UDD export as RDF data through D2R and a corresponding mapping definition

(this is a copy of a post I made on the debian-qa list)

I’ve been pursuing previous ideas about the use of Semantic Web standards for publication of facts about Debian using the UDD database as a source.

FYI, I have a running prototype (unfortunately only on my laptop at the moment, hopefully on the Web some day soon) which uses D2R, a server which can publish RDBMS tables as RDF documents, using a mapping description.

I’m documenting bits of what I’ve done in https://picoforge.int-evry.fr/cgi-bin/twiki/view/Helios_wp3/Web/UltimateDebianDatabaseToRDF should you be interested.

This first prototype allow to express queries such as :

SELECT DISTINCT ?b ?s WHERE {
?b vocab:bugs_package .
?b debbugs:hasSubmitter ?s
}

which returns a list of all bugs id and corresponding bug reporters, i.e. “bugs.submitters” (From) addresses on the package sympa in the form of links to resources like :

http://localhost:2020/resource/debbug/169102 - http://localhost:2020/resource/foafdebbug/Olivier+Berger+%3Colivier.berger%40it-sudparis.eu%3E

which each point to different RDF documents like :

http://localhost:2020/resource/debbug/169102 :
debbugs:bugUrl
debbugs:hasSubmitter db:foafdebbug/Olivier+Berger+%3Colivier.berger%40it-sudparis.eu%3E
debbugs:number 169102
debbugs:summary "sympa: Lack of exim configuration for pipe transport"
vocab:bugs_arrival "2002-11-14T16:48:05"^^xsd:dateTime
vocab:bugs_package db:package/sympa
vocab:bugs_severity debbugs:Important
rdf:type debbugs:Debbug
rdfs:label "Debian bug #169102"

and http://localhost:2020/resource/foafdebbug/Olivier+Berger+%3Colivier.berger%40it-sudparis.eu%3E :

rdf:type foaf:Person
foaf:mbox
foaf:mbox_sha1sum "6688a14521cd97db162af8f9757f2e2232300e50"
foaf:name "Olivier Berger"
and with inverse relationships :
debbugs:hasSubmitter db:debbug/169102
debbugs:hasSubmitter db:debbug/208203
...

This is obtained with a simple mapping of data from the bugs table in UDD to several ontologies like “foaf“, and “debbugs” (a fictive one).

When running such sparql queries, or navigating the URLs provided by D2R, the mapping is applied and corresponding SQL queries are done on the UDD database (like :

SELECT DISTINCT 1 FROM "d2r_bugsubmitters" WHERE
"d2r_bugsubmitters"."submitter" = 'Olivier Berger '

SELECT DISTINCT "d2r_bugs"."id" FROM "d2r_bugs" WHERE
"d2r_bugs"."submitter" = 'Olivier Berger '

Of course, this would be great if such a service was on the (Semantic) Web, with URI for resources becoming real URLs and which may then be navigated to discover links to other resources (like my other FOAF profiles searching for my foaf:mbox_sha1 on the Web, which would return http://www-public.it-sudparis.eu/~berger_o/foaf.rdf for instance).

I hope this will trigger some interest by UDD maintainers, and maybe we can continue discussing that, and maybe plan to have a D2R server alongside UDD on udd.debian.org some day.