Burghard B. Rieger:
Perception Based Processing of NL Texts.
Modeling discourse understanding as visualized learning in SCIP Systems.
In: Lotfi, A./ John, B./ Garibaldi, J. (eds.): Recent Advances in Soft
Computing (RASC-2002 Procedings). Nottingham (Nottingham Trent UP) 2002, pp. 506-511
Abstract
It is common practice in theoretical
linguistics, formal semantics and cognitive modeling to identify
real world entities with the (symbolic) structures that represent
them. Some of the problems in logics and linguistics that these
models encounter, are due to the (crisp) declarative
formats of (symbolic, compositional, propositional)
representations employed, and the (rule-based, modular,
deterministic) procedures chosen in processing language
entities (elements, structures, relations, functions, and
processes) whose meanings are specified model dependent via truth
conditions. In order to understand how natural languages (NL)
serve the purposes they do, it has to be investigated what makes a
sign stand for (or symbolize) something else. In doing so,
procedural and fuzzy approaches to modeling NL understanding have
devised some means to come to grips with the dynamics of cognition
as a multi-layered process of structure identification that allows to cope with the
variability and vagueness, adaptivity and learning, emergence and
plasticity of knowledge and understanding. Fuzzy modeling techniques
allow for (numerical, sub-symbolic, distributed,
non-propositional) formats whose (parallel, pattern-based,
quantitative) computation result in (the emergence of)
meanings which are the outcome rather than the presuppositions of
processing, and whose modeling is a form of realization
rather than simulation.
Semiotic Cognitive Information
Processing (SCIP) models are inspired
by information systems theory and concentrate on (natural
or artificial) system-environment situations whose knowledge-based
processing of information makes them cognitive ,
and whose sign and symbol generation, manipulation, and
understanding capabilities render them semiotic. SCIP
systems' ability comprises their performance in knowledge-based
information processing and representing its results, organizing
these representations by activating others from prior processing,
planning acts by selecting from such organized and represented
dispositions, and modifying them according to changing conditions,
results, and states of evolving system-environment adaptedness.
Based on NL structures, SCIP performance is a form of complex,
multi-resolutional information processing tied to (and even be
identified with) language understanding. Whenever such
cognitive processes are modeled as being based upon structures
whose representational status is not a presupposition to but a
result from an algorithmic processing, then these algorithms -
being able to initiate and modify the structures they are
operating on - may qualify as semiotic and part of
computational semiotics.
The perception based approach of SCIP systems to NL text
processing for discourse understanding is - like vison - part of an image generating semantics (BIGS for
Bild gebende Semantik) which complements the symbolic
(de)composition of propositional structures in traditional NL
semantics. Grounded in system-environment situations, BIGS
represents meanings as structured sets of perspectival relations
(dispositional dependencies) among new entities
(meaning points) which emerge in multi-layered vector space
mappings (corpus space, semantic space) from computation of
(patterns of syntagmatic and paradigmatic)
combinatorial constraints in (not necessarily NL) material
processed.
A prototype SCIP implementation as testbed for the
description and understanding processes covering
variable system-environment situations is presented to illustrate
the performance of a perception based, procedural approach to the
dynamics of semiotically grounded (natural language) meaning
constitution.
Full text
HTML Format
PDF Format
(112 Kb)
zurück
zu Aufsätze / back to Articles