Semantic query optimization : a data-driven approach
Alireza Shankaie · cIRcle (University of British Columbia) · 2009
The emergence of very large database systems over the last two decades has raised serious needs for more efficient query processing schemes. One of the proposed techniques for query optimization is Semantic Query Optimization (SQO), which is the subject of this thesis. One thing that distinguishes this technique from others is that it doesn't deal with low-level operations of the file system (e.g. block access scheduling, low-level indexing, etc). Here the objective is to alter the syntax of a query, without changing its semantics, in such a way that makes the query more efficient. In other words, by creating alternative queries, we aim at finding the alternative execution plan, which leads to the shortest execution time. We adopted a data-driven approach in the sense that we use the stored data to extract useful information (inference rules) that could be used later by a query processor to construct alternative queries. The rules are stored in a relational format in files that are called meta-database. The query optimizer searches through these meta-databases for relevant rules. We introduce the techniques that we have deployed to perform semantic query optimization in regards with our specific application. The results of experiments, as well as the amount of improvement achieved in each technique, are presented.