Abstract: Abstract: In order to talce into account both veracity and covering square of the holographic syntax rule, using a hierarchical model for knowledge expression and a combination of constants and variables for signing its characteristics, a kind of combinatorial matching model of correlated objects is induced. In allusion to the “combination blown-out” problem involving the model in rule matching search, a hierarchical intelligence search method based on decoupling the correlated objects is put forward. According to this method, at first, the data which do not satisfy the correlation constraints will be removed by the “closed-loop elimination method” and then only a simple sequential search needed for obtaining the problem solution. This method may avoid from any backtracking, and the time and space consumed by search are greatly reduced.

Key words: knowledge engineering, natural language understanding, holographic syntax rule, combinatorial matching model of correlated objects, hierarchical intelligence search by decoupling