Scriblio, Triplify and XMPP PubSub

6 views
Skip to first unread message

Joss Winn

unread,
May 17, 2009, 5:35:04 AM5/17/09
to scri...@googlegroups.com
Hello,

I've only just started to look at Scriblio but was wondering if
there's any mileage in some/all of the ideas I write about here:

http://joss.blogs.lincoln.ac.uk/2009/05/17/scriblio-triplify-and-xmpp-pubsub/

Is anyone experimenting with Scriblio in this way?

Cheers
Joss

Ross Singer

unread,
May 17, 2009, 9:26:22 AM5/17/09
to scri...@googlegroups.com
Joss, there's definitely mileage in this, although libraries are just
starting to dip their toes in RDF/linked data, there's currently quite
a few interesting projects turning up.

Right now we're addressing the idea of an RDF data model in the
Kochief project (http://kochief.googlecode.com/) and will be fleshing
out the vocabularies to use in the next couple of weeks. The idea is
that it wouldn't be limited to Kochief, but to triplifying any MARC
dataset (+enrichments).

Of course, there's also LIBRIS, from the National Library of Sweden,
who have already "gotten there":
http://libris.kb.se/

Scriblio seems like it would be pretty easy to replace MARCish with
ntriples (or RDF/JSON, or some internal representation), but I'm not
entirely sure how you would really make it act much differently than
it currently does... Conneg for different representations?

In my mind, the current way that Scriblio stores it's data is /very/
document-centric (and pretty static -- it being a big blob of text,
not very atomic), so this might need to be rethought in order to make
this work well.

Anyway, I'm interested (if you can't tell), so if anybody else is, let's try it.

-Ross.

Joss Winn

unread,
May 17, 2009, 12:04:54 PM5/17/09
to scri...@googlegroups.com
Great. I'll have a chat with colleagues in our Library and see about
getting our horizon records into Scriblio and setting up Triplify on a
single site. The XMPP stuff is still pretty experimental in WordPress
but there's a working plugin for publishing posts and comments via
PubSub and I've got a jabber server running at work, so who knows...
wordpress.com have it working. You can subscribe to realtime posts and
comments from their platform:

http://joss.blogs.lincoln.ac.uk/2009/05/05/xmpp-pubsub-on-wordpresscom/

http://en.wordpress.com/firehose/

I'm only just starting to get my head around the semantic web (and
XMPP), so any direction would be appreciated.

I know WordPress Multi User pretty well and work with it every day. I
kind of like the idea of a platform like wordpress.com where each site
is an OPAC and all the data is brought together for searching and
browsing across like it is on wordpress.com.

I don't see why Triplify couldn't then be applied to the WPMU database
to expose content from multiple OPACs as triples - there's a plugin
for WPMU that essentially brings all content from all sites together
into a single site, which is how you get this:

http://en.wordpress.com/tags/

I'll be in touch when we've got data into our Scriblio site and I've
had a look at the database tables.

Joss
--
Joss Winn
http://josswinn.org

Casey Bisson

unread,
May 18, 2009, 10:35:05 PM5/18/09
to scri...@googlegroups.com

On May 17, 2009, at 9:26 AM, Ross Singer wrote:

> Scriblio seems like it would be pretty easy to replace MARCish with
> ntriples (or RDF/JSON, or some internal representation), but I'm not
> entirely sure how you would really make it act much differently than
> it currently does... Conneg for different representations?

The behavior that I have mapped out is to resolve any data linkages as
the record is saved, store the link in the postmeta record, index the
textual representation in the facets, and do something fancy at
display time in the body of the record. That's not perfect, but it's a
start. Some additional logic could be added to update the indexing of
linked data other than at the time records are saved.


> In my mind, the current way that Scriblio stores it's data is /very/
> document-centric (and pretty static -- it being a big blob of text,
> not very atomic), so this might need to be rethought in order to make
> this work well.

You're not entirely wrong. Previous versions did indeed store records
as a big blob of text (structured html I exclaim), but the current
version stores the data as an array that can be easily manipulated at
display time. One application of linked data that it employs now is in
connecting to parent/child/other records (though only within the
current instance at the moment).

Joss Winn

unread,
May 19, 2009, 11:36:46 AM5/19/09
to scri...@googlegroups.com
I've started a discussion thread here to try to get some interest from
UK librarians.

http://mashedlibrary.ning.com/forum/topics/scriblio-wordpressmu

Ideally, we'll have a demo of Scriblio + Triplify to discuss at the
event on July 7th. We're working with Horizon which may slow the
process down. Perhaps we could import someone elses' records from III
for the demo ???

Joss

Casey Bisson

unread,
May 19, 2009, 11:50:46 AM5/19/09
to scri...@googlegroups.com

You're welcome to mine, I think the example values in the connector
will get you there. If not, here's an old screenshot:

http://www.flickr.com/photos/scriblio/1934135957/

Joss Winn

unread,
May 19, 2009, 12:03:03 PM5/19/09
to scri...@googlegroups.com
Excellent. Thanks. That will really help on the day. Paul and I are
going to spend Friday afternoon looking at all of this.
Reply all
Reply to author
Forward
0 new messages