[Humanist] 28.825 an argument & belief system

Humanist Discussion Group willard.mccarty at mccarty.org.uk
Tue Mar 17 07:40:41 CET 2015


                 Humanist Discussion Group, Vol. 28, No. 825.
            Department of Digital Humanities, King's College London
                       www.digitalhumanities.org/humanist
                Submit to: humanist at lists.digitalhumanities.org



        Date: Mon, 16 Mar 2015 12:17:43 +0100
        From: Tim Smithers <tim.smithers at cantab.net>
        Subject: Re:  28.819 an argument & belief system
        In-Reply-To: <20150315072307.9A021C29 at digitalhumanities.org>


Dear Dominic,

I too am curious about your argument and belief system.

From the mid to late 1980s, in the Department of Artificial
Intelligence (as it was then called), at Edinburgh University,
we worked on a series of Intelligent Knowledge Based Design
Support (IKBDS) systems .  (This was the day's of the Alvey
Programme, when AI was unusable, so the double speak term was
Intelligent Knowledge Based Systems.)

We built experimental IKBDSs to investigate how we might
support designers in intelligent ways.  Designing can be
understood as a kind of exploration, to discover possible
designs, and the arguments for why they are suitable designs;
an argument that depends upon starting assumptions, selected
and used data, and the inferences that join these together.

Designing, and the argument development that is thus a part of
designing, is typically a non-monotonic process: later changes
can reduce the number of valid inferences; inferred knowledge
doesn't just increase.  To support this basic characteristic of
designing we used (what were called) Truth Maintenance
Systems--sometimes called Belief Maintenance Systems--and one
in particular, called a Assumption-Based Truth Maintenance
System (ABTMS).  Essentially this kept a record of what each
inferred item depended upon, so that if any of it's
assumptions were withdrawn (by being changed or removed), it
could go though the current set of inferred items removing all
those now unsupported by the current assumption set.  We added
a "views" system to this, so that we could maintain different
sets of current assumptions, ie, design alternatives, or, in
your terms, different arguments.

In our IKBDSs, inferences were made by different specialist
"reasoning engines."  These were different for the different
kinds of designing we worked on: electro-mechanical design,
small molecule drug design, nuclear power plant maintenance
design, to illustrate that our ATMS was used in a range of
quite different design domains.

Today, our IKBDSs, with their explicit knowledge
representation schemes, multiple inference engines, ABTMS, and
Blackboard System architectures, are considered Old AI. No
Deep Learning here.  Still, though the ABTMSs were often
underpowered by the Sun Workstations we used back then, I
think this technique might still be made to do useful work for
the kinds of thing you describe for your argument and belief
system.

If you'd like to know more I could send some (old) papers
about what we did.

Best regards,

Tim

Tim Smithers
Independent Research Expert
Donostia / San Sebastián
The Basque Country

LinkedIn http://www.linkedin.com/in/timsmithers

 

> On 15 Mar 2015, at 08:23, Humanist Discussion Group <willard.mccarty at mccarty.org.uk> wrote:
> 
>                 Humanist Discussion Group, Vol. 28, No. 819.
>            Department of Digital Humanities, King's College London
>                       www.digitalhumanities.org/humanist
>                Submit to: humanist at lists.digitalhumanities.org
> 
> 
> 
>        Date: Sat, 14 Mar 2015 11:17:14 +0000
>        From: Martin Mueller <martinmueller at northwestern.edu>
>        Subject: Re:  28.812 an argument & belief system
>        In-Reply-To: <20150314071750.8C4B6BD5 at digitalhumanities.org>
> 
> 
> This argument & belief system stuff is very interesting. I think it's
> related to a presentation I heard at Oxford. On the other hand, I'm
> reading this post with the eyes of my colleagues in English and
> History, and I doubt whether this way of presenting it makes any sense to
> them whatever. Is it possible to describe in plain language what this
> thing does and what it is that you can do with it that you couldn't do
> before?
> 
> MM
> Martin Mueller
> Professor emeritus of English and Classics
> Northwestern University
> 
> On 3/14/15, 08:17, "Humanist Discussion Group"
> <willard.mccarty at mccarty.org.uk> wrote:
> 
>>                Humanist Discussion Group, Vol. 28, No. 812.
>>           Department of Digital Humanities, King's College London
>>                      www.digitalhumanities.org/humanist
>>               Submit to: humanist at lists.digitalhumanities.org
>> 
>> 
>> 
>>       Date: Fri, 13 Mar 2015 07:28:38 +0000
>>       From: Dominic Oldman <doint at oldman.me.uk>
>>       Subject: Re:  28.810 pubs on note-taking and on concerns for
>> digital	humanities
>>       In-Reply-To: <20150313062328.C0575BD6 at digitalhumanities.org>
>> 
>> 
>> This sounds very interesting and my poor knowledge of this area (Rhetoric
>> Studies) would benefit from reading this book [1].
>> 
>> However, it may be worth noting that I am implementing practically an
>> argument and belief system (within a BM project/team) using a
>> specialisation of the CIDOCCRMwhich makes it possible to make assertions
>> (beliefs) and convey arguments (observation, inference making and belief
>> adoption). We implement it in the same 'real world' knowledge
>> representation system used to represent 'canonical' data (although this
>> is not a pre-requisite), and means that we can openly and transparently
>> retain the full (connected) provenance of an argument and the assertions
>> that evidence it (back to an original premise) that may take place over
>> large periods of time  - and hopefully use computer reasoning to analyse
>> the discourse. We are developing a software tool at the moment.
>> 
>> This type of thing, to me, seems to address the Big Tent issue and is an
>> essential contribution to DH/HC epistemology that helps define what the
>> Tent is and what it is for.
>> 
>> Anyway, as long as it isn't too expensive I will purchase a copy. :-)
>> 
>> Cheers,
>> 
>> Dominic
>> 
>> PS. Happy to explain the KR system to anyone interested - Its called
>> CRMInf (Conceptual Reference Model Inference). Sorry if this is actually
>> off subject and I have misunderstood.






More information about the Humanist mailing list