On the separation of syntax
and semantics and the issue of (real time) pragmatics à
Wednesday, August 18, 2004
Substructural
Semantics
(this is a new page, currently being edited as discussions about the page are made in email
to make a comment
please sent an email to portal@ontologystream.com
8/18/2004 8:38 AM
Substructural Semantics and Q-SAR
The anticipatory web of encoded data
Substructural Semantics and
Q-SAR
One can empirically observe that there is regularly in how substructure, in the emergence of a whole, has a separate semantics from the semantics of the whole. In natural systems the meaning of the whole, or compound, is dependant on the function that the compound supplies to a complex environment, and not simply to the aggregated properties of the elements that compose the compound. This is controversial science, because (it is claimed by some) a full understanding of the phenomenon cannot be expressed in classical physics or classical mathematics. But observe that water is formed by chemically combining two gases. In more complex bio-chemistry, a theory of relationship between structure and function has not been provided, as yet, by science [1].
For example, chemistry largely depends on some degree of prediction of function for a compound from the atomic composition of the compound.
The formative construction of machine representation of human knowledge (called by most “ontology) must solve this stratification of semantics problem in real time. This is the nature of the Anticipatory Web architectures, that are being proposed for funding.
The use of Readware depends on a set of computed methods that assigns an incomplete meaning to text based on an examination of “letter semantics”. This assignment is similar in nature to incomplete assignments made to physicochemical properties of chemicals, and (I will claim) to the use of “semantic primitives” by Richard Ballard and others. Prueitt offers a generalized theory of semantic primitives.
Qualitative Structure Activity Relationship (Q-SAR) analysis is one type of predictive modeling paradigm in bio-chemistry.
Structure-activity
relationships and quantitative structure-activity relationships, collectively
referred to as (Q)SARs, are theoretical models that can be used to predict the
physicochemical and biological properties of molecules
A
structure-activity relationship (SAR) is a (qualitative) association
between a chemical substructure and the potential of a chemical containing the
substructure to exhibit a certain biological effect.
A
quantitative structure-activity relationship (QSAR) is a mathematical
model that relates a quantitative measure of chemical structure (e.g. a
physicochemical property) to a physical property or to a biological effect
(e.g. a toxicological endpoint).
Under
the current EU legislation for New and Existing Chemicals, the use of (Q)SARs
is limited, probably because there has been disagreement in the scientific and
regulatory communities over the applications of QSARs, and the extent to which
QSAR estimates can be relied upon.
However,
under the future REACH (Registration, Evaluation and Authorisation of
CHemicals) system, proposed by the Commission's White Paper on a Future Chemicals
Policy, it is anticipated that (Q)SARs will be used more extensively, in the
interests of time- and cost-effectiveness and animal welfare. In particular,
(Q)SARs are likely to play an important role in the assessment of chemicals
produced or imported in quantities between 1 and 10 tonnes, for which minimal
animal testing is foreseen by the White Paper. In principle, (Q)SARs could be
used for a number of purposes
in the implementation of legislation on chemical substances and products.
While
the EU Chemicals Policy raises a clear regulatory need for (Q)SAR models that
are scientifically valid and available to all stakeholders, it does not in
itself address current concerns about the validity and applicability of
(Q)SARs. It was therefore considered necessary to develop a framework for the independent
development, validation and dissemination of QSARs. The JRC is a suitable
organisation to coordinate such a framework, due to its recognised independence
from national and sectoral interests, and its established role in the provision
of scientific and technical support for the development and implementation of
EU legislation on chemicals
The European Chemicals Bureau
(ECB)
We cannot get into a full discussion of the state of the art in regards to the prediction of function from knowledge of structure. One should read between the lines in the ECB reports on the difficulties of this problem.
However, one of the core objectives of the Soviet efforts (1950 – 1994) in applied semiotics was exactly such a general theory, expressed as a software system. And of course, this software system needed to have an underlying notational system so that scientists might communicate in a common language about the efforts being made. A similar problem is being addressed before the European Chemicals Bureau. The Soviets developed quasi-axiomatic theory to provide this notational system. The collapse of the science institutions set this work aside. However the work has find expressions in various groups around the world. (A discussion about this is possible.)
The objective of Q-SAR theory and notation is to replace the absolute dependency on observation in the complexity sciences with some type of formalism and theory. So, for example, that the task of constructing machine ontology about some natural system, such as a social system, might be blessed with both a theoretical component and an observational component. The Cyc Corp theory is based on first order predicate logics and therefore, it is argued, is insufficient to this task.
We are, as it were, looking for “some type of mathematics” upon which to govern the experimental work being done.
One might be able to see that efforts are being made on various parts of the Q-SAR problem.
One “mistake” is to think that the induction of the integers and the consequence development of Hilbert mathematics provide complete models to the complex phenomenon.
Using the Hilbert mathematics and
the first order predict logics, we do not find complete models of the natural complexity
that is embedded in living systems. Some
mathematics does exists, in fact a lot of mathematics exists, that attempts to
model chaos and complexity.
Many scholars have written (publicly accessable books) about the limitations to Hilbert mathematics and first order predict logics (I. Prigogine “End of Certainty”; Penrose, “Shadows of the Mind”). There is recognition of the limitations, but not very much consensus about what happens next. Lev Goldfarb, who talks about the non-generic nature of the numerical model, addresses this mistake extensively. He points out that inductive informatics is only in its infancy.
Another mistake is the one made by the hard “Artificial Intelligence” community, in assuming that a science of living systems and natural intelligence is structurally/functionally simple and therefore can be expressed as properties of engineered silicon processors. I have talked about this problem more that I feel comfortable about. But again, I will say that when one gives up this AI mythology it is possible to begin to see how some of the artificial intelligence techniques can be placed within a proper context. For example, perceptions about the structure of computer data might be, I claim it is, a simple task and completely solvable in optimal ways using AI techniques. Distributes agents, natural language processing techniques, and data mining all make sense in this context.
But as discussed in the preceding bead, the formation of a complete semantics needs two things. The first is full sensitivity to the measurement of the real world in real time. Computers do not do this in anyway similar to how the biological system measures the world in real time. (We do not just use our sensory organs, as some type of simple input/output machine would do.) The second thing needed is the autopoiesis, or envelop separating a complex endophysics from a complex exophysics, as discussed by Maturana and Velara (“Tree of Knowledge, the biological origins of intelligence”, 1987) and others.
The anticipatory web of encoded data
In this section we will make as simple as possible exposition of stratified theory as expressed in a notational foundation for formative and differential ontology.
One of the diagrams used to talk about Q-SAR
The question of substructure semantics is vital to the understanding of stratified theory. The categorical invariances seen from the data mining process conducted over Orbs (Ontology referential bases), for which I have developed many tutorials about, is a measurement of the “atoms” in the data. These atoms are “placed” into the data by processes that depend on this regularity in order to express substructure aggregations that are expressed to fulfill vital functional roles in an environment. The environment might be a social discourse.
In humans, the anticipatory responses are top down expectations mediated by the frontal lobes and expressed by phase coherence in brain regions, primarily in the associational cortex and in the sub cortical regions of the human brain, called the limbic system (Prueitt, Levine and Prueitt, Pribram etc) .
Information states in the mind are experienced using awareness of perception and are acted on by cognition and emotions to bind together some subsetting function selecting some parts of the memory of invariance, as in textures, colors, sensations; and some specific set of anticipations, managed by the frontal lobes in conjunction with the associational cortex and limbic system.
The anticipatory Web of encoded data has the same architecture as outlined above. Thus the Figure used to talk about S-QAR is a figure that is used both in the Anticipatory Web technology and in discussions about the nature of the functions of the brain, and the functions of a social system, in the production and use of knowledge.
Where to from here? [76]