This is a proposal to start a rdf database server project at apache. New projects at apache start inside the Apache Incubator [1].
summary
TripleSoup is the simplest thing that you can do to turn your apache web server into a SPARQL endpoint.
[TripleSoup] will be an RDF [2] store [3], tooling to work with that database, and a REST [4] web interface to talk to that database using SPARQL [5], implemented as an apache webserver module.
Target: TLP
Sponsor: Incubator PMC
Champion: Leo Simons <leosimons@apache.org>
Mentors: Dirk-Willem van Gulik <dirkx@apache.org>,
Ben Hyde <bhyde@apache.org>,
Stefano Mazzocchi <stefano@apache.org>,
Leo Simons <leosimons@apache.org>
Resources: SVN: https://svn.apache.org/repos/asf/incubator/triplesoup/
Website: http://incubator.apache.org/triplesoup/
Jira: http://issues.apache.org/jira/browse/TRIPLES
Wiki: http://wiki.apache.org/triplesoup/
Mailing lists:
triplesoup-dev@incubator.apache.org
triplesoup-commits@incubator.apache.org
triplesoup-private@incubator.apache.org
Moderators: leosimons@apache.org
stefano@apache.org
dreid@apache.org
Initial committers:
Dave Beckett <dave@dajobe.org>, redland author
Dirk-Willem van Gulik <dirkx@apache.org>,
Ben Hyde <bhyde@apache.org>,
Stefano Mazzocchi <stefano@apache.org>,
Andrea Marchesini <baku@theveniceproject.com>, b store author
Alberto Reggiori <alberto@asemantics.com>, rdfstore author
David Reid <dreid@apache.org>,
Leo Simons <leosimons@apache.org>
Initial source: mod_sparql, commercial triple store,
existing open source triple store
Known risks: None
Technologies: c
Reference: http://wiki.apache.org/incubator/TripleSoupProposal
Proposal details
Technology (basics)
What is RDF? It is just about any kind of data, represented as triples of (subject, predicate, object), usually with a rich vocabulary describing the semantics of the data (with the vocabulary typically also encoded as triples).
This data has a representation as RDF/XML as well as using other formats such as N3, and a query language SPARQL for searching through it. See [6] for an overview.
So if it is just some data in some format, why does it need a special server? Because RDF data is fundamentally not constrained to a "file", and there often is no "resource identifier" that readily identifies something as a "document" which can be served up over HTTP.
So why the REST interface? RDF is one of the building blocks proposed for the "semantic web", and that's why a system that works well with/over HTTP is needed from the start.
Technology (concrete)
This is just an example. Imagine that there is an application "someapp" on the host foo.example.com which provides access to information about books, and you want to get a list of those books (their URIs) and the names of the books.
$ telnet foo.example.com 80
SELECT /someapp HTTP/1.0
Host: foo.example.com
Query-Language: http://www.w3.org/TR/2006/CR-rdf-sparql-query-20060406/
Accept: application/sparql-results+xml, rdf/xml, rdf/n3
PREFIX books: <http://example.org/book/>
PREFIX dc: <http://purl.org/dc/elements/1.1/>
SELECT ?book ?title
WHERE
{ ?book dc:title ?title }
HTTP/1.0 200 Ok
Content-Type: application/sparql-results+xml
Content-Length: 1234
<?xml version="1.0"?>
<sparql
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:xs="http://www.w3.org/2001/XMLSchema#"
xmlns="http://www.w3.org/2005/sparql-results#">
<head>
<variable name="book"/>
<variable name="title"/>
</head>
<results ordered="false" distinct="false">
<result>
<binding name="book">
<uri>http://example.org/book/book6</uri>
</binding>
<binding name="title">
<literal>Harry Potter and the Half-Blood Prince</literal>
</binding>
</result>
</results>
</sparql>
Connection closed by foo.example.com
$
It turns out there's only one book in the database in this example. (Sample data taken from http://www.sparql.org/
). David Reid has some code that does something not unlike this already [7], implemented as a httpd module, using the Redland library [11,12] as its backend store.
What would you use TripleSoup for?
It could be a backend for piggy bank [8].
- It could be a backend for the next version of wikipedia.
- It could be a backend for an "open" version of iTunes or IMDB.
- It could be the backend for the information management system of the
Dutch ministry of water management [9].
It could be the backend for projects.a.o [10] and similar applications.
- Most importantly, it could be a backend for dozens of useful new innovative
projects that no-one has envisioned yet.
The initial source
RDFstore is a standalone RDF storage system implemented as a C library, licensed under the ASL 1.1. It has perl bindings. Find its distribution at [15].
mod_sparql [7] is an in-development apache module that implements a SPARQL endpoint. It is licensed under the Apache License 2.0. It uses redland as a backend. The SVN repository can be found at [7].
B is an in-development storage backend for Redland implemented as a standalone C library. It is currently a closed source codebase. A code snapshot can be found at [16].
The initial committers
Dirk-Willem, Ben, Stefano, David and Leo are ASF members who hopefully need no introduction.
Dave Beckett is the primary author of the Redland RDF application framework.
Alberto Reggiori is the primary author of rdfstore, an rdf store developed by asemantics [13], which will be contributed to [TripleSoup]. He is a partner at asemantics.
Andrea Marchesini is the primary author of B, a storage backend for RDF developed at Joost [14], which will be contributed to [TripleSoup].
All initial committers have experience working on open source projects. They work for at least 5 different companies.
TripleSoup as an apache project
We think TripleSoup will have to reference dozens of specifications from the W3C (XML, RDF, OWL, SPARQL, their standards for URIs, and more) and from the IETF (HTTP, URL, URI, URN, and more), will make use of or integrate with quite a few existing open source projects (like the redland RDF libraries as well as apache apr&httpd). As such, it seems like TripleSoup should fit in really well at apache.
The responses we got from various members of the RDF and semantic web communities so far when discussing this proposal with them have all been quite positive, and we expect and hope there'll be quite a few people new to apache joining the project soon after it starts.
Most importantly, we think this project will be useful, innovative, and fun!
References
[1] http://incubator.apache.org/ [2] http://www.w3.org/RDF/ [3] these are often called "triple stores" [4] http://www.ics.uci.edu/~fielding/pubs/dissertation/rest_arch_style.htm [5] http://www.w3.org/TR/rdf-sparql-query/ [6] http://www.betaversion.org/~stefano/papers/ac2006.1.pdf [7] http://david-reid.com/repos/public/mod_sparql/ [8] http://simile.mit.edu/wiki/Piggy_Bank [9] http://www.wadi.nl/uk/ [10] http://projects.apache.org/ [11] http://www.librdf.net/ [12] http://svn.librdf.org/repository/ [13] http://www.asemantics.com/ [14] http://www.joost.com/ [15] http://rdfstore.sourceforge.net/downloads/RDFStore-0.51.tar.gz [16] http://opensource.joost.com/libb/