[Mulgara-dev] 7 Billion Node RDF Store

thomas thomas at stray.net
Mon May 7 13:21:34 UTC 2007


--On 4. Mai 2007 15:05:26 -0400 David Wood <dwood at softwarememetics.com> 
wrote:

> On 4 May2007, at 14:41, David Moll wrote:
>> their goal seems to be more of an RDF search engine than an RDF
>> database.
>
> I think that is an excellent summary and one that would apply to Garlik's
> (proprietary) store as well.
>
> Does anyone care to try to articulate the motivational differences
> between the two?  A correct characterization could help guide our future
> directions.

isn't that comparing apples with oranges? they build a store optimized for 
a certain use case, while mulgara is a general purpose store which has to 
balance different needs. well: which needs was the  question, right? ;-)

>  How much OWL/FOL should we invest in?  How much should rules
> play a part?


>  Do we continue to optimize for read at the expense of creating larger 
stores?

since things that aren't read more often than once possibly aren't worth 
the effort at all: yes

>  Do we start to federate right after XA2? Before?


in short, me preferences are:
- less scalability, instead federation
- more functionality


of course scalability is important but when it's traded against 
functionality i'd rather argue for more functionality and facilitating 
scalability through federation.
my wishlist:
- strong support for reification (a quint-store, or rather 4-and-a-half)
- strong integration of fulltext search
- support for long texts and blobs, webdav-integration

my prototypical user is not so much the big companies but knowledge 
workers, which have to handle more information than they can eat.
- integrating files, bookmarks, mails, schedules -
- seamlessly connecting personal stuff with stuff on the web with stuff of 
other people/partners/colleagues -
- combining browsing, querying, fulltext-search, autocategorization, ai -
- federating "personal" knowledge in a net of feeds -
that's my usecase for mulgara.

that for sure needs huge stores too, but presumably not as huge as the 
coporate or federal client would need. couldn't they be served with 
massively parallel rdf-stores?


mulgara already does a lot of what i need and i'm very thankfull for that!

hope that helps,
thomas


mailto:thomas at stray.net
http://stray.net




: accumulated wisdom
. early optimization is the root of many evil [donald e. knuth]
. if you've got a hammer every problem looks like a nail
. the difference between theory and practice is always greater
  in practice than it is in theory



More information about the Mulgara-dev mailing list