Linked Data Documentation - a challenge

[Richard Pope]

Can anyone rise to this?

http://memespring.co.uk/2011/01/linked-data-rdfsparql-documentation-challenge/

cheers,

Richard
--
/*
ric...@memespring.co.uk
memespring.co.uk
++44 7976730458
memespring (flickr/skype/etc)
memspr (aim)
*/

[Harry Metcalfe]

I can't, but I second that entire post. With some vitriol.

H

[Tim Green]

Another second.

I think it explains much of my fundamental suspicion of using SPARQL et
al. over REST interfaces/json dumps - even though I realise there are
patterns in that model that should be factored out. In particular I
couldn't imagine that had, for example, yournextmp.com used it at the
last election, it would have got anywhere near the amount of interest
from developers in using its data.

Unless I'm just misunderstanding it and it's just as easy as REST/json
for an end-user, which means it's a marketing/branding issue that puts
me off.

Has anyone made a Django application which can generate a SPARQL
interface directly from Django models? That'd be very useful, if it
makes sense.

-t

[Richard Pope]

If you would like to see this happen please pledge a few quids worth
of amazon vouchers to the first person to write the docs (as specced)
in the comments:

http://memespring.co.uk/2011/01/linked-data-rdfsparql-documentation-challenge/#comments

--

[Rufus Pollock]

On 14 January 2011 12:09, Richard Pope <ric...@memespring.co.uk> wrote:
> Can anyone rise to this?
>
> http://memespring.co.uk/2011/01/linked-data-rdfsparql-documentation-challenge/

Here, here!

[As some background: it took Richard and myself approximately *2+*
person-days to get a (full-featured) RDF data store working with
python on mac osx (Virtuoso in this case) and this only because Will
Waites has done a lot of work to a) make a python virtuoso driver and
[0] b) patch the standard pyodbc library [1] to use some of virtuoso's
special features. [2]]

My 2c: AFAICT there are no standard libraries for python, php, ruby
etc from the major open source rdf stores -- something rather
different from one sees with either standard RDBMS or NoSQL stuff.

If linked data and RDF really are going to get uptake in a broader
community and especially among web-devs this kind of thing really
needs to be sorted out (and realistically by people close to the
makes/vendors of the stores -- as happens with nosql ...).

Virtuoso, for example, seems a pretty impressive bit of kit and now
has an ubuntu package (>= lucid) but without client libraries it's a
non-starter for most people.

Regards,

Rufus

[0]: https://bitbucket.org/ww/virtuoso/
[1]: http://eris.okfn.org/ww/2010/10/pyodbc-spasql/pyodbc-2.1.8-virtuoso.tar.gz
-- see also https://bitbucket.org/ww/virtuoso/issue/3/failures-in-rdf-tests
[2]: An experience earlier in the year with 4store was similar -- with
one developer after a nightmare with compiling librasqal from source
(ubuntu version was too old), c-bindings-required python libraries we
simply gave up after about 1.5 days

[Dave Reynolds]

On Fri, 2011-01-14 at 12:09 +0000, Richard Pope wrote:
> Can anyone rise to this?
>
> http://memespring.co.uk/2011/01/linked-data-rdfsparql-documentation-challenge/

Can I second Matthew's comment on looking at Jeni's write ups of the
linked data API [1].

The aim there has been to make the data the government publishes as RDF
easily available REST/JSON style - so you can work with the data using
standard web tools without having to run your own RDF stack or learn
SPARQL.

We'd love feedback on whether that's providing a useful service or at
least heading in a helpful direction.

I appreciate that if you want to host & query your own data (and don't
want to use web-based RDF hosting sites) then that means running a
triple store. You are right that existing stores are not as well
packaged for PHP/Python/Ruby users as they should be. However, there is
some progress. For example, I may be biased but Andy Seaborne's Fuseki
[2] server seems easy to set up (if you are prepared to do a download)
and includes (server independent) command line tools for query/update
written in Ruby.

Cheers,
Dave

[1]
http://data.gov.uk/blog/guest-post-developers-guide-linked-data-apis-jeni-tennison
[2] http://openjena.org/wiki/Fuseki

[Keith Alexander]

On Fri, Jan 14, 2011 at 1:43 PM, Dave Reynolds
<dave.e....@gmail.com> wrote:
> On Fri, 2011-01-14 at 12:09 +0000, Richard Pope wrote:
>> Can anyone rise to this?
>>
>> http://memespring.co.uk/2011/01/linked-data-rdfsparql-documentation-challenge/
>
> Can I second Matthew's comment on looking at Jeni's write ups of the
> linked data API [1].
>
> The aim there has been to make the data the government publishes as RDF
> easily available REST/JSON style - so you can work with the data using
> standard web tools without having to run your own RDF stack or learn
> SPARQL.
>
> We'd love feedback on whether that's providing a useful service or at
> least heading in a helpful direction.
>

Feedback on this would be really good - has anyone here tried to use
and of the government linked data APIs?

> I appreciate that if you want to host & query your own data (and don't
> want to use web-based RDF hosting sites) then that means running a
> triple store. You are right that existing stores are not as well
> packaged for PHP/Python/Ruby users as they should be. However, there is

I've had installation issues with a few different RDF stores, but I'd
like to say that ARC[1] is easy to use if you are used to working with
MySQL and PHP. It's how I got started with RDF (a few years ago) from
a regular PHP developer background. You don't need to compile
anything, it doesn't depend on anything but PHP and MySQL, it's just a
PHP library, you give it the user,pass, and name of a MySQL database,
and in a couple lines of code you have a triple store that you can
programmatically load RDF into (from web or file system), query,
update, etc.

There is a debian package for it I believe (googling reveals
http://packages.debian.org/hu/source/experimental/php-arc), though I
have never tried it.

[1] https://github.com/semsol/arc2

[Stuart Harrison]

Funny you should mention this. I put together the first part of a two part simple SPARQL tutorial based on my learnings as a linked data newbie. Was supposed to go live on Monday, but I've published it now:

http://www.pezholio.co.uk/2011/01/a-beginners-guide-to-sparqling-linked-data-part-1/

Any thoughts welcome! :)

Cheers

[William Waites]

Hi Stuart, what might be useful are to follow two lines of thought from your
first (or second post) - what libraries to you use to make the sparql queries
and turn the results into something useful and how you write / delete data
into / from the store.

Also some background to Richard's original post. The environment that we
work in involves some level of "test driven development", so typically a
high-level test might be written and then the code that implements whatever
needs doing is written. So this will usually involve flushing some graphs,
loading fixture data into them and then running the actual code in question.

This type of development loop is not as easy as it should be. It is also not
really feasible if it depends on a remote store.

Cheers,
-w

--
William Waites
Email: wwa...@gmail.com
UK tel: +44 131 516 3563
UK mob: +44 789 798 9965

[Christopher Gutteridge]

I certainly can't find time to do the whole stack, but I've done how to
query SPARQL from PHP.
http://graphite.ecs.soton.ac.uk/sparqllib/
Admittedly my installer is very minimal *grin*.

I'm also trying to put together a guide for people who don't want to
become experts in RDF, just want to get up to speed quickly.
http://wiki.openorg.ecs.soton.ac.uk/index.php/Linked_Data_Guide_for_Newbies

--
Christopher Gutteridge -- http://id.ecs.soton.ac.uk/person/1248

/ Lead Developer, EPrints Project, http://eprints.org/
/ Web Projects Manager, ECS, University of Southampton, http://www.ecs.soton.ac.uk/
/ Webmaster, Web Science Trust, http://www.webscience.org/

[Kingsley Idehen]

All,

Here is a simple SPARQL guide (what, why, how). I also include a section
clarifying the simple 3-step guide re. using SPARQL with Virtuoso [1].
This isn't a challenge respond per se. as there are many profiles combos
to be clarified i.e., end-users, power-users, high level developers,
low-level developers.

For instance, going from CVS to Linked Data (hassle free) and then
querying via SPARQL (hassle free) is something that needs to come out of
this. I am happy to provide sample links and pointers to material from
the past and present re. the above. Hopefully, someone can then make a
generic (i.e., platform independent) guide for SPARQL exploitation.

Links:

1.
http://www.openlinksw.com/dataspace/kid...@openlinksw.com/weblog/kid...@openlinksw.com%27s%20BLOG%20%5B127%5D/1647

--

Regards,

Kingsley Idehen
President& CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen

[Christopher Gutteridge]

I've added it to the brief SPARQL section on the crash-course I'm
writing:
http://wiki.openorg.ecs.soton.ac.uk/index.php/Linked_Data_Guide_for_Newbies#Triple_Stores_and_SPARQL

Quick braindump Ideas;

I suggest when you do part 2 you add a hyper link from the bottom of part 1.

draw the graph next to the SPARQL to help people visualise it.
Picture:word value ratio and all that.

It would be nice if the page was wider so the long URIs didn't wrap. You
could put a 1000x1px image to force it, or you could make the example
frames scroll using css.

Indent inside {}

If you want it to be really good; highlight the syntax of the SPARQL
(and RDF). eg.
col1: keywords like SELECT,
col2: control chars like { ; . }
col3: variables ?foo ?bar
col4: uris and prefixed uris.
col5: literals
etc.

--