A graph-theoretic approach to indexing in object-oriented databases

Abstract

A graph theoretic approach to the path indexing problem is proposed. We represent the indexing relationships supported by indices allocated in the classes in the path in the form of a directed graph. All the previous approaches directly fit into the scheme and form a hierarchy of complexity with respect to the time required for selection of the optimal index configuration. Based on the general scheme, we develop a new approach to the path indexing problem exploiting the notion of visibility graph. We introduce a generalized nested inherited index, give algorithms for retrieval and update operations and compare the behavior of the new structure with previous approaches.