Leo,
I think the key word in your original reply to Rich was “greatly”. I believe this implicitly acknowledges that there will be some degree of differences in interpretation of the semantics of the respective participants. The key issue is what degree of difference in interpretation is material in any given shared context. My personal belief is that the answer is itself context dependent. In other words, the scope of the differences in shared meaning that are pragmatically tolerable depends on the context in which these differences exist, and the willingness of the participants in that shared context to execute a “discovery” protocol to uncover and resolve any differences that might be material to that shared context.
Using the earlier example of different meanings of “policy” that Matthew brought up, the life insurance company context influences the meaning of “policy”. That context has scope, so policy might mean something a bit different in a fire insurance or automobile insurance context, and it might have somewhat different meanings across different life insurance companies. And in other contexts, such as, say, cyber security or human resources, policy might mean something completely different. Even in a narrow institutional context of a single life insurance company, the definition of a life insurance policy might change with different customer/operational jurisdictions, like different states or countries, and there might be different types of policies, and different associations between customers and a policy. For example, a customer might be a person or a company, with differences depending on which it is. A policy with a company might be a “master” policy, with different employees of that company having “sub-policies” that belong to the context of the “master” policy. Individual employees might view their policy with the insurance company as their own private policy, but such a policy is not same as one that an individual might negotiation directly with the insurance company, other things being equal, because it is typically subject to the individual continuing to be an employee of the company with the “master” policy. So if someone tells you that they have a life insurance policy with, say, MetLife, the meaning is somewhat different if that policy is through an employer rather than directly between MetLife and the individual. This difference might not be material in many contexts, but if you are discussing retirement planning with a financial advisor, I can assure you that the difference is quite material in that context.
Hans Polzer
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Obrst, Leo J.
Sent: Sunday, February 19, 2012 5:21 PM
To: [ontolog-forum]
Subject: Re: [ontolog-forum] What goes into a Lexicon?
My theory does differ, I’m afraid.
Thanks,
Leo
Dear Leo,
You wrote:
How can we possibly understand what you are saying here, if we don’t greatly share common semantics and a common ontology?
Through the psychological process called “projection”. That is, we have each trained all our lives to read and understand other people’s words in our own terms. Notice that when one person talks to another, the listener rarely understands the utterance in exactly the same way as the speaker does. That is because the listener has different experiences which she cobbles together to make her subjective interpretation of the utterance.
To be plainer about it, we construct reality in our individual minds to match what we sense, and what we anticipate. We correlate that construction from components we have previously learned and stored. Once we have learned a new construction, we can use that as a component in future projections.
Scene analysis, as you probably remember, was found in part to be an activity of detecting corners in three dimensions, and then joining those up to match the observed scene in a familiar way. Escher’s many drawings of impossible figures demonstrate that phenomenon, as he meant to do by design. Each corner makes sense to us as a corner, but we individually integrate the corners into a more complex complete scene. Escher’s drawings show that the integrated scene might make no sense in reality, but our perception of the corners fits very nicely for each individual corner.
I think the same projective method is used for language. If you have read the web page title “evidence based linguistics” (I have lost the link, sorry), he paints a compelling case for conversations being an iterative reduction of ambiguity through passing linguistic hints from speaker to listener. He thinks syntax is a useful method for representing concepts in language, but that syntax is only useful in certain restricted constructions, not as a general method of human language analysis. Many other components are needed to let speaker and listener project their individual understanding of the concepts being discussed.
But that does NOT make their concepts equivalent. Each participant comes away from the conversation with a unique understanding, most likely a different understanding from the other speaker. How many times have you convinced someone of your point of view in a single conversation? I suggest it is a relatively rare event compared to the number of conversations about similar subjects you may have had with the same person.
That’s my theory, anyway, until more evidence is made available to me. Your theory may differ.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
Rich,
How can we possibly understand what you are saying here, if we don’t greatly share common semantics and a common ontology?
Thanks,
Leo
Dear Matthew, David and John,
Matthew wrote:
It is really rather easy to create an ontology of names, what they represent, and who uses them in what context.
The problem isn’t in ease of creation, but in consistent use by multiple people. On this list, we have consistently found that people vehemently disagree on the “proper” way to model concepts with ontologies. Our history of finding agreement indicates that it is very difficult to construct ontologies that many people agree on. Only very, very simple things such as Dublin Core have found widespread usage.
By contrast, David’s example of the policy number, which one would expect to be well understood within the insurance business, by his stated experience, isn’t so well understood. The problem is with the observers of any ontology; we don’t all see the same meaning in a given rendering of concepts.
After a lot of thought, and from years of discussion with well qualified people on this list, I have come to the conclusion that ontology is simply one way of rendering reality. Ontologies have not in general satisfied large groups of observers because every observer has a unique, distinct, and very complex model of those simple concepts which we throw around with variant lexicons.
The problem is that we don’t all have the same experience. It is our individual differences in experience that leads us to think in different sequences with different observations of the same phenomenon. I don’t see how that problem can be solved by monolithic ontological terms.
Instead, problem reduction methods such as top down structured programming, and packaging methods such as object oriented programming and client server architectures have made dramatically effective inroads toward constructing large software systems that mostly work as intended.
I suggest that ontologists should take the same viewpoint because it has worked so well there. Instead of insisting on a singular meaning, a singular ontology, a singular way of doing things, we should be more open to pluralities of ontology as just another art form that is juggled differently by each observer. In Chairman Mao Zedong’s terms, let a million ontologies flourish.
Singular ontology just hasn’t worked out and it looks like it never will.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
Dear David,
Of course.
As always I revert to my favorite, oft-repeated example... a life insurance company that found 70 different names for the core business concept (term?) "policy number." In a business environment where product lines are bought & sold, systems are custom built by different teams & packages are bought, the reality is there will be many (illogical) names for the same thingy.
MW: it all depends what you build you ontology to do. If what you want it to do is allow you to bring together lots of different names for the same thing, then there is no particular difficulty in that. It’s just that ontologists don’t tend to call those different names terms, but the common meaning they share. You just have to get over that, and adopt the local usage and carry on. It is really rather easy to create an ontology of names, what they represent, and who uses them in what context.
Regards
Matthew West
Information Junction
Tel: +44 1489 880185
Mobile: +44 750 3385279
Skype: dr.matthew.west
matthew.west@xxxxxxxxxxxxxxxxxxxxxxxxx
http://www.informationjunction.co.uk/
http://www.matthew-west.org.uk/
This email originates from Information Junction Ltd. Registered in England and Wales No. 6632177.
Registered office: 2 Brookside, Meadow Way, Letchworth Garden City, Hertfordshire, SG6 3JE.
At one point I entertained delusions that ontologies would help with this issue (one conceptual label = many physical labels). Obviously I no longer hope in that direction.