Fwd: IPFS-LD - Linked Data

[Dante-Gabryell Monson]

---------- Forwarded message ---------
From: S
Date: Sun, Feb 4, 2018, 04:34
Subject: Re: IPFS-LD - Linked Data · Issue #36 · ipfs/ipfs · GitHub
To: dante....@gmail.com <dante....@gmail.com>

any data can be addressed (URL) and transferred with either http or ipfs. ftp, sftp, and a million other protocols that serve various application niches. designating http:// or ipfs:// is what tells the software where and how to look.

 

on the other hand, semantic web is data which is *supposed to be* independent of not only the delivery method, but also the particulars of its representation hence the various 'serializations' such as N3, Turtle, RDFXML etc. but most importantly there are hidden representations, active in computer memory or databases, that a program will use while actively processing or querying RDF or any similar kind of graph data. here the programmer decides the particular data access modes that are most efficient and any inputs and outputs to the external environment must inevitably be adapted to that.

 

what really matters is that in the user interface, any portrayed symbols maintain their correct semantic grounding to the referred concepts, so that the user does not become confused or suffer any loss of clarity or introduce ambiguosity with what they are trying to do. think of the vast descriptive multilingual namespace of wikipedia pages which exists independently, but compatible with any system that wants to refer to its concepts - which can be addressed simply by one or more words.

 

whether this happens as a result of various combinations of IPFS, HTTP, JSON-LD, KQML, KIF, NAL, XML, CSV, etc. really shouldnt matter. the real semantic web apps should speak several of these and more and should decide which to use on their own according to its awareness of an application's context, and unify relevant details in a user experience in ways that don't emphasize any particular technology an provide implicit privilege over any other

 

RDF/OWL is in my opinion a dead-end no matter how anyone tries to spin it and co-opt it into some other system. ive shown examples of technologies pre-dating the so-called web which envisioned a more advanced paradigm of 'decentralized artificial intelligence' yet we all settled for hypertext pages served to obedient client 'browsers'.

 

https://www.elsevier.com/books/distributed-artificial-intelligence/unknown/978-1-55860-092-8

https://decentralized-ai.com/

 

i dont see how any kind of software isnt somehow 'semantic' or 'linking data' in some way. the real question remains unaddressed: how to improve and empower the actual user experience, not some particular way of maybe doing so.

 

 

03.02.2018, 20:37, "Dante-Gabryell Monson" <dante....@gmail.com>:

An old 3 year old comment

 

https://github.com/ipfs/ipfs/issues/36