[Humanist] 28.852 an argument & belief system

Humanist Discussion Group willard.mccarty at mccarty.org.uk
Thu Mar 26 07:29:50 CET 2015


                 Humanist Discussion Group, Vol. 28, No. 852.
            Department of Digital Humanities, King's College London
                       www.digitalhumanities.org/humanist
                Submit to: humanist at lists.digitalhumanities.org

  [1]   From:    Dominic Oldman <doint at oldman.me.uk>                       (88)
        Subject: Re:  28.844 an argument & belief system

  [2]   From:    Tim Smithers <tim.smithers at cantab.net>                    (29)
        Subject: Re:  28.844 an argument & belief system


--[1]------------------------------------------------------------------------
        Date: Wed, 25 Mar 2015 11:04:30 +0000
        From: Dominic Oldman <doint at oldman.me.uk>
        Subject: Re:  28.844 an argument & belief system
        In-Reply-To: <20150324061245.003D3CA0 at digitalhumanities.org>


Firstly with regard to Desmond's comment that his understanding is that
CIDOC CRM only deals with museum objects. This is a common misunderstanding
and a very common belief adoption. If we were using a digital argument and
belief system we could trace a vast and connected set of similar belief
adoptions back to an original observation and determine the weight of the
proposition within. This is the point. Many arguments go on for long periods
of time and are forgotten resulting in duplicated work and mistakes. I am
aware of threads of academic work that have been built up over long periods
of time that are then exposed as incorrect because of gaps in the academic
discourse. I make similar mistakes myself perhaps simply because I haven't
read a particular journal article.   

The email is therefore a useful example and I could say this (and encode
it): 

Desmond has an argument about the nature of the CIDOC CRM and its purpose
with a belief proposition derived from, “CIDOC CRM and any inference
system based on it would be a way of thinking/reasoning about metadata of
whole museum objects”. This belief could have been arrived at by reading
and interpreting the CIDOC CRM reference (an observation) or adopting the
belief of someone else. It could also be an assertion that is unqualified.

In the former (observation) he may make an inference from his proposition
and conclude that the CRM cannot represent the meaning of literary text or
indeed text. 

i.e. "CIDOC CRM deals with objects and I infer through simple deductive
logic (in conclusion) a belief that is doesn't deal with more conceptual
things. 

In the latter (adoption) he may also conclude the same belief but the
inference logic may be that he trusts the opinion of the person(s) from
which he adopted the belief, as an expert in this particular area.

Note: We can of course date Desmond's belief and record attribution so that
the belief is recorded for all time since forming a belief is a form of
activity to which we can sensibly attach temporal qualities and
participating actors.

My own belief is that Desmond's initial proposition is in fact false (the
same proposition but with a different belief value). I can add my own
propositions, for example, I can say that I believe the scope note of the
entity E73 Information Object is True, and do this with many other
non-physical entities within the CRM that, within a certain scope, address
text, including poems, sonnets and so on.I can therefore conclude a new
belief that the CRM deals with more than physical museum objects. In actual
fact it deals with vastly more that this.

However, this is an aside because the original note wasn’t about CIDOC CRM
which is a bit of a diversion. It was about a different ontology CRMInf.
(However, the fact that CRMInf is a specialisation of CIDOC CRM dealing with
propositional and information entities I could have used this as another
proposition to support my the previous argument and negation of Desmond's
belief). The CRMInf ontology deals with the specific problem of how to
create a digital provenance of facts (facts used in the wider sense of the
word) derived from the work of academics – and doesn’t really mention
physical objects. 

Surely I can make an argument about anything, including chains of symbols,
that I can clearly identify! The question is how best is this knowledge
represented from a computational point of view. 

It is true that Martin has selected a use case that may not have been in the
forefront of the mind of the creators (the main authors being an
archaeologist and a nuclear physicist by background). However, this matters
not as it was not designed for a particular narrow thing. If you agree in
principle that argumentation theory can be used to talk about text, then any
feedback about the relative pros and cons and other approaches which have
different agendas to the CIDOC CRM, are all very welcome. We can contribute
to a further development of this.  It is true that the CRM has a particular
agenda regarding the harmonization of information (through generalisations)
that may effect the extent to which it deals with this aspect, but this is a
feature not a fault.

In the museum world curators, who are also researchers and academics, albeit
in the realm of the real world, have been quite successful at
interpretation. The CIDOC CRM is designed primarily to deal with reality and
therefore, within CRMInf, a proposition (a line of literary text) should not
be taken as a proposition regarding reality. However, if this is clear, I
don’t see why CRMInf cannot be applied (in so far as it goes) to a sonnet.
But let's see.

A sonnet represents an interesting use case (partly because it is already
highly structured - right -  :-)  ). 

My question to Martin is, given the example above: What argument do you want
to make about sonnet 116? I note that there are many beliefs and arguments
documented in narrative, but as this is your domain and expertise it should
be your belief and argument which can be your own observation or adopting a
belief from someone else. So...

Do you want to make a new observation about the whole sonnet or individual
lines, etc? Do you want to adopt the beliefs of others? CRMInf would
consider a citation as an adoption of someone else’s belief. Do you want,
from an original premise (the sonnet) and other observations and belief
adoptions, to conclude a particular belief (an interpretation). The only
real limitation is that you make “honest propositions” (I quote from the
scope note of for I2 Belief) which as a minimum would require either reading
sonnet 116. 

I am happy to give it a go and if it doesnt provide a satisfactory result
(one that moves things along a bit) then it would be good to discuss further
requirements and develop this further.

D 

--[2]------------------------------------------------------------------------
        Date: Wed, 25 Mar 2015 08:41:45 +0100
        From: Tim Smithers <tim.smithers at cantab.net>
        Subject: Re:  28.844 an argument & belief system
        In-Reply-To: <20150324061245.003D3CA0 at digitalhumanities.org>


> On 24 Mar 2015, at 07:12, Humanist Discussion Group <willard.mccarty at mccarty.org.uk> wrote:
>
>       Date: Mon, 23 Mar 2015 06:24:45 +0000
>       From: Willard McCarty <willard.mccarty at mccarty.org.uk>
>       Subject: where arguments are defeated & beliefs prove wrong
>
> Following up on Martin's query to Dominic about his argument and belief
> system, asking for a demonstration of its success with Sonnet 116, I'd
> like to ask the opposite: an significant instance of its failing, in detail.
> Do we, unassisted or not, learn more by having our beliefs confirmed
> and arguments triumphant than by their defeat?
>
> Yours,
>  WM

Dear Willard,

You ask exactly the right question, and importantly so!

I would say this applies to lots of things, but, in AI (artificial
intelligence), from plenty of personal experience, looking for and
understanding failures of AI systems, and exploring the conditions that
provoke them, does too important things.

1 It tells us much much more about how well our system works, why, and under
what conditions, and

2 It keeps us more humble about what we claim it can do.

I miss this in most AI stuff I see talked about and often raved about these
days.

I guess I'm old fashioned, but to me, working this way is just a matter of
professional honesty.

Best regards,

Tim





More information about the Humanist mailing list