couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Parnell Springmeyer <ixma...@gmail.com>
Subject Re: CouchDB x RDF databases comparison
Date Mon, 11 May 2009 18:28:56 GMT
Yeah, I know about the JSON serialization of RDF - it has drawbacks though,
cardinality can't properly be expressed, as well as many other RDF features.
The primary structure of RDF can be captured by JSON though.

My key interest is serializing XML serialized RDF into the JSON objects and
using the proper libs to parse that into triples - N3 also wouldn't be bad,
however, N3 was designed more for humans, and since JSON doesn't like
carriage returns in it's data, that makes XML the more logical choice...

As for a SPARQL endpoint, I believe there could be a few useful hack-arounds
devised (such as this RDF hack-around), but, nothing will supplant a
properly built Erlang backed, distributed SPARQL endpoint/triple store. Only
problem there is Erlang's lacking XML libraries which makes building an RDF
parser that much more difficult.

I've been looking into using Redland and Erlang's C bridge to accomplish the
parsing task... I dunno, I'm still researching it - but all this research is
in my free time (which is stretched as it is) so it is slow moving.

Keep me updated as to what you come up with, last night I made some headway
with serializing XML into JSON objects within CouchDB and will begin doing
more comprehensive (ie: bloggable) experiments with rdfparse.js and the
eulersharp reasoner - may see what I can do with Python.

Another interesting thought would be to embed the Python interpreter in
CouchDB and use it as the primary view server - I like JSON as the storage
medium because it is simple and light weight and I like Python's strength as
a general purpose programming language with many mature libraries (including
RDF, hint hint).

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message