| 摘要 |
PURPOSE: An inference search method using a hypercube composed of bits is provided to offer an efficient inference search in high-capacity T-box and A-box. CONSTITUTION: N-number of property identifiers are assigned to n-number of properties, and n-number of property vectors, which are composed of n-number of bits corresponding to each identifier, are generated. M-number of class identifiers are assigned to m-number of classes. A triple structure between the class and the property is expressed by m-number of bit cubes composed of the property vectors for each class. If an instance search is requested, the property vectors stored in the bit cubes are checked for the instance and a class which the instance belongs to, a related class and property are obtained, and a related result is extracted from A-box. [Reference numerals] (AA) Question[Class list][Property list][Instance list]; (BB) Inference question; (CC,DD) Property identifier conversion; (EE) Class?; (FF) Instance?; (GG) Property identifier switch; (HH) Instance identifier switch; (II) Bit cube; (JJ) Bit matrix; (KK) Class reverse index; (LL) Instance reverse index; (MM) Inference result |
