HadoopSPARQL: A Hadoop-Based Engine for Multiple SPARQL Query Answering

摘要

An increasing amount of data represented using Resource Description Framework (RDF) have appeared on the Semantic Web. By September 2011, datasets from Linked Open Data had grown to 31 billion RDF triples, interlinked by around 504 million RDF links. As a consequence, it is extremely challenging to deal with the scalability issue of handling such large amount of semantic data. SPARQL is a standard query language for RDF datasets. There has been a lot of work to handle SPARQL queries. However, most of them only treat SPARQL as a transaction-based query language, and consider low latency query answering time as an important design requirement. Furthermore, the query engine processes one query at a time and concentrates on single query optimizations. Nevertheless, users can also use SPARQL in very different scenarios. For example, two users may submit queries to a dataset about publications at the same time. The first user wants to get a list containing all authors who publish at least one proceeding and at least one article while the second user wants to get a list containing all authors who publish at least one article but not necessarily publish a proceeding. Then, Query 1 and Query 2 in Fig. 1 are submitted at the same time.