This is a command line tool for converting XBRL documents to Resource Description Framework (RDF) format. It uses xCurator. It can be used together with secxbrl, which downloads the URLs of XBRL documents from SEC.

##Build

Building this tool requires Ant. Run ant -version to see if it is installed.

1. Check out the repo.
2. Run ant init to resolve dependencies
3. Currently the xCurator dependency cannot be resolved automatically. You need to build it yourself. See instruction here. Once built, copy the xcurator.jar to the lib directory.
4. Run ant clean dist to build afresh.
5. The runnable JAR file xbrl2rdf.jar is in dist directory

##Quick start

Follow the instructions for building. Once done, in the directory of xbrl2rdf.jar, run the following commands

mkdir tdb
java -jar xbrl2rdf.jar -d http://www.sec.gov/Archives/edgar/data/1326801/000132680114000007/fb-20131231.xml -h http://corpbase.org -m fb-20131231-mapping.xml -t tdb

This will download Facebook Inc.'s 2013 annual filing, and convert the document into RDF data. The RDF data is stored in a TDB, which is in the tdb directory. The entities and relations in the data are serialized into the mapping file fb-20131231-mapping.xml. The domain name of the RDF data is set to http://corpbase.org.

Local XBRL documents can also be used, just use local file path for -d argument.

##Serve data over HTTP

Fuseki can be used for serving the RDF data over HTTP through SPARQL endpoint. It also provides a nice web interface for running SPARQL queries.

For quick start, download Fuseki, then run the following command:

bash fuseki-server --loc /path/to/tdb/directory /corpbase

The web query interface can now be accessed at http://localhost:3030/control-panel.tpl, select /corpbase dataset.

##Query the data

The data can be queried directly through the SPARQL endpoint served by Fuseki server, either using ARQ or the web query interface provided by Fuseki server.

You can use the following SPARQL query to get the net income information in the filings.

prefix class: <http://corpbase.org/resource/class/>
prefix property: <http://corpbase.org/resource/property/>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>

select distinct ?name ?netincome ?start ?end
where
{
?a rdf:type class:xbrl.
?a property:EntityRegistrantName ?n.
?n property:value ?name.
?a property:NetIncomeLoss ?z.
?z property:value ?netincome.
?z property:context ?context.
?context property:entity ?contextentity.
filter not exists {?contextentity property:segment ?x}.
?context property:period ?period.
?period property:startDate ?start.
?period property:endDate ?end.
}
order by ?name

Apache Jena provides many other tools and libraries to work with RDF data, vist their website for more information.