SetJoinParallel Class Reference

#include <setjoin_parallel.h>

Public Member Functions
	SetJoinParallel (const std::vector< std::vector< ui > > &sorted_records, const std::vector< double > &recwt, const std::vector< double > &_wordwt, double _det, ui _maxHeapSize=0, bool _isWeightedComp=false)
	SetJoinParallel (const std::vector< std::vector< ui > > &work_records, const std::vector< std::vector< ui > > &query_records, const std::vector< double > &workwt, const std::vector< double > &querywt, const std::vector< double > &_wordwt, double _det, ui _maxHeapSize=0, bool _isWeightedComp=false)
	~SetJoinParallel ()=default
void	showPara () const
void	resizeData (std::vector< std::vector< ui > > &dataset)
void	reportTimeCost ()
void	reportLargestGroup ()
unsigned long long	getResultPairsAmount ()
void	mergeResults (std::vector< std::pair< int, int > > &finalPairs)
bool	overlapSelf (ui x, ui y, int posx=0, int posy=0, int current_overlap=0)
bool	overlapSelfIC (ui x, ui y, int posx=0, int posy=0, int current_overlap=0)
bool	overlapRS (ui x, ui y, int posx=0, int posy=0, int current_overlap=0)
bool	overlapRSIC (ui x, ui y, int posx=0, int posy=0, int current_overlap=0)
double	weightedOverlap (ui x, ui y)
double	weightedJaccard (ui x, ui y)
double	jaccard (ui x, ui y)
double	weightedCosine (ui x, ui y)
double	cosine (ui x, ui y)
double	weightedDice (ui x, ui y)
double	dice (ui x, ui y)
void	index (double threshold)
void	GreedyFindCandidateAndSimPairs (const int &tid, const int indexLenGrp, const ui rid, ui record_length, const std::vector< ui > &p_keys, const std::vector< ui > &od_keys, const std::vector< ui > &odk_st)
void	findSimPairsSelf ()
void	findSimPairsRS ()

Public Attributes
bool	ifRS = false
int	earlyTerminated [MAXTHREADNUM] = { 0 }
int	earlyTerminatedEmpty [MAXTHREADNUM] = { 0 }
SimFuncType	simFType {SimFuncType::JACCARD}
std::string	typeMap [3] = {"Jaccard", "Cosine", "Dice"}
double(SetJoinParallel::*	weightedFunc )(ui, ui) = nullptr
double(SetJoinParallel::*	normalFunc )(ui, ui) = nullptr
bool(SetJoinParallel::*	overlapFunc )(ui, ui, int, int, int) = nullptr
double	det
uint64_t	resultNum = 0
uint64_t	candidateNum = 0
uint64_t	listlens = 0
ui	maxIndexPartNum {0}
std::vector< std::vector< ui > >	work_dataset
std::vector< std::vector< ui > >	query_dataset
std::vector< double >	work_weights
std::vector< double >	query_weights
std::vector< double >	wordwt
std::vector< ui >	workEmpty
std::vector< ui >	queryEmpty
std::vector< ui >	workLength
double	coe {0.0}
double	coePart {0.0}
double	ALPHA {0.0}
ui	work_n {0}
ui	query_n {0}
ui	work_maxSize {0}
ui	work_minSize {0}
ui	query_maxSize {0}
ui	query_minSize {0}
ui	maxHeapSize {0}
std::vector< std::pair< int, int > >	result_pairs [MAXTHREADNUM]
std::vector< std::pair< int, int > >	emptyPairs [MAXTHREADNUM]
bool	isWeightedComp {false}
std::vector< WeightPair >	result_pairs_ [MAXTHREADNUM]
int	isHeap [MAXTHREADNUM] = { 0 }
double	index_cost
double	search_cost
double	hashInFind_cost [MAXTHREADNUM]
double	mem_cost [MAXTHREADNUM]
double	find_cost [MAXTHREADNUM]
double	alloc_cost [MAXTHREADNUM]
double	verif_cost [MAXTHREADNUM]
bool	flagIC {false}
std::vector< int >	grpIdA
std::vector< int >	grpIdB
std::vector< std::vector< int > >	groupA
std::vector< std::vector< int > >	groupB
std::vector< ui >	revIdMapA
std::vector< ui >	revIdMapB
std::vector< ui >	idMapA
std::vector< ui >	idMapB
double **	featureValueCache {nullptr}
int *	discreteCacheIdx {nullptr}

Constructor & Destructor Documentation

SetJoinParallel() [1/2]

SetJoinParallel::SetJoinParallel ( const std::vector< std::vector< ui > > & sorted_records, const std::vector< double > & recwt, const std::vector< double > & _wordwt, double _det, ui _maxHeapSize = 0, bool _isWeightedComp = false )

inline

SetJoinParallel() [2/2]

SetJoinParallel::SetJoinParallel ( const std::vector< std::vector< ui > > & work_records, const std::vector< std::vector< ui > > & query_records, const std::vector< double > & workwt, const std::vector< double > & querywt, const std::vector< double > & _wordwt, double _det, ui _maxHeapSize = 0, bool _isWeightedComp = false )

inline

~SetJoinParallel()

SetJoinParallel::~SetJoinParallel ( )

default

Member Function Documentation

cosine()

double SetJoinParallel::cosine ( ui x, ui y )

inline

dice()

double SetJoinParallel::dice ( ui x, ui y )

inline

findSimPairsRS()

void SetJoinParallel::findSimPairsRS

(

)

findSimPairsSelf()

void SetJoinParallel::findSimPairsSelf

(

)

getResultPairsAmount()

unsigned long long SetJoinParallel::getResultPairsAmount ( )

inline

GreedyFindCandidateAndSimPairs()

void SetJoinParallel::GreedyFindCandidateAndSimPairs	(	const int &	tid,
		const int	indexLenGrp,
		const ui	rid,
		ui	record_length,
		const std::vector< ui > &	p_keys,
		const std::vector< ui > &	od_keys,
		const std::vector< ui > &	odk_st )

index()

void SetJoinParallel::index

(

double

threshold

)

jaccard()

double SetJoinParallel::jaccard ( ui x, ui y )

inline

mergeResults()

void SetJoinParallel::mergeResults ( std::vector< std::pair< int, int > > & finalPairs )

inline

overlapRS()

bool SetJoinParallel::overlapRS ( ui x, ui y, int posx = 0, int posy = 0, int current_overlap = 0 )

inline

overlapRSIC()

bool SetJoinParallel::overlapRSIC ( ui x, ui y, int posx = 0, int posy = 0, int current_overlap = 0 )

inline

overlapSelf()

bool SetJoinParallel::overlapSelf ( ui x, ui y, int posx = 0, int posy = 0, int current_overlap = 0 )

inline

overlapSelfIC()

bool SetJoinParallel::overlapSelfIC ( ui x, ui y, int posx = 0, int posy = 0, int current_overlap = 0 )

inline

reportLargestGroup()

void SetJoinParallel::reportLargestGroup ( )

inline

reportTimeCost()

void SetJoinParallel::reportTimeCost ( )

inline

resizeData()

void SetJoinParallel::resizeData ( std::vector< std::vector< ui > > & dataset )

inline

showPara()

void SetJoinParallel::showPara ( ) const

inline

weightedCosine()

double SetJoinParallel::weightedCosine ( ui x, ui y )

inline

weightedDice()

double SetJoinParallel::weightedDice ( ui x, ui y )

inline

weightedJaccard()

double SetJoinParallel::weightedJaccard ( ui x, ui y )

inline

weightedOverlap()

double SetJoinParallel::weightedOverlap ( ui x, ui y )

inline

Member Data Documentation

alloc_cost

double SetJoinParallel::alloc_cost[MAXTHREADNUM]

ALPHA

double SetJoinParallel::ALPHA {0.0}

candidateNum

uint64_t SetJoinParallel::candidateNum = 0

coe

double SetJoinParallel::coe {0.0}

coePart

double SetJoinParallel::coePart {0.0}

det

double SetJoinParallel::det

discreteCacheIdx

int* SetJoinParallel::discreteCacheIdx {nullptr}

earlyTerminated

int SetJoinParallel::earlyTerminated[MAXTHREADNUM] = { 0 }

earlyTerminatedEmpty

int SetJoinParallel::earlyTerminatedEmpty[MAXTHREADNUM] = { 0 }

emptyPairs

std::vector<std::pair<int, int> > SetJoinParallel::emptyPairs[MAXTHREADNUM]

featureValueCache

double** SetJoinParallel::featureValueCache {nullptr}

find_cost

double SetJoinParallel::find_cost[MAXTHREADNUM]

flagIC

bool SetJoinParallel::flagIC {false}

groupA

std::vector<std::vector<int> > SetJoinParallel::groupA

groupB

std::vector<std::vector<int> > SetJoinParallel::groupB

grpIdA

std::vector<int> SetJoinParallel::grpIdA

grpIdB

std::vector<int> SetJoinParallel::grpIdB

hashInFind_cost

double SetJoinParallel::hashInFind_cost[MAXTHREADNUM]

idMapA

std::vector<ui> SetJoinParallel::idMapA

idMapB

std::vector<ui> SetJoinParallel::idMapB

ifRS

bool SetJoinParallel::ifRS = false

index_cost

double SetJoinParallel::index_cost

isHeap

int SetJoinParallel::isHeap[MAXTHREADNUM] = { 0 }

isWeightedComp

bool SetJoinParallel::isWeightedComp {false}

listlens

uint64_t SetJoinParallel::listlens = 0

maxHeapSize

ui SetJoinParallel::maxHeapSize {0}

maxIndexPartNum

ui SetJoinParallel::maxIndexPartNum {0}

mem_cost

double SetJoinParallel::mem_cost[MAXTHREADNUM]

normalFunc

double(SetJoinParallel::* SetJoinParallel::normalFunc) (ui, ui) = nullptr

overlapFunc

bool(SetJoinParallel::* SetJoinParallel::overlapFunc) (ui, ui, int, int, int) = nullptr

query_dataset

std::vector<std::vector<ui> > SetJoinParallel::query_dataset

query_maxSize

ui SetJoinParallel::query_maxSize {0}

query_minSize

ui SetJoinParallel::query_minSize {0}

query_n

ui SetJoinParallel::query_n {0}

query_weights

std::vector<double> SetJoinParallel::query_weights

queryEmpty

std::vector<ui> SetJoinParallel::queryEmpty

result_pairs

std::vector<std::pair<int, int> > SetJoinParallel::result_pairs[MAXTHREADNUM]

result_pairs_

std::vector<WeightPair> SetJoinParallel::result_pairs_[MAXTHREADNUM]

resultNum

uint64_t SetJoinParallel::resultNum = 0

revIdMapA

std::vector<ui> SetJoinParallel::revIdMapA

revIdMapB

std::vector<ui> SetJoinParallel::revIdMapB

search_cost

double SetJoinParallel::search_cost

simFType

SimFuncType SetJoinParallel::simFType {SimFuncType::JACCARD}

typeMap

std::string SetJoinParallel::typeMap[3] = {"Jaccard", "Cosine", "Dice"}

verif_cost

double SetJoinParallel::verif_cost[MAXTHREADNUM]

weightedFunc

double(SetJoinParallel::* SetJoinParallel::weightedFunc) (ui, ui) = nullptr

wordwt

std::vector<double> SetJoinParallel::wordwt

work_dataset

std::vector<std::vector<ui> > SetJoinParallel::work_dataset

work_maxSize

ui SetJoinParallel::work_maxSize {0}

work_minSize

ui SetJoinParallel::work_minSize {0}

work_n

ui SetJoinParallel::work_n {0}

work_weights

std::vector<double> SetJoinParallel::work_weights

workEmpty

std::vector<ui> SetJoinParallel::workEmpty

workLength

std::vector<ui> SetJoinParallel::workLength

---

The documentation for this class was generated from the following files:

• cpp/common/setjoin_parallel.h
• cpp/common/setjoin_parallel.cc