W3C home > Mailing lists > Public > public-lod@w3.org > June 2009

Re: Owning URIs (Was: Yet Another LOD cloud browser)

From: Sherman Monroe <sdmonroe@gmail.com>
Date: Mon, 1 Jun 2009 17:40:42 -0500
Message-ID: <e23f467e0906011540n6f2d7ff7xf9fececcec5e5821@mail.gmail.com>
To: Dan Brickley <danbri@danbri.org>
Cc: Kingsley Idehen <kidehen@openlinksw.com>, Daniel Schwabe <dschwabe@inf.puc-rio.br>, David Huynh <dfhuynh@alum.mit.edu>, Linked Data community <public-lod@w3.org>, semantic-web@w3.org
Dan,



> Basic idea in a nutshell is that SPARQL is great for data access, but there
> may be additional query-oriented data structures worth spec'ing based around
> the set-oriented navigation very nicely articulated by David Huynh in the
> Parallax screencast. And that if such a structure could be exchanged between
> systems we could hope that the navigational paradigm it supports could be
> found in various concrete UIs, and that the results of exploring data this
> way could become useful and standard artifacts in the public Web, rather
> than just bookmarks within some specific system.
>

I'm glad to hear this is moving along. I'd like to reiterate on the work on
the Faceted Browsing API [1] and hope these concepts make their way into
your proposal, as well as elicit comments from the community, as I feel this
area of research is one of the more imparative re. the Linked Data movement
at large because it deals with how people will experience the LD web. As a
developer who has dealt with the frustrations of client-side faceted
implementations (e.g. speed, scaling problems, non-RDF based), I've found
the approach taken by OpenLink addresses a lot of my demands.

I present <http://www.razorbase.com>razorbase [2] as an approach to facets,
and an implementation of the OpenLink Facets API (Disclaimer: This is based
on the current lod.openlinksw.com instance, which contains "dirty" entities.
There will be a newer clean-up version availible later this week which will
remove the weird characters, etc. Please don't allow it to distract from the
main point). If you click your browser's view source, you will see that the
client-side code is 100% plain XML, and simply uses a stylesheet to render
the HTML UI, so hardly any framework code had to be developed on my part.
This was key to allowing me to crank this demo out in literally 48 hours
(from first exposure to basic browser implementation). I've also prepared a
compare/contrast [3] of Razorbase and Parallax which shows that not only
does this API allow for duplication of all Parallax features, but allows
also for painless implementation of things such as mutual connections and
descendent connections. In the hands of people more skilled than I am in UI
design/implementation, etc, I believe that an API like this can blow the
door open on Linked Data.

The single most significant feature I believe is the structured XML
representation of the SPARQL query, which allows me to very easily create
the NL map you see when clicking the 'Your query:' link.

As for pivoting and set-based browsing in general, it's a very novel
paridigm that requires more study, espeically user-oriented studies. It's
easy to mis-apply old methods to this new UI approach, for example, how
Parallax creates a 1-dimensional (i.e. 2-directional) breadcrumb trail. This
is borrowed from WWW browsing, where the only directions from the "subject"
(i.e. web page loaded) is back and forward. But in a linked database, the
number of directions from a subject is equal to the number of possible types
of the links from that subject to its objects. So it's a truly n-dimensional
hyperspace. Thus, the 2D-breadcrumbs trail, while helpful, does not truly
orient user's current position in the database. Here is my attempt to
describe how navigating the linked dataspace would feel in real life, to
better bring out the point. Image a linked database as a building, and each
room contains a set of one or more resources matching a SPARQL criteria, and
along the walls of the room are portals leading to other rooms, where each
portal represents an RDF property those resources share. Now say I'm
browsing a social network database, and I'm in DanBrickley's room, and I
open and enter his *foaf:knows* portal. Now I am in *DanBrickley >>
foaf:knows* portal. The portals in here represent all the properties shared
by all the folks Dan knows. Where can I go? I can go back through the
DanBrickley portal, or through one of the property portals. Now here is the
magic introduced by the structured XML SPARQL query. Suppose that from here,
I want to enter the *foaf:interest* portal. In there, I see *semantic web*.
Now I kick out all the other resources in this room, so that only semantic
web remains. If I go back into the *foaf:interest* portal through which I
came, I now find this list of friends is narrowed to just those whose
interest is semantic web. The abilty to manipulate a room and have it affect
the state of all other rooms in my breadcrumbs, is something you can't
currently do with Parallax, because of it's 2D navigational path (David or
anyone, I make this statement after having tried witht he interface myself,
the browse all>> link on properties is the cloest thing I found). Now from
this room (the *friends whose interest is semantic web* room), I now go
through the *foaf:currentProject* portal. I have now forked my criteria (an
ability that was in Piggybank/Longwell, but missing from Parallax for some
reason). Here I find the *Umbel Project*. Because each resource also acts
like a portal (thanks to resource dereference :), I enter the *Umbel Portal*,
and now I'm in a new room inside a totally new building containing rooms
specialized around technology projects (a building represents a SPARQL
endpoint, linked database source, an RDF graph, etc). So the portals can
lead into/out of any building imaginable. Parallax only allows for
navigating inside one building, i.e. Freebase, a second short coming.

Not to pick on Parallax at all, it's outstanding work that must continue to
be pushed and improved, so I'm just here to whet the concepts it introduces.
I have found the above visualazation of the linked database helpful in my
work with linked data browsers, most recently razorbase. I've toyed with the
notion of a 3D linked data browser along these principles, but don't know if
that could actually be more useful than simple tables for large audiences.


[1]
http://virtuoso.openlinksw.com/dataspace/dav/wiki/Main/VirtuosoFacetsWebService
[2] http://www.razorbase.com
[3] http://www.slideshare.net/sdmonroe/razorbase-vs-parallax

-sherman


>
> cheers,
>
> Dan
>
>


-- 

Thanks,
-sherman

I pray that you may prosper in all things and be healthy, even as your soul
prospers
(3 John 1:2)
Received on Monday, 1 June 2009 22:41:15 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:21 UTC