Efficient Storage, Archiving, and Query Processing for Semantic Data in Smart Living
Abstract
Triplestores are the database managements systems for RDF data. Many triplestores exist and a lot of work has been done to find which of them perform better in specific use-cases. But just like any other database solution, a triplestore may perform better in some use-cases but worse in others. In this thesis, a criteria was devised to select the top three trending triplestores and benchmark their performance in a smart living scenario, as none was available. Unlike the previously developed benchmarks, these benchmarks were performed on raspberry pi and real data and queries were used instead of synthetically generated data and queries. The benchmarks showed that OpenLink Virtuo-so outperformed Jena TDB and RDF4J in inserting and deleting triples. For retrieving triples, the results were mix but in most cases Virtuoso outperformed the other triplestores especially when triples from a short time interval were queried. On on hand, Virtuoso's performance was found independent of dataset size whereas the performance of Jena TDB and RDF4J degraded with in-creasing dataset size. On the other hand, Virtuoso's performance degraded with increasing the time interval from which triples were queried whereas no such degradation was found for Jena TDB and RDF4J.
