
450 Analyzing and Securing Social Networks
Hadoop’s MapReduce functionalities. Our RDF query processing is composed of two main steps:
(i) preprocessing and (ii) query optimization and execution.
Preprocessing: To execute a SPARQL query on RDF data, we carried out data preprocessing
steps and stored the preprocessed data in HDFS. A separate MapReduce task was written to per-
form the conversion of RDF/XML data into N-Triples as well as for prex generation. Our storage
strategy is based on predicate splits (Husain et al., 2011).
Query execution and optimization: We have developed a SPARQL query execution and optimi-
zation module for Hadoop. As our ...