Not many people really get semiotics, and that is unfortunate. It's a beautifully simple and elegant philosophy. But I think one of the big problems holding back comprehension is all the bickering about terminology. Charles Peirce was notoriously bad about this, inventing cumbersome new words constantly and using them inconsistently. The problem continues to persist to this day. People get lost in all the "rheme" and "dicent" gibberish, and miss out on the fundamental strength and beauty of semiotics.
Computational semiotics lets us put all this aside, in the same way that building a machine ends all argument about what the machine is and what it's good for. A piece of software is always a formal logical construct of some kind, and the source code describes it precisely. So a big interest of mine is to apply semiotics to software engineering.
In this light, the problem becomes: What do you need to know in order to build a software implementation of semiotic knowledge representation? This is something that some very bright bulbs are working on, and I have my own ideas that I'd like to get into later.
But for now, if you are interested and want to get introduced to this world, I have to recommend a fantastic book by John Sowa called "Knowledge Representation: Logical, Philosophical, and Computational Foundations." Sowa is an excellent encyclopedist of the development of formal logic and semiotics, and his book presents it all in a very non-scary, approachable way. It's intellectually stimulating without being tedious or jargon-laden. Very cool book.
Details here:
www.jfsowa.com/krbook/
Gecko
Computational semiotics lets us put all this aside, in the same way that building a machine ends all argument about what the machine is and what it's good for. A piece of software is always a formal logical construct of some kind, and the source code describes it precisely. So a big interest of mine is to apply semiotics to software engineering.
In this light, the problem becomes: What do you need to know in order to build a software implementation of semiotic knowledge representation? This is something that some very bright bulbs are working on, and I have my own ideas that I'd like to get into later.
But for now, if you are interested and want to get introduced to this world, I have to recommend a fantastic book by John Sowa called "Knowledge Representation: Logical, Philosophical, and Computational Foundations." Sowa is an excellent encyclopedist of the development of formal logic and semiotics, and his book presents it all in a very non-scary, approachable way. It's intellectually stimulating without being tedious or jargon-laden. Very cool book.
Details here:
www.jfsowa.com/krbook/
Gecko
posted by:
|
San Diego
|
-
Re: Computation as a way to understand semiotics
Sun, December 7, 2003 - 8:34 PMThe trick is to throw down the gauntlet and promote semiotics from a nut-job obscure "philosophy" or interesting curiosity (as some would have it) and move the subject towards a concrete science. This means producing hypotheses that can be tested and validated (or not). This is the primordial step before computational models can be considered and then implemented in a computer science sort of way.
The lack of a hard science view on semiotics further confounds the confusing terminology. Why bother learning a bunch of "jibba jabba" (to quote the astute semiotician Mr. T) when no one can outline any positive conclusions? At that level one can get more traction reading Dr. Seuss.
Of course the double edged sword is that science requires precision, which unfortunately infers obtuse vernacular all too often. However, this can be mitigated by not falling into the academician fortress-building trap.
In some ways it would be desireable to leave all the archaic terminology behind and start with a simplified blank slate. In terms of complexity, Peirce was his own worst enemy. I realize we often see a serialized inner view of his evolutionary though process, but there never seems to have been an effort by him to consolidate his world view and refactor it in a manner that could be more easily consumed by all.
Let's compile a set of assertions that define "semiotics: the science" and refine those into working hypotheses. The computational models can then flow from there and be proven out (or not) and further refined. -
-
Re: Computation as a way to understand semiotics
Sun, December 7, 2003 - 10:31 PMJust to get the ball rolling, here's a starter kit of assertions:
- a sign is any thought, concept, idea or sensed experience
- signs are monads (embody no internal relational function), dyads (embody a relation of two other signs), or triads (embody a sign related to another sign via a third sign)
- relations more complex than triadic can be reduced to networks of triads, dyads or monads without destroying meaning
- Sign production occurs when any signs are related or when any monads are sensed
- When two or three signs can be related to produce meaning, a new sign embodying that relation is made. This is sign emergence
- Sign emergence is fractal: self-similar at any scale -
-
Re: Computation as a way to understand semiotics
Sun, February 22, 2004 - 1:52 PMI agree that a computational approach to semiotics may divest semiotics of its jargon-laden, speculative philosophy. Also, I agree that "Knowledge Representation: Logical, Philosophical, and Computational Foundations" is a hell of a good book.
A few quibbles:
Why refer to semiotic systems as being “fractal”, when “compositional” is probably more accurate descriptor. Fractal objects are recursively constructed and they exhibit levels of self-similarity. I can see the recursive construction, but what similarity is there between atomic signs and molecular signs, other than their being signs? A mathematically proscribed sameness would be helpful.
What exactly is a sign? Is it relationally established, or is it a metaphysical, determinant object? The use of ‘sign’ seems so general at times that vacuity threatens.
A great deal of work on reference, meaning, interpretation and translation has taken place in the analytic tradition apart from the semiotics of Saussure, Eco, Barthes, etc. Frege, Russell, Tarski, Carnap, Quine, Davidson, and others have all made interesting contributions that remain (largely) unsynthesized with semiotics. Additionally, the works of Chomsky, Shannon, Turing, Davis and other information theorists and theory of computation luminaries, have not been (to my admittedly limited knowledge) effectively incorporated into semiotics. To sum it up, what does semiotics have to offer these other intellectual traditions that work on the same issues?
These questions may sound unsympathetic, but they are not meant to be. I have a genuine interest in semiotics, but cannot yet see its contributions. -
-
Re: Computation as a way to understand semiotics
Wed, March 3, 2004 - 3:24 PMSigns have to be fractal because they are not only compositional (that is, a sign is the relation of other signs) but that signs are structurally triadic. In other words, it's a fractal because the "atomic" unit (a sign) has the same structure as the "molecular" unit (also a sign) and that signs are composed other signs.
What exactly is a sign? There are a lot of different ways to understand this term. Simply put, it's anything that has meaning. A concept, an idea, a claim, an observation, data, and so on. It's the grist of the computational mill. The meaning of a sign, it's metaphysical assertion, is established relationally.
To your last question regarding what semiotics has to offer, that depends on who you ask or what problem you're looking to solve. My interest is in the realm of logic, and semiotics has so much to offer here that it would take days to describe. Most of Charles Peirce's ~80,000 pages of writing have never been published, and much of this writing concerns logic.
For me, computational semiotics is about two things (which are really two sides to the same coin):
1. Taking what is now (effectively) a branch of philosophy and turning it into a hard science, using computers as the platform for empirical study.
2. Expanding the realm of useful, computationally tractable logical operations, extending computing machinery past data processing into real semiosis (sign processing).
Example: a decent algorithmic implementation of the logical operation known as abduction (hypothesis generation). No current software can do this very well, yet it's a cornerstone of semiosis. I'd love to have a computer that could spin hypotheses.
G -
-
Re: Computation as a way to understand semiotics
Wed, March 3, 2004 - 9:52 PMG Said: Signs have to be fractal because they are not only compositional (that is, a sign is the relation of other signs) but that signs are structurally triadic. In other words, it's a fractal because the "atomic" unit (a sign) has the same structure as the "molecular" unit (also a sign) and that signs are composed other signs.
Are you referring to the Sign vehicle (the form of the sign), Sense (the sense made of the sign) Referent (what the sign 'stands for') triad? Can you give me a specific example?
I should admit that I distrust systems that are too tidy. A triadic system unfolding in a crystalline iteration is such a too tidy system.
G Said: [A sign is] anything that has meaning.
Hmm, how does this differ from information theory and data structures, except in generality? At least information theory provides us with useful mathematics and a model of encoding, transmission and decoding that has led to significant applications. Data structures also have well-established mathematical properties and are (obviously) in widespread use.
I know that Pierce divided signs into three kinds: Icons, Indexes and Symbols. Icons are picture-like and refer due to similarity (e.g. maps), Indexes refer in virtue of being effected (e.g. smoke to fire), and Symbols refer due to a law or association (‘cat’ to actual cats). What I don’t understand is the significance of this classification. I know that we need to start somewhere and that somewhere may include initial vagaries, but when have they yet resolved into an application? Though, I suppose we could bring in isomorphism talk and even category theory to make Icons more precise.
G Said: The meaning of a sign, it's metaphysical assertion, is established relationally.
I am sympathetic with this statement because it aids in dispelling essentialism, but remain skeptical of sign talk.
G Said: Example [of an application]: a decent algorithmic implementation of the logical operation known as abduction (hypothesis generation). No current software can do this very well, yet it's a cornerstone of semiosis. I'd love to have a computer that could spin hypotheses.
As would I, but isn’t Pierce’s work on abduction quite separable from his work on semiotics? The idea that there is a theory that explains results is a high-level semantic procedure that may be treated without any significant concessions to semiotics (as in Induction by Holland et al and elsewhere).
I fear by the time you make semiotics computational will bear such little resemblance to semiotics that it will be, in effect, information theory, artificial intelligence, analytic philosophy of language, and the other established disciplines. Again, whence an original, practical contribution? -
-
Re: Computation as a way to understand semiotics
Thu, March 4, 2004 - 1:00 PMGood questions. Let me cherry pick:
"I should admit that I distrust systems that are too tidy. A triadic system unfolding in a crystalline iteration is such a too tidy system."
Fractals are anything but tidy. :)
"how does this differ from information theory and data structures, except in generality?"
Semiotics includes logical processes that are not (currently) amenable to computation. Such as abduction.
"Pierce divided signs into three kinds: Icons, Indexes and Symbols..."
Your confusion regarding Peircean semiotics is understandable, he was a poor writer and most of his fans, like William James, barely understood him. Here's the deal. Peirce identified the root of all conceivable category systems, which he called the Ontological Categories. He calls them different things in different contexts. Most folks know them as Icon/Index/Symbol but his best terms are much more abstract: Firstness/Secondness/Thirdness. Peirce defined them this way:
"Firstness is the mode of being of that which is such as it is, positively and without reference to anything else. Secondness is the mode of being of that which is such as it is, with respect to a second but regardless of any third. Thirdness is the mode of being of that which is such as it is, in bringing a second and third into relation to each other."
So a sign of the firstness category is the simplest kind of sign, with no relational characteristic. It is pure experience, a photon hitting your eye before your brain can even associate it with an image. Secondness is fact (and I would argue, the basis of most data representation in computer science). Thirdness is the fullest expression of the logic of relations: a sign related to another sign via a third sign.
"isn't Pierce's work on abduction quite separable from his work on semiotics?"
Nope. Peirce identifies three basic logical operations: induction, deduction, and abduction. These must operate on signs.
"I fear by the time you make semiotics computational will bear such little resemblance to semiotics that it will be, in effect, information theory, artificial intelligence, analytic philosophy of language, and the other established disciplines."
Your "fear" is what I would consider an exciting possibility of semiotics: the prospect of a hard science that unifies those fields, and more. You don't see any value to this kind of unification?
G -
-
Re: Computation as a way to understand semiotics
Thu, March 4, 2004 - 9:21 PMMy point, simply put, is that I do not see specific, applicable ideas suggested by semiotics that have not already been co-opted by other disciplines and are thriving without semiotics (e.g. abduction, codes). What is left of Semiotics seems to be of little use to someone programming a computer. Up to now, the heavy lifting has been done by mathematical (theory of computation, information theory) and engineering (computer design, robotics) disciplines, not Semiotics.
I do maintain that abduction is quite amenable to computation and that it functions well without recourse to things like Firstness, Secondness and Thirdness. Abduction has been done, but not very well, by programs that diagnose illnesses and the like. High quality inferences to the best explanation rely on a highly developed knowledge base and relatively sophisticated algorithms. But, as a matter of fact, abduction has been written about clearly and explicitly without reliance on the rest of Semiotics. For instance:
www.dis.uniroma1.it/~liberat...be-00.pdf
www.lia.deis.unibo.it/Staff/P...a99.pdf.
As to a potential unification of information theory, theory of computation, data structures, etc., it is exciting prospect. I would welcome such a synthesis, though I anticipate many difficulties, as these are quite distinct ways of treating the same object.
Why should we think that Semiotics will aid this endeavor? -
-
Re: Computation as a way to understand semiotics
Fri, March 5, 2004 - 8:53 AM"What is left of Semiotics seems to be of little use to someone programming a computer."
I disagree. There are a ton of ideas that I think have very promising applicability to computing logic.
"Up to now, the heavy lifting has been done by mathematical (theory of computation, information theory) and engineering (computer design, robotics) disciplines, not Semiotics."
Actually, much of the logic that traditionally forms the basis of computer science comes from semiotics: truth tables, ontologies/categories, the existential and universal quantifiers, existential graphs, and more.
"I do maintain that abduction is quite amenable to computation and that it functions well without recourse to things like Firstness, Secondness and Thirdness."
To that, your own next statement is the reply: "Abduction has been done, but not very well." And you're right, it hasn't been done well. Semiotics includes logical processes that may offer better ways to engineer abduction. I'm actually working on a paper right now that looks into this.
"As to a potential unification of information theory, theory of computation, data structures, etc., it is exciting prospect. I would welcome such a synthesis, though I anticipate many difficulties, as these are quite distinct ways of treating the same object."
Yes, there is much to be done.
"Why should we think that Semiotics will aid this endeavor?"
Because semiotics was designed to be an architectonic: a kind of root espistemological abstraction. Information theory, data structures, computer science, cognitive science, and so on are extremely valuable on their own... and features of semiotics (such as the ontological categories and logic of relations) can be used to base them all in a common logical architecture.
G -
-
This is the maximum depth. Additional responses will not be threaded.
Re: Computation as a way to understand semiotics
Fri, March 5, 2004 - 10:03 AM”Actually, much of the logic that traditionally forms the basis of computer science comes from semiotics: truth tables, ontologies/categories, the existential and universal quantifiers, existential graphs, and more.”
I think that this attribution is too broad, unless you mean to say that anything created or extended by Pierce is Semiotics. There is no doubt that Pierce was a brilliant logician, but a few attributions are a little off: Pierce, Frege and Schroder independently developed Truth tables. Quantifiers were knocking around well before Pierce and were formalized by Boole and Frege and refined by Peano, Pierce, Russell and Whitehead; ontologies and categories have been around at least since Aristotle, further developed by Liebniz and refined by countless thinkers, including Pierce.
I think the confusion I am experiencing originates in a disagreement of what Semiotics is. I took semiotics to be a philosophical study of signs as has been developed through Locke, Pierce, Saussure, and Eco. By contrast, you seem (I may easily be wrong here) to take Semiotics to be the basis for the study of anything that uses signs: mathematical logic, language, computing. Is this right? What do you take Semiotics to be (or what should it be)? Something like this?
www.dca.fee.unicamp.br/~gudwi...psemio/
I wish you the best of luck in developing abduction further and hope that you will share your ideas in this tribe.
John -
-
Re: Computation as a way to understand semiotics
Fri, March 5, 2004 - 4:32 PM"you seem (I may easily be wrong here) to take Semiotics to be the basis for the study of anything that uses signs: mathematical logic, language, computing. Is this right?"
Yep. Semiotics == the study of signs. Any signs.
This is a rough but accurate characterization:
en.wikipedia.org/wiki/Semiotics
"I wish you the best of luck in developing abduction further and hope that you will share your ideas in this tribe."
Thanks. I will once I'm ready. In the mean time, I appreciate the dialog you've helped to generate. I'd like to see more of it in this tribe.
G
-
-
-
Re: Computation as a way to understand semiotics
Mon, October 18, 2004 - 5:08 PMAll computational paradigms, including information theory, are severely incomplete as long as they restrict themselves to what semiotics sees as (merely) syntactic phenomena. The fact that semiotics can see this limitation which is largely invisible from within those paradigms themselves is already one point in its favor.
Also, it is semiotics that tells us that if any such paradigm can be completed (which may or may not be the case, since most of them were created by frogs at the bottom of a well), then we should be looking to semantics and pragmatics for its completion.
John says, "I do not see specific, applicable ideas suggested by semiotics that have not already been co-opted by other disciplines and are thriving without semiotics (e.g. abduction, codes). What is left of Semiotics seems to be of little use to someone programming a computer."
Absolutely. In other words, semiotics has been so efficient in producing concepts that were so extremely successful as to be co-opted wholesale into various client disciplines, that there are no useless ideas left over that semioticians might discuss which are of no import to their clients.
To take this enormous influence as a *discredit* to semiotics is like saying that mathematics should be discredited because every product of that discipline is either (a) useful, and therefore already co-opted by physics, enzymology or economics; or (b) useless, and therefore not worthy of further discussion.
John, do you believe that an approach to semiotics like the one I cited on the Thinking Signs thread is not really semiotics, not really useful, or both?
-
-
-
-
-
-
-
