Re: Cognitive AI CG update/Comment on the diagram

> On 15 Jan 2020, at 05:15, Paola Di Maio <paola.dimaio@gmail.com> wrote:
> 
> Dave
> 
> You have solicited a comment on the diagram, thanks for sharing
> 
> Although I am (marginally) familiar with cognitive architectures as they are described in 
> scholarly literature, I need to strain my brain (mor worrying, need processing time) to figure out what those labels mean, so a few words of explanation of that the terminology means and how ti is derived could help.
> Furthemore, some justification: perhaps also give some rationale as to how the diagram modifies the original 

The high level architecture and its major building blocks are explained in:

    https://www.w3.org/Data/demos/chunks/chunks-20200110.pdf

The rule engine inspired by ACT-R is explained in:

    https://www.w3.org/Data/demos/chunks/chunks-20191205.pdf

There are several books on ACT-R itself, see: http://act-r.psy.cmu.edu/ <http://act-r.psy.cmu.edu/>

In addition, there is a large body of research papers on the cognitive sciences that can be found via web search engines.

> ACTR (what are you doing that was not done before) and perhaps also why out of all the possible cognitive architecture, your CG choses that particular model
> PDM

Work on ACT-R has focused on specific research interests in Cognitive Science, and not in providing an application platform for industry needs. Chunks & Rules replaces the Lisp syntax with something easier for today’s programmers. The intention is to look more broadly at Cognitive Architecture, as ACT-R only focuses on the basal-ganglia circuit, and to prove how this broader approach can be applied to a wide range of topics, i.e. different kinds of machine learning, many different forms of reasoning, natural language dialogues, theory of mind, social interaction and so forth. 


> 
> 
> 
> On Wed, Jan 15, 2020 at 8:47 AM Paola Di Maio <paola.dimaio@gmail.com <mailto:paola.dimaio@gmail.com>> wrote:
> Congrats Dave and Ron
> 
> I may join when the works begin
> 
> The way I see it, every activity and goal of AI KR (in general as well as the CG)
> is aimed at supporting what you call CogAI.  As I mentioned, historically AI has always been çognitive'
> and I cannot image non cognitive AI . I d be interested to see what develops
> 
> Do let us know if there's anything this CG can do to support your work
> 
> PDM
> 
> On Wed, Jan 15, 2020 at 1:25 AM Dave Raggett <dsr@w3.org <mailto:dsr@w3.org>> wrote:
> The new W3C Cognitive AI Community Group will focus on practical considerations for building AI systems, inspired by what we've learned about the human mind and behaviour. This includes collaborating on defining and agreeing the details for a set of use cases and their implementation as Web-based demonstrators. If you are more interested in discussing philosophy or ethical policies for the application of AI, there are other groups that are better suited for such discussions. It would be interesting to discuss how the CogAI CG and the AIKR CG can complement each other.
> 
> We’ve created a GitHub repository and a guide to contributors, see:
> 
>     https://www.w3.org/community/cogai/ <https://www.w3.org/community/cogai/>
> 
> I would be interested in your thoughts on the cognitive architecture diagram and explanation at the bottom of the GitHub README file:
> 
>        https://github.com/w3c/cogai <https://github.com/w3c/cogai>
> Dave Raggett <dsr@w3.org <mailto:dsr@w3.org>> http://www.w3.org/People/Raggett <http://www.w3.org/People/Raggett>
> W3C Data Activity Lead & W3C champion for the Web of things 
> 
> 
> 

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