An Efficient Query Processing Algorithm on Object Databases with Multi-valued Attributes

Abstract

The signature-based indexing techniques support evaluating complex queries against all attributes of arbitrary classes in OODBs. However, the false drop is an intrinsic characteristic of signature-based techniques. The phenomenon becomes serious in OOBs With multi-valued attributes. In this paper, we propose to partition the schema graph for reducing false drop probability. Besides, we analyze quantitatively our approach and the multiindex technique (MX) in terms of storage cost and retrieval cost. The resukts show that our approach requires less storage overhead and achieves better performance in the average cast than MX.