Idiomatic Conceptualization and Contextual Concepts

This is a bit more detailed discussion of what I mean when I say idiomatic conceptualization and how that relates to contextual concepts.

First, I believe the proper approach to natural language processing (NLP) and information derivation is to NOT parse language as linguistic entities which have a direct correlation to an object and which may be logically manipulated, but rather to interpret language constructs as recognizable patterns which are translatable to contextual concepts.

These recognizable patterns, also known as idioms do not correlate directly to an object, that is, there is no 1:1 mapping from idiom to concept and no philosophical manipulation of idioms as each idiom is atomic unto itself, however, an idiom does provide enough information to the system to combine with context and stored knowledge to produce one or more concepts. It should be noted that these concepts are unique to the interpreting system and not universal among systems, though several systems may agree upon concept interpretation and encoding. To understand this better, think of the concept of the color red. It is not only based on context and environment, but consider how a color blind person stores this information internally versus a person who is not color blind. Do not discount this example as one of faulty input, it is the fact that two different systems may perceive the same sequence of events and derive entirely different information from this activity.

Words do not map to objects and idioms do not map to objects. There is no such thing as the word apple directly referring to an apple, or any word directly referring to any object. Certain idioms may include the concept of an apple but the word apple is not an idiom. To refer to an apple one is not referencing an unchanging object in undefined space for an infinite amount of time, but rather a concept at a certain place at a certain time with a certain atrophy and certain interactions with its environment, and in the case of the concept of an apple, that is really not very valuable information by itself as the apple is typically a part of a larger environment, which in reality may have very little to do with the fact that an apple is contained within it.

I mean this is not the way we think. It is an exercise in futility to just think 'unchanging apple (or anything) in a vacuum forever' because that is a meaningless concept; there is no value there. It is a philosophical contrivance, not a real thought with meaning and purpose. It is not worthy of being parsed or stored.

There is also no such thing as a universal concept or universal truth that is valid for an infinite amount of time. There are no absolutes. There are concepts in contexts which are unique to a given system at a given time; nothing more and nothing less. Systems exist in perceived environments which is nothing more than a collection of contextual concepts.

The set of stored concepts represent the knowledge of the system. The derivation of truth from these concepts may be thought of as the beliefs of the system. Given identical inputs, different systems may produce different belief function outputs. The same system may produce different belief function outputs at different times. There is no such thing as universal truth; truth is an agreed upon belief between one or more systems. Predictions may be made and information may be deduced but the only eternity is the present.

My Philosophical Inspirations

So due to the fact that I have had 3 wives and 4 sons to support, coupled with the desire to work in industry rather than academia, I have had to put my true passion of thinking machines on a back burner for nearly 25 years now, but since I am nearing retirement and limited financial stability I will begin to research this subject more frequently in the years until I pass.

This post is more a collection of authors who have influenced my thinking on this subject so if someone were to read some of my writings on this subject in the future, they will have a better understanding of where I am coming from and what/who has influenced my thinking in this area.

I guess the first book, which I basically stumbled upon over 30 years ago in the FAU library which got me started was by Russell and Whitehead and I believe it was called 'Mathematical Logic'.  From there I read a paper by, I believe David Robinson on the 'Theory of Resolution' (since Prolog was en vogue at the time for these sorts of things though my work at the time was done in LISP) and another by Putnam and Davis which I can't recall the name of, but each of these were too technical for me to understand and so I moved on to Alonzo Church and Lambda Calculus.

I then took a formal logic class with which I was extremely dissatisfied (Perch Charts and such) so from there I read 'Many Valued Logic' by Nicholas Rescher and then I advanced to different writings by Frege (more centering on his discssions of concepts rather than his formal mathematical work) and many texts on set and group theory. I revisited Russell, and I did some basic research on modal (especially temporal) logic which brought me to A.N. Prior and indirectly to Kripe's work regarding modal logic (the information if I recall was scant at best as was Prior's) and then I read Wittgenstein. Perhaps no one author has had a more profound influence on me than Wittgenstein, though McTaggart's Paradox and related writings by Q. Smith, Oaklander, etc fascinated me as I had not given the concept of time that much thought when considering formal systems and thinking machines. One last thing, is that since my mentor at the time was Marty Solomon from FAU and he was developing a system called Peirce based on the American philosopher, I also have been influenced (especially on the topic of concepts) by him also.

That pretty much sums it up. I mean I have read a bunch on formal math (Pascal, Euler, Galois, etc) and formal systems (Aristotle, Augustine, Mills, Turing) when needed but my true passion has always been on the intersection between philosophy, reality and mathematical logic and now that I am moving on to my system of contextual concepts (from my system of idiomatic conceptualization) I will lean heavily on Wittgenstein, Frege and anyone else I happen to uncover.

So to sum it up, I am a believer that reality is a very personal (Wittgenstein), difficult to share experience which will lend itself to a fuzzy (Frege) exchange at best, when an attempt to formalize that exchange (Church) is made. I believe a great starting point is many valued truth (Rescher) and modal systems (Krupke/Prior). So now, if you were to read my writings you would know where they are coming from. These authors and their works represent my foundational base and my philosophical inspirations.

Idiomatic Conceptualization

The belief that humans speak in idioms rather than structured linguistic elements, therefore the easiest way to parse natural language is to recognize idioms and map them to concepts.

The classic example is that when someone says "time flies", they are more likely referring to a shortage of time or perception that the universal heartbeat has increased its frequency, rather than to the concept of a clock flying through the air.

Idiomatic Conceptualization is simply a mapping of well understood word patterns to concepts. It does not attempt to define a concept.

The Beginnings of Contextual Concepts

If we assume the parsing of sentences can provide us with subject verb object (see idiomatic conceptualization) type structures to work with, a next logical question might be 'what do we do with this'? In other words what becomes of this datum?

We could catalog it and maybe save it in its original format (audio, for example) and we could even apply OCR and natural language processing against the OCR output and store it in a searchable format, but at the end of the day, there is a bigger question to be asked; what are we trying to create from this input and how do we intend to use that.

It is not a reach to summarize the answer to this question as 'we are trying to reduce the written/spoken word to objects'. A more technical explanation might be 'I am attempting to convert words to concepts'. So we are left to ask 'what is a concept'?

Also of interest is the question 'are concepts to be interpreted literally?' That is to say, if we can define the attributes and methods of a concept, do these remain constant over time or do even these core constituents change, and if they do change, what might cause them to change and how?

So when we think about a concept, perhaps what we are thinking about is a thing that changes its interpretation based upon the context it is presented in. I will call this 'contextual concepts', to indicate I believe that a concept changes based upon its context, and its context changes based upon the expression/evaluation of the concept, in time, within a given system.




We next ask, what is a context?