Changeset: 21c27a0ff296 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=21c27a0ff296
Modified Files:
monetdb5/extras/rdf/rdf_shredder.c
monetdb5/extras/rdf/rdfontologyload.c
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Merge two maximumCSs based on the computed similarity score between them.
diffs (truncated from 441 to 300 lines):
diff --git a/monetdb5/extras/rdf/rdf_shredder.c
b/monetdb5/extras/rdf/rdf_shredder.c
--- a/monetdb5/extras/rdf/rdf_shredder.c
+++ b/monetdb5/extras/rdf/rdf_shredder.c
@@ -223,7 +223,7 @@ tripleHandler(void* user_data, const rap
BUN bun = BUN_NONE;
BAT **graph = pdata->graph;
- printf("%s %s
%s\n",raptor_term_to_string(triple->subject),raptor_term_to_string(triple->predicate),raptor_term_to_string(triple->object));
+ //printf("%s %s
%s\n",raptor_term_to_string(triple->subject),raptor_term_to_string(triple->predicate),raptor_term_to_string(triple->object));
if (pdata->error > pdata->lasterror){
unsigned char* objStr;
int objLen;
diff --git a/monetdb5/extras/rdf/rdfontologyload.c
b/monetdb5/extras/rdf/rdfontologyload.c
--- a/monetdb5/extras/rdf/rdfontologyload.c
+++ b/monetdb5/extras/rdf/rdfontologyload.c
@@ -126,7 +126,9 @@ tripleHandler(void* user_data, const rap
parserData *pdata = ((parserData *) user_data);
//BUN bun = BUN_NONE;
//BAT **graph = pdata->graph;
+
printf("%s %s
%s\n",raptor_term_to_string(triple->subject),raptor_term_to_string(triple->predicate),raptor_term_to_string(triple->object));
+
pdata->tcount++;
return;
}
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
@@ -125,6 +125,25 @@ void addCStoSet(CSset *csSet, CS item)
}
static
+void addmergeCStoSet(mergeCSset *mergecsSet, mergeCS item)
+{
+ void *_tmp;
+ if(mergecsSet->nummergeCSadded == mergecsSet->numAllocation)
+ {
+ mergecsSet->numAllocation += INIT_NUM_CS;
+
+ _tmp = realloc(mergecsSet->items, (mergecsSet->numAllocation *
sizeof(CS)));
+
+ if (!_tmp){
+ fprintf(stderr, "ERROR: Couldn't realloc memory!\n");
+ }
+ mergecsSet->items = (mergeCS*)_tmp;
+ }
+ mergecsSet->items[mergecsSet->nummergeCSadded] = item;
+ mergecsSet->nummergeCSadded++;
+}
+
+static
CSrel* creataCSrel(oid csoid){
CSrel *csrel = (CSrel*) malloc(sizeof(CSrel));
csrel->origCSoid = csoid;
@@ -719,6 +738,16 @@ void freeCSset(CSset *csSet){
free(csSet);
}
+static
+void freemergeCSset(mergeCSset *csSet){
+ int i;
+ for(i = 0; i < csSet->nummergeCSadded; i ++){
+ free(csSet->items[i].lstProp);
+ }
+ free(csSet->items);
+ free(csSet);
+}
+
static
CSset* initCSset(void){
CSset *csSet = (CSset*) malloc(sizeof(CSset));
@@ -729,6 +758,16 @@ CSset* initCSset(void){
return csSet;
}
+static
+mergeCSset* initmergeCSset(void){
+ mergeCSset *mergecsSet = (mergeCSset*) malloc(sizeof(mergeCSset));
+ mergecsSet->items = (mergeCS*) malloc(sizeof(mergeCS) * INIT_NUM_CS);
+ mergecsSet->numAllocation = INIT_NUM_CS;
+ mergecsSet->nummergeCSadded = 0;
+
+ return mergecsSet;
+}
+
/*
static
void freeCS(CS *cs){
@@ -770,6 +809,95 @@ CS* creatCS(oid csId, int numP, oid* buf
return cs;
}
+static
+void mergeOidSets(oid* arr1, oid* arr2, oid* mergeArr, int m, int n, int
numCombineP){
+
+ int i = 0, j = 0;
+ int pos = 0;
+
+ while( j < m && i < n )
+ {
+ if( arr1[j] < arr2[i] ){
+ mergeArr[pos] = arr1[j];
+ pos++;
+ j++;
+ }
+ else if( arr1[j] == arr2[i] )
+ {
+ mergeArr[pos] = arr1[j];
+ pos++;
+ j++;
+ i++;
+ }
+ else if( arr1[j] > arr2[i] ){
+ mergeArr[pos] = arr2[i];
+ pos++;
+ i++;
+ }
+ }
+ if (j == m && i < n){
+ while (i < n){
+ mergeArr[pos] = arr2[i];
+ pos++;
+ i++;
+ }
+ }
+
+ if (j < m && i == n){
+ while (j < m){
+ mergeArr[pos] = arr1[j];
+ pos++;
+ j++;
+ }
+ }
+
+ assert(pos == numCombineP);
+ /*
+ printf("pos = %d, numCombineP = %d\n", pos, numCombineP);
+
+ for (i = 0; i < m; i++){
+ printf(BUNFMT " ", arr1[i]);
+ }
+
+ printf("\n");
+ for (i = 0; i < n; i++){
+ printf(BUNFMT " ", arr2[i]);
+ }
+
+
+ printf("\n");
+ for (i = 0; i < pos; i++){
+ printf(BUNFMT " ", mergeArr[i]);
+ }
+
+ printf("\n");
+ */
+
+
+}
+
+static
+mergeCS* mergeTwoCSs(CS cs1, CS cs2, int numCombineP, int support, int
coverage){
+
+ mergeCS *mergecs = (mergeCS*) malloc (sizeof (mergeCS));
+ mergecs->id1 = cs1.csId;
+ mergecs->id2 = cs2.csId;
+ mergecs->lstProp = (oid*) malloc(sizeof(oid) * numCombineP);
+
+ if (mergecs->lstProp == NULL){
+ printf("Malloc failed. at %d", numCombineP);
+ exit(-1);
+ }
+
+ mergeOidSets(cs1.lstProp, cs2.lstProp, mergecs->lstProp, cs1.numProp,
cs2.numProp, numCombineP);
+
+ mergecs->numProp = numCombineP;
+ mergecs->support = support;
+ mergecs->coverage = coverage;
+
+ return mergecs;
+
+}
static
str printFreqCSSet(CSset *freqCSset, oid* csSuperCSMap, BAT *freqBat, BAT
*mapbat, char isWriteTofile, int freqThreshold){
@@ -889,6 +1017,52 @@ str printFreqCSSet(CSset *freqCSset, oid
return MAL_SUCCEED;
}
+
+static
+str printmergeCSSet(mergeCSset *mergecsSet, int freqThreshold){
+
+ int i,j;
+ FILE *fout;
+ char filename[100];
+ char tmpStr[20];
+ int ret;
+
+ str propStr;
+ char* schema = "rdf";
+ int nummergecs;
+
+ nummergecs = mergecsSet->nummergeCSadded;
+
+ if (TKNZRopen (NULL, &schema) != MAL_SUCCEED) {
+ throw(RDF, "rdf.rdfschema",
+ "could not open the tokenizer\n");
+ }
+
+
+ strcpy(filename, "mergeCSFullInfo");
+ sprintf(tmpStr, "%d", freqThreshold);
+ strcat(filename, tmpStr);
+ strcat(filename, ".txt");
+
+ fout = fopen(filename,"wt");
+
+ for (i = 0; i < nummergecs; i++){
+ mergeCS cs = (mergeCS)mergecsSet->items[i];
+
+ fprintf(fout, "MergeCS %d: "BUNFMT " and " BUNFMT
"\n",i,cs.id1, cs.id2);
+ for (j = 0; j < cs.numProp; j++){
+ takeOid(cs.lstProp[j], &propStr);
+ fprintf(fout," %s\n", propStr);
+ }
+ fprintf(fout, "\n");
+ }
+
+ fclose(fout);
+
+ TKNZRclose(&ret);
+ return MAL_SUCCEED;
+}
+
/*
* Hashing function for a set of values
* Rely on djb2 http://www.cse.yorku.ca/~oz/hash.html
@@ -1171,6 +1345,40 @@ static int isSubset(oid* arr1, oid* arr2
}
/*
+ * Use Jaccard similarity coefficient for computing the
+ * similarity between two sets
+ * sim(A,B) = |A B| / |A U B|
+ * Here each set contains distinct values only
+ * */
+
+static
+float similarityScore(oid* arr1, oid* arr2, int m, int n, int *numCombineP){
+
+ int i = 0, j = 0;
+ int numOverlap = 0;
+
+ while( i < n && j < m )
+ {
+ if( arr1[j] < arr2[i] )
+ j++;
+ else if( arr1[j] == arr2[i] )
+ {
+ j++;
+ i++;
+ numOverlap++;
+ }
+ else if( arr1[j] > arr2[i] )
+ i++;
+ }
+
+ *numCombineP = m + n - numOverlap;
+
+ return ((float)numOverlap / (*numCombineP));
+}
+
+
+
+/*
static
void printCS(CS cs){
int i;
@@ -1187,7 +1395,7 @@ 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){
+void getMaximumFreqCSs(CSset *freqCSset, oid* csSuperCSMap, BAT* coverageBat,
int* superCSCoverage, BAT* freqBat, int* superCSFrequency, int numCS, int
*nMaxCSs){
int numFreqCS = freqCSset->numCSadded;
int i, j;
@@ -1224,6 +1432,8 @@ void getMaximumFreqCSs(CSset *freqCSset,
//printCS( freqCSset->items[i]);
}
}
+
+ *nMaxCSs = numMaxCSs;
printf("Number of maximum CSs: %d / %d CSs \n", numMaxCSs, numCS);
/*
@@ -1268,7 +1478,39 @@ void getMaximumFreqCSs(CSset *freqCSset,
*/
}
_______________________________________________
checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
