  |
The Virtual Associative Network (VAN) model of human cognition (Yufik, 1998) provides insight into how humans comprehend and ascribe meaning in the face of massive and dynamic data input and gives rise to a VAN-inspired computational technology to address processing and integration of diverse information streams for automated discovery and extraction of structure (concepts and conceptual relations). Issues for consideration include: A) Automated techniques for structure recognition/identification within closed and open formed information/data. B) Distributed techniques for structure classification and naming. C) Ontology development, representation, and maintenance techniques. D) Common interfacing solutions for distributed structure discovery. E) Structure discovery engine architectures. F) Structure event notification and command/control techniques for both dynamic data gathering management and resource management in a complex, emerging, environment.
VAN computations constitute what can be defined as an Engine for Structure Discovery. Specifically, VAN algorithms perform the following:
- Represent temporal regularities in multivariable data streams as topological regularities in a spatial structure (not to be confused with structured data).
- Identify emerging stable components in the spatial structure.
- Identify spatial structure transformations caused by variations in the data stream over time.
- Obtain globally coherent spatial structure partitioning into cohesive and weakly coupled (minimally interdependent) components.
Data streams can be produced by signal sources, data repositories (e.g., via scanning) or both. The ultimate purpose is to capture a representation of the underlying organization in the environment from which the streams emanate, in the form of distinct entities and entity groups (concepts). More precisely, the purpose can be defined as “environment disambiguation” obtained in a hierarchy of representational levels containing each a minimal number of maximally independent components (entities, groups of entities, groups of groups, etc.).
The VAN model claims that environment disambiguation is the main function of the human brain and the foundation on which the entire apparatus of human intelligence is built, including language, reasoning, and other higher cognitive functions. The model was developed based on the available neuropsychological evidence as to how this function is carried out in biological neuronal structures. Combining this evidence with mathematical speculations resulted in the algorithms proposed here as the Engine for Entity Discovery.
|
|
