Performance evaluation of attribute-based tree organization

Abstract
A modified version of the multiple attribute tree (MAT) database organization, which uses a compact directory, is discussed. An efficient algorithm to process the directory for carrying out the node searches is presented. Statistical procedures are developed to estimate the number of nodes searched and the number of data blocks retrieved for most general and complex queries. The performance of inverted file and modified MAT organizations are compared using six real-life databases and four types of query complexities. Careful tradeoffs are established in terms of storage and access times for directory and data, query complexities, and database characteristics.

This publication has 6 references indexed in Scilit: