[Colloq] PhD Thesis Proposal - Panfeng Zhou, Tuesday 5/3, 10am

[Rachel Kalweit]

College of Computer and Information Science
presents a PhD Thesis Proposal by:
Panfeng Zhou

Title:
Indexing Multi-dimensional Data with the hB-pi Tree


Tuesday, May 3, 2005
10:00am
366 West Village H

Abstract

This thesis proposal presents the proposed research on multi-dimensional 
indexing based on the hB-pi-tree. The R-tree family is the most popular 
multi-dimensional index method. But the R-tree has overlaps and its 
index page fanout decreases rapidly as the dimension increases. In 
addition, the R-tree has poor concurrency performance. The hB-pi-tree 
does not have overlaps and its index page fanout is insensitive to the 
dimension increases. The hB-pi tree also has an efficient concurrency 
control algorithm. We propose an improved hB-pi-tree which can index the 
empty and spare space (with few objects inside) more efficiently. We 
also propose methods to improve the performance of the hB-pi tree for 
non-point data sets.

Advisor:   Prof. Betty Salzberg
Committee: Prof. Gene Cooperman
                      Prof. Donghui Zhang
                      Prof. George Kollios