Abduction in AI and Computer Science?

69 views
Skip to first unread message

Michael DeBellis

unread,
Mar 9, 2020, 6:11:32 PM3/9/20
to ontolog-forum
I'm currently auditing a fascinating seminar at Berkeley on Semiotics and Information Theory. Mostly we are focusing on C.S. Peirce although we've also explored other theories such as Shannon's Information Theory.  As we were discussing abduction, the history of the idea, how it compares with induction and deduction, etc. someone asked me about the uses of Abduction in AI and computer science. He said that he had heard there was some interesting work in this area. I did a quick Google search and didn't find anything, I'm wondering if anyone knows of any significant work in this area?

Michael

Jon Awbrey

unread,
Mar 9, 2020, 7:02:19 PM3/9/20
to ontolo...@googlegroups.com, Michael DeBellis
Michael,

At dinner now, but here's a survey of blog and wiki links for the time being:

Survey of Abduction, Deduction, Induction, Analogy, Inquiry
https://inquiryintoinquiry.com/2017/03/08/survey-of-abduction-deduction-induction-analogy-inquiry-%e2%80%a2-1/

Regards,

Jon

inquiry into inquiry: https://inquiryintoinquiry.com/
academia: https://independent.academia.edu/JonAwbrey
oeiswiki: https://www.oeis.org/wiki/User:Jon_Awbrey
facebook page: https://www.facebook.com/JonnyCache

Michael DeBellis

unread,
Mar 9, 2020, 7:04:33 PM3/9/20
to ontolog-forum
Interesting. Thanks!

Jon Awbrey

unread,
Mar 9, 2020, 7:20:31 PM3/9/20
to Ontolog Forum
Michael,

Here's a bit on early discussions by McCulloch and early work by Pople, et al.

Abduction, Deduction, Induction, Analogy, Inquiry : 23
https://inquiryintoinquiry.com/2016/05/01/abduction-deduction-induction-analogy-inquiry-23/

Regards,

Jon

Adrian Walker

unread,
Mar 9, 2020, 7:24:45 PM3/9/20
to ontolog-forum
Hi Michael & All,

Michael asked About abduction.

The Executable English system [1] explains results with an English form of each derivation step. 

When an expected result is missing, it provides an explanation showing one or more assumptions that can make
the result derivable.

Thus given  the rule

some-person is a man
-----------------------------
that-person is mortal

and the fact

this-person is a man
================
Fred

the system concludes that Fred is mortal

If asked whether Socrates is mortal the system explains

missing: Socrates is a man
---------------------------------------
not shown:  Socrates is mortal


So, a form "of abduction is what you asked for is not derivable, but it would be if these facts were added."

Interestingly, this topic overlaps with planning.

Cheers,  -- Adrian

Adrian Walker
Executable English LLC  
San Jose, CA, USA
(USA) 860 830 2085 (California time)
[1] https://www.executable-english.com



On Mon, Mar 9, 2020 at 3:11 PM Michael DeBellis <mdebe...@gmail.com> wrote:
I'm currently auditing a fascinating seminar at Berkeley on Semiotics and Information Theory. Mostly we are focusing on C.S. Peirce although we've also explored other theories such as Shannon's Information Theory.  As we were discussing abduction, the history of the idea, how it compares with induction and deduction, etc. someone asked me about the uses of Abduction in AI and computer science. He said that he had heard there was some interesting work in this area. I did a quick Google search and didn't find anything, I'm wondering if anyone knows of any significant work in this area?

Michael

--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info/
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/c01a569c-7cc1-4890-984b-bc6e5f23f880%40googlegroups.com.

Jon Awbrey

unread,
Mar 9, 2020, 7:52:37 PM3/9/20
to Ontolog Forum
Michael, All ...

All through 1995 I worked on a graduate project in systems engineering at Oakland University developing my ideas about
Inquiry Driven Systems. I wrote a project report on Peirce's treatments of analogy and inquiry that included a
discussion of the logical inferences involved in the abductive and deductive steps. Here's a part of that:

Functional Logic : Inquiry and Analogy
https://oeis.org/wiki/Functional_Logic_%E2%80%A2_Inquiry_and_Analogy

Regards,

Jon

Jerry Hobbs

unread,
Mar 10, 2020, 2:02:15 AM3/10/20
to ontolo...@googlegroups.com
Michael, 

Hobbs et al., “Interpretation as Abduction””, Artificial Intelligence,  1993,

— Jerry


On Mar 9, 2020, at 3:11 PM, Michael DeBellis <mdebe...@gmail.com> wrote:

I'm currently auditing a fascinating seminar at Berkeley on Semiotics and Information Theory. Mostly we are focusing on C.S. Peirce although we've also explored other theories such as Shannon's Information Theory.  As we were discussing abduction, the history of the idea, how it compares with induction and deduction, etc. someone asked me about the uses of Abduction in AI and computer science. He said that he had heard there was some interesting work in this area. I did a quick Google search and didn't find anything, I'm wondering if anyone knows of any significant work in this area?

Michael

John F. Sowa

unread,
Mar 10, 2020, 3:12:03 AM3/10/20
to ontolo...@googlegroups.com
Jerry,

Thanks for the reference.  I was planning to dig out some articles from the GOFAI days (Good Old Fashioned AI) about abduction.  But you cited one of the classics, which also cites Peirce as the inspiration:

JH> Hobbs et al., "Iterpretation as Abduction", Artificial Intelligence, 1993,  https://www.isi.edu/~hobbs/interp-abduct-ai.pdf

And for anyone who may be interested in abduction, Google  the two words 'perception'  and 'abduction' for 4 million hits about how perception depends on abduction for filling in the gaps that are missing from the visible (or auditory, tactile, olfactory...) fragments.

To see why abduction is necessary for perception, note the attached cartoon.  The millions of photos used to train DNNs to recognize cars rarely include a car in a garage with its hood up.  But anybody who has has ever seen a car engine and knows anything about getting it fixed can interpret the cartoon  correctly, relate it to the dialog, and (most important) laugh..

John.

shoe23mar15.jpg

Jon Awbrey

unread,
Mar 10, 2020, 8:00:50 AM3/10/20
to ontolo...@googlegroups.com, Adrian Walker
Adrian,

Re: > Interestingly, this topic overlaps with planning.

Exactly.
Resolving a surprise through an explanation
and solving a problem through a plan of action
are dual species of inquiry in general.

This is one of the themes at the top of my work on Inquiry Driven Systems.
See, for example, the "statement of research interests" I was asked to
submit for my application to grad school back in the early 90s.

Prospects for Inquiry Driven Systems

* https://oeis.org/wiki/User:Jon_Awbrey/Prospects_for_Inquiry_Driven_Systems#Topos_:_Rudiments_and_Immediate_Resources

Regards,

Jon

On 3/9/2020 7:23 PM, Adrian Walker wrote:
> Hi Michael & All,
>
> Michael asked About abduction.
>
> The Executable English system [1] explains results with an English form of
> each derivation step.
>
> When an expected result is missing, it provides an explanation showing one
> or more assumptions that can make
> the result derivable.
>
> Thus given the rule
>
> *some-person is a man*
> *-----------------------------*
> *that-person is mortal*
>
> and the fact
>
> *this-person is a man*
> *================*
> *Fred*
>
> the system concludes that Fred is mortal
>
> If asked whether Socrates is mortal the system explains
>
> *missing: Socrates is a man
> ---------------------------------------
> not shown: Socrates is mortal *

Ronald Stamper

unread,
Mar 10, 2020, 11:05:52 AM3/10/20
to 'Rob Rovetto' via ontolog-forum

to Michael de DeBellis

 

Dear Michael

You mentioned ”a fascinating seminar at Berkeley on Semiotics and Information Theory.”  You were “wondering if anyone knows of any significant work in this area? I rejoiced to hear about it because I began to argue the releance of semiotics to IS analysis and design in the 1960s when the UK steel industry asked me to look for a solution to high failure rate of ICT applications.  I quickly noted

    1. all ISAD training came from IT mareters, and

    2. the  “information” was a concept lamentably poorly understood for several reasons, for example:

            a.  computers were sold as “information processors” when they can only be token processors;

            b.  the still-popular DIKW hierarchy of computers distilling these magical fluids: mediaeval rubbish!

            c.  misleadingly calling Shannon’s theory of signal transmission “Information theory”, instead of “A narrowly-focused theory about one important aspect of information”.

thereby deflecting attention from the wider collection fo problems.  I created new courses on ISAD giving equal weight to IT and social organisation and began to look for a sound science of Organisations as the Real Information Systems, beginning by demanding: Show me some information and its properties.  

 

That led me to Peirce, of course, and the 2 millennia of relevant work preceding him.  These rich resoources produced a set of six groups of properties of sign-tokens:  3 technical – physical, empirical (incl. Shannon’s) and syntactic; PLUS 3 intrinsically human – semantic, pragmatic and social.  Those 6 shaped my courses and served as a Semiotic Framework for my book, Information (1973 Batsford, London and Wiley, New York). 

            After 6 cycles, the government’s NCC with BCS turned them into the national programme (with no acknowledgement). I moved to the London School of Economic’s team creating IS as a new discipline and I began a research programme, essentially on, but not called, Organisational Semiotics, funded by 2 research councils and IBM.   

            The result is an ISAD methodology, MEASUR, for specifying a BIS or any institution as a system of social norms incluing those governing semantic and pragmatics but in a LEGally Orientated Language, a strict formalism based on statute law.  So, when the business IS is speciied semiotically, you have also specified the ICT application to support, in a language that can be understood by people and interpreted by computer. 

            Early adopters of MEASUR have clocked up 50+ years of cost-reducing, highly adaptable ICT operating better-designed organisations.

            I learned a sharp lesson in scientific progress (see Th. Kuhn’s famous book) when the professor of computer science chairing my presetation of these ideas in 1979, stopped the meeting to tell me I “was leading us all into a philosophical bog”.  In that bog, I had already pulled out Blockchain (25 years before Bitcoin) and all the tools necessary for a Semantic Web, 22 years before Sir T B-L et al’s famous Sci Am paper.

            So I rejoice at your news.  UC Berkley have entered my favourite philosophical bog,  In the IFIP community, 10 years of discussion in its TC8;1 task group FRISCO, allowed me to make the case for Organisational Semiotics, which now appears among IFIP’s events.

            The potential for this scientific field is exciting.  Sadly, it seems to cause more irritation than thought in the mechanistically-inclined thinkers.  We need to unify the technical and social aspects and semiotics offers a basis.

            My book Information (subtitle: in Business and Admin) is out of print.  If there is interest, I can make it available.

            Although MEASUR is now risk-free, I’m looking for more daring organisations willing to put it to the test. Having discovered that new ideas only get recognition as a threat, therefore to be suppressed, if they call for a new paradigm.  Th Kuhn had made that clear, but I never expected the fierce reaction of the devotees to the Information Flow paradigm (UML etc.) which FW Taylor introducd in the 1890s.  Our new, Information Paradigm, acknowledges the social norms, which, like a magnetic force, determine what info should flow where.

Regards to you an fellow Ontologger

Ronald

           

“” “

 

 

 

Mostly we are focusing on C.S. Peirce although we've also explored other theories such as Shannon's Information Theory.  As we were discussing abduction, the history of the idea, how it compares with induction and deduction, etc. someone asked me about the uses of Abduction in AI and computer science. He said that he had heard there was some interesting work in this area. I did a quick Google search and didn't find anything, I'm wondering if anyone knows of any significant work in this area?

 

Michael

 

 

Simons Institute for the Theory of Computing
121 Calvin Lab #2190
UC Berkeley
Berkeley, CA 94720-2190

 

On 9 Mar 2020, at 22:11, Michael DeBellis <mdebe...@gmail.com> wrote:

I'm currently auditing a fascinating seminar at Berkeley on Semiotics and Information Theory. Mostly we are focusing on C.S. Peirce although we've also explored other theories such as Shannon's Information Theory.  As we were discussing abduction, the history of the idea, how it compares with induction and deduction, etc. someone asked me about the uses of Abduction in AI and computer science. He said that he had heard there was some interesting work in this area. I did a quick Google search and didn't find anything, I'm wondering if anyone knows of any significant work in this area?

Michael

Michael DeBellis

unread,
Mar 11, 2020, 2:04:30 PM3/11/20
to ontolog-forum
I've been having some medical issues lately, nothing serious, just things that took a lot of time and made me too tired to do any work so just getting back to this now. What an amazing amount of information, thanks very much to everyone who has posted. I agree with Ron's comment about Shannon absolutely, it's one of the first things we went over in the seminar that essentially Shannon is about communication not the actual semantics of what is communicated (which is what a theory of Information should actually be about). The professor for the class is Terrance Deacon, I recommend his books The Symbolic Species: The Co-evolution of Language and the Brain  and Incomplete Nature: How Mind Emerged from Matter very highly. Also, one of the books we read for the seminar was by James Gleick called The Information: A History, a Theory, a Flood  This is more a historical overview, not very deep (and he makes a few errors in the way he describes Gödel) but still an interesting and easy book for a historical perspective from Aristotle to Meme theory. Thanks again to everyone who posted, I have lots to read.

Michael

Jon Awbrey

unread,
Mar 15, 2020, 3:40:23 PM3/15/20
to ontolo...@googlegroups.com, Michael DeBellis
Michael, All,

There are numerous threads here (on the List) and there (on my blog)
tracing the integral relationships among information, inquiry, and
the three types of inference. Here's a sample:

* Ontolog : Abductive Inference, Concept Formation, Hypothesis Formation
https://groups.google.com/d/topic/ontolog-forum/8mhlH2YBoA0/overview

* Inquiry: Survey of Abduction, Deduction, Induction, Analogy, Inquiry
https://inquiryintoinquiry.com/2017/03/08/survey-of-abduction-deduction-induction-analogy-inquiry-%e2%80%a2-1/

* Ontolog : Pragmatic Semiotic Information
https://groups.google.com/d/topic/ontolog-forum/WsdtHpkRvbI/overview

* Inquiry : Survey of Pragmatic Semiotic Information
https://inquiryintoinquiry.com/2019/01/08/survey-of-pragmatic-semiotic-information-%e2%80%a2-4/

* Ontolog : { Information = Comprehension × Extension } Revisited
https://groups.google.com/d/topic/ontolog-forum/TEyKMCnwy2c/overview

* OEIS Wiki : Information = Comprehension x Extension
https://oeis.org/wiki/Information_%3D_Comprehension_%C3%97_Extension

Regards,

Jon

On 3/9/2020 6:11 PM, Michael DeBellis wrote:
Reply all
Reply to author
Forward
0 new messages