Topology Labeling: An Indexing Structure to Find Complex Relationships within Ontologies

Abstract

Abstract. Semantic Web technologies have contributed mainly to organize the knowledge and to search about this organized knowledge. One of the most important and complex kinds of search is to know if two entities are related within an ontology. These are called Semantic Associations, which have been classified using ρ operators: ρ-path, ρ-join and ρ-iso. Then, a ρ-query will solve any of them.Studies about this area offer low performance execution times, while others increase the performance with pre-processing using complex structures in memory.We focus on design of a simplified representation of the ontology that facilitates the graph traversal and reduces the algorithms complexity to solve these operators, starting from the first of them: ρ-path. We propose a topology labeling: we create a tree structure and identify each node with an interval index besides the level of root dependence. We will leave some space within the interval to manage future ontology modifications.To validate this technique we will create ontologies of the order of thousands to 10000 nodes and a framework test with the implementation of the technique proposed in this paper. We will implement other techniques in order to compare execution times and performance.