Changeset: b384ccba763e for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=b384ccba763e
Modified Files:
monetdb5/extras/rdf/hashmap/hashmap.c
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
monetdb5/extras/rdf/rdfschema.mal
Branch: rdf
Log Message:
Add frequent Characteristic sets to a set.
- This is done while putting CS into a hashmap.
Each CS will be check for the frequency (support),
if the support of a CS >= a threshold value, it will be added to the set
diffs (244 lines):
diff --git a/monetdb5/extras/rdf/hashmap/hashmap.c
b/monetdb5/extras/rdf/hashmap/hashmap.c
--- a/monetdb5/extras/rdf/hashmap/hashmap.c
+++ b/monetdb5/extras/rdf/hashmap/hashmap.c
@@ -65,6 +65,7 @@ static char intsetcmp(int* key1, int* ke
* arr1 has m members, arr2 has n members
* */
+/*
static int isSubset(int* arr1, int* arr2, int m, int n)
{
int i = 0, j = 0;
@@ -90,7 +91,7 @@ static int isSubset(int* arr1, int* arr2
else
return 1;
}
-
+*/
/*
* Return the integer of the location in data
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
@@ -41,12 +41,85 @@ static void copyIntSet(int* dest, int* o
}
}
-static void putaCStoHash(map_t csmap, int* buff, int num, oid *csoid){
+
+static
+void addCStoSet(CSset *csSet, CS item)
+{
+ void *_tmp;
+ if(csSet->numCSadded == csSet->numAllocation)
+ {
+ csSet->numAllocation += INIT_NUM_CS;
+
+ _tmp = realloc(csSet->items, (csSet->numAllocation *
sizeof(CS)));
+
+ if (!_tmp){
+ fprintf(stderr, "ERROR: Couldn't realloc memory!\n");
+ }
+ csSet->items = (CS*)_tmp;
+ }
+ csSet->items[csSet->numCSadded] = item;
+ csSet->numCSadded++;
+}
+
+static
+void freeCSset(CSset *csSet){
+ int i;
+ for(i = 0; i < csSet->numCSadded; i ++){
+ free(csSet->items[i].lstProp);
+ }
+ free(csSet->items);
+ free(csSet);
+}
+
+static
+CSset* initCSset(void){
+ CSset *csSet = malloc(sizeof(CSset));
+ csSet->items = malloc(sizeof(CS) * INIT_NUM_CS);
+ csSet->numAllocation = INIT_NUM_CS;
+ csSet->numCSadded = 0;
+
+ return csSet;
+}
+
+/*
+static
+void freeCS(CS *cs){
+ free(cs->lstProp);
+ free(cs);
+}
+*/
+
+static
+CS* creatCS(int subId, int numP, int* buff){
+ CS *cs = malloc(sizeof(CS));
+ cs->lstProp = (int*) malloc(sizeof(int) * numP);
+
+ if (cs->lstProp == NULL){
+ printf("Malloc failed. at %d", numP);
+ exit(-1);
+ }
+
+ copyIntSet(cs->lstProp, buff, numP);
+ cs->subIdx = subId;
+ cs->numProp = numP;
+ cs->numAllocation = numP;
+ return cs;
+}
+
+/*
+ * Put a CS to the hashmap.
+ * While putting CS to the hashmap, update the support (frequency) value
+ * for an existing CS, and check whether it becomes a frequent CS or not.
+ * If yes, add that frequent CS to the freqCSset.
+ *
+ * */
+static void putaCStoHash(map_t csmap, int* buff, int num, oid *csoid, char
isStoreFreqCS, int freqThreshold, CSset *freqCSset){
oid *getCSoid;
oid *putCSoid;
int err;
int* cs;
int freq;
+ CS *freqCS;
cs = (int*) malloc(sizeof(int) * num);
if (cs==NULL){
@@ -64,12 +137,20 @@ static void putaCStoHash(map_t csmap, in
(*csoid)++;
}
- else
+ else{
+ if (isStoreFreqCS == 1){ /* Store the frequent CS to the
CSset*/
+ if (freq == freqThreshold){
+ freqCS = creatCS(*getCSoid, num, buff);
+ addCStoSet(freqCSset, *freqCS);
+ }
+ }
free(cs);
+ }
}
+
static void putPtoHash(map_t pmap, int value, oid *poid, int support){
oid *getPoid;
oid *putPoid;
@@ -171,7 +252,7 @@ static void getStatisticCSsBySupports(ma
/* Extract CS from SPO triples table */
str
-RDFextractCS(int *ret, bat *sbatid, bat *pbatid){
+RDFextractCS(int *ret, bat *sbatid, bat *pbatid, int freqThreshold){
BUN p, q;
BAT *sbat = NULL, *pbat = NULL;
BATiter si, pi; /*iterator for BAT of s,p columns in spo table
*/
@@ -184,6 +265,7 @@ RDFextractCS(int *ret, bat *sbatid, bat
int* buff;
int INIT_PROPERTY_NUM = 50000;
int maxNumProp = 0;
+ CSset *freqCSset; /* Set of frequent CSs */
buff = (int *) malloc (sizeof(int) * INIT_PROPERTY_NUM);
@@ -199,6 +281,8 @@ RDFextractCS(int *ret, bat *sbatid, bat
/* Init a hashmap */
csMap = hashmap_new();
+ freqCSset = initCSset();
+
numP = 0;
curP = 0;
@@ -206,7 +290,7 @@ RDFextractCS(int *ret, bat *sbatid, bat
bt = (oid *) BUNtloc(si, p);
if (*bt != curS){
if (p != 0){ /* Not the first S */
- putaCStoHash(csMap, buff, numP, &CSoid);
+ putaCStoHash(csMap, buff, numP, &CSoid, 1,
freqThreshold, freqCSset);
if (numP > maxNumProp)
maxNumProp = numP;
@@ -232,11 +316,14 @@ RDFextractCS(int *ret, bat *sbatid, bat
}
/*put the last CS */
- putaCStoHash(csMap, buff, numP, &CSoid);
+ putaCStoHash(csMap, buff, numP, &CSoid, 1, freqThreshold, freqCSset );
if (numP > maxNumProp)
maxNumProp = numP;
-
+
+
+ printf("Number of frequent CSs is: %d \n", freqCSset->numCSadded);
+
/*get the statistic */
getTopFreqCSs(csMap,20);
@@ -248,6 +335,9 @@ RDFextractCS(int *ret, bat *sbatid, bat
BBPreclaim(pbat);
free (buff);
+
+ freeCSset(freqCSset);
+
hashmap_free(csMap);
*ret = 1;
diff --git a/monetdb5/extras/rdf/rdfschema.h b/monetdb5/extras/rdf/rdfschema.h
--- a/monetdb5/extras/rdf/rdfschema.h
+++ b/monetdb5/extras/rdf/rdfschema.h
@@ -24,19 +24,27 @@ rdf_export str
RDFSchemaExplore(int *ret, str *tbname, str *clname);
rdf_export str
-RDFextractCS(int *ret, bat *sbatid, bat *pbatid);
+RDFextractCS(int *ret, bat *sbatid, bat *pbatid, int freqThreshold);
rdf_export str
RDFextractPfromPSO(int *ret, bat *pbatid, bat *sbatid);
-typedef struct SubProps
+typedef struct CS
{
int subIdx; //Id of subject
int* lstProp; //List of properties' Ids
int numProp;
int numAllocation;
-} SubProps;
+} CS;
-SubProps* subPropSet;
+#define INIT_NUM_CS 100
+
+typedef struct CSset{
+ CS* items;
+ int numCSadded;
+ int numAllocation;
+} CSset;
+
+CSset *freqCSs;
#endif /* _RDFSCHEMA_H_ */
diff --git a/monetdb5/extras/rdf/rdfschema.mal
b/monetdb5/extras/rdf/rdfschema.mal
--- a/monetdb5/extras/rdf/rdfschema.mal
+++ b/monetdb5/extras/rdf/rdfschema.mal
@@ -22,9 +22,9 @@ command rdfschemaexplore(tbname:str, cln
address RDFSchemaExplore
comment "Explore the schema information from input table e.g., SPO in RDF";
-command rdfextractCS( sbat:bat[:any_1,:oid], pbat:bat[:any_2,:oid] ) :void
+command rdfextractCS( sbat:bat[:any_1,:oid], pbat:bat[:any_2,:oid],
freqThreshold:int ) :void
address RDFextractCS
-comment "Extract Characteristic sets from SPO table";
+comment "Extract Characteristic sets from SPO table. While extracting CSs, get
the frequent CSs";
command rdfextractPfromPSO(pbat:bat[:any_1,:oid], sbat:bat[:any_2,:oid]):void
address RDFextractPfromPSO
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
