Changeset: 3e92a59a232e for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=3e92a59a232e
Modified Files:
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Remove mappings from CS to SuperCS including the frequency and the coverage
maps.
diffs (truncated from 586 to 300 lines):
diff --git a/monetdb5/extras/rdf/rdfschema.c b/monetdb5/extras/rdf/rdfschema.c
--- a/monetdb5/extras/rdf/rdfschema.c
+++ b/monetdb5/extras/rdf/rdfschema.c
@@ -85,13 +85,21 @@ static void initArray(oid* inputArr, int
}
-static void initIntArray(int* inputArr, int num, oid defaultValue){
+
+static void initcsIdFreqIdxMap(int* inputArr, int num, int defaultValue, CSset
*freqCSset){
int i;
for (i = 0; i < num; i++){
inputArr[i] = defaultValue;
}
+
+ for (i = 0; i < freqCSset->numCSadded; i++){
+ inputArr[freqCSset->items[i].csId] = i;
+ }
+
}
+
+
static void initCharArray(char* inputArr, int num, char defaultValue){
int i;
for (i = 0; i < num; i++){
@@ -348,12 +356,34 @@ void printCSrelSet(CSrel *csrelSet, char
}
+static
+oid getMaxCSIdFromCSId(oid csId, int* csIdFreqIdxMap, CSset *freqCSset){
+
+ int freqIdx;
+ oid maxCSoid;
+
+ freqIdx = csIdFreqIdxMap[csId];
+ if (freqIdx != -1){ //A freqCS
+ if (freqCSset->items[freqIdx].type == MAXCS){
+ maxCSoid = freqCSset->items[freqIdx].csId;
+ }
+ else
+ maxCSoid =
freqCSset->items[freqCSset->items[freqIdx].parentFreqIdx].csId;
+ }
+ else{
+ maxCSoid = BUN_NONE;
+ }
+
+ return maxCSoid;
+}
+
/*
* Show the relationship from each CS to maximumFreqCSs
* */
+
static
-str printCSrelWithMaxSet(oid* csSuperCSMap, CSrel *csrelToMaxSet, CSrel
*csrelFromMaxSet, CSrel *csrelBetweenMaxSet, CSrel *csrelSet, char *csFreqMap,
BAT* freqBat, int num, int freqThreshold){
+str printCSrelWithMaxSet(CSset *freqCSset, int* csIdFreqIdxMap, CSrel
*csrelToMaxSet, CSrel *csrelFromMaxSet, CSrel *csrelBetweenMaxSet, CSrel
*csrelSet, char *csFreqMap, BAT* freqBat, int num, int freqThreshold){
int i;
int j;
@@ -379,11 +409,11 @@ str printCSrelWithMaxSet(oid* csSuperCSM
// Merge the relationships to create csrelToMaxSet, csrelFromMaxSet
for (i = 0; i < num; i++){
- maxCSoid = csSuperCSMap[csrelSet[i].origCSoid];
+ maxCSoid = getMaxCSIdFromCSId(csrelSet[i].origCSoid,
csIdFreqIdxMap,freqCSset);
if (csrelSet[i].numRef != 0){
for (j = 0; j < csrelSet[i].numRef; j++){
- if (csSuperCSMap[csrelSet[i].lstRefCSoid[j]] !=
BUN_NONE){
-
addReltoCSRelWithFreq(csrelSet[i].origCSoid,
csSuperCSMap[csrelSet[i].lstRefCSoid[j]], csrelSet[i].lstPropId[j],
csrelSet[i].lstCnt[j], csrelSet[i].lstBlankCnt[j], &csrelToMaxSet[i]);
+ if
(getMaxCSIdFromCSId(csrelSet[i].lstRefCSoid[j],csIdFreqIdxMap,freqCSset) !=
BUN_NONE){
+
addReltoCSRelWithFreq(csrelSet[i].origCSoid,
getMaxCSIdFromCSId(csrelSet[i].lstRefCSoid[j], csIdFreqIdxMap,freqCSset),
csrelSet[i].lstPropId[j], csrelSet[i].lstCnt[j], csrelSet[i].lstBlankCnt[j],
&csrelToMaxSet[i]);
}
}
@@ -393,8 +423,8 @@ str printCSrelWithMaxSet(oid* csSuperCSM
// Else, use its csoid
if (maxCSoid != BUN_NONE){
for (j = 0; j < csrelSet[i].numRef; j++){
- if
(csSuperCSMap[csrelSet[i].lstRefCSoid[j]] != BUN_NONE){
- addReltoCSRelWithFreq(maxCSoid,
csSuperCSMap[csrelSet[i].lstRefCSoid[j]], csrelSet[i].lstPropId[j],
csrelSet[i].lstCnt[j],csrelSet[i].lstBlankCnt[j], &csrelFromMaxSet[maxCSoid]);
+ if
(getMaxCSIdFromCSId(csrelSet[i].lstRefCSoid[j], csIdFreqIdxMap,freqCSset) !=
BUN_NONE){
+ addReltoCSRelWithFreq(maxCSoid,
getMaxCSIdFromCSId(csrelSet[i].lstRefCSoid[j], csIdFreqIdxMap,freqCSset),
csrelSet[i].lstPropId[j], csrelSet[i].lstCnt[j],csrelSet[i].lstBlankCnt[j],
&csrelFromMaxSet[maxCSoid]);
}
else{
addReltoCSRelWithFreq(maxCSoid,
csrelSet[i].lstRefCSoid[j], csrelSet[i].lstPropId[j], csrelSet[i].lstCnt[j],
csrelSet[i].lstBlankCnt[j], &csrelFromMaxSet[maxCSoid]);
@@ -462,7 +492,6 @@ str printCSrelWithMaxSet(oid* csSuperCSM
fclose(fout1);
fclose(fout1filter);
- /*------------------------*/
strcpy(filename2, "csRelationshipBetweenMaxFreqCS");
sprintf(tmpStr, "%d", freqThreshold);
@@ -475,11 +504,11 @@ str printCSrelWithMaxSet(oid* csSuperCSM
// Merge the csrelToMaxSet --> csrelBetweenMaxSet
for (i = 0; i < num; i++){
- maxCSoid = csSuperCSMap[csrelToMaxSet[i].origCSoid];
+ maxCSoid = getMaxCSIdFromCSId(csrelToMaxSet[i].origCSoid,
csIdFreqIdxMap,freqCSset);
if (csrelToMaxSet[i].numRef != 0 && maxCSoid != BUN_NONE){
for (j = 0; j < csrelToMaxSet[i].numRef; j++){
-
assert(csSuperCSMap[csrelToMaxSet[i].lstRefCSoid[j]] ==
csrelToMaxSet[i].lstRefCSoid[j]);
- addReltoCSRelWithFreq(maxCSoid,
csSuperCSMap[csrelToMaxSet[i].lstRefCSoid[j]], csrelToMaxSet[i].lstPropId[j],
csrelToMaxSet[i].lstCnt[j],csrelToMaxSet[i].lstBlankCnt[j],
&csrelBetweenMaxSet[maxCSoid]);
+
assert(getMaxCSIdFromCSId(csrelToMaxSet[i].lstRefCSoid[j],
csIdFreqIdxMap,freqCSset) == csrelToMaxSet[i].lstRefCSoid[j]);
+ addReltoCSRelWithFreq(maxCSoid,
getMaxCSIdFromCSId(csrelToMaxSet[i].lstRefCSoid[j], csIdFreqIdxMap,freqCSset),
csrelToMaxSet[i].lstPropId[j],
csrelToMaxSet[i].lstCnt[j],csrelToMaxSet[i].lstBlankCnt[j],
&csrelBetweenMaxSet[maxCSoid]);
}
}
}
@@ -519,6 +548,8 @@ str printCSrelWithMaxSet(oid* csSuperCSM
return MAL_SUCCEED;
}
+
+
static
void printSubCSInformation(SubCSSet *subcsset, BAT* freqBat, int num, char
isWriteTofile, int freqThreshold){
@@ -779,10 +810,10 @@ void freeCS(CS *cs){
*/
#if STOREFULLCS
static
-CS* creatCS(oid csId, int numP, oid* buff, oid subjectId, oid* lstObject)
+CS* creatCS(oid csId, int numP, oid* buff, oid subjectId, oid* lstObject, char
type, int parentfreqIdx, int support, int coverage)
#else
static
-CS* creatCS(oid csId, int numP, oid* buff)
+CS* creatCS(oid csId, int numP, oid* buff, char type, int parentfreqIdx, int
support, int coverage)
#endif
{
CS *cs = (CS*)malloc(sizeof(CS));
@@ -798,7 +829,6 @@ CS* creatCS(oid csId, int numP, oid* buf
cs->numProp = numP;
cs->numAllocation = numP;
/*By default, this CS is not known to be a subset of any other CS*/
- cs->parent = BUN_NONE;
#if STOREFULLCS
cs->lstObj = (oid*) malloc(sizeof(oid) * numP);
if (cs->lstObj == NULL){
@@ -809,6 +839,14 @@ CS* creatCS(oid csId, int numP, oid* buf
cs->subject = subjectId;
//printf("Create a CS with subjectId: " BUNFMT "\n", subjectId);
#endif
+
+ cs->type = type;
+
+ // This value is set for the
+ cs->parentFreqIdx = parentfreqIdx;
+ cs->support = support;
+ cs->coverage = coverage;
+
return cs;
}
@@ -1000,7 +1038,7 @@ void mergeTwomergeCS(mergeCS *mergecs1,
}
static
-str printFreqCSSet(CSset *freqCSset, oid* csSuperCSMap, BAT *freqBat, BAT
*mapbat, char isWriteTofile, int freqThreshold){
+str printFreqCSSet(CSset *freqCSset, BAT *freqBat, BAT *mapbat, char
isWriteTofile, int freqThreshold){
int i;
int j;
@@ -1037,7 +1075,7 @@ str printFreqCSSet(CSset *freqCSset, oid
takeOid(cs.csId, &subStr);
- printf("CS " BUNFMT " (Freq: %d) | Subject: %s |
ConsistsOf " BUNFMT " \n", cs.csId, *freq, subStr, csSuperCSMap[cs.csId]);
+ printf("CS " BUNFMT " (Freq: %d) | Subject: %s |
Parent " BUNFMT " \n", cs.csId, *freq, subStr,
freqCSset->items[cs.parentFreqIdx].csId);
for (j = 0; j < cs.numProp; j++){
printf(" P:" BUNFMT " --> \n", cs.lstProp[j]);
}
@@ -1064,18 +1102,18 @@ str printFreqCSSet(CSset *freqCSset, oid
takeOid(cs.subject, &subStr);
- fprintf(fout,"CS " BUNFMT " (Freq: %d) | Subject: %s |
ConsistsOf " BUNFMT " \n", cs.csId, *freq, subStr, csSuperCSMap[cs.csId]);
+ fprintf(fout,"CS " BUNFMT " (Freq: %d) | Subject: %s |
FreqParentIdx %d \n", cs.csId, *freq, subStr, cs.parentFreqIdx);
// Filter max freq cs set
- if (csSuperCSMap[cs.csId] == cs.csId){
- fprintf(fout2,"CS " BUNFMT " (Freq: %d) |
Subject: %s | ConsistsOf " BUNFMT " \n", cs.csId, *freq, subStr,
csSuperCSMap[cs.csId]);
+ if (cs.type == MAXCS){
+ fprintf(fout2,"CS " BUNFMT " (Freq: %d) |
Subject: %s | Parent " BUNFMT " \n", cs.csId, *freq, subStr, cs.csId);
}
for (j = 0; j < cs.numProp; j++){
takeOid(cs.lstProp[j], &propStr);
//fprintf(fout, " P:" BUNFMT " --> ",
cs.lstProp[j]);
fprintf(fout, " P:%s --> ", propStr);
- if (csSuperCSMap[cs.csId] == cs.csId){
+ if (cs.type == MAXCS){
fprintf(fout2, " P:%s --> ", propStr);
}
@@ -1095,14 +1133,14 @@ str printFreqCSSet(CSset *freqCSset, oid
}
fprintf(fout, " O: %s \n", objStr);
- if (csSuperCSMap[cs.csId] == cs.csId){
+ if (cs.type == MAXCS){
fprintf(fout2, " O: %s \n", objStr);
}
}
fprintf(fout, "\n");
- if (csSuperCSMap[cs.csId] == cs.csId){
+ if (cs.type == MAXCS){
fprintf(fout2, "\n");
}
}
@@ -1390,9 +1428,9 @@ oid putaCStoHash(CSBats *csBats, oid* ke
//Handle the case when freqThreshold == 1
if (isStoreFreqCS ==1 && freqThreshold == 1){
#if STOREFULLCS
- freqCS = creatCS(csId, num, key, subjectId, buffObjs);
+ freqCS = creatCS(csId, num, key, subjectId, buffObjs,
FREQCS, -1, 0,0);
#else
- freqCS = creatCS(csId, num, key);
+ freqCS = creatCS(csId, num, key, FREQCS,-1,0,0);
#endif
addCStoSet(freqCSset, *freqCS);
}
@@ -1412,9 +1450,9 @@ oid putaCStoHash(CSBats *csBats, oid* ke
if (isStoreFreqCS ==1 && freqThreshold == 1){
#if STOREFULLCS
- freqCS = creatCS(csId, num, key, subjectId,
buffObjs);
+ freqCS = creatCS(csId, num, key, subjectId,
buffObjs, FREQCS,-1,0,0);
#else
- freqCS = creatCS(csId, num, key);
+ freqCS = creatCS(csId, num, key,
FREQCS,-1,0,0);
#endif
addCStoSet(freqCSset, *freqCS);
}
@@ -1434,9 +1472,9 @@ oid putaCStoHash(CSBats *csBats, oid* ke
//printf("FreqCS: Support = %d, Threshold %d
\n ", freq, freqThreshold);
if (*freq == freqThreshold){
#if STOREFULLCS
- freqCS = creatCS(csId, num, key,
subjectId, buffObjs);
+ freqCS = creatCS(csId, num, key,
subjectId, buffObjs, FREQCS,-1,0,0);
#else
- freqCS = creatCS(csId, num, key);
+ freqCS = creatCS(csId, num, key,
FREQCS,-1,0,0);
#endif
addCStoSet(freqCSset, *freqCS);
}
@@ -1633,40 +1671,37 @@ void printCS(CS cs){
* Here maximum frequent CS is a CS that there exist no other CS which
contains that CS
* */
static
-void getMaximumFreqCSs(CSset *freqCSset, oid* csSuperCSMap, BAT* coverageBat,
int* superCSCoverage, BAT* freqBat, int* superCSFrequency, int numCS, int
*nMaxCSs){
+void getMaximumFreqCSs(CSset *freqCSset, BAT* coverageBat, BAT* freqBat, int
numCS, int *nMaxCSs){
int numFreqCS = freqCSset->numCSadded;
int i, j;
int numMaxCSs = 0;
- oid tmpCSId;
+ int tmpParentIdx;
int* coverage;
int* freq;
printf("Retrieving maximum frequent CSs: \n");
for (i = 0; i < numFreqCS; i++){
- if (freqCSset->items[i].parent != BUN_NONE) continue;
+ if (freqCSset->items[i].parentFreqIdx != -1) continue;
for (j = (i+1); j < numFreqCS; j++){
if (isSubset(freqCSset->items[i].lstProp,
freqCSset->items[j].lstProp,
freqCSset->items[i].numProp,freqCSset->items[j].numProp) == 1) {
/* CSj is a subset of CSi */
- freqCSset->items[j].parent =
freqCSset->items[i].csId;
- csSuperCSMap[freqCSset->items[j].csId] =
freqCSset->items[i].csId;
+ freqCSset->items[j].parentFreqIdx = i;
}
else if (isSubset(freqCSset->items[j].lstProp,
freqCSset->items[i].lstProp,
freqCSset->items[j].numProp,freqCSset->items[i].numProp) == 1) {
/* CSj is a subset of CSi */
- freqCSset->items[i].parent =
freqCSset->items[j].csId;;
- csSuperCSMap[freqCSset->items[i].csId] =
freqCSset->items[j].csId;
+ freqCSset->items[i].parentFreqIdx = j;
break;
}
}
/* By the end, if this CS is not a subset of any other CS */
- if (freqCSset->items[i].parent == BUN_NONE){
+ if (freqCSset->items[i].parentFreqIdx == -1){
numMaxCSs++;
- csSuperCSMap[freqCSset->items[i].csId] =
freqCSset->items[i].csId;
//printCS( freqCSset->items[i]);
}
}
@@ -1674,46 +1709,44 @@ void getMaximumFreqCSs(CSset *freqCSset,
*nMaxCSs = numMaxCSs;
printf("Number of maximum CSs: %d / %d CSs \n", numMaxCSs, numCS);
_______________________________________________
checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
