Database Reference
In-Depth Information
Database Management Systems Architecture
Relational Database Management Systems are mainly built from these components (Connolly & Begg,
2005): Storage Manager, responsible for data blocks interchange between disc and main memory;
Transaction Manager, responsible for maintaining the system data integrity ensuring concurrency and
recovery from system failures; Query Processor (or Query Engine) , responsible for gathering a data
manipulation or querying statement, expressed in SQL, to transform it into a series of requests regarding
the database, guarantying semantics and performance constraints.
The Query Engine is composed of inner modules that work in this way: The Parser takes the SQL
statement, verifies the syntax and searches in the Catalog for objects definitions and tables and gener-
ates an abstract data structure called a query tree . The Planner-Optimizer takes the query tree , translates
it into a relational algebra tree , applies a process to obtain a good execution plan that is a relational
algebra tree in which each operator is annotated with the chosen physical mechanisms for evaluating it
(Connolly & Begg, 2005), such as: sequential scan, index scan, bitmap heap scan, nested loop join, hash
join and merge join. The Executor processes the execution plan building the result set.
In order to support fuzzy queries as native feature in an RDBMS, in this chapter, we propose the
extension of a query engine. To the best of our knowledge, the present work constitutes the first proposal
to extend a query engine for processing SQLf queries at the core of an RDBMS.
SCOPE DELIMITATION
SQLf is the most complete fuzzy extension to SQL due to the diversity of fuzzy queries that allows the
extension of all SQL constructions with Fuzzy Sets. SQLf Data Definition Language (SQLf-DDL) al-
lows inside its syntactic structure the following fuzzy terms :
Atomic Predicates interpreted by Fuzzy Sets (we call fuzzy predicates ).
Modifiers build predicates by Fuzzy Set transformations.
Comparators as fuzzy binary relations.
Connectors as operation over membership values.
Quantifiers represented as Fuzzy Sets over natural numbers.
Fuzzy terms allows building fuzzy conditions that can be used in SQLf anyplace where standard SQL
allows a boolean logic condition. Thus, SQLf is a rather complex language.
Processing model that we propose in this chapter may be applied to all SQLf querying features.
Nevertheless, due to complexity of this language, we delimit the scope of our actual implementation to
the following characteristics.
Fuzzy Data Definition
Selected subset of SQLf allows the definition of fuzzy terms as shown in Box 1. We will call here SQLF-
DDL this set of SQLf statements.
Search WWH ::




Custom Search