Fuzzy Querying Capability at Core of a RDBMS - Advanced Database Query Systems - page 181

Box 11.
CREATE FUZZY PREDICATE LOWAVAIL ON 0 .. 10000 AS (
INFINITE, INFINITE, 1000, 1500
);
CREATE RELATIVE QUANTIFIER most_of AS (
0.5, 0.75, 1.0, 1.0
);
CREATE COMPARATOR '~' ON 1 .. 100 AS (
'y-5','y','y','y+5'
);
CREATE COMPARATOR '<<' ON 1 .. 100 AS (
INFINITE, INFINITE,'y/3','y'
);
CREATE COMPARATOR '>>' ON 1 .. 100 AS (
'y','y*3', INFINITE, INFINITE
);
CREATE COMPARATOR 'cerca' on 'nation' AS (
'(ALGERIA,ETHIOPIA)/0.1', '(ALGERIA,KENYA)/0.2',
'(ALGERIA,MOROCCO)/0.3', '(ALGERIA,MOZAMBIQUE)/0.4',
'(ARGENTINA,BRAZIL)/0.9', '(ARGENTINA,CANADA)/0.1',
'(ARGENTINA,PERU)/0.5', '(FRANCE,GERMANY)/0.7',
'(FRANCE,ROMANIA)/0.5', '(FRANCE,RUSSIA)/0.1'
);
Box 12.
SELECT p_name, s_name
FROM part, partsupp, supplier
WHERE (p_partkey=ps_partkey AND ps_suppkey=s_suppkey)
AND ps_availqty '<<' LOWAVAIL
INTERSECT
SELECT p_name, s_name
FROM part, partsupp, supplier
WHERE (p_partkey=ps_partkey AND ps_suppkey=s_suppkey)
AND ps_supplycost '~' affordable;
onstrate the feasibility of the proposed design. We used the principal characteristics of SQLf according
to SQL2 and SQL3 standards.
Accordingly to benchmark test using TPC Benchmark™H, the proposed design was validated through
functionality tests. With statistical analysis over 56 queries the implementation proved scalability and
performance compared to classical queries. Main result is the fact that fuzzy querying with our proposed
strategy has not a significant impact in query processing time. This behavior is due to fuzzy query pro-