Database Reference
In-Depth Information
Figure 3. Fuzzy Query Tree Derived with push Projection
DATA STRUCTURES
This section presents main data structures that we must implement in order to give support of fuzzy query
processing at core of a RDBMS. This data structures are given here according to actual PortgreSQL
structures and code. We have extended PostgreSQL querying engine for processing fuzzy queries in
SQLf. We name PosgresSQLf our extension.
Fuzzy Terms Catalog
The fuzzy predicates, comparators, connectors, modifiers and quantifiers are stored in the PostgreSQLf´s
catalog. We create a fuzzy catalog with various tables as a subset of the full catalog; this data is very
important to process the fuzzy query.
The catalog structure named pg_fuzzpred will store the fuzzy predicates' name domain and parameters
of the membership function (extension, expression or trapezoidal). It is defined in Box 3.
We store the name of the fuzzy modifier, the power, name of the expression, left term and right term
and type of modifier. Such information is contained in the catalog structure p g_fuzzymod (see Box 4).
Fuzzy comparators will be stored in a catalog structure named pg_fuzzy comp. It will contain the
name, the domain (discrete or real), support and core (if the membership function is trapezoidal), the
extension membership function and type of comparator (see Box 5).
We define also a new catalog structure pg_fuzzyconn. It is intended to store the name of the fuzzy
connector and the expression associated as shown in Box 6.
Finally, we will store fuzzy quantifiers' definitions in the catalog structure pg_fuzzyquan. Stored
definition comprises the name, the type and the Fuzzy Set describing the fuzzy quantifier (see Box 7).
