Linked Data
(And what it can do for the JS developer)
© 2016 CC-BY-SA Rein van ‘t Veer
Rein.van.t.Veer@geodan.nl
Rein.vanVeer@kadaster.nl
Rein.vant.Veer@vu.nl
Rein van ‘t Veer
Been working as pro JS developer for a year, but been hacking away for a few years
Brought several identities:
My employer is Geodan
I am seconded to the National Cadastre, where, {you do what?}
I am a researcher at the Free University of Amsterdam, embedded at Cadastre
Today I’m giving a Semantic Web / Linked Data pres, with a twist
Most of the semantic web stuff can be done with JS {most, not all}
So what’s the SW?
What Is
Semantic Web?
A movement for changing the web from a collection of documents to a global database of semantically rich information sources; a collection of strategies for gathering and utilizing knowledge.
Linked Data?
The language of the semantic web, highly interoperable, expressed in the Resource Description Framework (RDF)
The web is mostly unstructured. We can structure it from text -> knowledge
This knowledge part is pretty central. Knowledge is something you can reason about.
A lot of the intelligence to do with the SW is about reasoning over knowledge.
We’ll get into that.
So…
What is RDF, what do you mean by ‘semantic’, what’s the interoperability, how is Linked Data a ‘language’, what collection of strategies is in the semantic web, what global database are you blabbering about?
I could tell you
But then I’d have to… you know
Keep you here for several weeks
So… we’re going to do linked data by example. With a twist. Serverless.
So, I’m not going to do a talk where I explain all this stuff.
Why not: it’s a ‘framework’ - http, vocab/onto, graph analysis;
it supports varying use cases, from taxonomies to product SEO
So real catch of this talk: no ‘classical’ impression. It’s what I like about the SW
We’re making our own social network, with some network analysis.
Our own little social network
Actually brings people together
The only social network without privacy issues
Its USP: tell in what circle of acquaintances you participate
{fill in first row}
Semi-structured data: table
It has table headers. It’s json parsed from csv to keys and values. Semi-structured.
{second row}
We have a directed graph, from linked data
It allows us to reason over it
{third row}
Hey, we have a cyclical graph!
We did actual network analysis on the graph to make an application feature
So how does this work?
SPARQL
Who knows him/herself through three degrees {slash} of separation?
This (client side) deviates. SW is usually heavy infrastructure. Multi-node stuff.
Just a start of what you can do. Like SQL you can aggregate, count, subquery. It’s even more expressive than SQL.
This JS impl. is limited in the standard. Normally property paths can be expressed through +, * and {}.
With the used library, you can’t. There’s no support (yet, that’s only a matter of time).
Actual SPARQL
Who knows him/herself through three degrees {slash} of separation?
This (client side) deviates. SW is usually heavy infrastructure. Multi-node stuff.
Just a start of what you can do. Like SQL you can aggregate, count, subquery. It’s even more expressive than SQL.
This JS impl. is limited in the standard. Normally property paths can be expressed through +, * and {}.
With the used library, you can’t. There’s no support (yet, that’s only a matter of time).
Semantics
{
"@subject": "name",
"@type": "foaf:Person",
"@context": {
"foaf": "http://xmlns.com/foaf/0.1/",
"@base": "http://mysocialnetwork.org/id/",
"name": "foaf:name",
"link": {
"@id": "foaf:knows",
"@type": "@id"
}
}
}
The semantics: strategy called JSON-LD - LD for JSON. It’s a sane way of expressing semantics for JSON
In this case, it provides a context for all JSON documents or SN relationships
We saw something called FOAF {explain}
JSON-LD
The JSON-LD is just JSON with some context
Some credits
This presentation was made using
Rein.van.t.Veer@geodan.nl
Rein.vanVeer@kadaster.nl
Rein.vant.Veer@vu.nl