RDF / Linked Data

[finnp]

Hey,

I was just trying to wrap my head around how to use RDF with dat.

Here are a few points that I came up with that seem to be relevant?

• How to use dat as a graph database? (levelgraph, also https://github.com/mcollina/levelgraph-jsonld)
• How to provide a RDF/JSONLD front-end on top of dat? (Does this need a graph database?)
• How to integrate with existing linked data platforms? Like marmotta

@almereyda @wikier It would help if you could describe your use cases to know what is important for you.

Best,
Finn

[wikier]

Sure. I'll prepare something in the following days.

[balupton]

Any status on this?

[wikier]

Not really, sorry, I did not have time for even look to it...

If we'd like to integrate dat on top of any Linked Data platform, I have some questions to focus the work: what would be the main features required? where should I start to look in to plug it in?

[okdistribute]

After some convo last night, I think it'd be cool to interface with Linked Data discovery platforms. Perhaps, for starters, in the package.json for the dat, you can include an RDF-compatible mapping between a local column and a remote column . For example, I could have a dataset with citycode, population, and map the column citycode to wikipedia entries for cities, for instance..

If we put that dat into a discovery platform, like dathub or ckan, you can imagine being able to traverse all of the links that people have reported.

[jbenet]

Please please please!

And, check out JSON-LD, which is also linked data and has all the RDF power but in very clean, simple json.

—
Sent from Mailbox

On Fri, Aug 21, 2015 at 1:35 PM, Karissa McKelvey
notifications@github.com wrote:

After some convo last night, I think we'd want to interface with Linked Data discovery platforms. Perhaps, for starters, in the package.json for the dat, you can include an RDF-compatible mapping between any of your columns and other columns. For example, I could have a dataset with citycode, population, and map the column citycode to another dataset's column on the web.

If we put that dat into a discovery platform, like dathub or ckan, for instance, you can imagine being able to traverse all of the links that people have reported.

Reply to this email directly or view it on GitHub:
#20 (comment)

[todrobbins]

Following...

[almereyda]

I'm pinging everyone associatable from my perspective, as I feel highly insecure about some of the statements below and would like to invite for a fair portion of critique here.

---

Please allow me to reemphasize the TransforMap community story first, before sketching our interest in using dat with RDF data, as requested.

TransforMap

a commoning economics community

Within an - as far as I can tell - unique topology, actors of the emerging field of Commoning economics initiated what was then called Mapping all alternatives. A pretty representative set of organised and practitioning opponents against the global hegemony of capital, though reaching from Commons and Social Solidarity Economy, Transition Networks or Degrowth academics to the Collaborative Economy and a so-called Global Ecovillage Network. The list would just be too long, as it's still a very polymorphous interest group.
The point is: It is not the usual kind of collaboration, where existing organisational resources are being deployed for productivity. We are only another Commons in its infancy of self-conception.

One of our shortest descriptions reads as follows:

TransforMap provides an online-platform to map the myriads of alternatives to the dominant economic model. It gives to everybody the opportunity to map all initiatives, communities, projects, worker-owned, self-managed or (at least) democratically organised companies dedicated to meeting people's needs and serving the common good. It will visualize all places, spaces and networks that work on fostering cooperation and deepening human relationships through (co-)producing, exchanging, contributing and sharing, for a free, fair and sustainable world.
TransforMap invites all the existing mappings to cooperate for an open solution to map the plenty of alternatives based on a common taxonomy, Free Software and a standardised APIs and published under an Open Data License.

While we made it easy for groups and individuals to express their intent in working with and for us, being such an advanced narrative of civic self-organisation brings a higher viscosity and lower comprehensibility to the process. Thus we depend on cleverly interfering and resonating with fitting value flows inside the free and open source and open data ecosystems.

I'm writing this two weeks after turning 30, perfectly broke, hungry, living in a painful post-socialistic debt which before '89 we didn't imagine could be possible. And I am in a good mood, happy; as I have probably never been. @WardCunningham told me once in a @federated-wiki Hangout one needs a so-called day job to gain resources and stability for working on open source software, especially in these experimental environments. I don't accept this. Why I consider working on TransforMap a full time occupation which doesn't allow to get tainted by the corporate world, because it is just too important for itself. For it will help us to make the day without money.

A brief history of Linked Open Data and the Federated Social Web

Of course I am confused. When Linked Open Data turns out to be a dialect of quantum mechanics and a Federated Social Web is the unixoid ideal of the « old » Web 1.0 still existing next to the centralised social silos, SoLiD is of course a slap into the face of what the Internet (here: protocol family) is for the Web. Even Sir @timbl himself cannot deny the flourishing diversality, as opposed to universality, of furious hacks and any non-HTTP-based RX/TX protocols.

But with Information Centric Networks like NDN or IPFS around the corner, while schemes and patterns further get drastically intermingled, infrastructural developments often remain within or circling around the frameworks of hierarchical organisational structures, namely States, Corporations and Academia, thus networked organizations. Yet the alternating world of resistants often doesn't fit within those prescribed social conglomerates and precariously nourishes and promotes downscaling economies of need, not greed. As informality and low-to-none resources come together, vulnerable organized networks intrinsically rely on distributed trust models in their everyday practice.

Our task is - as for many - to communicate one specific part of the complexity of the world precisely and caringly back to the Public Domain. In doing so we are strictly bound to collaborative work modes and often highly experimental collective governance constructs. A definition of the semi-permeable fringes is often doomed to fail, lacking explicit articulation of and agreement upon the shape of the sociotechnical artefacts to be expected from what one calls TransforMap. Do you have a test for that?

Particular technological constitutive outsides to this process are OpenStreetMap and Wikidata, both which represent more the autodidact and exploratory way of patching their Commons, next to more formal work done by OGC and W3C. Being fully aware of the joint Spatial Data on the Web WG (@philarcher1) and what's been done over at the Social Web WG, it is often hard for me to justify remaining in a slow pace and to argue for seperated concerns between a federation of Commons, instead of falling into the productivists' trap while implementing another incompatible silo.

"We can always implement decentralisation later."

The role of the dat ecosystem

Where I see a fair chance to finally approach a real-time-to-offline distributed semantic WebGIS, with containerised and interconnected applications for the masses available to collectively host shared nanoservices infrastructures in small data scenarios, others tie their expectations to a reinvention of RDF or another singleton infrastructure for geodata. Additionally, we need a way to catalogue and intersect (:arrow_upper_right: top geosemantics references) the still unconnected Many Many Maps of activist activity on the ground.
Data transformation pipelines are usually heavy, to be kind. They don't have to be.

A UUID is the least common denominator which we could agree upon, which is pretty close thinking TransforMap as an ICN. Evolving the (non existing!) stack out of Semantic MediaWiki's, Wikibase's and OSM's implicit legacy is my crucial interest to integrate our work into the redecentralised web; in direct opposition to the other core developers, who favour staying within known realms without wasting too much resources into interoperability research.

In my perception, dat can play the role of a mediator here. In citing @michielbdejong at our proclamatory Mapping MMM session:

Q How should we publish Linked Data?
A Just publish JSON-LD and that's Linked Data!

If I could just put all the practically unstructured data from these diverse maps for quick versioned republishing into dats, aggregation and updating maintained by dathub, discoverable by dps, automatically transmogrified into some simple archival format expressed in JSON-LD, I'd immediately have access to the data in the form of triples to start the collective ontology modelling process from.

And could even skip CKAN. Do you see how I am avoiding to use any PHP-based backends? These don't count anymore as collectively customizable code for me.

If we add a thin LDP layer on top of this, we'd be happy to use the additional readwrite-distributed-authentication-patterns. But happynessand beauty arrives from small ideas, a comparable JSON-LD representation of any dat could be one of these.

In the end, IMHO this thing is bigger than INSPIRE, as it involves a greater diversity of actors, happens closer to basic human's needs and even explicitly intends to help spreading the word for alternative approaches of fulfilling them. We are federating tempospatial data which is highly social.

Marmotta, for instance, could then serve as a full-blown Linked Data backend for more advanced postprocessing, eventually mirroring to something like Strabon for tempospatial indexing. But here be dragons; I haven't touched any of this, yet. Also the collective ontology modeling part is considerably still pretty vague, but writing OWL won't do for most of our users, which we still consider in primary responsability for whatever they can conclude with our aid. But dats could carry the imprints of their interface interaction and provide provable, versioned datastores for a high number of (unstructured) schemes to be mapped in a logically consistent, non-contradictoric way (@bblfish) onto known vocabularies.

---

Is this kind of what you were asking for @finnp?

Any conclusions derived thanks to numerous individual conversations with @keikreutler, @elf-pavlik, @rhiaro, @chron0, @kirilind, @mozboz, @Karissa, @pierreozoux, @fosterlynn, @bhaugen, @marianamalta, @ahdinosaur, @rynomad, @species, @thoka, @AdrienLab, @gandhiano, @jaensen and many others.

[todrobbins]

Won't Merkle DAG (https://github.com/mafintosh/merkle-dag and https://github.com/mafintosh/dat-graph) be part of the solution to these kinds of use cases? @mafintosh @maxogden