Analyzing Linked Open Data with R Background The idea of statistical data analysis has never been more popular, from Nate Silver's book The Signal and the Noise: The Art and Science of Prediction to industry trends such as Big Data. Data itself comes in a vast variety of models, formats, and sources. Users of the R language are familiar with CSV and HDF5 files, ODBC databases, and more. Linked Data Against this sits Linked Data, based on a graph structure: simple triple statements (entity-attribute-value or EAV) using HTTP URIs to denote and dereference entities, linking pools of data by means of shared data and vocabularies (ontologies). For example, a photography website might use an entity for each photo it hosts, which, when dereferenced, displays a page-impression showcasing the photograph with other metadata surrounding it. This metadata is a blend of the WGS84 geo ontology (for expressing a photo's latitude and longitude) and the EXIF ontology (for expressing its ISO sensitivity). By standardizing on well-known ontologies for expressing these predicates, relationships can be built between diverse fields. For example, UK crime report data can be tied to Ordnance Survey map/gazetteer data; or a BBC service can share the same understanding of a musical genre as Last.FM by standardizing on a term from the MusicBrainz ontology. The lingua franca of Linked Data is RDF, to express and store the triples, and SPARQL, to query over them. Worked Example The challenge: to explore the United Kingdom's population density using data from DBpedia. Collating the Data We start by inspecting a well-known data-point, the city of Edinburgh. The following URLs and URI patterns will come in handy: http://dbpedia.org/sparql - the DBpedia SPARQL endpoint against which queries are executed http://en.wikipedia.org/wiki/Edinburgh - a page in Wikipedia about Edinburgh dbpprop:title - a CURIE, identifying the title attribute within the dbpprod namespace (http://dbpedia.org/property/), thereby assigning a title ("The City of Edinburgh"). http://dbpedia.org/resource/Edinburgh - an identifier for a DBpedia resource, comprised of data automatically extracted from the corresponding Wikipedia page; if you view it in a Web browser, it redirects to: http://dbpedia.org/page/Edinburgh - a human-readable view of the DBpedia resource, showing its attributes and values On examining the last of these, we see useful properties such as dbpedia-owl:populationTotal 495360 (xsd:integer) This tells us that dbpedia-owl:populationTotal is a useful predicate by which to identify a settlement's population. (Note: we do not need to search for entities of some kind of "settlement" type; merely having a populationTotal implies that the entity is a settlement and choosing the wrong kind of settlement - stipulating "it has to be a town or a city" would risk losing data such as villages and hamlets.) Looking further down the page, we see the two properties: geo:lat 55.953056 (xsd:float) geo:long -3.188889 (xsd:float) Again, we do not need to know the type of the entity; that it has a latitude and longitude is sufficient. So far, we have some rudimentary filters we can apply to dbpedia: to make a table of latitude, longitude, and corresponding population. Finally, we can filter it down to places in the UK, as we see the property: dbpedia-owl:country dbpedia:United_Kingdom Constructing the SPARQL Query We can build a SPARQL query using the above constraints, thus: prefix dbpedia: <http://dbpedia.org/resource/> prefix dbpedia-owl: <http://dbpedia.org/ontology/> SELECT DISTINCT ?place ?latitude ?longitude ?population WHERE { ?place dbpedia-owl:country <http://dbpedia.org/resource/United_Kingdom> . ?place dbpedia-owl:populationTotal ?population . ?place geo:lat ?latitude . ?place geo:long ?longitude . } ORDER BY ?place LIMIT 100 and the resultset looks like: