Re: Open source release of "Stralo", a new Linked Data platform

On Thu, Oct 19, 2017 at 9:19 AM, Jean-Marc Vanel <jeanmarc.vanel@gmail.com>
wrote:

> I see what you mean, and I saw that your application exposes the triples
> as RDFa.
> I  have more questions:
>
>    - can your application import triples from the LOD (and display,
>    reason, etc) ?
>
> Yes, though for now, this is only implemented in the Fact-block (
https://github.com/republic-of-reinvention/com.beligum.blocks.imports.fact).
A "fact" in Stralo is actually a "formal statement", defined by a
(server-side) RDF mapping.
See this page as an example:
http://www.mot.be/resource/Bakehouse/zwaen-1-voeren-belgium?lang=en
There, you can see the page is about a Bread oven (the MOT keeps an
inventory of existing and lost bread ovens) and the fact "City" references
an external ontology (Geonames). I'm attaching a screenshot of what the
admin mode of that page looks like: we implemented the links to external
ontologies as autocomplete boxes: users just start typing and the backend
of the autocomplete-box communicates with one or more LOD ontologies. When
the users selects something from the autocomplete, the URI of that external
resource is inserted into the RDFA and the remote RDF triples are copied
into the local triplestore (for safety).

When you click the resource (see example above), the server decides what to
do with it (in this case, redirect to the geonames page).

We haven't implemented any reasoning yet, but there's a full SPARQL
endpoint API (not public/RESTful yet), so it's on our TODO.

>
>    - a standard SPARQL database is not present, and the Stralo site
>    mentions  "[instead of] storing all information in a triplestore
>    <https://en.wikipedia.org/wiki/Triplestore>, we have built in native
>    support for machine learning, grid computing, distributed storage, etc" ;
>    can you tell us more ?
>
> Yes, there is. The triple store is stored in
"storage/pages/triplestore/data/" and is a vanilla Sesame native sail
(rdf4j-sail-nativerdf), so everything eventually ends up in an
"old-fashioned" triplestore with a nice SPARQL endpoint. On top, we added a
specialized indexer (currenly based on Lucene) to speed things up. I know
there's a Sesame (or RDF4j) sail for that, but we decided to write our own.


>
> I didn't see a Fact-block here :
> https://github.com/republic-of-reinvention/com.beligum.
> blocks.core/tree/master/src/main/java/com/beligum/blocks
>

That's right. Every block has it's own repository, so you can cherry pick
which ones you want to include in your project and which not. Most of them
(currently all of them, actually) are grouped in "the pack" for easy
management, see
https://github.com/republic-of-reinvention/com.beligum.blocks.imports.pack


I hope this starts to answer some questions. We're busy writing the docs, I
hope to put new ones online soon, but needs to structure it all a bit first.

all the best,

b.