7 Using Inductive Learning to Generate Rules for Semantic Query Optimization

Semantic query optimization can dramatically speed up database query answering by knowledge intensive reformulation. But the problem of how to learn the required semantic rules has not been previously solved. This chapter presents a learning approach to solving this problem. In our approach, the learning is triggered by user queries. Then the system uses an inductive learning algorithm to generate semantic rules. This inductive learning algorithm can automatically select useful join paths and attributes to construct rules from a database with many relations. The learned semantic rules are eeective for optimization because they will match query patterns and reeect data regularities. Experimental results show that this approach learns suucient rules for optimization that produces a substantial cost reduction. 17.1 Introduction This chapter presents an approach to learning semantic knowledge for semantic query optimization (SQO). SQO optimizes a query by using semantic rules, such as all Maltese seaports have railroad access, to reformulate a query into a less expensive but equivalent query. For example, suppose we have a query to nd all Maltese seaports with railroad access and 2,000,000 ft 3 of storage space. From the rule given above, we can reformulate the query so that there is no need to check the railroad access of seaports, which may save some execution time. Many SQO algorithms have been developed (Hammer and Average savings from 20 to 40 percent using hand-coded knowledge are reported in the literature.