Changeset: 8407c931a2e1 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=8407c931a2e1
Modified Files:
monetdb5/extras/rdf/rdfparams.c
monetdb5/extras/rdf/rdfparams.h
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Compute average precision
diffs (122 lines):
diff --git a/monetdb5/extras/rdf/rdfparams.c b/monetdb5/extras/rdf/rdfparams.c
--- a/monetdb5/extras/rdf/rdfparams.c
+++ b/monetdb5/extras/rdf/rdfparams.c
@@ -29,6 +29,8 @@
int dimensionFactor;
float ontologySimThreshold;
+int upperboundNumTables;
+float generalityThreshold;
void createDefaultParamsFile(void){
@@ -37,7 +39,8 @@ void createDefaultParamsFile(void){
paramFile = fopen("params.ini", "wt");
fprintf(paramFile, "dimensionFactor 3\n");
- fprintf(paramFile, "ontologySimThreshold 0.8\n");
+ fprintf(paramFile, "ontologySimThreshold 0.75\n");
+ fprintf(paramFile, "upperboundNumTables 1000");
fclose(paramFile);
}
@@ -64,7 +67,21 @@ void readParamsInput(void){
ontologySimThreshold = atof(value);
printf("ontologySimThreshold =
%f\n",ontologySimThreshold);
}
+ else if (strcmp(variable, "upperboundNumTables") == 0){
+ upperboundNumTables = atoi(value);
+ printf("upperboundNumTables = %d\n",
upperboundNumTables);
+ }
}
}
+
+ if (upperboundNumTables != 0){
+ generalityThreshold = (float) 1 / (float)upperboundNumTables;
+ printf("generalityThreshold = %f\n",generalityThreshold);
+ }
+ else{ //default
+ generalityThreshold = 0.001;
+ }
+
+
}
diff --git a/monetdb5/extras/rdf/rdfparams.h b/monetdb5/extras/rdf/rdfparams.h
--- a/monetdb5/extras/rdf/rdfparams.h
+++ b/monetdb5/extras/rdf/rdfparams.h
@@ -33,6 +33,8 @@
extern int dimensionFactor;
extern float ontologySimThreshold;
+extern int upperboundNumTables;
+extern float generalityThreshold;
rdf_export void
createDefaultParamsFile(void);
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
@@ -3088,7 +3088,7 @@ void updateParentIdxAll(CSset *freqCSset
#if USE_LABEL_FINDING_MAXCS
/*
- * * Return 1 if there is semantic evidence against merging the two CS's,
this is the case iff the two CS's have a hierarchy and their common ancestor is
too generic (support above IMPORTANCE_THRESHOLD).
+ * * Return 1 if there is semantic evidence against merging the two CS's,
this is the case iff the two CS's have a hierarchy and their common ancestor is
too generic (support above generalityThreshold).
* */
static
char isEvidenceAgainstMerging(int freqId1, int freqId2, CSlabel* labels,
OntoUsageNode *tree) {
@@ -3132,7 +3132,7 @@ char isEvidenceAgainstMerging(int freqId
level++;
}
- if (tmpNode->percentage >= IMPORTANCE_THRESHOLD) {
+ if (tmpNode->percentage >= generalityThreshold) {
// have common ancestor but it is too generic --> there is
semantic evidence against merging the two CS's
return 1;
} else {
@@ -4494,7 +4494,7 @@ char isSemanticSimilar(int freqId1, int
*/
- if (tmpNode->percentage < IMPORTANCE_THRESHOLD) {
+ if (tmpNode->percentage < generalityThreshold) {
//printf("Merge two CS's %d (Label: "BUNFMT") and %d
(Label: "BUNFMT") using the common ancestor ("BUNFMT") at level %d (score:
%f)\n",
// freqId1, labels[freqId1].name, freqId2,
labels[freqId2].name,tmpNode->uri, i,tmpNode->percentage);
oid classOid;
@@ -8972,6 +8972,7 @@ void computeMetricsQ(CSset *freqCSset){
int tblIdx = -1;
CS cs;
int totalCov = 0;
+ float totalPrecision = 0.0;
float Q = 0.0;
int i;
int curNumMergeCS = countNumberMergeCS(freqCSset);
@@ -8990,11 +8991,14 @@ void computeMetricsQ(CSset *freqCSset){
weight[tblIdx] = (float) cs.coverage * (
fillRatio[tblIdx] + refRatio[tblIdx]);
//weight[tblIdx] = (float) cs.coverage * (
fillRatio[tblIdx]); //If do not consider reference ratio
totalCov += cs.coverage;
+ totalPrecision += fillRatio[tblIdx];
Q += weight[tblIdx];
}
}
printf("Performance metric Q = (weighting %f)/(totalCov %d * numTbl %d)
\n", Q,totalCov, curNumMergeCS);
+ printf("Average precision = %f\n",(float)totalPrecision/curNumMergeCS);
+ //printf("Average precision = %f\n",(float)totalPrecision/totalCov);
Q = Q/((float)totalCov * curNumMergeCS);
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
@@ -235,7 +235,8 @@ typedef struct SubCSSet{
#define INIT_NUM_CS 1000
#define SIM_THRESHOLD 0.6
#define SIM_TFIDF_THRESHOLD 0.75
-#define IMPORTANCE_THRESHOLD 0.001 //This is used when merging CS's by common
ancestor
+//#define IMPORTANCE_THRESHOLD 0.001 //This is used when merging CS's by
common ancestor
+ // Replace by generalityThreshold =
1/(upperboundNumTables)
#define COMMON_ANCESTOR_LOWEST_SPECIFIC_LEVEL 2
//#define MIN_PERCETAGE_S5 5 // Merge all CS refered by more than
1/MIN_PERCETAGE_S6 percent of a CS via one property
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
