[PragmaticWeb] Ideas for a possible framework

Mon Oct 10 21:43:26 CEST 2016

(Apologies for cross posting. Still don't know which would be the right
community for this)

Updated draft (incomplete). Open to more suggestions, helped me a lot in
the past. Thanks in advance,

Sebastián Samaruga.

On Oct 2, 2016 12:03 PM, "Jim Whitescarver" <jimscarver at gmail.com> wrote:

> Sebastián,
>
> The problem of wrappers that unify heterogeneous sources is universal.  A
> survey of tools and languages for this purpose is needed.  This is a hard
> problem often requiring a human brain to resolve.
>
> The framework you suggest models human brain processes, which can extend
> human memory with vast datasets logically connected.  I like it.
>
> A DARPA proposal I worked on planned employing machine learning which may
> be applied to learning about the data and the state of knowledge and rules
> of the questioner recognizing fields by their content and resolving
> ambiguities by asking intelligent questions knowing the questioner,
> according to their rules.  Your framework might provide scaffolding for
> such an effort.
>
> FreeTrust.org in DivvyDAO.org proposes crowd sourcing oncologies for
> security, identity, trust and cooperation employing Bayesian learning and
> maximum entropy with deep learning ultimately.
>
> This focus is because I believe security and identity are most essential
> to humanity in the realm of the online society faced with information war.
> We are responsible for keeping ahead of our adversaries
>
> Others might bring other classes of ontologies into the fray and we may
> begin to reify the semantic web vision of augmented human intelligence.
> The dream has been slow in coming.  I suggest we learn to collaborate
> better.
>
> Best,
>
> Jim
>
> On Sat, Oct 1, 2016 at 8:47 PM, Sebastian Samaruga <ssamarug at gmail.com>
> wrote:
>
>> The point is that, in fact, I'm not leaving the simplicity and beauty of
>> single RDF quads but I try to reify everything in a model, from SPOs to
>> triples, into a metamodel which allows for augmenting this reification(s)
>> with useful metadata which is, in turn, encoded into an RDF quads model
>> useful for type, relationship and behavior inference.
>>
>> Sebastián.
>>
>> On Oct 1, 2016 9:24 PM, "Timothy Holborn" <timothy.holborn at gmail.com>
>> wrote:
>>
>>> You've cross-posted to a bunch of communities i'm not directly linked-to
>>> and whilst i understand the context of having a community discussion; i
>>> don't really understand what additional functionality you are attempting to
>>> obtain via what appears to be far more complex modelling than the relative
>>> beauty / simplicity of triples/quads.  Have you tried to build the
>>> functional outcome that you are trying to define a solution to, using
>>> existing linked-data systems?
>>>
>>> It's worth nothing that the development of what is termed 'RDF' or
>>> 'Linked Data' in its simple form; has been a work of art produced by many
>>> notable individuals over a period that extends to decades, but moreover,
>>> embodies close to 20 years work, patent related IP rationalisation and a
>>> great many other complex feats 'ticked off' through a structure that may
>>> perhaps be considered 'too simple' by the initiated. Like all good things...
>>>
>>> Whilst i never like to suggest ideas put-forward do not have some merit
>>> in some form (even when i can't identify it) i do wonder whether you might
>>> be better off working with existing projects to identify how the sorts of
>>> things you are trying to achieve; may be done achieved collaborative
>>> efforts, with others.
>>>
>>> Hereafter; a few pointers.
>>>
>>> Kingsley has an array of materials online which in-turn fit into a
>>> solution you can test.
>>>
>>> http://www.slideshare.net/kidehen/
>>> https://www.youtube.com/user/kidehen/videos
>>>
>>> Some of these videos outline functionality that provides
>>> interoperability between RDBMS and RDF.
>>>
>>> https://github.com/solid/solid  is a project that looks to
>>> decentralise.  If you are interested in building an APP, i'm sure they'd be
>>> interested in more helpers.  It is likely important though that you are
>>> able to develop the app.
>>>
>>> http://linda.epu.ntua.gr/ is a neat little tool that helps you easily
>>> remap data (ie: CSV data) into RDF.
>>>
>>> http://vowl.visualdataweb.org/webvowl/index.html#iri=http://
>>> lov.okfn.org/dataset/lov/vocabs/schema/versions/2016-08-09.n3 is an
>>> example of a tool that can help to visualise ontologies.
>>>
>>> https://cse.google.com/ helps perform queries based on structured data.
>>>  http://lod-cloud.net/  is a bunch of structured data.
>>>
>>> Semantic Reasoning, et.al. leads to knowledge about
>>> https://www.w3.org/TR/sparql11-query/  which in-turn leads me
>>> suggesting you play with http://json-ld.org/playground/  and/or
>>> http://linkeddata.github.io/rdflib.js/example/people/social_book.html as
>>> an old example that should help.
>>>
>>> Beyond that;
>>>
>>> It takes time to understand what has already been made, why it's been
>>> made that way, and how to contribute.  https://twitter.com/WebCivics
>>> /status/492707794760392704 i use often as an 'intro' piece.
>>>
>>> If some particular function exists that is not supported by what is
>>> made, let us know.  However i'm really not sure what the underlying
>>> principles are to the way in which you are trying to find purposeful means,
>>> at present.
>>>
>>> The 'build a solid app' strategy may be a really good way to further
>>> demonstrate your ideas, IMHO.
>>>
>>> hope something noted above is useful for you.
>>>
>>> Tim. Holborn.
>>>
>>>
>>> On Sun, 2 Oct 2016 at 03:23 Sebastian Samaruga <ssamarug at gmail.com>
>>> wrote:
>>>
>>>> Those sets (and classes) are my ontology. Consider like I'm reifying
>>>> subjects, predicates, objects and triples into sets and 'calculating' its
>>>> kinds and this allows for schema less data sources (plain RDF triple
>>>> sources) type, relationships and behavior inference. That's why I bother
>>>> with metamodels, because I don't rely with source data coming with an
>>>> schema or ontology and I have to build or infer one and link and merge it
>>>> with existing ones. Then, the metamodels allow, for example, to build a LDP
>>>> or other protocol service from the schema less sources by means of the
>>>> inferred metadata. The whole document explains how this is intended to be
>>>> implemented.
>>>>
>>>> Best Regards,
>>>> Sebastián Samaruga.
>>>>
>>>> On Oct 1, 2016 10:02 AM, "Timothy Holborn" <timothy.holborn at gmail.com>
>>>> wrote:
>>>>
>>>> +1
>>>>
>>>> On Sat., 1 Oct. 2016, 10:16 pm Martynas Jusevičius, <
>>>> martynas at graphity.org> wrote:
>>>>
>>>> Sebastian,
>>>>
>>>> I've said this before and I'll say it again: why do you need to build
>>>> a (meta)model above RDF? Kind, SubjectKind, Dimension etc. -- why is
>>>> all this stuff necessary?
>>>>
>>>> Do not attempt to extend RDF, and drop the UML/object-oriented models.
>>>> Instead, work *within* RDF: use triples to store data, and use OWL
>>>> ontologies, classes, properties, datatypes etc. to model your domain.
>>>>
>>>> Those are the only things you need. Show us your ontologies, then you
>>>> will get better responses. You can try some of these ontology editors:
>>>> http://protege.stanford.edu/
>>>> http://www.cognitum.eu/semantics/FluentEditor/
>>>> http://www.topquadrant.com/tools/modeling-topbraid-composer-
>>>> standard-edition/
>>>>
>>>>
>>>> Martynas
>>>>
>>>> On Sat, Oct 1, 2016 at 5:20 AM, Sebastian Samaruga <ssamarug at gmail.com>
>>>> wrote:
>>>> > (Apologies for cross posting / over posting)
>>>> >
>>>> > Hi, I'm currently a software student and developer. Since I've meet
>>>> semantic
>>>> > related technologies development about twelve years ago I've been
>>>> revolving
>>>> > with the idea that a framework could be built that could ease building
>>>> > semantic business applications as they are frameworks for Java and
>>>> > relational databases.
>>>> >
>>>> > A lot of time passed. Now many big players offer solutions that
>>>> somehow rely
>>>> > on semantics for their work. And although this could seem strange,
>>>> here in
>>>> > Buenos Aires I couldn't find anyone really interested in the area,
>>>> being in
>>>> > academia or places I've worked in.
>>>> >
>>>> > So, having no one to share my thoughts with, I'm frequently publishing
>>>> > documents to this list(s) hoping for some kind of peer's feedback.
>>>> Sorry if
>>>> > this aren't the right lists or I'm off topic. I send my attachment as
>>>> a PDF
>>>> > document. Anyone willing to comment in the original just ask me for
>>>> the
>>>> > Google Docs link.
>>>> >
>>>> > Note: I've sent this draft before but in a very early version state. I
>>>> > invite anyone interested in reading to see the last section
>>>> (Dashboards).
>>>> > Maybe I'm wrong but I think there is a lot of innovation that may be
>>>> done
>>>> > regarding that subject (sorry for the poor diagrams :--)
>>>> >
>>>> > Best Regards,
>>>> > Sebastián Samaruga.
>>>>
>>>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.spline.inf.fu-berlin.de/pipermail/pragmaticweb/attachments/20161010/d6afe418/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Spec.pdf
Type: application/pdf
Size: 543041 bytes
Desc: not available
URL: <http://lists.spline.inf.fu-berlin.de/pipermail/pragmaticweb/attachments/20161010/d6afe418/attachment-0001.pdf>