Linked Data

02 Nov 2011

See also: IRC log


about 10 000 people according to Ian.
Alexandre Bertails
Alexandre Bertails


who's here? why?

Virginie Galindo, from Gemalto

<scribe> ... new member of W3C, I'm here to know more about Linked Data

jeff from @@: we're interested in exploring Linked Data

scribe: maybe a layer on top of what we already have

<scribe> scribe: Alexandre Bertails

<scribe> scribenick: betehess

Chingteng from Academia Sinica Taiwan

MArk from SAP: I work with people defining SemWeb platform, hence Linked Data

@@ coming from Russia: just started a Linked Data project in Russia

scribe: Higher School of Economics

@@ from EMC: Linked Data evangelist

scribe: re: Open Linked Data, we don't really have anything to contribute, but we have customisers who are interested to publish data using thaSemantic Web technologies

<schunter> What is linked data?

Alex Milowski, XML Processing WG, interested in Open Scientific data

Mary from Mark Logic: interested for several reasons. we have several customers using Linked Data

ashok, Oracle: part of the TAG, participate to RDB2RDF

scribe: I have questions about this Linked Data. If it's not RDF, how do we find things?
... we don't have SPARQL

Arnaud Le Hors, IBM: Linked Data standards lead

scribe: used to be a research topic for a while, this has changed since data became a critical part of our architecture
... one interest is the use of LD for data integration
... for us, it's at the architecture level
... eg. you expose a bug on the Web as a Web resource
... it's RDF + http

<Cdavisafc> S/@@ from EMC: Linked Data evangelist/Cornelia Davis from EMC: "link your data" evangelist within EMC/

gkellogg_: I'm a IE on RDF
... one of the editors for JSON-LD, special case for RDF

<olyerickson> I'm John Erickson, Dir. WebSciOps at TWC RPI ...participant on W3C GLD WG, eGov IG

gkellogg_: one the numerous syntaxes for LD

john erickson, RPI: project leader for linked gov data

<gkellogg_> JSON-LD requirements: http://www.json-ld.org/requirements/latest/

scribe: involved in US data gov
... oversee our infrastructure re: 5 stars

<Arnaud> small (but important) correction: we use LD for *application* integration

<EricAxel> Eric Franzon from http://www.semanticweb.com - providing news & information about the business of Semantic Technologies and Linked Data, and providing educational resources: conferences, webcasts, podcasts, etc.

scribe: interested in data mashups, we also so evangelism

Eric Franzon: run SemanticWeb.com

scribe: we're also bejing the SemWeb conference

<FabGandon> Fabien Gandon, Senior Researcher INRIA, Vice Leader Edelweiss research team on semantic web and social web graphs, http://fabien.info


HadleyBeeman: working @@@@ in UK

@@ from @@: very interested in Linked Data. we'll be W3C Member

<HadleyBeeman> I'm CEO of LinkedGov, a not-for-profit initiative to clean, link and reformat all UK government/public data. I'm also open data lead for government's Technology Strategy Board and on the Government Linked Data working group

@@ from KDDI: interested with data mining with Linked Data

<gkellogg_> Gregg Kellogg, on RDF WebApps WG, also Microdata to RDF transformation and one of the editors of JSON-Ld

Satoru Takagi from KDDI: interested in Linked Multimedia data

tlr: from W3C

<ihilerio> Israel Hilerio, Microsoft IE team - curious about Linked data/semantic web

Yosuke Funahashi from Tomo-Digi Corporation: consultant for Japan gov on creating better disaster information system, hence interested by Linked Data

jean-francois Abramatic, IBM: also part of AB at W3C

scribe: I see lot of promises in Linked Data

<olyerickson> For Reference: TBL's Linked Data Principles http://www.w3.org/DesignIssues/LinkedData.html

(open) linked data / semantic web

@@: when we explored that in the JSON, we saw that with LD, the URIs are dereferencable

scribe: as opposed to simple URIs

arnaud: they *can* be, they don't have

@@: this became an issue in JSON-LD

scribe: so it's really a practice
... I posted a Linked Data article
... LD describe data relationasips between documents
... then, you use http uri, so that you can look up
... 3rd, when you look up, you provide usefull information
... no RDF involved
... it took our groups some time to come with that

arnaud: not that different from timbl's article

olyerickson: yes, but you use standards

arnaud: there is no official definition of this
... that's a concept, a term
... that's a problem for the industry


olyerickson: by using LD, you have easy and quick data integration
... coming from the first point
... if I use http uri, it's even more easy

@@: when you guys put all this data in triple store, is this linked data?

olyerickson: we create linked data, in the sense we harvest gigabytes of gov data, turn them in triples (not linked)
... then we have process of linking terms

<stakagi> Is "human-understandable data" a criteria of linked-data?

olyerickson: to provide meaning
... then we see if this data can be combine with other data
... the first step is timbl's first star

[ olyerickson speaking of in-links and out-links ]

Zakimm take agenda 5

[ timbl speaking about the 5 stars ]

timbl: there are other stars, eg. when you use open license

@@: we see also premium data, when data with quality can be sold

arnaud: it's like enterprise linked data
... in which term doesn't really matter

timbl: you might not want to put your data on the Web

olyerickson: the LD approachcan facilitate data integration

timbl's five stars of Linked Data

<FabGandon> e.g. of the five star scale http://lab.linkeddata.deri.ie/2010/star-scheme-by-example/

olyerickson: if you standardize on that in your organization, you can more or less easily link to external data

who's here? why?

betehess: is there one page with all the questions answered?

timbl: it's in many places
... sometimes in wikis

<olyerickson> see also http://linkeddata.org

timbl: I'm interested to know what should be in it

Arnaud: if you go to @@ on w3 website, you see some info about LD

<olyerickson> See also: http://linkeddata.org

Arnaud: but w3c should do a better job

<alexmilowski_> Here's the UK's data.gov.uk definition: http://data.gov.uk/linked-data

Arnaud: in my case, it's too difficult to speak about that internally
... for example, we would need one single slide
... or it's too difficult to have a discussion

@@: if you look at uk and us, they have different definitions of linked data

scribe: different ways to do it
... and what about SPARLQ endpoint?
... do we need that for linked data?

Arnaud: when you talk about LD, is it about al that data out there, or the technology?

olyerickson: good point
... people understand the concept, but where does it actually live?

Arnaud: on the Web :-)

olyerickson: if you go at RPI, we have a SPARQL endpoint, but also data available to download

betehess: can we define SPARQL?

olyerickson: pronounced sparkle
... it's an RDF query language and a protocol

<stakagi> Probably, SPARQL will be difficult for ordinary person

olyerickson: it's like SQL for relational data

timbl: but SQL does not tell you how to send it to the wire

olyerickson: your endpoint can be an http url
... and then for example feed your bolg

@@: you might query a sparql endpoint that doesn't contain *your* data

FabGandon: endpoint vs triple store

@@: so a SPARQL endpoint is something that understands SPARQL

<stakagi> I think that RDF/XML and SPARQL make Linked Data difficult for people.

olyerickson: yes, just a point of control that speaks the SPARQL protocol

<HadleyBeeman> stakagi In my experience, it's more the distributed model of triples that does most people's heads in

timbl: it could be called web service

FabGandon: yes, there is a soap binding for the sparql protocol

<timbl> stakagi?

olyerickson: you can have queries from the SPARQL endpoints, other APIs, etc.

<Zakim> stakagi, you wanted to ask of "human-understandable data" is a criteria of linked-data?

<timbl> What do you mean, human-understandable?

HadleyBeeman: depends on the data
... if it follow the structure in English
... you don't necesserally what it is about

FabGandon: that's a point I usually make when I speak about that
... for ex, people don't document their documents
... title can be a book title, or person title

HadleyBeeman: how does it compare to link on the web and pages?
... it's just we can speak about something out there, nothing more
... in gov, there is fear that it's gonna be misunetrepted, analysed, ...
... you loose control of your data
... it's more difficult than with web pages

timbl: I'm interested in patents
... eg. a pattern language
... for example, something is exposing data using PHP

<FabGandon> see also http://ontologydesignpatterns.org

timbl: you can also expose data from a database with a simple script

<olyerickson> See also http://patterns.dataincubator.org/book/ "Linked Data Patterns" (book by Leigh Dodds et.al.)

timbl: also, there is the pattern where to talk to people to align your ontologies
... eg. OGP
... schemas.org is another example
... what the UK did: we looked at the other ontologies our there
... the most important one
... when we have to produce a new one, we would first to it internally, then we bring it to w3c
... the uk gov does it well
... anti pattern: we release an ontology without speaking with anybody
... ontology alignment is another pattern
... would be great to write down all these patterns

betehess: propose to summarize tonight with:
... 1. subjects we talked about
... 2. need for w3c to write one page with definition re: (open) linked data, sem web, etc.
... 3. w3c to start one page about patterns to implement Linked Data

olyerickson: in addition to papers, there are several book that emerged

Minutes formatted by David Booth's scribe.perl version 1.136 (CVS log)
$Date: 2011/11/02 22:59:49 $