com.bigdata.rdf.sparql.ast

Class BSBMQ5Setup

java.lang.Object

com.bigdata.rdf.sparql.ast.BSBMQ5Setup

public class BSBMQ5Setup
extends Object

Problem setup for BSBM Q5.

Analysis of BSBM Q5

The following predicates all join on #product:

• p0
• p2
• p4
• p5

The predicates (p3 and p5) do not directly join with any of the other predicates (they do not directly share any variables). In general, a join without shared variables means the cross product of the sources will be materialized and such joins should be run last.

However, in this case there are two SPARQL FILTERs ( c1 and c2) which (a) use those variables (BSBMQ5Setup#origProperty1 and BSBMQ5Setup#origProperty2); and (b) can constrain the query. This means that running the predicates without shared variables and applying the constraints before the tail of the plan can in fact lead to a more efficient join path.

Field Summary

Fields

Modifier and Type	Field and Description
BigdataLiteral	_120
BigdataLiteral	_170
FilterNode	c0 FILTER (productInstance != ?product)
Set<FilterNode>	C0 c0 attaches when any of [p0,p2,p4,p6] runs for the 1st time.
FilterNode	c1 FILTER (?simProperty1 < (?origProperty1 + 120) && ?simProperty1 > (?origProperty1 - 120))
Set<FilterNode>	C1 c1 attaches when 2nd of (p3,p4) runs (in any order).
FilterNode	c2 FILTER (?simProperty2 < (?origProperty2 + 170) && ?simProperty2 > (?origProperty2 - 170))
Set<FilterNode>	C2 c2 attaches when 2nd of (p5,p6) runs (in any order).
FilterNode[]	constraints The constraints on the join graph.
Set<FilterNode>	NA no constraints.
StatementPatternNode	p0 ?product rdfs:label ?productLabel .
StatementPatternNode	p1 productInstance bsbm:productFeature ?prodFeature .
StatementPatternNode	p2 ?product bsbm:productFeature ?prodFeature .
StatementPatternNode	p3 productInstance bsbm:productPropertyNumeric1 ?origProperty1 .
StatementPatternNode	p4 ?product bsbm:productPropertyNumeric1 ?simProperty1 .
StatementPatternNode	p5 productInstance bsbm:productPropertyNumeric2 ?origProperty2 .
StatementPatternNode	p6 ?product bsbm:productPropertyNumeric2 ?simProperty2 .
BigdataURI	product53999
BigdataURI	productFeature
BigdataURI	productPropertyNumeric1
BigdataURI	productPropertyNumeric2
ProjectionNode	projection
QueryRoot	queryRoot
BigdataURI	rdfsLabel

Constructor Summary

Constructors

Constructor and Description
BSBMQ5Setup(AbstractTripleStore store)

Method Summary

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

rdfsLabel

public final BigdataURI rdfsLabel

productFeature

public final BigdataURI productFeature

productPropertyNumeric1

public final BigdataURI productPropertyNumeric1

productPropertyNumeric2

public final BigdataURI productPropertyNumeric2

product53999

public final BigdataURI product53999

_120

public final BigdataLiteral _120

_170

public final BigdataLiteral _170

queryRoot

public final QueryRoot queryRoot

projection

public final ProjectionNode projection

p0

public final StatementPatternNode p0

?product rdfs:label ?productLabel .

p1

public final StatementPatternNode p1

productInstance bsbm:productFeature ?prodFeature .

p2

public final StatementPatternNode p2

?product bsbm:productFeature ?prodFeature .

p3

public final StatementPatternNode p3

productInstance bsbm:productPropertyNumeric1 ?origProperty1 .

p4

public final StatementPatternNode p4

?product bsbm:productPropertyNumeric1 ?simProperty1 .

p5

public final StatementPatternNode p5

productInstance bsbm:productPropertyNumeric2 ?origProperty2 .

p6

public final StatementPatternNode p6

?product bsbm:productPropertyNumeric2 ?simProperty2 .

c0

public final FilterNode c0

FILTER (productInstance != ?product)

c1

public final FilterNode c1

FILTER (?simProperty1 < (?origProperty1 + 120) && ?simProperty1 > (?origProperty1 - 120))

Note: The AND in the compound filters is typically optimized out such that each of these is represented as its own FilterNode, but I have combined them for the purposes of these unit tests.

c2

public final FilterNode c2

FILTER (?simProperty2 < (?origProperty2 + 170) && ?simProperty2 > (?origProperty2 - 170))

Note: The AND in the compound filters is typically optimized out such that each of these is represented as its own FilterNode, but I have combined them for the purposes of these unit tests.

constraints

public final FilterNode[] constraints

The constraints on the join graph.

NA

public final Set<FilterNode> NA

no constraints.

C0

public final Set<FilterNode> C0

c0 attaches when any of [p0,p2,p4,p6] runs for the 1st time.

C1

public final Set<FilterNode> C1

c1 attaches when 2nd of (p3,p4) runs (in any order).

C2

public final Set<FilterNode> C2

c2 attaches when 2nd of (p5,p6) runs (in any order).

Constructor Detail

BSBMQ5Setup

public BSBMQ5Setup(AbstractTripleStore store)