Multi-Dimension Sandpile Space: a new theory of representation and reasoning for experience knowledge

摘要

Fuzzy experience knowledge cannot be explained quantitatively with mathematical formula or certain rules. In order to solve this problem, a new theory of Multi- Dimension Sandpile Space (MDSS) was proposed in this paper with analogy to the Sandpile Model in self-organization theory. MDSS theory has the ability of dealing with uncertain or fuzzy knowledge by combining the virtues of fuzzy logic theory and neural networks, the characteristic of continual learning of case-based reasoning (CBR), and avoided the unceasing inflation of the size of information in database when using CBR. Finally, the simulation results of an example are given to show the change of the size of database by using MDSS theory.