Notes All That SPARQLs is not gold

From ActiveArchives
Jump to: navigation, search

Notes 09-11-2011

Jargon heavy; alphabet soup

dbpedia being centric

from a manifesto to technical realities

Why is it worth all the fuzz? A lot of mechanics! Look at it as a space of writing; finding a balance.

W3C (Tim Berners Lee) image: locating semantic web / linked data as part of a set of technologies and tools, standards and practices. Between web technologies and web technologies. Different practices, systems of thought. From hypertext-fiction to Artificial Intelligence; linked data as a space where different worlds meet. If you think of it in terms of writing, what does it means.

An example: Ted Nelson Computer Lib, Tempus, US, 1974. 1986. Saw himself as a computer counter culture figure. 'Hyper'text - as a writer he was frustrated with linearity of writing.

"A file Structure for the Complex, the Changing and the Indeterminate" Reprinted in New Media Reader

Writing about the problem of IBM-ers that think CS is clean, hierarchical, clear. Digression is essential; linking is important as a writing tool.

Why Ted Nelson Hates The Web (it got everything wrong ;-))! Berners Lee left out the fact that links are two-way; as a writer he was interested in attribution; financially and as a form or authoring. Micro-payment et all. How to cite something you disagree with? It is a dummed down version of a great idea. HTML, one-way linking was actually a good decision; now time to move on.

Nelson not shy of commerce; he wanted to be able to set the terms. Whose notion of giving are we accepting? What do we mean by sharing? Professional archivists understand this as a problem)

Cached on Google does not guarantee long lived access.

Fragility ... archiving is also about the choice of representation; some more stable than others. Spreading is a way to stabilize.

'centralising tendencies': Keeping the technology simple. The archive in the pocket, on an usb stick as html-files.

Video as an example of a potentially fragile media.

Europeana project ( )

Linked Data: The Good Parts!

Watching Video Web 3.0 by Kate Ray.

Aggregate news in a way that matters to the user

How do I find the right file?

the links mean something

A Little structure goes a long way

Fighting about the baby that is called the semantic web

The web is many things, the semantic web too

Systems that can reason

Once we have enough data on the web ...

Not about specific applications

We can not imagine yet what that will mean

Schism: splitting knowledge from writing; to disembody it. The split is part of the problem; why it ends up as enormous.

The world is messy, but there is an interesting space to describe.

Language is of course problematic; knowledge production is part of it.

Getting excited about data structures ... touching truth

I the problem a problem of scale?

dbpedia becoming central as a problem?

efficiency of a centralised data storage. There is a role for specialized indexes. Pull together specialized data. Create 'webrings' between smaller localized data sets

Freebase, "A social database about things you know and love, spanning millions of topics in thousands of categories." by Google with AA, search in the webrings of users of AA (Constant, Fundacio Tapies, Oral Site)

striving for the final information, encyclopedia, scientific publications...

semantic web: people want to meet, connect their data. How do you make transactions? f.ex. Wikipedia as a mutual ground

start in a very local manner your own encyclopedia and then join with others in a very refined explicit way, instead of starting from one central place, instead of big datasets - unique/true encyclopedia.

conflicting data about the date of birth of Tchaikovsky!

what is dbpedia?

it's a german attempt to extract knowledge from infoboxes of wikipedia (5-6 years ago - January 2007). in wikipedia your resources are encyclopedia entries about a topic. in dbpedia, you are able to talk directly about a topic and idenitfy directly tchaikovsky and share info directly with other humans or machines via rdf

if you want someone to link to your website you have to ask permission in France

dbpedia Live:

"La Gaîté lyrique explore les cultures numériques sous toutes leurs formes"

"Tout lien simple ou par hypertexte est strictement interdit, sauf un accord écrit exprès du directeur de la publication."

The freedom of language, to talk about things ... on the web, this is done through making a link. But the lawyers are coming, telling us "you can not link to this" (i.e. you can not talk about this).

Different theories of reference on the web possible

You can have your own ontology

Uri (or rather the resource it identifies!) = social object; one way to talk about a thing. URL :

every semantic meaning is defined by a URL or URI. "dbprop:placeOfBirth" ( )

Patrick J. Hayes, Harry Halpin: In Defense of Ambiguity Abusing the relation of identity relations. They are so strong, that they are not usable

When owl:sameAs isn’t the Same: An Analysis of Identity

Links on the Semantic Web, Harry Halpin, Patrick Hayes

How to talk about the same topic but specifying that these are different approaches of the same topic?

semantic web is in a crisis phase (already 10 years of existence)

'Links with flavor' Links with descriptive tag, kind of "smart links"

exercise now! (almost)

 sudo apt-get install redland-utils raptor-utils

download rdf

rapper g11.rdf --output turtle


Abstract model V.S serializations (ways of writing RDF, hence mediations).

RDF : three elements are linked together, most of the time they are resources identified by URIs (but there ere blank nodes and literals too). RDF/XML: the first serialization, is not very legible so we're shifting to Turtle (Terse RDF Triple Language: (this is a W3C working draft to make Turtle a recommendation).

Graphviz + SPARQL

Tagging as a way of writing, as part of writing, not on top or afterwards.

google's powerforce is training data.

select ?s ?p ?o
from <Lewis_Carroll.n3>
where {
?s ?p ?o .
select ?s ?p ?o
from <Lewis_Carroll.n3>
where {
?s dcterm:subject ?o .

Error! Instead, you have to declare the namespace by putting the URI of the dcterm:subject relation (the resource!).

select ?s ?o
from <Lewis_Carroll.n3>
where {
?s <http:/> ?o .
select ?o
from <Lewis_Carroll.n3>
where {
 ?o .

prefix rdfs: <>
prefix <>
from <Lewis_Carroll.n3>
where {
:Lewis_Carroll rdfs:label ?l . 
select ?o ?label
where {
    dbpedia:Lewis_Carroll dcterm:subject ?o .0
 ?o rdfs:label ?label .   

Dbpedia: a data repository that became a (messy but shared!) vocabulary.


  • Michael Murtaugh - computer scientist, F/LOSS Video Lab, Active Archives Video Wiki
  • Myriam Van Imschoot - dramaturg, performer, writer, Oral Site
  • Peter Westenberg - member of Constant, artist, F/LOSS Video Lab, interventions in public space
  • Nicolas Maleve - member of Constant, data activist, GenderArtNet
  • Maria Ptqk - Researcher, curator, GenderArtNet
  • Alex Leray - programmer, designer; Oral Site, Active Archives Video Wiki
  • Alexandre Monnin - PhD - Paris 1, INRIA, Head of Web and Metadata Research at Centre Pompidou Research and Innovation Institute (IRI)
  • Bart Magnus -VTi (Vlaams Theater Instituut); Travelogue, Archipel
  • Marie Wocher - Graphic designer, student at PZI (Networked Media)
  • An Mertens - member of Constant, storyteller, remixing of stories
  • Linda Valdés - Fundacio Antoni Tapies, working on Ars Combinatoria
  • Camille Pageard - art historian, art critic, interested in history of the book & publication as social medium
  • Stéphanie Vilayphiou - designer, OSP, Active Archives Video Wiki, works on Oral Site
  • Femke Snelting, member of Constant, graphic design & many more
  • Dries Moreels - BAM, Visual arts in Flanders, since 2008 fan of RDF; wants to open up data. Archipel project manager: archipelago!
Personal tools