C*-tree: A Multi-Dimensional Index Structure for Resource Space Model

Abstract

Resource Space Model (RSM) is a semantic data model for specifying, organizing, and retrieving resources based on classification of resources. An efficient index mechanism is critical for its implementation. However, traditional index mechanisms are not well suitable to resource spaces because the spaces are discrete and hierarchical. This paper proposes a new multidimensional index structure C*-tree. It employs boxes to represent resources so that the containment of boxes can reflect the hierarchical structure of resources. The metric function defined in the C*-tree is able to gather close resources in the index structure. To avoid overlap between bounding boxes, a new restricted bounding box is further introduced. Moreover, the insertion algorithms based on the data partition approach can fully utilize characteristics of RSM. The C*-tree is scalable and efficient on both the number of dimension and the amount of data for different distributions.