Friday, October 05, 2007

Evolution of Semantic Systems (ESS 2007) in Jena, Oct 5th, 2007 - Day 01

Yesterday evening, the Symposion on Evolution of Semantic Systems (ESS 2007) started with a welcome dinner in Jena. This symposium attempts to foster a better understanding of semantic systems from an evolutionary perspective. As the conference's abstract says, any complex system in nature and culture is using information for the development of internal organisation and the control of functional mechanisms. Without the storage, transmission, and processing of information, the evolution of complexity would have been impossible. Furthermore, information always has a semantic dimension, which can be described in terms of meaning, sense, reference, and function and thuis, the study of semantics must play a decisive role in the analysis of complex systems. The Symposion on 'Evolution of Semantic Systems' gatheres a large variety of scientific disciplines ranging from philosophy over computer science, bio-informatics, didactics, and linguistics and we are looking forward to many rather interesting invited talks and discussions.

I'm not quite sure, whether I will have WLAN access in the university main building. Nevertheless, I'm recording and commenting the conference's programme. So, if live-blogging is not possible, I will post an entire report in the evening.

This time I have a very difficult job to summarize and comment the talks I have heard because they come from completely unfamiliar areas. Esp. in philosophy I can follow to a certain degree, but soon I'm lost completely (e.g. if it comes to discussion different views of people who the philosophers in the audience are of course familiar with...but I do lack the semantics....). Anyway, I'll try to do my best and hope not to make too many mistakes.......

Bernd-Olaf Küpers: The Evolution of Semantic Systems - Background
"nothing makes sense in biology except in the light of (evolution) information"
Life = matter + information

Opening with these two quotes the first talk in the symposion started to give some background information on the symposion's main subject.
So what is information?
First, let's consider the syntactic dimension of information, as e.g. in a dna sequence. There, the principle difference between physical and biological systems becomes obvious: boundaries or constraints. While in physical systems the organization of the system does not depent on its boundaries , i.e. it has contingent boundaries (e.g. gas molecules in a gas tank do not change their organization, if walls of gas tank are changed), in biological systems boundaries and constraints are non-nontingent. Therefore, small changes will have a significant effect on the entire system (e.g. exchanging a nucleotide in a DNA sequence).
On the other hand let's consider the semantic dimension of information, i.e. content & meaning. There, we might refer to similarities between human language and genetic code (as being a molecular language, where nucleotides are referring to single characters, nocleotide sequences to morphemes, genes to words, and so on ....). We discover that both human and molecular language have common features in syntax and structure, such as linear sequence, unidirectionality, aperiodicity, recursivity, and hierarchical structure. And on the level of semantics and functions, we identify common features, such as contingency and context dependency.
Language seems to be a basic principle of complex systems in nature. Starting with Shannon's (statistical) theory of information, we add properties such as novelty, pragmatics, complexity, to conclude that 'meaningful informationÖalways refers to 'information of maximum complexity'.

In the second part of the talk, structural sciences (how reality can be strucutred) as a new branch of science that has evolved over the last 50 years is addressed (e.g. with single topics ranging from system theory even to chaos theory, including philosophy, physics, or computer science).
Structural science overlaps (in its borders) with the traditional (natural) sciences and humanities. It's objects of cognition are structures and relationships, the methods of cognition causal analysis and analogical form, bound by structural laws as well as boundaries and constraints.
This is leading to the foundation of the Frege Center for Structural Sciences at the University of Jena with its first project: The Evolution of Semantic Systems

In the second presentation Dagfinn Follesdal from Stanford University was talking about 'The Emergence of Reference'. Forst he addressed several of the important questions, which have been previously posted on the symposion's homepage.

  1. What are paradigmatic examples of semantic systems? E.g., formal languages, natural languages, etc....A System can be defined as a structure, that is, aset of entities (nodes) with relations dewfined between them. Syntax and semantics are dependent on each other, phonology finds the distinctions, where they matter for semantics.

  2. Semantic vs nonsemantic systems -- is the a line of demarcation between semantic and non-semantic systems?
    A system becomes a semantic system, when it becomes endowed with meaning.
    A structure becomes a language, if it is used for communication (e.g. a slide rule is a semantic system (uses cmore complicated mathematical structure...instead of multiplying numbers one adds its exponents)..but not the physical system itself is semantic, but the system under a specific interpretation).
    A system becomes semantic by use in publicly obvservable ways (behaviouristic view) (therefore, as e.g., a neural network is no semantic structure...they do not become endowed with meaning through a public process).


We speak of Holism, if changes in one part of the structure tend to affect to entire structure. The semantic interpretation always is affected by new knowledge. Frege distinguished two branches of semantics: theory of sense and theory of reference. All three basic kinds of expressions, singular terms, general terms and sentences all function in the same way: sense determines reference. Objects have three features that are crucial for reference: (1) transcendence (2) change over time and (3) fallibility (we may have false beliefs about objects).

Next, Rüdiger Inhetveen and Bernhard Schieman form the University of Erlangen/Nürnberg were giving a presentation on Talking about Structures. So the basic question was 'what are structures and what are they used for...?' (-> Eulers Könisberger Brückenproblem....instead of walking, Eurler transfered the problem to a structure....here, a graph) Thus, structures (more or less complex) may be used for mapping between different systems of different complexity (at least as far as I have undrstood the talk).

Christoph von der Malsburg from the University of Frankfurt / University of Southern California was next with 'About the Organization of Semantic Systems'. First, he spoke about invariant object recognition, the cognitive model of the brain and more general, how does the brain work? For illustration, he presented the following example:
the Binding Problem: consider a blue square and a red triangle. Then in your brain are two sets of neurons, one for considering if a colour is blue or red and another one identifying quares and triangles. But, to get bot together and to recognize which shape has what colour, you are going to need more neurons (thats the binding problem)....which again leads to new binding problems. Are there universal principles of the evolution and organization of semantic systems? -> Yes!!
Then, he also referred to the basic principle questions that have been posed previous to the symposion: What are paradigmatic examples of semantic systems? Formal languages of course....only if you decypher symbols into instances...they don't have meanings by themselves. Programming languages are also not semantic by themselves, but creation and interpretation of natural language requires lots of semantics. Finally, Malsburg concluded that social systems are nothing but a higher level version of the brain....

Today's last presenter was Barry Smith from the University of Buffalo on 'The Evolution of Biomedical Ontologies'. He started his talk with a reference to 'the olden days', when people had difficulties in comparing units of measuring lengths. The situation completely changed, when in June 22, 1799 the Paris meter and kilogramm had been proposed, which startet an evolutionary process that lead to the development of well known SI base units. SI is an ontology and in the same time a controlled vocabulary. It is the achievement of the SI system that makes scientific results comparable. On the other hand, controlled vocabularies are indispensible for scientific advance. Anyway, defining ontologies in biology is much more complex than defining a system of measurement such as SI. There, the Gene Ontology starting in 1999 served as a starting point for the development of obo( open biomedical ontologies). Obo is a shared portal for (so far) 58 ontologies.