Querying

Abstract

While machine-readable datasets are published primarily for software agents, automatic data extraction is not always an option. Semantic Information Retrieval often involves users searching for the answer to a complex question, based on the formally represented knowledge in a dataset or database. While Structured Query Language (SQL) is used to query relational databases, querying graph databases and flat Resource Description Framework (RDF) files can be done using the SPARQL Protocol and RDF Query Language (SPARQL), the primary query language of RDF, which is much more powerful than SQL. SPARQL is a standardized language capable of querying local and online RDF files, Linked Open Data (LOD) datasets, and graph databases; constructing new RDF graphs, based on the information in the queried graphs; adding new RDF statements to or deleting triples from a graph; inferring logical consequences; and federating queries across different repositories. SPARQL can query multiple data sources at once, to dynamically merge the smaller graphs into a large supergraph. While graph databases often have a proprietary query language (often based on or extending SPARQL), most publicly available datasets have a SPARQL endpoint, from which you can run SPARQL queries. As for the developers, many Semantic Web software tools provide a SPARQL application programming interface (API) for programmatic access.