and the reasoning performance has obvious advantages over other data stream reasoning systems.