关键词: MMDR算法:符号概率推导链, 正则规则, 复杂适应系统, 证券交易市场仿真

Abstract: MMDR(Machine Method for Discovering Regularities) is a method in Data Minning field which can get knowledge from data. In this paper, author designed a wide-first approximative regularity dual layer growing algorithm to implement MMDR, and introduced how to apply this algorism in stock price forecasting and building individual models in stock market simulation with the rules produced in the process that the algorithm runs.

Key words: MMDR, chains of semantic probabilistic inference, regularity, complex adaptive system, stock market simulation