Re: Connectionists vs Reductionists, Hard vs. Soft or Hard + Soft KR, Re: neural networks being purported as KR?

> On 29 Jul 2019, at 18:25, Stephen D. Williams <sdw@lig.net> wrote:
> 
> Yes: I think for most people most of the time, defining a concept always results in paragraphs of natural language.  This implies that, beyond has-a, is-a, etc., it is difficult to express the ideas of a concept without relying on the complex web of understanding of other concepts.  And that, beyond ontologies like foaf, we don't have enough shared context of our webs of understanding to express many concepts in a definitive or directly machine understandable way.
> 
> If anyone has a better definition and grounding for the concept of concepts, please enlighten me.  It is difficult to research completely based on such a common and widely used word.

This reflects the dichotomy between a reductionist approach to meaning and a relativist approach [1]. The former tries to define meaning in terms of a small set of core assumptions using logical principles. The latter gives up on overall logical consistency to instead define meaning locally in the context of particular queries. The relativist approach lends itself to words having fuzzy meanings that depend upon the context, and which shift over time to suit changing needs. This may be abhorrent to the logically minded, but appeals to a more statistically oriented mindset, and to implementation of symbolic reasoning on top of the noisy sub-symbolic spiking neural machinery of the brain.

The Semantic Web so far has focused on the reductionist perspective with its emphasis on logical deduction. That’s fine for some purposes, but we now need to explore the relativist perspective. The first phase is to make use of rule languages to define meaning operationally. The second is to introduce sub-symbolic information to better mirror well established characteristics of human memory. The third phase is to use natural language to define lessons to teach concepts and skills rather than programming them directly, which doesn’t scale effectively.

The rise in popularity of Property Graph databases shows that industry is already moving in this direction. Businesses are finding that they can mostly do what they need with software using query languages like Gremlin and GraphQL, whilst rule languages are popular for business logic.

[1] https://plato.stanford.edu/entries/relativism/ <https://plato.stanford.edu/entries/relativism/> which includes the following extract:

> Relativism has been, in its various guises, both one of the most popular and most reviled philosophical doctrines of our time.

Dave Raggett <dsr@w3.org> http://www.w3.org/People/Raggett
W3C Data Activity Lead & W3C champion for the Web of things