A) Elements of Information:
Bits and forms can both be seen as basic elements in information processing. Software structures are technically based on bits, but the forms of George Spencer-Brown are just as simple. Both show a minimalistic information structure, consisting of one macrostate (bit or cross) and two microstates (bits: on/off; form: inside/outside). While the two microstates of the bit have the same likelihood of being selected, this is very different in the form, where the unmarked state has no definite limit, and neither does its likelihood.
This difference in definition and size of their microstates can explain the different behaviour in information processes between form and bit. Information is –according to C. A. Shannon and C. F. von Weizsäcker – basically linked to entropy. Entropy is defined as the amount of information present in the microstate and missing in the macrostate. Thus, by definition, the two microstates of the bit carry the information of log2(2) = 1 bit. Entropy and information content in the continence of a form, however, are always higher, due to the unequal size of the two spaces around the cross.
B) Natural Language Processing:
In our work on the interpretation of medical texts, we had to accept the open space of reality and found the same situation of micro- and macrostate. The primary data (input) of text is always richer in information than the interpretation that is aimed at. We had to accept that we would have to deliberately reduce information in order to obtain interpretation results. This can be seen as the fundamental paradox of information - to lose information (of the microstate) in order to acquire the desired information (in the macrostate). Which information must be omitted? The selection to look for, as an entropy process, cannot be defined by the input data alone, but needs the active role of an interpreter. For NLP, the interpreter program consists of an extended system of concepts and high-level rules, which can be described as an active and autonomous interpretive system – the subject.
Humans and NLP programmes both face the need to reduce data and deal with a limited internal information space. This drew us away from conventional logic, FOL in particular, to a non-monotonic logic (NML), which is suitable for bidirectional processing and enabled us to integrate time as an active, innate element of logic.
Each real interpretive system (IS) can be seen as a continence in reality, its border with the outside as a cross in GSB formalism. The interpretation results of the IS, on the other hand, are a further continence inside the IS and they reconstruct the outside reality according to the intentions and world view of the IS. Such information-reduced reconstruction of reality inside the subject is well described by the constructivists of the 1990s, like H. von Förster, G. Bateson, F. J. Varela and many others.
C) Conclusions
Because the information handling (= reduction) is always directed by the goals of an interpreter, different interpreters can achieve different results that represent different goals and world views. This affects NLP more than commonly expected. It makes it necessary to work with changing classifications and formal ontologies and represents a further reason for moving from a static logic (FOL, DL) to a dynamic one. A dynamic logic is better suited to real situations and naturally incorporates paradoxes and time, like the forms of George Spencer-Brown do. The active interpreter (IS) highlights the roles of the subject and of time in formal logic.
Hans Rudolf Straub, originally a physician, worked as medical computer scientist as of 1986. He specialised in the field of NLP and developed the method of concept molecules and the theory of interpretive systems (IS) and their communication - influenced by ideas from Sowa, Peirce, Bateson, Shannon, von Förster, Prigogine and Weizsäcker. His team is currently developing a software system for the graphical representation of forces and time relationships in the interplay of arguments, opinions and perspectives.
More info: [ Ссылка ]
