Spanning Tree Representations of Graphs and Orders in Conceptual Structures

Graphs and partial orders are fundamental to conceptual structures theory. Conceptual graphs are used in the knowledge base and canonical basis as well as for type and conceptual relation deenitions. Partial orders are used to specify the associations among types and relations , as well as graphs in the generalization hierarchy. The speed with which basic operations on these structures can be achieved will have a pronounced eeect on eeciency. In this paper we explore the use of spanning trees as underlying representations of graphs and partial orders, and their eeect on storage and operational eeciency. The simple structure of trees can be exploited to improve matching and joins of graphs (through normalization), to reene search algorithms on orders, and for taxonomic encoding. Although this inquisition is preliminary in nature, we provide some useful techniques for gaining eeciency through the use of spanning trees, including an implementation with a special form of feature term.