--
You received this message because you are subscribed to the Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to topbraid-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/e694589f-d060-4bc6-b595-da8fe2a273a4o%40googlegroups.com.
On 26/06/2020 03:31, Irene Polikoff wrote:
Hi Richard,
We are glad to hear you are enjoying this new feature. 6.4 is the first release of this feature and it was added relatively late in the release cycle which is why it is currently marked as experimental.
We plan on extending the active data shapes support and its use within EDG even further for the next release. And we want to provide more documentation. With this, the experimental tag will definitely go away before the year end.
Let's see if Holger has something else to add.
For an example of what's coming next (even in the 6.4 final
release), see http://datashapes.org/active/import.html
which adds a new importer framework for JSON, Spreadsheets and XML
files (or any text-based format really). It does so relatively
effortlessly, through the flexibility of having a general-purpose
programming language integrated with a simple RDF graph API. And
the mapping from SHACL shapes to JavaScript classes will hopefully
make it easy for newcomers to get started.
The changes.txt log of 6.4 includes some details of the state and what's on the horizon:
- Feature: TBS-2949: First preview of the experimental Shape
Script feature. Included for feedback from early adopters.
Known limitations/unfinished pieces:
- Only server-side JavaScript execution (neither
Python, nor R, nor client-side JS execution)
- Limited system class API (rdfs_Class, sh_NodeShape
etc). These are currently injected from a "magic" file
system.shapes.ttl but for all this to function
properly, that really should be owl:imported into all Ontologies.
- No proper API for iterative queries yet, i.e. you can
either fetch all property values or use SPARQL as-is.
- Buttons to interrupt scripts may sometimes leave
unclosed iterators.
- No real debugger/console yet. You can use print(...)
to print to System.out though.
- Only superficial integration with existing SWP
modules for now. These should become just JS functions.
- Almost none of the well-known SPARQL functions from
namespaces like teamwork, spif etc have been migrated to
sh:Functions yet,
and therefore don't show up in their JS prefix
object.
- May create too many copies of internal data
structures and GraalVM engines, e.g. not reuse the ontology across
workflows.
Meanwhile we appreciate anyone's feedback (positive or negative) to fine tune and further improve this work.
For example, would anyone be interested in Python support?
Thanks,
Holger
For those on the mailing list who have not yet heard about Active Data Shapes, please see http://datashapes.org/active/. The video linked below is rather long. If you want to go directly to the part of the video that introduces active data shapes, fast forward to 40:15.
Regards,
Irene
On Jun 25, 2020, at 1:13 PM, Richard Dijkstra <richarddi...@gmail.com> wrote:
Dear all,
I have looked at the webinar of Holger about shapes http://www.lotico.com/index.php/The_Many_Shapes_of_SHACL and have been experimenting prompt with the script editor and used dot notation for updating triples! A great exprience.
The state of this feature of active data shapes is experimental but works already perfectly. Can you say more about the expectation when the tag experimental is removed?
Richard D
--
You received this message because you are subscribed to the Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to topbraid-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/e694589f-d060-4bc6-b595-da8fe2a273a4o%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to topbraid-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/87646424-0E96-4D57-8BCB-CAB8CF8C6FA9%40topquadrant.com.
Hi Rob (or others),
as you say, one of the benefits of a Python mode of Active Data Shapes would be that 3rd party Python libraries could be used within TopBraid. Compared to an external process this might lead to a more intimate integration. Instead of serializing data from triples into text files and sending them across to the 3rd party engines, the data could remain where it is (in the TB database) and the algorithms could potentially just incrementally fetch the info that they need.
I have raised an internal dev ticket to investigate such an
integration, potentially for the version that will be out by the
end of the year (tentatively called 7.0). I would appreciate input
on specific tools that make this more concrete. What APIs would we
need to expose so that the data can be more seamlessly by exposed
to those libraries, e.g. do they have "interfaces" that we can
then implement?
Holger
To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/110973ba-eefe-4892-b13c-1b56209b4ca3n%40googlegroups.com.