A new hybrid diagrammatic reasoning system
Shuang-Shuang Yan, Ning Zhao, Shizhong Liao · 2013
Diagrams are a straightforward knowledge-representation approach to reveal topological and geometrical relations. In contrast to symbolic reasoning with surrogate representation, diagrammatic reasoning uses direct representation. Therefore, combining surrogate and direct representations is a natural approach to diagrammatic reasoning. In this paper, we design and implement a hybrid reasoning system that combines diagrammatic and symbolic reasoning. Statically, the system consists of two components, the graphic perception module and the constraint solving module. The former extracts geometric and spatial information from diagrams, and the latter accomplishes symbolic reasoning. Dynamically, the system works as a production system. In this way, we construct a hybrid diagrammatic reasoning system that reserves the characteristics of diagrams and maintains the effectiveness of symbolic reasoning.