[Top] [All Lists]

Re: [ontolog-forum] The DIKW Hierarchy issue(s)

To: edbark@xxxxxxxx, "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Ronald Stamper <stamper.measur@xxxxxxxxx>
Date: Sat, 20 Jun 2009 14:24:54 +0100
Message-id: <8D21CB62-A4F7-47DE-8471-8334A23D8ECF@xxxxxxxxx>
Dear Ontologgers    (01)

Long ago, teaching information systems at the London School of  
Economics, I insisted that my students take care using the vague terms  
“data”, “information” and “knowledge” and I joked that some nutter  
would soon be writing a paper about wisdom as the next level of  
analysis.  The following day, just such a person stopped me in  
Kingsway with a religious pamphlet containing a flowchart that  
included wisdom as one of its flows; a week later an accounting  
journal embarked on the same route to enlightenment.  What I taught  
instead was a kind of hierarchy based on semiotics and it has turned  
out to be quite popular among alumni who use it in practice.    (02)

I’ll explain briefly.  A sign is concrete and makes an ideal starting  
point as a primitive concept for an empirical science of information.   
Depending upon the things you want to do with signs and relevant  
operational procedures, you can find many different, precise meanings  
for the term “information”, each of them capturing some but never all  
the aspects of information belonging to our common sense treatment of  
that abstract notion.  I gathered together these ideas from the  
literature in my 1973 book “Information”; I had expected them to be  
common knowledge by now in the information systems field.  To my  
dismay they are not and, consequently I have to read works that have  
not progress beyond the stage in our scientific domain that  
corresponds with mediaeval notions of physics: the DIKW hierarchy is a  
good example.  I feel pangs of guilt at not keeping that book in print.    (03)

It seems obvious to me and, I’m sure to all ontologgers (so I  
apologise if I sound naïve), that, to function effectively, any  
information system must handle signs correctly on six major levels  
(levels?), where the grand abstractions – information, meaning,  
relevance etc – have their own appropriate precise definitions.    (04)

PHYSICAL – where one operates on particular things that can be used as  
sign-tokens to stand for other things.  Here you measure amounts of  
information in numbers of tokens and all the other grand abstractions  
relate to physical properties of signs.    (05)

EMPIRIC – where one deals with populations or streams of sign-tokens  
with concern for all the problems that interested Shannon, and where  
the relative frequency is a key empirical property and information and  
other abstraction can mostly be reduced to its terms.    (06)

SYNTACTIC – that concerns structures of sign-types and of  
manipulations performed on them; Bar Hillel and Carnap defined  
information in several ways based on a logical concept of probability  
but called these semantic measures.    (07)

             Most of the effort to build a science of information has  
been devoted to these three technical layers to the sad neglect of the  
three essentially human and social layers.  Even when our colleagues  
do venture into them, they attempt to do so in the spirit of their  
technical work (and I suspect that applies to some in our community),  
so they usually strive, in the spirit of AI, to keep human beings  
outside the frame.  But people are indispensible    (08)

             In my own work, I have concentrated on the following  
three human and social levels but always aiming at enough formal rigor  
to enable default solutions on the technical levels to be arrived at  
automatically.  This strategy works but it raises a chain of most  
interesting problems.  Those levels are:    (09)

SEMANTIC – where we investigate how sign-types can stand for other  
things; this is just one – perhaps the most important – of the  
meanings of meaning.  The notion of information can be given several  
different definitions suited to this level: the use of subjective  
probability seems appropriate here but ontology (metaphysical sense)  
can also contribute solutions.  To use a finite vocabulary to stand  
for things in a potentially infinite reality calls for constant  
renegotiation of meanings for new purposes.    (010)

PRAGMATIC – where investigations concern particular illocutionary acts  
using sign-tokens.  The use of sign-tokens to produce socially  
significant effects suggests definitions of information related to the  
potential scope of these effects.  Formal or mechanical systems cannot  
supply intentions.    (011)

SOCIAL – where we are concerned with the perlocutionary acts that  
result in changes in the attitudes or the norms (= knowledge) held by  
one or more interpreters of particular intention-bearing sign-tokens.   
This level is where signs metaphorically inform minds as the potter  
materially informs the clay on his wheel.  Information concerns the  
extent of this alteration of people’s attitudes and norms.    (012)

Each of the second group of three levels can be investigated in the  
best traditions of empirical science.  (Oh that I had another life to  
devote to them!)  Of special interest is the problem of how to link  
the powers of formal methods to the essentials of human conduct.   
Signs cannot have meaning or intention without involving people and  
signs have no value until they produce changes of attitudes in  
people.  To retreat into formal methods is just to ignore the  
semantic, pragmatic and social aspects of signs.    (013)

I have formed a love-hate relationship with ontolog.  I love the  
interesting ideas that the discussions throw up from time to time; I  
hate giving way to displacement activity when I should be writing up  
the work I have done. Perhaps, despite withdrawal symptoms, I’d better  
opt out of ontolog.    (014)

Am I wrong to suspect that many ontologgers would like to keep messy  
people and social problems at arms length or totally apart from the  
work they are doing?  I also feel alienated from so called ontologies  
that are largely generic-specific structures that express cognitive  
norms and have little to do fundamentally with the nature of what  
exists.    (015)

I have the comfort that our methods do work so extraordinarily well in  
practical information systems analysis, design and implementation that  
they are likely to be accepted eventually.  Theoreticians will then be  
tempted to examine the ideas behind the methods.    (016)

Saying such things invites criticism. As a disciple of Popper, with  
whom I overlapped briefly at LSE, that is intentional.  Might I  
provide you with a better target by reworking some standard case study  
like those in Appx C of John Sowa’s 2000 KR book?  Please give me a  
challenge in such concrete terms.    (017)

Ronald Stamper    (018)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (019)

<Prev in Thread] Current Thread [Next in Thread>