- From: Sebastian Samaruga <ssamarug@gmail.com>
- Date: Tue, 20 Jun 2017 16:08:19 -0300
- To: semantic-web@w3.org, public-rww <public-rww@w3.org>, pragmaticweb@lists.spline.inf.fu-berlin.de
- Message-ID: <CAOLUXBuBs1j+7w0sT0f0b_PC2oOmFxuu1hJnhdvUyNeXKnS-7w@mail.gmail.com>
Hi, Sorry, I apologise again if I'm bothering someone with all this stuff. I know it's not gotten read a lot. But if someone does, I'm just looking forward for some feedback. I just try to explain the idea of RDF being the platform of (layers) of metamodels, each one addressing one of: identity alignment and merge, attribute and links discovery, and ordering / sorting of knowledge. All this leveraging the separation of concerns as regarded in the OOP design pattern: DCI (Data, Contexts, Interactions: https://en.m.wikipedia.org/wiki/Data,_context_and_interaction) which, with the help of metamodel abstractions (Services) enables the possibility of a declarative style of programming (see attached document). Its funny that when I was developing the models I was struggled with the ways to encode RDF Quads components (CSPO identifiers) in a way compatible with Machine Learning representations such as those of Google TensorFlow. Well, it comes up that the metamodels I've developed forms a 3 by 3 matrix of 'quad tensors' (D, C, I by Schema, Rels, Order) being D, C, I instances of Schema, Rels and Order. If this is right maybe it should be easier to develop TF models wrapped around Nodes that should enable ML over this representations. My problem is the following: I can't mention Semantic Web (not even this project) with friends, colleagues, partners or whatever without getting shout off. And I really don't know if this list are the right place either. So, feel free to tell me whatever your opinions are or forward me to another lists. Thanks in advance, Sebastián.
Attachments
- application/pdf attachment: Business.pdf
Received on Tuesday, 20 June 2017 19:09:55 UTC