Heuristic algorithms for path determination in a semantic network

Abstract

The authors present two heuristic algorithms for determining traversal paths in a semantic network which models an object-oriented database. The first algorithm is an extension of Dijkstra's shortest path algorithm, and it identifies the most likely interpretation of an incomplete specified query. The second algorithm finds all possible interpretations of the query and ranks them in order of likely interpretations. Cost assignment for different paths is based on a set of heuristic rules which allows costs to be dynamically determined during a path traversal. The algorithms have been implemented and are in use in a graphics interface developed for an object-oriented knowledge base management system.<>