A Partitioning and Index Algorithm for RDF Data of Cloud-Based Robotic Systems

Abstract

Robotic systems generally employ resource description framework (RDF) to express heterogeneous data coming from different sensors. With the access of more terminals, the RDF volume in robotic systems is becoming larger and larger, posing new significant challenges to the storage and retrieval of RDF data. This paper proposes a star-based partitioning and index algorithm for RDF data of robotic systems. First, we construct a two-hop star structure by MapReduce and HDFS, and get a coarsened weighted graph. Next, a balance partitioning algorithm is used to divide the weighted graph. After partitioning, a compressed and linked S-tree index is proposed to improve the query efficiency. Experiments are executed on benchmark and real data sets to evaluate the studied partitioning and index methods. Results show that our partitioning method has a lower replication ratio, and a better load balancing performance, so our method is efficient for star query and competitive in complex query.