Re: Draft on some weird KR ideas

First, thanks for taking time on reading and commenting.

Of course that you regarding the concepts exposed as a 'strategy',
or another way to encode triples is pretty much enough for me, as I
intend to analyse the whole 'triple' subject in a different manner. 

Fundamentally I try to understand the triple paradigm in a functional

What I also try to understand, is, in one way, the 'natural' meaning
that could exist in many triple relationships on the real world, that could
carry some
'functional' logic embedded in them and the mappings they provide.

This 'mapping' could seem useful in scenarios like development of functional
languages structures, where extension of one of the three kind of entities,
meaningful relationship between them and their roles, by each mapped
concept, could allow to programaticaly or declaratively handle instances of
the three kind of entities in an ontology in which, for example, one can
incrementally add statements and retrieve state of affairs of the current
state or snapshot, or any previous moment in time.

The main usefulness of such functionality could be in the development of
'classifiers' like the ones used in machine learning tools (Weka). I think
this is not currently addressed by triples / triple stores as the ones
available now, but limited to special file format inputs.

And, finally, another objective held is the possible encoding of knowledge
concepts, triples or whatever, in a way where an algebra could be developed
to encode such statements and to infer, or calculate, in an algebraic
manner, new or related knowledge. In a first attempt, given the encoding of
predicates in which the possibility of a given situation is known, one could
infer (calculate) what kind of further actions, predicates, are possible,
over given possible/available subjects/objects (this concept I've borrowed
from workflow techniques).

Also, in a BI fashion, this kind of ontology would allow to ask what kind of
previous state is/should be necessary for a given actual or later state,
what kind of actual state could be carried given a previous one, etc.

Sure I'll try to push this further, however I can. Mainly I'll try to
develop some kind of tool for use the desired technique as an internal
storage format, with interaction with other KR serialization mechanisms. 

But the main goal should be to encode or 'set up' triples in a functional
manner, allowing for an algebra to be developed over them, and allowing for
data classification/aggregation to be only a matter of 'map'. Once one could
take some dataset, determine kind of ETL/MDM rules, and apply the whole
process, the whole ontology should arrange because of functional similarity,
membership function application, quasi automatically.


Adam Saltiel-2 wrote:
> Confused.
> What is weird about this? Peirce is the father of KR and Ontology or,
> rather
> of central to Pragmatist Philosophy and frequently quoted in the KR
> literature. I don't know enough, but isn't this what already is? If not
> how
> is it different?
> If it is different you should push it further as you are surely describing
> what *should* be meant by KR and RDF.
> But, so far, it just seems like a strategy for setting up triples and
> their
> treatment, nothing special.
> What have I missed?
> I really am very confused.
> Adam
> On 29 July 2010 08:01, SebXama <> wrote:
>> Hi,
>> Anyone who has the time or is willing to read something rare, I'm
>> trying to take a different approach to knowledge representation (KR). I
>> don't
>> know if it is new or even useful stuff neither
>> I'm trying to develop further those ideas to find useful applications
>> (if there is one at all!)
>> Regards,
>> Sebastián Samaruga
>> --
>> View this message in context:
>> Sent from the - semantic-web mailing list archive at

View this message in context:
Sent from the - semantic-web mailing list archive at

Received on Saturday, 31 July 2010 01:59:05 UTC