• Service and datasets
• Data examples
• Graphs
• Prefixes
• Referenced namespaces
• SPARQL coverage and protocol
• Hints

This is a public SPARQL endpoint driven by open source Virtuoso  and accessible from https://shared.semantics.cancer.gov/sparql. It is meant to provide an integrated view of the terminology and data elements produced and/or published by the Cancer Data Standards Registry and Repository (caDSR) and Enterprise Vocabulary Services (EVS) projects in the Semantic Infrastructure group. We anticipate that eventually it will include various other datasets of utility to the NCI.

The endpoint is fronted by this web-site which provides a SPARQL query editor  for ad hoc queries to facilitate testing queries and data exploration, documentation (this page), sample queries, and various file downloads.

The web server running this web site also examines queries sent to the endpoint for various issues. This has two consequences:

1. The rules might on some cases prevent valid queries from executing.
• If you believe that your query is valid, please contact us at NCIAppSupport@nih.gov, we will review your query and either rewrite it or edit our rules to allow your query to run.
2. Add to the response time, which will depend on the query itself and the size of the resultset and can add between 20 and 500 msecs to the response.
• If this is an issue for you, and your application runs from within the NCI network or externally on behalf of an NCI program, please contact us at NCIAppSupport@nih.gov to be safelisted. We will contact you for additional information.

The current datasets in the endpoint are:

• A subset of the data in caDSR, primarily the Data Elements sans some administrative information (no forms are included). Monthly XML exports from the caDSR are taken and converted to RDF and loaded into the quadstore; the monthly export taken for processing is from the first day of the month.
• The NCI Thesarus (NCIt) in a flat RDF representation. Monthly distributions of the NCIt in OWL DL are processed and the flat RDF generated. This RDF consists of a conversion of class expressions to simple assertions so that, with the exception of owl:Axioms, blank nodes are eliminated.

We are not currently making the RDF available but could periodically post a downloadable version to serve as an example of the dataset(s). Please let us know if this is of interest to you via email at NCIAppSupport@nih.gov.

In addition to the support email above, you can post requests or issues in the "shared-si-issues"  umbrella Github repository created for this purpose. Some of the repositories for this project are private, so this issue repository covers all aspects of the project.

• Do not include comments or carriage returns in your query when accessing the endpoint directly. Comments can be included in the SPARQL query editor, but only up to a point, the query could wind up being rejected by this service.


• Include prefixes in your query, but do not include a prefix for XMLSchema if you need to cast a value to a float (this is a bug), e.g.

  
  prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
  prefix xsd: <https://www.w3.org/2001/XMLSchema#>
  select ?labelA ?labelB ( xsd:float(strlen(?labelA)) / xsd:float(strlen(?labelB)) as ?fx)
  from ...
  

• Identify the datasets (graphs) that you want to query using the FROM, FROM NAMED or GRAPH keywords, for instance

  
  select ?x ?y ?z
  from <graphname>
  where ...
  

• Include a limit on the resultset no matter how small you expect it to be, e.g.

  
  select ?x ?y ?z
  from <graphname>
  where { ?x ?y ?z }
  limit 1000
  

  Use of the limit keyword is not currently being enforced, but it will be in the future.
  • If there's a need to generate a large resultset (>10000 rows), you can page with order by... limit... offset...